Little bit of confusion in the comments of this video so hopefully I can clear it up here. volt-amps and cooling is what you give a CPU, heat is what you get. No harm can come out of checking your temperatures while you are gaming, and making sure they are, say below 70c. Heat also causes combustion, which is why it's so important to keep it at low as possible in the everyday PC. Ultimately though, on the a320 (without changing our cooling solution), we can reduce these temps by undervolting, and setting in a max temp limit (which also reduces the clock speeds to levels that consume less voltages too). So I was working more in reverse order in this video. However the video is designed to help people without much knowledge on the subject too. As for the aggressive voltages, AMD should release a bios update, with the option to select a more conservatively tuned CPU, as we could only really undervolt -.075mv before we reached instability on this setup.
What I used to do is lower the cpu clock speed from 4.2 to 3.6 ghz in ryzen Master with just a couple of clicks and I would drop 20c right of the bat, but now that I have a 6700xt I feel like I’m giving up a significant amount of performance by doing that
@DEA GT idk man the wraith prisms pretty solid for what it is. A bit noisy but it keeps my friends 3900x cool stock. That's pretty awesome to cool a 12 core chip
Looks an awful lot like AMD got greedy and harvest more bad modes and clocked them high enough to just fit into a 3600 chip. I.E theses cores really belong in a lower end CPU.
TjMAX of server parts is lower with a reason. Also i have a thinkpad, it doesn;t allow the CPU to go higher then 72c. Most consumer stuff just pushes parts too hard cause they are expecting a lower duty cycle... and if it dies outside warranty it's not their issue anyways.
Exactly .. my Noctua NH-U9B cooled my Intel Core 2 Quad Q9550 (socket 775) for 13 years .. now it is cooling my brand new AMD Ryzen 5 5500, thanks to the 8$ AM4/AM5 socket adaptor I bought from Noctua .. this cooler was a great, great investment I'm telling you ..
@@brokeandtired it's mainboard makers defaulting the core voltage .05 - .15v too high for the chips. More voltage makes higher temps and the voltage is what's killing the chip, it will slow down when it's too hot it won't do anything if it's being shocked to death.
Very true, about 5 years ago I had a I7 8700K based system built. It came with a huge Gigabyte tower cooler. The system is still motoring along nicely, the only recent changes I had done to it, was to have a RTX 3060 replace my trusty 1070ti and a extra 1TB SSD put in. The hottest my CPU has ever gets is around the 70 degrees C mark, and that is when doing a heavy work load.
Interesting to see more being looked into this. Greg has had quite a few defective 3600s come through on fix or flop so it’s interesting to see it be more widespread than just there.
My 9900k required a helluva lot of cooling to keep it from hitting 95 to 100c. I have a 5600x and 7700x as well now that I built last year. My 7700x doesn't go over 80 with my 280mm AIO in my Corsair 4000D Airflow.
Back in 2020, I immediately noticed my 3600's temps were very high for my taste, even with a good air cooler (arctic 34 duo), like spiking instantly to low 70s and reaching high 70s during a sustained high load. But it turned out those were pretty much normal out of the box thermals. The solution was simply undervolting, no temp limit needed. I use CTR for fine tuning and OC but anyone can do it easily from the bios as you pointed out. Great vid as always Bryan!
undervolting, or buying a very basic AIO could resolve issues too. but I guess thats a moot suggestion because buying an AIO for a 3600 would have been excessive for a budget 6 core cpu.
@@pauls4522 IDK.... I put the Thermalright Peerless Assassin with a Ryzen 5700x and I could not pay it enough for it to go over 60c. Thermals were in the 50's once I used a voltage offset.... and that's running Cinebench r23 that uses the CPU much more than any game will. The Arctic eSports cooler is okay but not very good. It's more ideal for those old i7 processors like the 6700k. Those may have had a 91w TDP, but they'd use maybe 50w while gaming and 65w under stress.
Stuff like this is why I like your channel. Getting the most out of older hardware, and the breakdown of the BIOS settings really gives knowledge to someone like me that is still somewhat new to this wonderful hobby. Can't tell you how intimidating some of the Asus bios settings are to me. I'm a Weiner and afraid to mess with a majority of them. But I have a little more confidence moving forward now thanks to videos like these. Thanks again!
Same here, though I was using hyper 212 evo on one of the systems. Both died after 3 years with only 2 months separating them even though they had different motherboards and cooler.
My best advice for longevity is to just cap your fps, it doesn't even have to be an aggressive cap just at the higher end of your avg, or at refresh for non competitive titles. It will feel the same in game but your components will stop running at 100%, and temperatures can drop quite drastically. Giving it that little bit of breathing room really helps.
I heard on gamers nexus that some am4 boards use too high a voltage to make their board look faster than the competition, i think that might be at play here as well for some people?? It was a pretty big deal
Well, I did watch Buildzoid’s video about chip degradation, only high and low temperature cycle will affect gpu (solder balls crack), and cpu are highly affected by electro migration due to high voltage or the silicon quality itself. Notice that, overclocking a cpu with liquid nitrogen can often kill your cpu too. Therefore, I have to disagree with Tech Yes City if he says high temperature is the cpu killer.
@@nhozdien5058 The solder ball issue is only really a big problem when the underfill is wrong for the temperature variations it will see, 100% it's the voltage killing the chips temperature is protected against the chip will slow itself down for heat, it will allow itself to die via over voltage every single time though. The high voltage is what's causing at least some of the high temps as well.
@@nhozdien5058 tech yes city is actually right because this has been a thing for very long time: voltage x amps is what kills the CPU but temperature is a factor of how quickly that CPU will die internal resistance of transistors rises with temperature which brings up the amps,voltage and with that wattage in order for CPU to not have stability issues this was always the thing you had to do with ryzen CPUs because they ran hot as hell with stock coolers and asked for both decent undervolts and cooling solutions
@@xthelord1668 The resistance of copper does increase as temperature goes higher. However, silicon decreases its electrical resistance as temperature increases. In the beginning of the video, Brian kinda gave me that vibe of saying high temperature was the CPU killer. This was the reason why I said I would disagree if it was his implication in the beginning. As the video went, I saw his conclusion and couldn't disagree.
Going into AMD CBS and changing the TDP/PPT/EDC/etc settings is the best solution (other BIOS settings might also be needed to prevent the BIOS from ignoring the PPT). That is typically better than messing with voltages or frequencies because single-core / fewer-cores operation won't be impeded as much. Just lower the package power limit to something reasonable. I do this on all my machines, even the high-end ones, because otherwise they use double the power just to squeeze out another 10% in performance or so, which gets into silly-land pretty quickly. The thermal temperature limit in the BIOS is not as good a setting because it is a far more indirect and slow servo. PPT is a very fast servo.
Reminder to everyone that your Ryzen 3000 is probably not that old and that the warranty is three years. Check if you can RMA if yours goes bad. If you are in Australia, you may be able to argue that the lifespan is too short with the ACCC if your CPU fails shortly after the expiration of the warranty period.
I'm using this CPU since its introduction to the market, and quickly I have realised that you can optimise it with Ryzen Master too: I can set whatever TDP target I like (usually 55W for transcoding, and 65W for gaming). Also set the "CPU Boost override" to its maximum 200Mhz value (so it drives speeds that much higher with the same voltage). This setup works very well even for the stock cooler :)
@@raven4k998 Both are valid advice! :) And I do follow them: I have a Brocken 3 for a few years now (bought for the 2700 originally) , and I always turn on vsynch (I just prefer it that way, and don't care much about the potential latency increase). And yet I still use Ryzen Master the way I described above: in games there is no speed loss (because it is never 100% on all 12 threads), and in transcoding, I can live with the 10-15% speed loss (if that) for much lower temps & silence :)
In my experience, undervolt it from 1.4V to 1.2V do the trick. It seems the latest BIOS update make the CPU consume the power more than it actually needs and more power=more heat.
A lot of board manufacturers tell you which is the highest BIOS you should use for a particular CPU, so that's more than likely your issue. Read the support pages don't just go updating your BIOS.
A lot of motherboard vendors violate AMD's standards and overvolt the shit out of CPU's in order to try to win a few FPS in performance reviews. My Gigabyte TRX40 Aorus Master killed two Threadripper 3960x CPU's back in 2019 when it was new, at bone stock settings. This is a huge, well known problem and the Motherboard vendors keep doing it on purpose, it is not an unintentional bug, and it will damage CPU's. AMD needs to crack down on the board vendors. Some (but not all) motherboards vendors have responded to criticism by updating their BIOS to correct this issue, so the first thing people should do is to check for a newer BIOS. It might have been fixed.
First thing I noticed when I made my 3600 setup was high CPU temps! Even at idle. Doing some research I found that Mobos by default set core voltages very high. Mine was 1.4v at the time. You need to manually set the max core voltage ( or offset) and find the sweet spot. Mine is set at 1.25 volts (over clocked) and it give max performance while keeping cool.
Truth, most boards seem to have the vcore about .15v over what the chip actually needs to run, that excess voltage is producing excess heat and lots of it. Heat isn't the problem the chip will protect itself if it's too hot, it won't protect itself if it's being electrocuted to death slowly by over voltage though.
I dropped my v-core 1.21 and set clocks to 4.2 on my 3800x. At default the cpu fan ramped up whenever opening a browser neverming gaming, annoyed the hell out of me.
As an owner of a Ryzen 5 3600 (and A320 mb) this has got me intrigued - admittedly I've got an aftermarket air/fan cooler. I haven't seen any kind of temps like this. Thanks for this investigation. Love your content :)
Your vcore was peaking at over 1.45v that is WAY too high for that silicon, these things aren't burning up they are being electrocuted to death by MB manufacturers trying to eek out more overclock headroom with higher voltages. What you did was not so much an under-volt as it was a fix of the motherboard's default over-volt.
The thing is that it appears that 3600s actually need that much voltage. My 3600 also runs at about the same voltage but the problem is that I can only undervolt by about 12.5mv before I start getting BSODs while gaming (in easy to run games at 120 FPS). For reference, my Vega 64 runs at an undervolt of -150mv pretty much without losing performance. I suspect the worst dies were turned into 3600s as yields on the 8 core Ryzen 3000 chiplet should have been pretty high. So 3600s need a ton of voltage to operate correctly.
Yeah, being electrocuted and reddit and any other places are not filled with unhappy customers that their CPU and mb are frying like crazy. It's almost like amd engineers knew how to design a CPU properly.
Bro, those voltages that you showed are CRAZY! My 3600 runs with 975mv while 3750mhz , temps - 50° max in cinebench r20. Undervolt is a MUST! At the same time my second machine with 5700x runs 1008mv, 4250mhz, 55° Saving that silicon for the children to inherit it lol
You should down volt the 5800x, that's what I do, it decrease it running temp by like 10-20 celsius with no drop in performance keeping all the core at 4.6ghz. The 5800x has the same voltage as the 5900 but with 2/3 of the cores so each core has way too much current going through it for no reason since the cores are also the same. A 5600 has 65 tdp for 6 cores, a 5900 has 105 tdp for 12 cores, the 5800 has 105 tdp for 8 cores. That's 10.8 tdp per core for the 5600, 8.75 for the 5900 and 13.125 for the 5800.
