The results are basically always the same , larger amounts of ram of course decent speed ram , raises the lows up and keeps them higher.. Do i need 64gb of 3600 ram? Not really... Does it give a slight performance benefit? Yes.
I really hope you do a complete Ryzen and memory overclocking tutorial. These new CPU's aren't the same as older ones. It'll be awesome for someone upgrading to this new stuff from hardware from 5 or 6 years ago.
tweaking memory and the infinity fabric is the most important part of overclocking and tweaking for better performance for sure on all ryzen cpu's......i gone from 6900 score in ciniR20 to 9800 just tweaking ram to 3800 mhz cl16 timings with the infinity running 1900 mhz...half the speed of ram always for best performance ...remember Weird Ale's song .....its all about the Pentium...lmao......its all about the infinity fabric now ....lmao
@@Tallnerdyguy The 8350 came 4 GHz stock. 5.5 stable on all cores is pretty decent for a non-professional build. He was 17 at the time and using a Noctua air cooler.
@@IndyMiraaga Depends on silicon lottery. I can't get mine above 4.5. No matter how much I try. FYI I'm talking about the 8350 and I use water. It's literally the rig I'm still running as I type this. Damn thing just won't go beyond 4.5. (Yes I'm poor and can't afford current gen CPUs. I'm trying though. But my build is around 3K and with 2 children, 3K is ALOT for a computer.)
@14:13 "I challenge anyone to say they can see a difference of 142FPS and 150FPS" What about frame times, 1% lows, and 0.1% lows? That's where the real gain is.
Also, that's crossing back and forth across the 144 sync threshold on a 144 hz monitor, so it WILL be noticable, either stay over 144 or stay under it, but don't float back and forth, causes weird frame time inconsistency.
This is what I would like to see is the 1% and 0.1% as that would seem to be where you would really notice the difference as if the RAM can't keep up fast enough with moving files as needed that can introduce stuttering.
I hate these stupid tests, lets test the RAM with CPU and GPU benchmarks :'( RAM tuning makes bigger differece with "lower end" CPUs, for the best experience (low latency) not equal with highest benchmark number, everybody should use 4 stick of dual ranked memory with the tightest timings. It's 16GB for DDR3 and 32GB for DDR4. First of all they should test 3 type of CPU from AMD, because of the different layout of the chips, 3300X, 3600, 3950X. Secondly, they sould test multi player games, I would say at least with 20 player on the map. (Only Steve from HW Unboxed do some multi player testing) Just launch Fortnite and you can see the difference in the lobby where 100 player present in a restricted area. EDIT: here is an older benchmark on a Ryzen 1600 CPU ua-cam.com/video/JtVRs_Q1ngQ/v-deo.html
@@tcclaviger Why would frame time inconsistency be problematic when the fps fluctuates between 130-150? Present an argument please, because you sound like you just heard and read from here and there that 'frame time inconsistency is baaad', ignoring the actual reason that causes it to be a bad thing.
Now that’s some funny stuff right there, been building intel for years. Just finished my first ryzen build maybe 3 hours ago. An hour later this comes out, thanks Jay!!!
9:40 I own both a 3950X system and a 3990X system. In Ryzen Master, you can set clocks by core, by CCX, by CCD, or all core, as well as turn entire CCDs off. The issue (in my admittedly somewhat limited testing) is that you cannot set the voltage per CCD-at least, not through Ryzen Master. So all that fancy undervolting we might be able to do (for example, my 3990X is stable with up to a .075V offset and the clocks otherwise left to auto, and I can run my 3950X at 1.10625 and still hit 4150MHz all-core) is wasted when we have to give the okay to 1.3V+ just so a single CCD can hit that higher clock. It would be really nice if Ryzen Master let us set per-core or per-CCD voltage, that way we could set our programs' affinities so we have a low heat, decent clock on most of the CPU, but then have our 4-8 cores set up to be a little higher clocked for gaming. More granularity is always better, if the user takes the time to understand what it is they might be tweaking of course! As far as the ideal RAM clock speed, Steve made a really detailed video a few weeks or months back (what even is time?) that explained how the uclock, mclock, and fclock like to be in ratios with each other, and pushing fclock past 1800 whacks out the uclock to be 2:1 instead of 1:1 (basically cuts its clock in half), which robs you of...I want to say 9ns. But of course depending on your kit, if you just push even harder on your mclock and tighten the timings even more, you can still overcome this drawback. For normal people who want to spend their time using their computer instead of tweaking it (both of which endeavors are valid and good fun), 3200C14 or 3600C16 are some pretty attainable sweet spots currently.