Interesting video Bryan. This is actually pretty disappointing if it's the case (higher temps), given their relatively young age. Many CPU's in the past (eg. Intel 14nm++++ coffee lake; comet lake) would frequently hit these temps and not fail prematurely, so this is something AMD needs to look into - especially if the chips aren't thermal throttling and staying within AMD's design window. It seriously makes me wonder if it's a long term high voltage issue (eg. certain manufacturers over-volting by default), or if it's related to increased rates of degradation of the TSMC 7nm node at extended elevated temperatures/voltages... The 2080Ti you previously tested with is significantly more powerful than the RX 6600, so you would expect temps would have been equal/higher running that card.
From my experience the Precision Boost Overdrive settings under the tweaker settings on Asus aren't always doing its job at disabling it and you'd need to disable it under one of the AMD tabs in the advanced settings. The extreme overvolting with PBO has been a thing since first gen ryzen and has caused voltages of ~1.45V back then as well. Disabling it always had massive thermal and efficiency improvements while also lowering the voltage to more reasonable levels on every single ryzen system I have built so far (including anything from 1000 series to 5000 series), so I'm always disabling it because that little bit of extra performance doesn't seem worth the thermals and unhealthy voltages imho. I'm surprised that nobody really talked about that yet and, while I didn't test with any 7000 series CPU myself, I'm pretty sure that's what is mostly leading to the high temps everyone reported on the 7000 series launch. If anyone wants to verify that with a 7000 series CPU feel free to respond, am curious to see if it is indeed still the case.
yes exactly. average users and gamers do not need pbo. my 3600 with pbo on went as high as you said upto 1.45V and i didn't see any performance gain in gaming (by see i mean notice it, you can obviously compare with a benchmark but the negligible gains do not justify having pbo). with pbo disabled my 3600 would need just under 1v to operate to its max 3600mhz base clock and i could even lower the wraith stealth prism fan speed to 50% and would never need more for it to run cool and super quiet and very low power consumption too (even if i loaded all cores 100% for stress test it would stay around 70-80 degrees with fan still at 50%) the 3600 can easily handle even games like cyberpunk, red dead 2, etc... at 1440p with my 3060ti i had most high settings and gamed at 60 capped fps. i went not with the 5700x cause i gave my mother my good old 3600 but it's still great cpu for gaming
In my experience, the Ryzen 3000 series have usually run pretty cool and have been easy to cool. I am wondering if the TIM has started failing after a couple years of use. As they are soldered, its not exactly easy to check. Need Roman to pull one of these excessively hot CPU's apart.
Put a good cooler on it and change your fan profile in the bios , keep an eye and set a max voltage if needed but I find temps are fine with a good board and cooler
If the voltage is too high, and it objectively is on a good deal of motherboards at default. Keeping it cool won't save it, the temps aren't what's hurting it the excess voltage is.
@@bdhale34 what are you talking about? AMD itself has said that those 1.4v spikes are normal, if you set Manually those 1.4v everytime it's obvious you're going to hurt your CPU Keeping it cool makes it last longer and Will save it
@@bdhale34 need good vrm with ryzen chips i have 4 x 3600 and 2 5600x cpu and voltages and temps are stable, max temps under load are 72 deg c and max voltages are 1.35-1.4v on the 3600's , the 5600x have slightly less voltages out of the box applied and subsequently run a little cooler, also all my chips are under hyper 212 air coolers and they do the job perfect, the std ryzen coolers are junk for any intensive purpose.remember amd put out a statement regarding voltages in the ryzen line and operating temps when chips were released. good luck :-)
I think the problem lies with stock voltages rather than high temperatures (although the two are correlated somewhat). On my own Ryzen 5 3600 + CVN X570M platform, the motherboard will push 1.475V into the CPU by default, and I think that is the main cause of this degradation. I had got it tuned in to run at 0.86V and 3.4Ghz by default, and manually going into the Ryzen master to push it to 4.2Ghz all core with 1.16V sometimes, and so far, its overclocking potential hasn't degraded much from the three years of use while temps look great. I think the issue lies with the difference between the earlier batches of Ryzen 5 3600 and later ones, which uses a more refined 7nm process that require much less voltage to run at the same clocks. Motherboard manufacturers want their board to work out of the box with all those Zen 2 chips, regardless of their silicon quality, and that could be the reason behind the aggressive voltages, but when the chips themselves improve to a point that it don't need that, it just became a factor that drove the power and temperature up and efficiency down, while eating away at the lifespan (TSMC recommends a max of 1.3V for these chips, and some motherboards go way past that). Those unstable Ryzen 5 3600 that you have might be a result of a normal computer user leaving the motherboard bios to regulate the voltages delivered to the CPU, and over time it just tears away at the silicon quality.
I think the only thing you forgot to mention was that if they use a Prism there is an option switch on the bottom side to change the fan speed to High.
@@NotANOVERPRICEDGOUGINGTECHFAN But your AIOs might be running quieter. You might need to adjust the fan speeds on your AIO / motherboard settings to increase air flow?
@Gamer's Always Get Expensive Scrap Wafers! The Wraith Prism is a down firing cooler. So it might just be blowing air onto one of the temp sensor zones.
I'd be using hwinfo64 to check the power reporting deviation, this was an issue early on with motherboards shoving more voltage than reported. This was for the most part solved with BIOS updates but i'm sure there'll be countless boards running theses chips out there that were never updated to solve this problem.
Checked mine out of curiosity. In Cinebench R23 it got 9173 for multi and 1231 for single and temps were around 65°C. I do have slightly better than stock cooling though (BeQuiet! Shadow Rock 2). I bought the 3600 in March 2020, so I've had it for almost 3 years
I noticed this back in 2020 with my 3600 CPU, The stealth cooler just wasn't up to the task so i replaced it for a prism rgb and also undervolted it slightly, the temp were night and day difference!
I've just ordered one of these from amazon, gutted to find there are issues with them an hour after clicking pay. The bundle is coming direct from ADMI as a preassembled thing, so do you think it might come with settings changed if they know there are issues?
i see that you tested this ryzen 5 3600 with an a320 board and these problems appears, have you tried using a higher tier board like a b450 to see if the problem is still there?
I have a 3600 from late 2019 as well paired with a Mugen 5 cooler. It's undervolted with a negative voltage offset set at -0.125 and while gaming it's always between 45c and 55c max. Pretty impressive indeed.
I noticed that 5600X with the included default cooler would go up to 90 degrees regularly when gaming. I just opted to buy a big Be Quiet cooler and after that the CPU never got over 67 degrees even under 100% load when compressing and decompressing files for half an hour.
Personally I have never built a gaming PC with a stock cooler, I always recommend upgrading the cooling for gaming PCs as stock coolers are never intended for cooling CPUs while gaming. If the person is on a limited budget I usually recommend they wait and save some more money for a better cooler or downgrade a peripheral or 2 so they can afford it, it is allot easier and cheaper to upgrade peripherals later than replace failed CPUs. There are allot of good budget coolers out there like the Snowman you recommend (I like the Cryorig coolers) so why not upgrade right from the start.
I've had a variety of 3000 and 5000 Ryzens and an offset undervolt between minus 0.050V and minus 0.075V is almost compulsory for sane temperatures. Never used hard temperature limits though.
Changed from a 2600 to a 5600x both using a Dark Rock 4. Pretty impressive, the 5600x with curve optimized OC performs so much better while staying cooler than the 2600. Absolute worthy upgrade. While being on it, I also repasted my GPU that was running extremely hot looking at the Hot Spot temperature. Dropped the °C by 20-30° depending on the game. Feels like an entirely new PC x)
@@spreadinglove648 This won't help if the mbd BIOS is consistently pushing too much voltage aswell though, which may be more the relevant issue here. However, without a larger sample size and in addition testing with a decent B450, it's hard to be certain.
I bought a few wraith prisms and they are great for 3700X and below . I just like the design and the looks . Blows some air around the socket and Vrms too .
Takeaways from this is not just efficiency tuning, but also, do not skimp on your CPU cooler - basically any 20-30$ cooler will vastly outperform Prism and Spire. Don't go over the Hz of your screen FPS wise, unless you're pushing for absolute minimum latency, it's inefficient.
This is what I do with my 3600. Just an offset of -0.072 mV and the temps are in control. Also using a tower cooler helps. Still using the deepcool gammaxx 400.
@@nedjimb0 dude i took mine off when i upgraded from an i5-10400f to my i5-11500, and one of the plastic mounts finally snapped and broke :( best cpu cooler i will ever find for for less than $25, can't really find the LGA1200 version for cheap anymore. Ended up getting a Cooler Master 212 Black Edition cuz its all that my local Best Buy will stock for non-AIO's lmao
@@dc-gm7lu Damn, on my ryzen 3 1200 pc with an OC to 3.8ghz @ 1.3v, it fine. My brothers pc with a 3600x (or 3700x, cant recall) is probably fine, as I havent heard complaints and my laptop with a 5600H has been fine, the problems in it are other things, lol
Death Stranding requires AVX instruction set. AMD’s on the compute aren’t great at AVX instruction set. Its an intel patent and explains why it was running hot. AMD’s version of AVX and AVX-2 is actually an emulated version of intels instructions. So the CPU will naturally run hotter as it decompresses 64bit and 128bit instructions. Intel wasnt even able to perfect AVX and AVX-2 until 6th generation cpu’s with efficiency cores. Prior to efficiency cores 4790’s and previous gen i7’s would either run hot or with first gen icore simply wont render.
AVX 64 bit compression, and AVX-2 128 bitcompression elevates cpu VOLTAGE. Because x86 processors actually compute system ram in 32 bit chunks. AVX 64bit compression means 2 sets of 32 bit instruction are simultaneously being sent to the cpu for a 64bit compression total. Instead of in 32bit individual instruction sets. This means voltage has to increase 30 to 40% in mili volts to send a compressed 64bit instruction. AVX-2 being 128 compression increases mili volts upto 70%. Since system ram is required to send larger instruction packages. There for the reality is AMD cannot efficiently optimize AVX and AVX-2. As unlike intel AMD doesn’t utilize efficiency core optimization.
I find that undervolting is almost mandatory with some motherboards, i had a msi b450 board and it overshot quite a bit with the voltages and i quickly came to realize that undervolting was way more rational than attemting to oc that cpu since vdrop control was terrible and the ryzen 2600 did oc like shit so i would end up in the high 70's only to barely get 4ghz with a 240mm aio. I have found that ryzen really likes to work around 1.1-1.2v cpu vcore so if you see that spike to over 1.3 or even hit 1.4v then definitely try to undervolt because that is just too much
Question my cpu voltage is at 1.1v never oc mine and i have a r5 5600x. i have the stock cooler and repasted it with arctis mx-5, whenever i play wz2 the temps goes upto 70degrees and voltage at 98%, is that normal?
I just leave it at auto because at the time thats what the recommendation was. Less silicon degrade that way because static voltage and manual pinned oc is somehow worse than letting it control its voltage and clocks when idle or under load. When things are set to manual all of the auto features are disabled, at least on ryzen 3000. Which at the time was different than intel and much simpler honestly.