My question is why do some of yall stuff run better on lower power when I have gone the opposite direction qnd see my stuff running world's smoother at a higher than stock power limit Why is this
A slight FPS boost is always nice, but the big difference I've seen/noticed when increasing RAM and FCLK was with **FRAME PACING** (frame times) Meaning, it elliminated most of the stutters and inconsitencies that I got when my 3800x ran with default RAM and InfFab speeds. The ration of MCLK to FCLK speed is/should actually always be 1:1 as DDR (double data rate) 3600Mhz means that it actually runs at 1800Mhz. The reason why I mention this is because of overclocking the FCLK. The maximum InfFab speed is 1867Mhz (DDR3733), past that it won't run 1:1 syncronous mode which will **hurt** performance.
I like Steve's videos for the granular data and info, I like Linus' videos for the wild tests and random content, I like Jay's videos for a mix of both worlds.
Just to clarify the bit about the FCLK (Infinity Fabric clock) at 8:26 and 12:04 - you do NOT want to run FCLK at half the speed of the memory, you want to run it at the same speed as the memory (1:1 ratio). Keep in mind the effective clock speed of DDR memory will always be double the memory clock speed shown in BIOS because it is double data rate. Just bringing this up so people won't go and set their FCLK at 1:2 by accident for 3600 speed or lower. So, like Jay says at 11:53, this is why the general consensus is that 3600 speed memory is the sweet spot for Ryzen. Most (if not all) Ryzen CPUs will do 1800 MHz FCLK for a 1:1 ratio with DDR4-3600 memory. As to 14:29 - you absolutely can run 4200 memory if you want on Ryzen without any additional effort (assuming XMP/DOCP for your memory kit works at that speed on your motherboard without additional tweaking) but you would probably be looking at setting FCLK to 1050 MHz in 1:2 ratio and lose some or all of the performance difference. And it costs a lot more, so probably not worth it.
Would have been good to see a comparison of performance with different ram timings as well, to see if there's more gains to be had from low latency ram, or just outright ram clock speed, and where the best balance of those two is.
Best case scenario after fiddling with Ryzen for the first time was lock my cores to 4.3 and set faster RAM to 3600 with tighter timings and run 1:1 with IF. PBO was always inconsistent even under water for my 3700x. I find it comparable to my 7800x locked at 4.8GHz in gaming while also having 2 extra cores.
The f clock automatically sets itself at one half the memory speed at up to 3600 MHz on Ryzen 3000 processors. The f clock goes up to 2000 MHz on 5000 series processors so 4000 MHz is the sweet spot. You can overclock the f clock on Ryzen 3000 series to 2000 MHz and the 5000 series to 2200 MHz but that involves undervolting the cpu and overvolting the I/o die. Way too complicated for most people.
5-6 % overall performance improvement on tightly timed bdie, but 0% in some tasks and more than 10% in others, it's all over the place. The IF matters more than me freq. 14-14-14-28-42 3733 vs 3200 DOCP CL14.
Makes huge diff to minimums in some games. Like 20%+ in some. Averages in those same games only go up maybe 5% tho so people think it's not a big deal. Running samsung bdie 3600c16 at c14 with pretty much every timing tuned.
Hynix c die is just as good. I got 3200mhz cl16 hynix c die and it ocd to 3800 cl16. I saw my fps in modern warfare go from 130fps to 250fps was kind of insane and i didnt expect gains like that.
@@yoltsbp cjr is junk compared to samsung bdie it scales well frequency wise when you throw voltage at it but timing wise its mediocre at best. That being said samsung isn't worth the premium. I own hynix djr 3600c16 kit and samsung bdie 3600c16 kit.
February 22, 2020 i run my Ryzen 7 3700x Plus Msi x570 atx + Rtx 2060 8gb Super & 32gb 3000 Ram with Xmp On = Rock -N- Roll 👾 Zero issues 6 months later 🤘🏽 Thanx for Posting 🇵🇹🇵🇹🇵🇹🇵🇹
Ill never forget the time my girlfriends friend was setting up a cellphone for us I said "please dont hit my credit a million times", her response was "dont worry, Im a ferpessional"
3800 C14 Fclock 1900 is very much doable on some CPUs. Most can do 3733/1866. Custom timings also make a big deal on 1% lows even more than outright FPS.
That background really took me back, I was so confused for like 10 seconds trying to wrap my head around how you got the testing suite to work on that, then I saw the taskbar lmao, great video!
When I started doing memory overclocking on my b350 board, if I locked it up with bad settings I had to remove the memory completely, clear the CMOS, boot it without the memory in the slots, then turn it off and reinstall the memory, then it would boot with default memory settings.
Like how you summed up the ending. Can you maybe make a series where you go over all of the words/terminology(specs), so that a noob can maybe understand what your saying. Love the channel.
Please do a video about timings on ryzen as well. Becasue some people say you can get hella better performance with that. Oh I forgot. Great video Jay. :)
As some one who watches AHOC as well I have to agree, plus Cinebench is not memory sensitive and is a good example of why memory speed can and should be tuned speed wise more for your workloads as you can save money on your RAM that way. Plus 3200 is actually a JDEC spec so you can find kits that do it out of the box with out setting XMP at least when dealing with Ryzen not sure on Intel as they have not been where I've been paying attention as I'm a core heavy user. Just ask my poor Ryzen 5 3600 that has seen more full core loading than most of them will see :D
@@justjoeblow420 i dont think a single kit sold so far has a 3200 JEDEC profile yet. Kits only recently started shipping with 2400 and 2666 JEDEC profiles... Anything above that is 99% XMP profiles.
TheHavocInferno there are plenty of sticks that you can buy, and I mean PLENTY that you can buy with JEDEC 3200. I would know it too given I literally get an email from JEDEC every morning about random ram stuff. The majority of JEDEC 3200 sales come from OEM sticks. Actually, you can buy some JEDEC 3200 B die from Samsung right now.
I wouldn't say 3600 is a 'sweet spot' at all. It is for Ryzen 3000 if you don't manually OC your FCLK cuz it maxes out at 1800mhz by default. That said 3600 cl16 is a great kit of memory for out of the box XMP/DOCP. There's a lot of 3600 CL 18 kits out there which is technically slower than 3200 CL15
Put 3200mhz in my x570 with ryzen 3600. XMP was loose cl16. I manually changed it to 3600mhz tight cl18. Boost was easily measurable across the board. Highly recommend 3600mhz
Amd ryzen master, you can specifically target cores to overclock, I oc 2 of my cores to 4.4 and the other 4 at 4.3 on my 3600. It's fun to mess with too.
That's weird, I am running a 4.3 Ghz all core oc using ryzen master at 1.225V , anything above 1.3V makes my processor heat up like crazy in prime 95. Have you ever run prime95 for stress testing or just cinebench?
Perspent is the percentage performance you gain for the percentage money spent. Spend $500 and get 500 performance points Spend $1000 and get 1100 performance points This gives it a +10 Perspent value.
Great video Jay. I have been having issues with my ram for a while where I couldn’t get it to run at 3200mhz even with docp on and all the turbo boost and core enhancement, etc. Thanks to this video I now know to go in and set the Fclock which I shall try later when I’m home from work. Thanks again for the great content you’re putting out.
@Tinkering With Tech I’ve ran 4 different Ryzen cpus since zen1 launched and never had a single issue. It’s cute when people with bad analytical skills make broad generalizations about computer parts they just don’t know how to use.
Been running a Ryzen 3600 X with fclock @ 1800 MHz on 3600 ddr4 Viper cl14 & core enhancement Since January 2020, a week after I built it. Good to know it already doing the best.
Usually with ad spaces I'm like oh cool I'll avoid that company because if they need ads they probably suck but microcenter is my favorite place. It's better than buying online and if I have to buy online it's nice to know that I'm supporting a cool store and business. Also nice vid Jay lol
Fabric clock isn't half of memory speed when it's half of *reported* RAM frequency. DDR means Double Data Rate, which means RAM talks twice per clock. That is shown as double actual frequency. Fabric clock is exactly RAM freq when it's "half", probably also why it gives the best performance because they are in sync.
Yes memory timings video please!! I've been into building computers since 2013 and I've overclocked everything I can in all those computers except memory
It depends on where you are. Mayfield Heights, OH has a BUNCH of motherboards, not sure about the budget CPUs, though. The main issue I'm seeing GLOBALLY would be power supplies. I can't find a single decent 650W+ Gold that isn't price gouged to fucking smithereens.