As a long term 3600 user, I found that the poor cpu, under stock settings, tends to get blasted with way more voltage than it needs. Undervolting should be the first step. If you have a limited bios, I'd recommend Clock Tuner Ryzen to set some p1 and p2 profiles. You should be able to have your cake and eat it too. I'm also aware that my cpu is a golden sample (I think it's a downbinned 3600xt), so I wont share the exact settings I use as most 3600's wont be able to keep up. However a cautious starting point would be: P2 (trigger at 10% cpu usage) 4200 MHz @ 1300 mV P1 (trigger at 50% cpu usage) 4000 MHz @ 1150 mV CTR has an automatic idle state it should default to if cpu usage is below the (trigger usage) value for P2. The values are very rough guide lines, and might be too aggressive still. Or you'll find more headroom to raise clocks further or to lower the voltage. Having a better cooler will also improve the speeds for any given voltage, or allow you to lower the voltages. CTR has a "tune" button that's tempting to press, but for me it has never produced any beneficial results. It seems to think my cpu can hit 4.7 GHz. It cannot. Maybe you'll have more luck, but it takes forever to run.
This is why i always use manual overclock over auto, every motherboard pumps crazy voltages into CPU. For example on my Ryzen 3600 on auto it would hit 1.4/1.45v while on manual overclock i find it stable at around 1.25, the temps are like 20c lower.
Good call to at least check. I'm running PBO and i just checked and i'm not getting anywhere near 1.4V, like it tops out at 1.33V for super short time, actually staying around 1.3V. Maybe that's an issue with particular types of mainboards? Lots of lower end boards have old offset-only VRMs which don't really have precise voltage control. Though i'm running Cezanne die (Ryzen 5500) and MSI B450M Pro VDH. So of course the results aren't transferable anyhow.
PBO was actually a huge deal for me. After turning it off my voltages dropped a lot and so did temperature. I guess YMMV. Anyways, I undervolted to 1.25, overclocked to 4.2 all core and I have my very old NH-U12P on it. Now it doesn't crack 70°C even under sustained 100% load.
These temperatures were always present with the absolutely INADEQUATE Wraith Stealth. As someone who is still running a 3600 since 2019 I have always recommended people to get a better cooler than the included one even a 20 USD tower cooler keeps the 3600 safely under 70 C in 100% usage on stock settings. It always really bothered me how many tech channels (don't remember if TYC did that too tbh) and people on the internet played up the 3600's included cooler as adequate and an advantage over Intel at the time, the reality is that those coolers should have been used at most for a couple weeks until you get a cooler if you can't get one right away.
If you are wondering why the cpu is so hot out of the box, it’s because of the voltage on ryzen 3rd generation boost was only achieved by excessive voltage in excess of 1.4v and it has been a problem ever since launch and amd claims it’s fine, but in reality it’s not and not only causes high temperatures but also causes cpu degradation. In my personal experience working with this CPU, and other third generation CPUS is the best bet you can do is a manual overclock and then setting the voltage as low as it can go before it crashes. For example, I took my friend’s Ryzen 5 3600 and manually overclocked it to 4ghz locked and then got the voltage down as low as 1.36v and his temps on the stock cooler dropped by 17c which is incredible and will really keep the cpu going for years to come.
This is a great video on so many levels for the AMD community! I have ALWAYS underclocked AMD equipment. CPU and GPU. I've always got lower temps and higher performance due to the clock map where it detects that the temps are lower therefore it will increase the frequency as long as it's stable. I haven't found a need to limit my equipment thermally, but undervolting alone is fantastic. I even do the same to my ram and run vigorous tests to make sure it can still pass memtest. Example, if my ram can run 3600Mhz CL16 @ 1.2v, but crashes at 1.15v, I'll set it to 1.25v for some overhead stability. Saving me from needing to run 1.35 rated voltage. This will reduce the heat on the motherboard & case, as well as the load on your motherboard and PSU. Every little bit counts. I used to be of the mindset of just brute-forcing things to max and ramping fans up, but after reaching points of diminishing returns and reducing life expectancy, it just isn't worth it anymore.
Why modern CPUs are failing? Because of many factors. Some were working in bad conditions (very high overclock, high voltage, bad cooling). Bad working conditions can damage or degrade CPU (it is called an electron migration and such a CPU will not work properly with default clock or voltage). In some cases paste below cover (IHS) is dry and degraded. Such problems I saw with many CPUs (many intel models), not only with Ryzens.
Haven't had any issues with my 3600. Ran it with pbo on, auto oc on and i think a small under volt. It was on a b450 tomahawk(non max) with a cooler master ml240 aio. Replaced it with a 5800x3d last month. Will keep an eye on the temps when I repurpose the 3600.
Like other people mentioned, the problem is the voltage that some mobos apply on them. Usually way past what one would call redline for these CPUs. The solution is RMA, but if it was me i'd seriously consider upgrading because the 3600 aged really poorly, it can't get 60 FPS on Kingdom Come deliverance max settings or No Man's Sky and has mediocre lows in games like Elden Ring and i doubt it'll handle these newer PS5 games (Hogwards's legacy?) well with how close in performance it is to the consoles, at least if you care about open world games which it won't handle very well.
@@jnightmare2386 I call BS on that modded skyrim claim there, quite sure it'll drop below 60 at any point where draw calls reach the neighborhood of 7500 which with heavily modded skyrim is everywhere, even with a 12600k it struggles to keep 60 at 9k draw calls, as long as you use a grass mod + dyndolod and ENB you're not running 60FPS on skyrim with a 3600, only in interiors. And MH i can break the 75FPS mark on a 3400G with max settings so that's not exactly a demanding title. Witcher 3? Maybe DX11 witcher 3, a PS4 game from 7 years ago, which i can also hit 120 FPS on a 3400G. The titles i mentioned are WAY more demanding than those you did, apples to oranges.
These days simply using a better cooler will give you extra performance. If you buy a good brand like Noctua you will be able to use that cooler into the future as well. $120 isn't that much for a multi-generational cooler that will handle anything you throw at it.
Been using mine for idk, maybe 2/3 years and not a single problem, MSI B550M Bazooka is the motherboard, still stock cooler, yeah sometimes pushes a bunch to 80/90 degrees, but not single problem, been wanting to upgrade to a Ryzen 7 5600X and AMD Radeon 7600XT
If you are a 60fps player i recommend locking your fps to 60 in Nvidia control panel or Amd relive. The result of that is you gonna have a better gaming experience because the frame-time is gonna be more stable and your cpu is gonna have an easier time working because you are not asking a lot from it.
I had problems with performance bias set to auto (does a small Overclock to the CPU) but seems to be having a problem on that cpu or bad bios thermal throttling (lack of it in automatic mode)
I want to say a few things. OC is not always OC. Different manufactorers have different boost values in their bioses. Some are higher, some lower. There was a commotion in the past for CPU reviews, that had an extra boost on certain boards, thus giving different results.
I bought a 3600 this year (2024) for a budget PC. It came with the Wraith Stealth cooler and it idle at 35 Celsius and rises to 55-65 when gaming, reaching a maximum of 77 when running Cinebench.
My ryzen 5 3600 regularly hit 80-90 degrees with the wraith stealth since day one. Bought an aftermarket cooler and applied an undervolt, allowing the CPU to stay under 75 degrees at max load (50-60 degrees during gaming sessions). Happy I made this decision, because performance remains unaffected after 3+ years.
Well, reduce the voltage, raise amperage and watts ( PPT / SOC / TDC / EDC ) limit all the way to the max and you will have the best performance possible with low temps. Do this via AMD Ryzen Master, it's very easy. My R5 3600 is completely 200% stable on all games at 1.125V 4.2GHz, the temps never go above 55°, this cpu is very good for real. But the default settings are completely garbage for some reasons, very high temps and the amperage/watts limit is way too low and is always at 100% limiting the performance of the CPU, but at least AMD give us the software to change that easily.
This one is pretty interesting to me. I have a 3600 that has never worked at stock settings. It was a very hard issue to diagnose, because I have tried so many things, and it tended to present as more of a GPU or memory issue, but after thorough testing, I found it did not work at default settings. I eventually had it stable for over a year at 4ghz all core locked, but eventually it started getting worse and getting less stable. This was across 4 motherboards, 4 GPUs, and 3 ram kits. This was a launch 3600 I got as a gift, so it took me a long time to determine it was a CPU fault, as there were a bunch of bios issues at launch.
i have a ryzen 5 3500 and an RTX 2070 super... is that ok?.... i am scared now - SOMEONE HELP ME Also: running memory at 3200 with DOCP (XMP) the memory's official rated speed Edit: i checked 100% util CPU temp - in Davinci video render, adobe PP and AE video render and cinebench - temperatures sit on 84-86 degree (i do thermal paste changes frequently - arctic Mx4) also gaming hits max 75-76 degree.... i have PBO on at 10x scalar (200MHz) max cpu power draw is 60-65w and GPU power draw is 230-240W max i am also running a wraith stealth (smallest cooler of the stocks)
I loved my 3600x because it was my first solid cpu. My first pc had a fx 6300 and while it could play esports titles, it was lacking in newer more demanding games. The 3600x was a breath of fresh air compared to that and it treated me well from 2020-2022. I ended up going with a 5700x recently and while that wasn't as big of a jump as the fx 6300 to the 3600x was, it still was a solid upgrade because its a power efficient 8 core chip and runs cool. I just got a noctua nh-u12s redux a couple days ago to pair with it, and my temps and noise is way better than my wraith spire cooler was on my 3600x. My 3600x will always have a place in my heart though. I will probably find a friend or relative to give it to so it can have a new life instead of sitting on a shelf
I started with 2013's AMD FM2+ socket, cpu A-10 6800k (total garbage) Than got the Ryzen 3600 when it launched. And upgrades that to a 5700x last week :)
@@TheTryingDutchman oh wow i can only imagine the feeling you got when you first booted up the 3600! I could tell a difference right away by just scrolling around in windows. Its crazy how much AMD cpus have come in just the past 5-10 years. AM4 has had great longevity!
I got the 3600 when it first came out in 2019 and it was a very buggy experience for a few months. I spent hours, all day long, just testing, bench marking and adjusting settings. I managed to get my Ryzen 3600 at a 3900mhz all core clock at 1.25v and stayed there. With basic cpu coolers and with everything left to default/automatic, the CPU would shoot up to 85 - 90c because of precision boost overdrive going absolute insane. The voltage spikes on PBO was super dangerous. I saw spikes up to 1.48v when PBO would activate and cause the CPU to overheat. Im happy at 3900mhz all core at 1.25v. The out of the factory default settings are faulty and volatile. Even 4 years later after updating to the latest mobo bios and amd chipset drivers, the default PBO is very problematic.
Mine did degrade with a decent cooler, so i think it's more then a temperature issue. I think it's just an combination of trash bin dies + high voltages. Your CB23 scores are on the low side tho, i get even at 3600mhz (turbo disabled) 8400 multi core and 9300 with turbo enabled.