Ryzen 3 3200G is probably the most overrated CPU I have seen. It costs as much as Ryzen 5 1600, but is MUCH worse. Not only in game performance, 1600 is 6c 12t CPU which gives you better framerate on lower usage. I would say the best budget CPUs are Ryzen 5 1600, 1600af, 2600, Ryzen 3 3100x. Eventually I3 9100F
@@MrDutch1e 3300x is basically non existent right now, can't get em anywhere! I had this planned on a custom build and had to bite the bullet on 3600xt because it was all they had in newer amd cpus with relative or better performance, although of course, way less price/performance value for me sadly which is probably why it was available! 😂
I would like to see a new test with the new Ryzen 5000. The fps improvement for faster RAM should be higher, since the clock speed of zen 3 is higher as well I think? Also need a new gaming rig and kind of confused on which RAM to by. But speed should matter since with their new gpu in combination to access the faster DDR6 graphic memory they also boost performances, or am I mistaken?
Hello Jay, oc by ccx , normally 4 options check with hwinfo64 what cores go faster 4.7ghz etc. On bios you will oc by ccx and not all core. Ccx0 44.75 ccx1 44.50 etc just an example. Voltage don't go over 1.38v be careful with the LLCs so it won't go over 1.4v. my 3900x like this does 7850 on cinebenq with a x370 gaming Pro carbon. Hope it helps
So I was able to overclock my ram from 3200 to 3600 at CL14 with tight timings using the DRAM calculator that's been floating around for a while. It was a significant performance boost and it's literally never crashed or locked up in about a year's time now. The question I have is how do I know that I'm not getting errors? Not every error causes a crash, but I am assuming it could cause it to have to recalculate something if fails a check or something.
Jayz, what's the relationship and impact of CL against a motherboard's QVL. Eg. A 3600MHz RAM with lower CL but not on the motherboard's QVL vs a 3600MHz RAM with higher CL but is on the motherboard's QVL. Does it even matter for a RAM to be on the QVL? Can I take a non-QVL RAM with lower CL (eg. CL16) and use "DRAM Calculator for Ryzen" to try and tighten the timings, or is it still faster to use a QVL-certified RAM that's maybe CL18 or CL19?
Looking back at his uploads what I think it might be is the fact that he's uploaded like 3-4 videos in the past year with "ram speed effecting gaming" in his titles.. I guess they weren't specifically about ryzen but I'm pretty sure his test bench is ryzen now anyways lol
I took some cheap corsair dom 3200mhz 16-18-18 up to 3600mhz 18-19-19 and my score went up a lot . My fps in games also noticeably up. I feel frequency/infinity fabric boost is more noticeable than when I did 3200mhz with tighter timings. X570 3600 cpu
I’m planning on building my first pc. So far my parts list is the R5 3600, rtx2060, b450 tomahawk max, 16gb 3600 g skill ram, 250gb m.2 nvme ssd, and 1 tb Seagate barracuda hard drive. What are the best choices for a case and powersupply for my build, since I don’t know where to choose from in terms of best cooling and such. I’m mainly looking for help on what case since ik psu r in low stock and once they get back up I can find one
It's the exact opposite. Faster RAM raises your avg fps because it removes some throughput bottlenecking from the top % of your fps where your CPU load is minimal. For example looking mostly at a skybox in a game will see your fps dramatically increase as the CPU is allowed to run free, and with faster RAM you will get higher fps there. However, when then CPU itself is under a heavy load (as Jay showed with Cinebench) the effect of RAM speed becomes negligible as the bottleneck moves to compute instead of throughput. That is to say - if you only want a stable fps (e.g. you're capping it somehow for G-sync or Freesync), RAM speed has very little effect on your experience.
Not fast memory but low latency memory. Memory at 3600 with FCLK at 1800 is the sweet spot, but Ryzen wants low timings to improve performance, unlike Intel. This video was kinda pointless imho, it told nothing about how to improve Ryzen gaming performance. It just said what we all know already and kinda misled people into thinking that memory timings are inconsequential for Ryzen as they are for Intel.
Alright...so what about a 16GB kit of DDR4 3600 CL14. More specifically talking about the Teamgroup T-Force Xtreem ARGB kit. Would the faster speed and really low latency be a huge/noticeable boost for Zen 2?