And people called me crazy for tuning my 3600 to under 70C under load lol... "It's FINEEE at 1.45V, it's fineee at 85C..." No, it's not. My Intel i5 6500 I had before ran at a little over 1V full tilt and did not exceed 65C, ever. And that was on 14nm+++++++, no reason a 7nm product like the 3600 would need 1.4V+ to reach turbo speeds. Yes, AMD pushed their CPUs above their sweet spot for a few % more, but they did it with every gen, resulting in the house-fire hazzards that are the 7000 series. Here are some examples : - tuned a R5 1600 for a friend - from 1.35V core / 1.2V SOC down to 1.27V core / 1.05V SOC - same speed, XMP stable, ran ice cold after that - under 55-60C at all times. - my old 3600 - my sample was probably the worst one in history , was not able to reach turbo speeds like ever, ran at 85C under load with Wraith MAX and 55C idle - L-O-L. Default ran at 1.45V and about 1.3V+ under load (wtf). After some tuning, was able to use a -0.05V (50mV) offset and tinkered with the RAM speed and voltages - result was under 70C under load. Any higher offset resulted in instability and/or losing performance. Called it a day and lived with it. To this day, this is the crappiest CPU sample I have ever purchased. I even regretted not getting a 1600AF instead fml... - now living with a 5600X and besides the speed advantage, it is VERY VERY power efficient, took a -50mV undervolt and never goes above 60-65C. Same Wraith MAXX as before. Donno if karma is real and I got a golden sample or the 5000 series are much better by design, or the 3600s were all crap, or all of the above, but I am satisfied. From my research over the net when I was trying to subdue my raging 3600, I can tell you all of them take AT LEAST -50mV (-0.05V) offset without any disadvantages or clock stretching. Some people reported up to -75mV on a 3600 and up to -125mV (!) on a 3700. So it depends on the sample that you got. Start with -0.05V and go up until you see performance degradation or instability. See how far you can lower that Vcore. Every little bit helps. From all that fiasco, I will ride my 5600X to whenever it can carry me and then buy Intel again. I am sick and tired of tinkering with AMD CPUs out of the box just so they don't burst into flames, it is ridiculous. OR, I can buy one of their APUs - they are actually another story - monolithic, sip power and almost as fast as the regular Ryzen CPUs. And have none of the overheating problems. I have 3 of them at work (4650G, 5600G and 4600G) and they are quiet little beasts. We'll see what AMD has to offer in a few years time.
A little undervolting and good cooling can make your CPU last for decades and you are so right about the low-end coolers not cutting it as software demands more from the CPU. Even my old daily driver - just used for email UA-cam etc. has had a cooler upgrade. It was just an i5 with stock cooler and I noticed the temps were starting to get quite high. Now all is good and cool again. And for all you guys out there who do not clean your rig regularly, remember dust = heat!
@@ausfoodgarden The game in question in this video. Death Stranding is an AVX instruction game. Which means requiring 64 bit compression from (2) 32 bit buss lanes from the system ram requires more milivoltage to the CPU. Lowering the clock speed of the cpu is the only thing you can do with AVX instruction titled games. Compromising the voltage in AVX games causes a loss to CPU texture rendering:
CPUs should have built-in thermal protection and should not fail even if the cooling is insufficient. It appears that AMD sacrificed reliability for benchmark scores.
don't blame the cooler, I'm running a 5600x with stealth cooler. CB R23 temps never go above 75c(no I didn't change throttle temps). Something else is wrong in your setup. I had a 3600 in an earlier build and it was also not running hot. Probably that cheap crap motherboard you have.
By default, my Ryzen 7 5800X was consistently hitting above 80C, sometimes ekeing towards 85C, where idle temps were at an elevated 50C - 60C. I thought that it was really obnoxious, and I did my due diligence in mounting the cooler correctly. Disabling PBO2 barely made a dent, but long story short almost everything was tied to Windows' power management settings and I just forced Windows to use the "Balanced" power plan, on top of just disabling PBO2. I don't think I made a manual thermal limit, but after setting everything that I think could made a huge difference, it went back down to the usual 25C - 32C idle temperatures, and never went above 45C - 50C (depending on season) when gaming. I have a Cryorig H5 Universal, and it should be more than capable of keeping that thing tamed. Seeing Ryzen 70xx series processors have 90C at idle as the "norm" was incredibly outrageous and inane of AMD to do. This is absolutely wrong. This is a huge signal to every other consumer that AMD is just continuing to, by default, overclock and maybe overvolt the shit out of their chips just to meet a specific performance improvement goal and appease their investors, and it isn't helping that most motherboard manufacturers tend to go a bit crazy with voltage settings, too. This is "fun" for overclockers, but the target audience should really be towards the average consumer. "90C is below the maximum operating temperature of 100C, therefore it's okay" is a *bad* excuse to make when selling a chip to most people, because that's when most thermal throttling should occur.
I just built another new PC for a customer with a Ryzen 5600 with a stock Ryzen cooler. First thing I did was test performance and heat levels, and noticed that already the CPU was at 60 degrees at idle. I then did a benchmark with Heaven and under load the CPU temp didnt even move away from the 60s. I finally pulled the Heat sink off the CPU, and although I had applied a fresh coat of thermal paste on the CPU, I redid the thermal paste and put it on extra thick. Voila! The idle temp dropped from 60 to 40. That's the first time I had to use an extra large amount of thermal paste on a CPU. Even my new 5800X3D works great with standard amount of thermal paste, but its cooled by a Cooler Master Hyper 212 Black tower cooler. Bottom line is, if your using a stock cooler, don't be stingy with the thermal paste.
I think your unit is defective and that's that. I have a Ryzen 5 3600 with 74 max temp in synthetic tests, furmark etc. I do have some good airflow in the case and the CPU has an Arctic tower cooler
I had a similar problem when I bought my 3600 back in 2020... Even with a hyper 212 cooler it would idle near 60°C and instantly jump to 80s and 90s when under any load, getting the fan to ramp up and down all the time. Within the first week I learned and did all the same tests you did and some more, settled with an undervolt, running 4,1 Ghz at 1,12v and dropped temperatures to 70s even under stress, no performance loss or clock stretching. I too had the cpu being fed up to 1,4v by default, but I don't think that's the real reason for the excess heating as the cooler should've been able to handle 80w just fine. After some research, I believe it was a bad batch with poor die to ihs conductivity or uneven surface, as some people who lapped the cpu got massively lower temperatures without messing with voltages, so the heat was accumulating too fast in the cpu without being passed to the heatsink.
I have a Ryzen 3600 with a Stealth cooler on an Asus Strix B450-F mobo. (RTX 3080 if anyone is curious.) It's still on default settings as far as I know and I have no problems.
I find it becoming absurdly loud for what it is. The cooler and fans go rampant like crazy every single time up to v1.412 volts then it downclocks again constantly due to PB. I just disabled Precision Boost and keep it like this , plays all games fine just not as good as it could but i don't need to hear my case fans and my cooler running like a jet engine it has as much noise as a old PS4.
Well, if you use a manufacturing process that is mainly developed for apple's iphones with a lifespan of 5 years and planned redundancy, what do you expect?
TBH, AMD had upped the base core voltages on all their CPUs that are Zen 2 or newer - I still remember seeing 3600X reviews back before corona hit (and I learned how to tweak hardware) and thinking "Hmm... Nah, it's too damn hot, my 2600 is a more sane option". 5000X and 7000X series processors basically have the same symptoms (which luckily were remedied in the non-X versions to which I can attest as a R5 5600 owner) and the fix is basically the same that you showed (undervolt + TDP limit).
I just turn off the "core performance boost", that means it won't go beyond 3,6ghz. I don't notice the minimal performance difference, however I do notice the huge difference in heat and noise output. Currently I'm running a simple Noctua tower cooler, the fan never goes above 400rpm. Temperatures during 100% load test do not go beyond 55˚C. I'm using the silencio 550 case, a case often referred to as "the GPU owen"
Little bit of confusion in the comments of this video so hopefully I can clear it up here. volt-amps and cooling is what you give a CPU, heat is what you get. No harm can come out of checking your temperatures while you are gaming, and making sure they are, say below 70c. Heat also causes combustion, which is why it's so important to keep it at low as possible in the everyday PC. Ultimately though, on the a320 (without changing our cooling solution), we can reduce these temps by undervolting, and setting in a max temp limit (which also reduces the clock speeds to levels that consume less voltages too).
So I was working more in reverse order in this video. However the video is designed to help people without much knowledge on the subject too.
As for the aggressive voltages, AMD should release a bios update, with the option to select a more conservatively tuned CPU, as we could only really undervolt -.075mv before we reached instability on this setup.
@@charlesg5085 Thats very overkill for the 3600, the stock cooler will do just fine.
love the top gear retrowave track in the video big fan of that game series. crazy how good the utilization is on ryzen now its awesome
What I used to do is lower the cpu clock speed from 4.2 to 3.6 ghz in ryzen Master with just a couple of clicks and I would drop 20c right of the bat, but now that I have a 6700xt I feel like I’m giving up a significant amount of performance by doing that
Factory coolers have always been rubbish!;-)
@DEA GT idk man the wraith prisms pretty solid for what it is. A bit noisy but it keeps my friends 3900x cool stock. That's pretty awesome to cool a 12 core chip
Investing a good cooler could extend your CPU lifespan and it can probably sit in your next rig then.
Looks an awful lot like AMD got greedy and harvest more bad modes and clocked them high enough to just fit into a 3600 chip. I.E theses cores really belong in a lower end CPU.
TjMAX of server parts is lower with a reason. Also i have a thinkpad, it doesn;t allow the CPU to go higher then 72c. Most consumer stuff just pushes parts too hard cause they are expecting a lower duty cycle... and if it dies outside warranty it's not their issue anyways.
Exactly .. my Noctua NH-U9B cooled my Intel Core 2 Quad Q9550 (socket 775) for 13 years .. now it is cooling my brand new AMD Ryzen 5 5500, thanks to the 8$ AM4/AM5 socket adaptor I bought from Noctua .. this cooler was a great, great investment I'm telling you ..
@@brokeandtired it's mainboard makers defaulting the core voltage .05 - .15v too high for the chips. More voltage makes higher temps and the voltage is what's killing the chip, it will slow down when it's too hot it won't do anything if it's being shocked to death.
Very true, about 5 years ago I had a I7 8700K based system built. It came with a huge Gigabyte tower cooler. The system is still motoring along nicely, the only recent changes I had done to it, was to have a RTX 3060 replace my trusty 1070ti and a extra 1TB SSD put in. The hottest my CPU has ever gets is around the 70 degrees C mark, and that is when doing a heavy work load.
Interesting to see more being looked into this. Greg has had quite a few defective 3600s come through on fix or flop so it’s interesting to see it be more widespread than just there.
Not only 3600 in his videos
DEFECTIVE* roughly translated to 'some goofy twat killed his cpu coz he thought OC was a perk XD
ryzen 7000 series with their 95c temperature 24/7 enters the chat.
This is fine.
Amd says it’s normal but I call bs lol 😂 they won’t last
@@cppctek What about Intel's 13th gen 100*C out of the box? 🤔
My 9900k required a helluva lot of cooling to keep it from hitting 95 to 100c. I have a 5600x and 7700x as well now that I built last year. My 7700x doesn't go over 80 with my 280mm AIO in my Corsair 4000D Airflow.
Those AMD reps are lying their underwear's off.
Back in 2020, I immediately noticed my 3600's temps were very high for my taste, even with a good air cooler (arctic 34 duo), like spiking instantly to low 70s and reaching high 70s during a sustained high load. But it turned out those were pretty much normal out of the box thermals.
The solution was simply undervolting, no temp limit needed. I use CTR for fine tuning and OC but anyone can do it easily from the bios as you pointed out. Great vid as always Bryan!
undervolting, or buying a very basic AIO could resolve issues too. but I guess thats a moot suggestion because buying an AIO for a 3600 would have been excessive for a budget 6 core cpu.
@@pauls4522 IDK.... I put the Thermalright Peerless Assassin with a Ryzen 5700x and I could not pay it enough for it to go over 60c. Thermals were in the 50's once I used a voltage offset.... and that's running Cinebench r23 that uses the CPU much more than any game will.
The Arctic eSports cooler is okay but not very good. It's more ideal for those old i7 processors like the 6700k. Those may have had a 91w TDP, but they'd use maybe 50w while gaming and 65w under stress.
@pauls4522 a budget 240 is like a hundred bucks, I ran a corsair i110 on a 3600 for years and it never went past 50° under load
Stuff like this is why I like your channel. Getting the most out of older hardware, and the breakdown of the BIOS settings really gives knowledge to someone like me that is still somewhat new to this wonderful hobby. Can't tell you how intimidating some of the Asus bios settings are to me. I'm a Weiner and afraid to mess with a majority of them. But I have a little more confidence moving forward now thanks to videos like these. Thanks again!
If you're intimidated by Asus you havent seen the barebones Gigabyte ones 🤣
I have a 3600x and it suddenly died after 4 years of usage. Have a liquid cooler kraken z73 though. Thanks for this vdo.
Same here, though I was using hyper 212 evo on one of the systems. Both died after 3 years with only 2 months separating them even though they had different motherboards and cooler.
My best advice for longevity is to just cap your fps, it doesn't even have to be an aggressive cap just at the higher end of your avg, or at refresh for non competitive titles. It will feel the same in game but your components will stop running at 100%, and temperatures can drop quite drastically. Giving it that little bit of breathing room really helps.
I dunno man. I've seen that video where Der8auer overclocked for 4125 hours (6 months) straight & there was almost no degradation.
👍
still run in a 1060 6gb. still struggles to get 60 fps. Sadge
the stock cooler on the 3600 is just garbage. the stock one from the 1600 was way better
I always enable Vsync. My system sounds like a jet taking off until I enable Vsync then it sounds like an RC jet taking off 😂
I heard on gamers nexus that some am4 boards use too high a voltage to make their board look faster than the competition, i think that might be at play here as well for some people?? It was a pretty big deal
Yeah, keeps the CPU happy and boosted up, it's acting like an involuntary overclock.
Well, I did watch Buildzoid’s video about chip degradation, only high and low temperature cycle will affect gpu (solder balls crack), and cpu are highly affected by electro migration due to high voltage or the silicon quality itself. Notice that, overclocking a cpu with liquid nitrogen can often kill your cpu too. Therefore, I have to disagree with Tech Yes City if he says high temperature is the cpu killer.
@@nhozdien5058 The solder ball issue is only really a big problem when the underfill is wrong for the temperature variations it will see, 100% it's the voltage killing the chips temperature is protected against the chip will slow itself down for heat, it will allow itself to die via over voltage every single time though. The high voltage is what's causing at least some of the high temps as well.
@@nhozdien5058 tech yes city is actually right because this has been a thing for very long time: voltage x amps is what kills the CPU but temperature is a factor of how quickly that CPU will die
internal resistance of transistors rises with temperature which brings up the amps,voltage and with that wattage in order for CPU to not have stability issues
this was always the thing you had to do with ryzen CPUs because they ran hot as hell with stock coolers and asked for both decent undervolts and cooling solutions
@@xthelord1668 The resistance of copper does increase as temperature goes higher. However, silicon decreases its electrical resistance as temperature increases.
In the beginning of the video, Brian kinda gave me that vibe of saying high temperature was the CPU killer. This was the reason why I said I would disagree if it was his implication in the beginning.
As the video went, I saw his conclusion and couldn't disagree.
Going into AMD CBS and changing the TDP/PPT/EDC/etc settings is the best solution (other BIOS settings might also be needed to prevent the BIOS from ignoring the PPT). That is typically better than messing with voltages or frequencies because single-core / fewer-cores operation won't be impeded as much. Just lower the package power limit to something reasonable. I do this on all my machines, even the high-end ones, because otherwise they use double the power just to squeeze out another 10% in performance or so, which gets into silly-land pretty quickly.
The thermal temperature limit in the BIOS is not as good a setting because it is a far more indirect and slow servo. PPT is a very fast servo.
With my current 5800X, capping the power just gave more performance.
It'll try and check this out, any more tips or steps on how our do it?
Reminder to everyone that your Ryzen 3000 is probably not that old and that the warranty is three years. Check if you can RMA if yours goes bad.
If you are in Australia, you may be able to argue that the lifespan is too short with the ACCC if your CPU fails shortly after the expiration of the warranty period.
dont be ridiculous. this video was stupid. 3600s are not failing at all.
@@p4radigm989 It is ridiculous to let people know they have a warranty if their CPU fails? lol
@@MarcoGPUtuber no, but they are not failing any more than any other CPUs if cooled improperly. This Tech YES video was clickbait BS.
@@p4radigm989 So why are you bothering me about it then?
@@MarcoGPUtuber because you are my son. I need to protect you from fake news
I'm using this CPU since its introduction to the market, and quickly I have realised that you can optimise it with Ryzen Master too: I can set whatever TDP target I like (usually 55W for transcoding, and 65W for gaming). Also set the "CPU Boost override" to its maximum 200Mhz value (so it drives speeds that much higher with the same voltage). This setup works very well even for the stock cooler :)
or just buy a better cooler and get more from your 3600🤣🤣🤣 or turn on vsync🤣🤣🤣
@@raven4k998 you want him to spend more money? STFU!
@@raven4k998 Both are valid advice! :) And I do follow them: I have a Brocken 3 for a few years now (bought for the 2700 originally) , and I always turn on vsynch (I just prefer it that way, and don't care much about the potential latency increase). And yet I still use Ryzen Master the way I described above: in games there is no speed loss (because it is never 100% on all 12 threads), and in transcoding, I can live with the 10-15% speed loss (if that) for much lower temps & silence :)
exactly get an ag400 or something similar for $20 and can use on future higher end cpus that will make quick work of cooling an r5 3600
In my experience, undervolt it from 1.4V to 1.2V do the trick. It seems the latest BIOS update make the CPU consume the power more than it actually needs and more power=more heat.
same i just put the voltage to - offset and temps dropped.
... mine will insta crash with anything more than 1.1v
A lot of board manufacturers tell you which is the highest BIOS you should use for a particular CPU, so that's more than likely your issue.
Read the support pages don't just go updating your BIOS.
A lot of motherboard vendors violate AMD's standards and overvolt the shit out of CPU's in order to try to win a few FPS in performance reviews. My Gigabyte TRX40 Aorus Master killed two Threadripper 3960x CPU's back in 2019 when it was new, at bone stock settings.
This is a huge, well known problem and the Motherboard vendors keep doing it on purpose, it is not an unintentional bug, and it will damage CPU's. AMD needs to crack down on the board vendors.
Some (but not all) motherboards vendors have responded to criticism by updating their BIOS to correct this issue, so the first thing people should do is to check for a newer BIOS. It might have been fixed.
You destroyed 2 3960x ?
First thing I noticed when I made my 3600 setup was high CPU temps! Even at idle. Doing some research I found that Mobos by default set core voltages very high. Mine was 1.4v at the time. You need to manually set the max core voltage ( or offset) and find the sweet spot. Mine is set at 1.25 volts (over clocked) and it give max performance while keeping cool.
Truth, most boards seem to have the vcore about .15v over what the chip actually needs to run, that excess voltage is producing excess heat and lots of it. Heat isn't the problem the chip will protect itself if it's too hot, it won't protect itself if it's being electrocuted to death slowly by over voltage though.
Interesting. Mine idles just in the bios at 45-50 degrees and it always annoyed me. Will need to try this when I get a chance.
This!
My game mode on the mobo auto set it to 1.45v.
It got ruined a year in, and i kept bsoding every game
I swear these mobo manufacturers out here deliberately trying to fry our CPUs so we have to go out and buy new ones! 😅
I dropped my v-core 1.21 and set clocks to 4.2 on my 3800x. At default the cpu fan ramped up whenever opening a browser neverming gaming, annoyed the hell out of me.
As an owner of a Ryzen 5 3600 (and A320 mb) this has got me intrigued - admittedly I've got an aftermarket air/fan cooler. I haven't seen any kind of temps like this. Thanks for this investigation. Love your content :)
yeah try upgrading your cooler for better temps simple🤣
Omg same 😅 I'm using an Hyper 212 hope I'm fine.
Same
Your vcore was peaking at over 1.45v that is WAY too high for that silicon, these things aren't burning up they are being electrocuted to death by MB manufacturers trying to eek out more overclock headroom with higher voltages. What you did was not so much an under-volt as it was a fix of the motherboard's default over-volt.
The thing is that it appears that 3600s actually need that much voltage. My 3600 also runs at about the same voltage but the problem is that I can only undervolt by about 12.5mv before I start getting BSODs while gaming (in easy to run games at 120 FPS). For reference, my Vega 64 runs at an undervolt of -150mv pretty much without losing performance. I suspect the worst dies were turned into 3600s as yields on the 8 core Ryzen 3000 chiplet should have been pretty high. So 3600s need a ton of voltage to operate correctly.
Yeah, being electrocuted and reddit and any other places are not filled with unhappy customers that their CPU and mb are frying like crazy. It's almost like amd engineers knew how to design a CPU properly.
Bro, those voltages that you showed are CRAZY! My 3600 runs with 975mv while 3750mhz , temps - 50° max in cinebench r20.
Undervolt is a MUST!
At the same time my second machine with 5700x runs 1008mv, 4250mhz, 55°
Saving that silicon for the children to inherit it lol
if you remember at launch, we had problems with high voltages on 3600 and zen2. mine was doing like 1.4v. it was fixed by agesa from amd later on.
Strange, my 3600 was a great CPU and was always cool, I only recently upgraded to a 5800x and was surprised how much hotter it was.
ditto with my old 3700x and my new 5900x , using the same noctua nh-d14 the difference is huuge
I have a NH-D14 and went from a 3600 to a 5800x3d and get the same temps lol! After using curve optimizer of course
You are comparing a 6-core cpu with a Single-CCD 8-core cpu
You should down volt the 5800x, that's what I do, it decrease it running temp by like 10-20 celsius with no drop in performance keeping all the core at 4.6ghz. The 5800x has the same voltage as the 5900 but with 2/3 of the cores so each core has way too much current going through it for no reason since the cores are also the same. A 5600 has 65 tdp for 6 cores, a 5900 has 105 tdp for 12 cores, the 5800 has 105 tdp for 8 cores. That's 10.8 tdp per core for the 5600, 8.75 for the 5900 and 13.125 for the 5800.
Damn I wanted to upgrade to a 5800x :(
Ryzen CPUs have a history of being over volted. Always under volt them.
Interesting video Bryan.
This is actually pretty disappointing if it's the case (higher temps), given their relatively young age. Many CPU's in the past (eg. Intel 14nm++++ coffee lake; comet lake) would frequently hit these temps and not fail prematurely, so this is something AMD needs to look into - especially if the chips aren't thermal throttling and staying within AMD's design window. It seriously makes me wonder if it's a long term high voltage issue (eg. certain manufacturers over-volting by default), or if it's related to increased rates of degradation of the TSMC 7nm node at extended elevated temperatures/voltages...
The 2080Ti you previously tested with is significantly more powerful than the RX 6600, so you would expect temps would have been equal/higher running that card.
From my experience the Precision Boost Overdrive settings under the tweaker settings on Asus aren't always doing its job at disabling it and you'd need to disable it under one of the AMD tabs in the advanced settings.
The extreme overvolting with PBO has been a thing since first gen ryzen and has caused voltages of ~1.45V back then as well. Disabling it always had massive thermal and efficiency improvements while also lowering the voltage to more reasonable levels on every single ryzen system I have built so far (including anything from 1000 series to 5000 series), so I'm always disabling it because that little bit of extra performance doesn't seem worth the thermals and unhealthy voltages imho. I'm surprised that nobody really talked about that yet and, while I didn't test with any 7000 series CPU myself, I'm pretty sure that's what is mostly leading to the high temps everyone reported on the 7000 series launch. If anyone wants to verify that with a 7000 series CPU feel free to respond, am curious to see if it is indeed still the case.
yes exactly. average users and gamers do not need pbo. my 3600 with pbo on went as high as you said upto 1.45V and i didn't see any performance gain in gaming (by see i mean notice it, you can obviously compare with a benchmark but the negligible gains do not justify having pbo).
with pbo disabled my 3600 would need just under 1v to operate to its max 3600mhz base clock and i could even lower the wraith stealth prism fan speed to 50% and would never need more for it to run cool and super quiet and very low power consumption too (even if i loaded all cores 100% for stress test it would stay around 70-80 degrees with fan still at 50%)
the 3600 can easily handle even games like cyberpunk, red dead 2, etc... at 1440p with my 3060ti i had most high settings and gamed at 60 capped fps. i went not with the 5700x cause i gave my mother my good old 3600 but it's still great cpu for gaming
I'll check this, thanks for spending the time to write the comment, appreciate ya
In my experience, the Ryzen 3000 series have usually run pretty cool and have been easy to cool. I am wondering if the TIM has started failing after a couple years of use. As they are soldered, its not exactly easy to check. Need Roman to pull one of these excessively hot CPU's apart.
I'm thinking the same thing. It's probably expanding and contracting a lot
Put a good cooler on it and change your fan profile in the bios , keep an eye and set a max voltage if needed but I find temps are fine with a good board and cooler
If the voltage is too high, and it objectively is on a good deal of motherboards at default. Keeping it cool won't save it, the temps aren't what's hurting it the excess voltage is.
@@bdhale34 temperature also doesn't help. It impact semiconductor life since...well, since first transistor ever made.
@@bdhale34 what are you talking about? AMD itself has said that those 1.4v spikes are normal, if you set Manually those 1.4v everytime it's obvious you're going to hurt your CPU
Keeping it cool makes it last longer and Will save it
@@bdhale34 Not true, AMD itself stated their stock voltage powerspikes are safe. However, who knows if they lied.
@@bdhale34 need good vrm with ryzen chips i have 4 x 3600 and 2 5600x cpu and voltages and temps are stable, max temps under load are 72 deg c and max voltages are 1.35-1.4v on the 3600's , the 5600x have slightly less voltages out of the box applied and subsequently run a little cooler, also all my chips are under hyper 212 air coolers and they do the job perfect, the std ryzen coolers are junk for any intensive purpose.remember amd put out a statement regarding voltages in the ryzen line and operating temps when chips were released.
good luck :-)
I think the problem lies with stock voltages rather than high temperatures (although the two are correlated somewhat). On my own Ryzen 5 3600 + CVN X570M platform, the motherboard will push 1.475V into the CPU by default, and I think that is the main cause of this degradation. I had got it tuned in to run at 0.86V and 3.4Ghz by default, and manually going into the Ryzen master to push it to 4.2Ghz all core with 1.16V sometimes, and so far, its overclocking potential hasn't degraded much from the three years of use while temps look great. I think the issue lies with the difference between the earlier batches of Ryzen 5 3600 and later ones, which uses a more refined 7nm process that require much less voltage to run at the same clocks. Motherboard manufacturers want their board to work out of the box with all those Zen 2 chips, regardless of their silicon quality, and that could be the reason behind the aggressive voltages, but when the chips themselves improve to a point that it don't need that, it just became a factor that drove the power and temperature up and efficiency down, while eating away at the lifespan (TSMC recommends a max of 1.3V for these chips, and some motherboards go way past that). Those unstable Ryzen 5 3600 that you have might be a result of a normal computer user leaving the motherboard bios to regulate the voltages delivered to the CPU, and over time it just tears away at the silicon quality.
Be careful with Ryzen master, the voltage set there is actually about 30mV higher than when set in Bios, so your 4.2Ghz at 1.16V, is more like 1.19V.
Greg Salazar also has some videos running to similar issues with Ryzen 3000 series
I think the only thing you forgot to mention was that if they use a Prism there is an option switch on the bottom side to change the fan speed to High.
Forgot about that, thanks!
@@NotANOVERPRICEDGOUGINGTECHFAN But your AIOs might be running quieter. You might need to adjust the fan speeds on your AIO / motherboard settings to increase air flow?
@Gamer's Always Get Expensive Scrap Wafers! The Wraith Prism is a down firing cooler. So it might just be blowing air onto one of the temp sensor zones.
@Brian Babin Which motherboard was this with?
@@techyescity my system integrator stress test my cpu and never reached 100C, but when I run it, it does reach it
I'd be using hwinfo64 to check the power reporting deviation, this was an issue early on with motherboards shoving more voltage than reported. This was for the most part solved with BIOS updates but i'm sure there'll be countless boards running theses chips out there that were never updated to solve this problem.
Checked mine out of curiosity. In Cinebench R23 it got 9173 for multi and 1231 for single and temps were around 65°C. I do have slightly better than stock cooling though (BeQuiet! Shadow Rock 2). I bought the 3600 in March 2020, so I've had it for almost 3 years
Thanks for posting your numbers, those temps are great.
There is an option named CPB in BIOS which is ON by default.
Turn it off if too hot or too loud, then CPU will maintain at 3.6GHz quietly.
I noticed this back in 2020 with my 3600 CPU, The stealth cooler just wasn't up to the task so i replaced it for a prism rgb and also undervolted it slightly, the temp were night and day difference!
Yeahhhh 3600 + prism gang
I've just ordered one of these from amazon, gutted to find there are issues with them an hour after clicking pay. The bundle is coming direct from ADMI as a preassembled thing, so do you think it might come with settings changed if they know there are issues?
the snowman cooler sure had a wobbly looking fan when you showed it with the fan spinning
i see that you tested this ryzen 5 3600 with an a320 board and these problems appears, have you tried using a higher tier board like a b450 to see if the problem is still there?
I have a 3600 from late 2019 as well paired with a Mugen 5 cooler. It's undervolted with a negative voltage offset set at -0.125 and while gaming it's always between 45c and 55c max. Pretty impressive indeed.
bro do u live in a cold country
@@annaaffkhan just buy gammax gte v2 white deep cool
Yes pretty impressive, you just need undervolting and a $60 cooler to keep it relatively cool lol
I noticed that 5600X with the included default cooler would go up to 90 degrees regularly when gaming. I just opted to buy a big Be Quiet cooler and after that the CPU never got over 67 degrees even under 100% load when compressing and decompressing files for half an hour.
@@WarbandGames I try to avoid liquid cooling when possible but for high end stuff it's kind of necessary.
Personally I have never built a gaming PC with a stock cooler, I always recommend upgrading the cooling for gaming PCs as stock coolers are never intended for cooling CPUs while gaming. If the person is on a limited budget I usually recommend they wait and save some more money for a better cooler or downgrade a peripheral or 2 so they can afford it, it is allot easier and cheaper to upgrade peripherals later than replace failed CPUs. There are allot of good budget coolers out there like the Snowman you recommend (I like the Cryorig coolers) so why not upgrade right from the start.
I've had a variety of 3000 and 5000 Ryzens and an offset undervolt between minus 0.050V and minus 0.075V is almost compulsory for sane temperatures. Never used hard temperature limits though.
Changed from a 2600 to a 5600x both using a Dark Rock 4. Pretty impressive, the 5600x with curve optimized OC performs so much better while staying cooler than the 2600.
Absolute worthy upgrade.
While being on it, I also repasted my GPU that was running extremely hot looking at the Hot Spot temperature. Dropped the °C by 20-30° depending on the game. Feels like an entirely new PC x)
In the US, the Thermalright Assassin is often on Amazon for $17-25 even for the RGB versions. I use it on a lot of PC builds.
was going to say, the assassin 120 is a much better deal and you don't have to wait for it to ship from china.
8:32 3.6GHz at 1.4V on 3600? No wonder those temps went that high!
All you need to do now is just put a good cooler to be honest
yea so the CPU dosen't overheat
@@spreadinglove648 This won't help if the mbd BIOS is consistently pushing too much voltage aswell though, which may be more the relevant issue here. However, without a larger sample size and in addition testing with a decent B450, it's hard to be certain.
I bought a few wraith prisms and they are great for 3700X and below . I just like the design and the looks . Blows some air around the socket and Vrms too .
Takeaways from this is not just efficiency tuning, but also, do not skimp on your CPU cooler - basically any 20-30$ cooler will vastly outperform Prism and Spire.
Don't go over the Hz of your screen FPS wise, unless you're pushing for absolute minimum latency, it's inefficient.
Settings in PBO did nothing
Because they only apply if PBO is on
This is what I do with my 3600. Just an offset of -0.072 mV and the temps are in control. Also using a tower cooler helps. Still using the deepcool gammaxx 400.
Gammaxx 400 has been one of the best budget tower coolers for over half a decade now.
Have one of those cooling my 11700f running 4.4 indefinite boost, never sees over 72c, amazing cooler for the price.
@@nedjimb0 dude i took mine off when i upgraded from an i5-10400f to my i5-11500, and one of the plastic mounts finally snapped and broke :( best cpu cooler i will ever find for for less than $25, can't really find the LGA1200 version for cheap anymore. Ended up getting a Cooler Master 212 Black Edition cuz its all that my local Best Buy will stock for non-AIO's lmao
@@nedjimb0 im on the gammaxx 300 on mine and my brothers rig, cools just fine and is quite cheap.
@@dc-gm7lu Damn, on my ryzen 3 1200 pc with an OC to 3.8ghz @ 1.3v, it fine.
My brothers pc with a 3600x (or 3700x, cant recall) is probably fine, as I havent heard complaints and my laptop with a 5600H has been fine, the problems in it are other things, lol
Death Stranding requires AVX instruction set. AMD’s on the compute aren’t great at AVX instruction set. Its an intel patent and explains why it was running hot.
AMD’s version of AVX and AVX-2 is actually an emulated version of intels instructions. So the CPU will naturally run hotter as it decompresses 64bit and 128bit instructions.
Intel wasnt even able to perfect AVX and AVX-2 until 6th generation cpu’s with efficiency cores.
Prior to efficiency cores 4790’s and previous gen i7’s would either run hot or with first gen icore simply wont render.
AVX 64 bit compression, and AVX-2 128 bitcompression elevates cpu VOLTAGE.
Because x86 processors actually compute system ram in 32 bit chunks.
AVX 64bit compression means 2 sets of 32 bit instruction are simultaneously being sent to the cpu for a 64bit compression total. Instead of in 32bit individual instruction sets.
This means voltage has to increase 30 to 40% in mili volts to send a compressed 64bit instruction.
AVX-2 being 128 compression increases mili volts upto 70%. Since system ram is required to send larger instruction packages.
There for the reality is AMD cannot efficiently optimize AVX and AVX-2. As unlike intel AMD doesn’t utilize efficiency core optimization.
I find that undervolting is almost mandatory with some motherboards, i had a msi b450 board and it overshot quite a bit with the voltages and i quickly came to realize that undervolting was way more rational than attemting to oc that cpu since vdrop control was terrible and the ryzen 2600 did oc like shit so i would end up in the high 70's only to barely get 4ghz with a 240mm aio.
I have found that ryzen really likes to work around 1.1-1.2v cpu vcore so if you see that spike to over 1.3 or even hit 1.4v then definitely try to undervolt because that is just too much
Question my cpu voltage is at 1.1v never oc mine and i have a r5 5600x. i have the stock cooler and repasted it with arctis mx-5, whenever i play wz2 the temps goes upto 70degrees and voltage at 98%, is that normal?
@@kidskid8210 As long as it is 1.3v or lower and temps are normal it should be fine. If yours is at 1.1 that seems pretty good
I just leave it at auto because at the time thats what the recommendation was. Less silicon degrade that way because static voltage and manual pinned oc is somehow worse than letting it control its voltage and clocks when idle or under load. When things are set to manual all of the auto features are disabled, at least on ryzen 3000. Which at the time was different than intel and much simpler honestly.
As a long term 3600 user, I found that the poor cpu, under stock settings, tends to get blasted with way more voltage than it needs. Undervolting should be the first step. If you have a limited bios, I'd recommend Clock Tuner Ryzen to set some p1 and p2 profiles. You should be able to have your cake and eat it too.
I'm also aware that my cpu is a golden sample (I think it's a downbinned 3600xt), so I wont share the exact settings I use as most 3600's wont be able to keep up.
However a cautious starting point would be:
P2 (trigger at 10% cpu usage) 4200 MHz @ 1300 mV
P1 (trigger at 50% cpu usage) 4000 MHz @ 1150 mV
CTR has an automatic idle state it should default to if cpu usage is below the (trigger usage) value for P2. The values are very rough guide lines, and might be too aggressive still. Or you'll find more headroom to raise clocks further or to lower the voltage. Having a better cooler will also improve the speeds for any given voltage, or allow you to lower the voltages.
CTR has a "tune" button that's tempting to press, but for me it has never produced any beneficial results. It seems to think my cpu can hit 4.7 GHz. It cannot. Maybe you'll have more luck, but it takes forever to run.
Locking the cpu to 45w is a brute force approach but it does work.
This is why i always use manual overclock over auto, every motherboard pumps crazy voltages into CPU. For example on my Ryzen 3600 on auto it would hit 1.4/1.45v while on manual overclock i find it stable at around 1.25, the temps are like 20c lower.
Good call to at least check. I'm running PBO and i just checked and i'm not getting anywhere near 1.4V, like it tops out at 1.33V for super short time, actually staying around 1.3V. Maybe that's an issue with particular types of mainboards? Lots of lower end boards have old offset-only VRMs which don't really have precise voltage control.
Though i'm running Cezanne die (Ryzen 5500) and MSI B450M Pro VDH. So of course the results aren't transferable anyhow.
Motherboard BIOS should have 3 standard settings. "Load optimized" "Load Minimum" and "Load Manufacturer Spec"
PBO was actually a huge deal for me. After turning it off my voltages dropped a lot and so did temperature. I guess YMMV. Anyways, I undervolted to 1.25, overclocked to 4.2 all core and I have my very old NH-U12P on it. Now it doesn't crack 70°C even under sustained 100% load.
These temperatures were always present with the absolutely INADEQUATE Wraith Stealth. As someone who is still running a 3600 since 2019 I have always recommended people to get a better cooler than the included one even a 20 USD tower cooler keeps the 3600 safely under 70 C in 100% usage on stock settings.
It always really bothered me how many tech channels (don't remember if TYC did that too tbh) and people on the internet played up the 3600's included cooler as adequate and an advantage over Intel at the time, the reality is that those coolers should have been used at most for a couple weeks until you get a cooler if you can't get one right away.
So it's just the temps not anything other to major right and what cooler do you think I should use for the r5 3600 ?
You would think they would pair up the new Ryzen 5 7600 with a spire but no they have to give the bare bones.
naw it's not the cooler...
@@dyyylllaannn You do have to admit though it is part of problem it’s not the best compares to the spire or prism.
@@scarfaceReaper I can recommend the SilentiumPC Fera 5 if it is available in your country. Arctic freezer 34 eSports is also good.
If you are wondering why the cpu is so hot out of the box, it’s because of the voltage on ryzen 3rd generation boost was only achieved by excessive voltage in excess of 1.4v and it has been a problem ever since launch and amd claims it’s fine, but in reality it’s not and not only causes high temperatures but also causes cpu degradation. In my personal experience working with this CPU, and other third generation CPUS is the best bet you can do is a manual overclock and then setting the voltage as low as it can go before it crashes. For example, I took my friend’s Ryzen 5 3600 and manually overclocked it to 4ghz locked and then got the voltage down as low as 1.36v and his temps on the stock cooler dropped by 17c which is incredible and will really keep the cpu going for years to come.
you could run that cpu at max temp 24/7 for a decade and it would be fine
My 3600 is still going strong that I built for a grandson's gaming PC. No issues - air cooled, Hyper 212.
This is a great video on so many levels for the AMD community!
I have ALWAYS underclocked AMD equipment.
CPU and GPU.
I've always got lower temps and higher performance due to the clock map where it detects that the temps are lower therefore it will increase the frequency as long as it's stable.
I haven't found a need to limit my equipment thermally, but undervolting alone is fantastic.
I even do the same to my ram and run vigorous tests to make sure it can still pass memtest.
Example, if my ram can run 3600Mhz CL16 @ 1.2v, but crashes at 1.15v, I'll set it to 1.25v for some overhead stability. Saving me from needing to run 1.35 rated voltage.
This will reduce the heat on the motherboard & case, as well as the load on your motherboard and PSU. Every little bit counts.
I used to be of the mindset of just brute-forcing things to max and ramping fans up, but after reaching points of diminishing returns and reducing life expectancy, it just isn't worth it anymore.
Why modern CPUs are failing? Because of many factors. Some were working in bad conditions (very high overclock, high voltage, bad cooling). Bad working conditions can damage or degrade CPU (it is called an electron migration and such a CPU will not work properly with default clock or voltage). In some cases paste below cover (IHS) is dry and degraded. Such problems I saw with many CPUs (many intel models), not only with Ryzens.
Haven't had any issues with my 3600. Ran it with pbo on, auto oc on and i think a small under volt. It was on a b450 tomahawk(non max) with a cooler master ml240 aio. Replaced it with a 5800x3d last month. Will keep an eye on the temps when I repurpose the 3600.
Huge help! Thank you so much!💯
Like other people mentioned, the problem is the voltage that some mobos apply on them. Usually way past what one would call redline for these CPUs. The solution is RMA, but if it was me i'd seriously consider upgrading because the 3600 aged really poorly, it can't get 60 FPS on Kingdom Come deliverance max settings or No Man's Sky and has mediocre lows in games like Elden Ring and i doubt it'll handle these newer PS5 games (Hogwards's legacy?) well with how close in performance it is to the consoles, at least if you care about open world games which it won't handle very well.
I agree it's the cheaper mb and running in high ambient TURN ON THE AC
i havent had any problems with mine playing games plays skyrim heavly moded monster hunter witcher 3 all highest
@@jnightmare2386 I call BS on that modded skyrim claim there, quite sure it'll drop below 60 at any point where draw calls reach the neighborhood of 7500 which with heavily modded skyrim is everywhere, even with a 12600k it struggles to keep 60 at 9k draw calls, as long as you use a grass mod + dyndolod and ENB you're not running 60FPS on skyrim with a 3600, only in interiors. And MH i can break the 75FPS mark on a 3400G with max settings so that's not exactly a demanding title. Witcher 3? Maybe DX11 witcher 3, a PS4 game from 7 years ago, which i can also hit 120 FPS on a 3400G. The titles i mentioned are WAY more demanding than those you did, apples to oranges.
These days simply using a better cooler will give you extra performance. If you buy a good brand like Noctua you will be able to use that cooler into the future as well. $120 isn't that much for a multi-generational cooler that will handle anything you throw at it.
3600s are dropping like flies from what I've seen. Usually after 2 years of usage.
I've had mine since 2019 and never have had a single problem
you mean dropping what, which things?
Been using mine for idk, maybe 2/3 years and not a single problem, MSI B550M Bazooka is the motherboard, still stock cooler, yeah sometimes pushes a bunch to 80/90 degrees, but not single problem, been wanting to upgrade to a Ryzen 7 5600X and AMD Radeon 7600XT
If you are a 60fps player i recommend locking your fps to 60 in Nvidia control panel or Amd relive. The result of that is you gonna have a better gaming experience because the frame-time is gonna be more stable and your cpu is gonna have an easier time working because you are not asking a lot from it.
I had problems with performance bias set to auto (does a small Overclock to the CPU) but seems to be having a problem on that cpu or bad bios thermal throttling (lack of it in automatic mode)
I want to say a few things. OC is not always OC. Different manufactorers have different boost values in their bioses. Some are higher, some lower. There was a commotion in the past for CPU reviews, that had an extra boost on certain boards, thus giving different results.
I bought a 3600 this year (2024) for a budget PC. It came with the Wraith Stealth cooler and it idle at 35 Celsius and rises to 55-65 when gaming, reaching a maximum of 77 when running Cinebench.
My ryzen 5 3600 regularly hit 80-90 degrees with the wraith stealth since day one. Bought an aftermarket cooler and applied an undervolt, allowing the CPU to stay under 75 degrees at max load (50-60 degrees during gaming sessions). Happy I made this decision, because performance remains unaffected after 3+ years.
Well, reduce the voltage, raise amperage and watts ( PPT / SOC / TDC / EDC ) limit all the way to the max and you will have the best performance possible with low temps. Do this via AMD Ryzen Master, it's very easy. My R5 3600 is completely 200% stable on all games at 1.125V 4.2GHz, the temps never go above 55°, this cpu is very good for real.
But the default settings are completely garbage for some reasons, very high temps and the amperage/watts limit is way too low and is always at 100% limiting the performance of the CPU, but at least AMD give us the software to change that easily.
This one is pretty interesting to me. I have a 3600 that has never worked at stock settings. It was a very hard issue to diagnose, because I have tried so many things, and it tended to present as more of a GPU or memory issue, but after thorough testing, I found it did not work at default settings. I eventually had it stable for over a year at 4ghz all core locked, but eventually it started getting worse and getting less stable. This was across 4 motherboards, 4 GPUs, and 3 ram kits. This was a launch 3600 I got as a gift, so it took me a long time to determine it was a CPU fault, as there were a bunch of bios issues at launch.
I always use stock speed at default settings and I maybe get this CPU in February with a320 hdv r4.0. I probably expect no issue.
i have a ryzen 5 3500 and an RTX 2070 super... is that ok?.... i am scared now - SOMEONE HELP ME
Also: running memory at 3200 with DOCP (XMP) the memory's official rated speed
Edit: i checked 100% util CPU temp - in Davinci video render, adobe PP and AE video render and cinebench - temperatures sit on 84-86 degree (i do thermal paste changes frequently - arctic Mx4) also gaming hits max 75-76 degree.... i have PBO on at 10x scalar (200MHz) max cpu power draw is 60-65w and GPU power draw is 230-240W max
i am also running a wraith stealth (smallest cooler of the stocks)
I loved my 3600x because it was my first solid cpu. My first pc had a fx 6300 and while it could play esports titles, it was lacking in newer more demanding games. The 3600x was a breath of fresh air compared to that and it treated me well from 2020-2022. I ended up going with a 5700x recently and while that wasn't as big of a jump as the fx 6300 to the 3600x was, it still was a solid upgrade because its a power efficient 8 core chip and runs cool. I just got a noctua nh-u12s redux a couple days ago to pair with it, and my temps and noise is way better than my wraith spire cooler was on my 3600x. My 3600x will always have a place in my heart though. I will probably find a friend or relative to give it to so it can have a new life instead of sitting on a shelf
I started with 2013's AMD FM2+ socket, cpu A-10 6800k (total garbage)
Than got the Ryzen 3600 when it launched.
And upgrades that to a 5700x last week :)
@@TheTryingDutchman oh wow i can only imagine the feeling you got when you first booted up the 3600! I could tell a difference right away by just scrolling around in windows. Its crazy how much AMD cpus have come in just the past 5-10 years. AM4 has had great longevity!
@@Bdot888 yup, the difference was huge, and am4 was/is indeed an awesome platform! Enjoy :D
I got the 3600 when it first came out in 2019 and it was a very buggy experience for a few months. I spent hours, all day long, just testing, bench marking and adjusting settings.
I managed to get my Ryzen 3600 at a 3900mhz all core clock at 1.25v and stayed there.
With basic cpu coolers and with everything left to default/automatic, the CPU would shoot up to 85 - 90c because of precision boost overdrive going absolute insane. The voltage spikes on PBO was super dangerous. I saw spikes up to 1.48v when PBO would activate and cause the CPU to overheat.
Im happy at 3900mhz all core at 1.25v. The out of the factory default settings are faulty and volatile. Even 4 years later after updating to the latest mobo bios and amd chipset drivers, the default PBO is very problematic.
Mine did degrade with a decent cooler, so i think it's more then a temperature issue. I think it's just an combination of trash bin dies + high voltages.
Your CB23 scores are on the low side tho, i get even at 3600mhz (turbo disabled) 8400 multi core and 9300 with turbo enabled.
And people called me crazy for tuning my 3600 to under 70C under load lol... "It's FINEEE at 1.45V, it's fineee at 85C..." No, it's not. My Intel i5 6500 I had before ran at a little over 1V full tilt and did not exceed 65C, ever. And that was on 14nm+++++++, no reason a 7nm product like the 3600 would need 1.4V+ to reach turbo speeds. Yes, AMD pushed their CPUs above their sweet spot for a few % more, but they did it with every gen, resulting in the house-fire hazzards that are the 7000 series.
Here are some examples :
- tuned a R5 1600 for a friend - from 1.35V core / 1.2V SOC down to 1.27V core / 1.05V SOC - same speed, XMP stable, ran ice cold after that - under 55-60C at all times.
- my old 3600 - my sample was probably the worst one in history , was not able to reach turbo speeds like ever, ran at 85C under load with Wraith MAX and 55C idle - L-O-L. Default ran at 1.45V and about 1.3V+ under load (wtf). After some tuning, was able to use a -0.05V (50mV) offset and tinkered with the RAM speed and voltages - result was under 70C under load. Any higher offset resulted in instability and/or losing performance. Called it a day and lived with it. To this day, this is the crappiest CPU sample I have ever purchased. I even regretted not getting a 1600AF instead fml...
- now living with a 5600X and besides the speed advantage, it is VERY VERY power efficient, took a -50mV undervolt and never goes above 60-65C. Same Wraith MAXX as before. Donno if karma is real and I got a golden sample or the 5000 series are much better by design, or the 3600s were all crap, or all of the above, but I am satisfied.
From my research over the net when I was trying to subdue my raging 3600, I can tell you all of them take AT LEAST -50mV (-0.05V) offset without any disadvantages or clock stretching. Some people reported up to -75mV on a 3600 and up to -125mV (!) on a 3700. So it depends on the sample that you got. Start with -0.05V and go up until you see performance degradation or instability. See how far you can lower that Vcore. Every little bit helps.
From all that fiasco, I will ride my 5600X to whenever it can carry me and then buy Intel again. I am sick and tired of tinkering with AMD CPUs out of the box just so they don't burst into flames, it is ridiculous. OR, I can buy one of their APUs - they are actually another story - monolithic, sip power and almost as fast as the regular Ryzen CPUs. And have none of the overheating problems. I have 3 of them at work (4650G, 5600G and 4600G) and they are quiet little beasts. We'll see what AMD has to offer in a few years time.
A little undervolting and good cooling can make your CPU last for decades and you are so right about the low-end coolers not cutting it as software demands more from the CPU.
Even my old daily driver - just used for email UA-cam etc. has had a cooler upgrade. It was just an i5 with stock cooler and I noticed the temps were starting to get quite high.
Now all is good and cool again.
And for all you guys out there who do not clean your rig regularly, remember dust = heat!
Undervolting is not what you want to do. Death Stranding is an AVX instruction game which demands higher voltage for 64 bit compression.
@@boating2strokenovice726 What? I'm not talking about a single mediocre game. I'm saying that it's the best way to prolong the life of your system.
@@ausfoodgarden The game in question in this video. Death Stranding is an AVX instruction game. Which means requiring 64 bit compression from (2) 32 bit buss lanes from the system ram requires more milivoltage to the CPU.
Lowering the clock speed of the cpu is the only thing you can do with AVX instruction titled games. Compromising the voltage in AVX games causes a loss to CPU texture rendering:
CPUs should have built-in thermal protection and should not fail even if the cooling is insufficient. It appears that AMD sacrificed reliability for benchmark scores.
I’ve had to replace my old 3600X when I had that Lel, they shipped me a new one for free
don't blame the cooler, I'm running a 5600x with stealth cooler. CB R23 temps never go above 75c(no I didn't change throttle temps). Something else is wrong in your setup. I had a 3600 in an earlier build and it was also not running hot. Probably that cheap crap motherboard you have.
By default, my Ryzen 7 5800X was consistently hitting above 80C, sometimes ekeing towards 85C, where idle temps were at an elevated 50C - 60C. I thought that it was really obnoxious, and I did my due diligence in mounting the cooler correctly. Disabling PBO2 barely made a dent, but long story short almost everything was tied to Windows' power management settings and I just forced Windows to use the "Balanced" power plan, on top of just disabling PBO2. I don't think I made a manual thermal limit, but after setting everything that I think could made a huge difference, it went back down to the usual 25C - 32C idle temperatures, and never went above 45C - 50C (depending on season) when gaming. I have a Cryorig H5 Universal, and it should be more than capable of keeping that thing tamed.
Seeing Ryzen 70xx series processors have 90C at idle as the "norm" was incredibly outrageous and inane of AMD to do. This is absolutely wrong. This is a huge signal to every other consumer that AMD is just continuing to, by default, overclock and maybe overvolt the shit out of their chips just to meet a specific performance improvement goal and appease their investors, and it isn't helping that most motherboard manufacturers tend to go a bit crazy with voltage settings, too. This is "fun" for overclockers, but the target audience should really be towards the average consumer.
"90C is below the maximum operating temperature of 100C, therefore it's okay" is a *bad* excuse to make when selling a chip to most people, because that's when most thermal throttling should occur.
My 4800H laptop would hit 110c sometimes 😅. No problems with it after 2 years.
Big reason why intel with stock cooler working no issues
I just built another new PC for a customer with a Ryzen 5600 with a stock Ryzen cooler. First thing I did was test performance and heat levels, and noticed that already the CPU was at 60 degrees at idle. I then did a benchmark with Heaven and under load the CPU temp didnt even move away from the 60s. I finally pulled the Heat sink off the CPU, and although I had applied a fresh coat of thermal paste on the CPU, I redid the thermal paste and put it on extra thick. Voila! The idle temp dropped from 60 to 40. That's the first time I had to use an extra large amount of thermal paste on a CPU. Even my new 5800X3D works great with standard amount of thermal paste, but its cooled by a Cooler Master Hyper 212 Black tower cooler. Bottom line is, if your using a stock cooler, don't be stingy with the thermal paste.
I think your unit is defective and that's that. I have a Ryzen 5 3600 with 74 max temp in synthetic tests, furmark etc. I do have some good airflow in the case and the CPU has an Arctic tower cooler
I had a similar problem when I bought my 3600 back in 2020... Even with a hyper 212 cooler it would idle near 60°C and instantly jump to 80s and 90s when under any load, getting the fan to ramp up and down all the time. Within the first week I learned and did all the same tests you did and some more, settled with an undervolt, running 4,1 Ghz at 1,12v and dropped temperatures to 70s even under stress, no performance loss or clock stretching.
I too had the cpu being fed up to 1,4v by default, but I don't think that's the real reason for the excess heating as the cooler should've been able to handle 80w just fine.
After some research, I believe it was a bad batch with poor die to ihs conductivity or uneven surface, as some people who lapped the cpu got massively lower temperatures without messing with voltages, so the heat was accumulating too fast in the cpu without being passed to the heatsink.
I have a Ryzen 3600 with a Stealth cooler on an Asus Strix B450-F mobo. (RTX 3080 if anyone is curious.) It's still on default settings as far as I know and I have no problems.
Stock coolers were never great, on both Intel and AMD. NEVER!
If you can't pay £30 for an aftermarket cooler, you should not buy a PC. End of story.
I didn't watch all video, but I'm pretty sure it's the A320's shitty VRM that are either failing or damaging the 3600... have you tried a Bx50?
I find it becoming absurdly loud for what it is. The cooler and fans go rampant like crazy every single time up to v1.412 volts then it downclocks again constantly due to PB.
I just disabled Precision Boost and keep it like this , plays all games fine just not as good as it could but i don't need to hear my case fans and my cooler running like a jet engine it has as much noise as a old PS4.
love this channel, you provide such interesting info!
So........ the CPU's arent really failing and this is a video to say that a better cooler and under volting can lower temps. cool story.
Was a clever way to edit 2022 at 4:02. Haha the attention to detail. +1 for re-recording just the pronunciation of 3.
Well, if you use a manufacturing process that is mainly developed for apple's iphones with a lifespan of 5 years and planned redundancy, what do you expect?
So the moral of the story is, spend the $20 for a better air cooler.
TBH, AMD had upped the base core voltages on all their CPUs that are Zen 2 or newer - I still remember seeing 3600X reviews back before corona hit (and I learned how to tweak hardware) and thinking "Hmm... Nah, it's too damn hot, my 2600 is a more sane option".
5000X and 7000X series processors basically have the same symptoms (which luckily were remedied in the non-X versions to which I can attest as a R5 5600 owner) and the fix is basically the same that you showed (undervolt + TDP limit).
you just gimped the cpu no reason to stop it from hitting thermals it's supposed to when using Cinebench especially
been running my R5 3600 at 4.2-4.3ghz, sometimes undervolted for 3yrs now. 0 problems
This wouldn't happen on intel CPUs right? Or not as worse as this? I knew AMD was always the "cheap" one.
kind of extensive dont u think? just get a cheap tower cooler for 30 bucks with at least 150 tdp and ure good to go
I just turn off the "core performance boost", that means it won't go beyond 3,6ghz. I don't notice the minimal performance difference, however I do notice the huge difference in heat and noise output. Currently I'm running a simple Noctua tower cooler, the fan never goes above 400rpm. Temperatures during 100% load test do not go beyond 55˚C. I'm using the silencio 550 case, a case often referred to as "the GPU owen"