Jay FCLK it's actually on a 1:1 ratio if you put it at "half" ram speed, base ram speed it's exactly the half of what you see (DDR = Double Data Rate), so that's actually 1:1, and that's why you have the best results with such settings. And no, CL timings DOES COUNT. At least in gaming. I have my ram at 3733 with a mix of cl14/15 with very tight subtimings, and difference with cl16 and relaxed sub timings is it there. The only downside of oc ram too much, it's you will get really higher temps on the cpu, because you just increase the cpu ipc by having way faster memories, but temps really spikes up. If I put my ram at 3200mhz cl14 (stock xmp/docp profile) I have something like 10 Celsius LESS on my cpu, and spikes (mainly in idle) are way less prominent. I tried also with 3733mhz but cl16 timings, and temps were always lower than with very tight timings. My 2 cents? If you have a good cpu cooler, leave zen 2 do what he does (boost, or turn on pbo with some UV, but be careful, or you will get core stretching), and play a LOT with rams, it's where you will get the major boost, especially in gaming.
I am still seeing a lot of 3200mhz 8x2 sticks on sale for around $60 -- G-Skill, Crucial, etc. Just sign up for newsletters from etailers and you will be notified when parts go on sale. I doubt DDR4 will see a big price increase when they need to clear stock for DDR5 in 2021.
linus did same video awhile back and nothing happened with prices so doubt they will now unless stock runs out.. and seeing as all the MOBOs are gone unless u buy the highend ones they may be next on the list. I did however upgrade from 16g 2933 tridents to 32g of 3600 after they went on sale
I don't know about you but ordering multiples of twin packet sticks (ie x2 packs of two sticks/pkt for 16GB DDR4 @3600MHz) saves a coupla hundred, compared to 4-pack 'matched' ram. Its much like Crossfire and SLI to me. The bet I make is to stack twin stick packs in the A1 B1 +A2 B2 dimm slot order so the electronic abstraction sees each matched ram pair in their own channel (if I have that right). **Can't wait to finish my build: now 64GB of 3600MHz ddr4 richer, just arrived. G.Skill Ripjaws V Black, in a Black Corsair Crystal 280X mATX case with its glass top pane sitting on stacked nylon washers and the grommets upside down for extra standoff height, main glass side panel grommets reversed to set panel flush against case, front panel dust filter removed. I have an MSI Radeon RX 5700XT Mech otw (this model is only 232mm long). Almost everything going on inside my case is black, except the brushed aluminium heatsinks of the Asrock X570M Pro4 mobo, and its white printed stuff, and the Ryzen 9 IHS. Not a fan of pretty lights. I might custom build a puppet theatre stage curtain-esque drape to the glass side panel .. I'll think it over.
I love how, after so many years of being succesfull, you still take the time to explain stuff most of your viewers know. (Like how DDR4 out of the box runs, free benchmarking programs, difference between Intel CPU's) I know you know the majority of your viewers know, but some don't (me :-) )
R5 3600X here with DDR4 running at 3733Mhz and Fabric Clock set to 1866Mhz (OC). Most Zen 2 chips will be able to overclock the Fabric Clock to this and if you plan to upgrade to upcoming Zen 3 (aka Ryzen 4000) it makes sense to invest in higher frequency RAM. Also don't forget to tweak your subtimings too using Ryzen DRAM Calculator as it makes a noticeable difference.
I have 32oomhz memory but I bought the ASUS Prime X370 when I got my Ryzen 7 all those years ago only for the board to claim it runs 3200mhz but becomes unstable over 2600mhz. I will NEVER buy any asus product again. Nor will I buy nvidia for lying when i bought my gtx970. PC companies are the worst.
Why does no one increase the priority to high on the Cinebench 20 process while running a benchmark? Constant and pretty much max scores every time. Backgrund tasks don´t bother me anymore! Do I miss something here?
The JEDEC DDR4 SPD specification allows speeds other than 2133 (both slower and faster); so non-overclock 2400, 2666, etc is quite possible. The standard includes examples up to 3200. It's just a combination of manufacturing and marketing that has made most DDR4 modules ship with SPD configured as 2133 (and XMP for the real rated speed after binning).
Isn't there a ram divider that kicks in after certain ram speeds that essentially reset memory performance? You need to test 3200, ~3400, 3600, ~3700, ram speeds as well, and check when the dividers kick in. There are certain ram speeds that will peak performance at a particular divider, but then drop after a new divider kicks in (ie. 3600 is peak of one divider, but a new divider kicks in at 3700 and performance drops down a bit).
I was disappointed that this video didn't address the concerns of timings. There is a substantial lack of information on this but the few channels and sites that test it seem to show significant difference in performance in a variety of CPUs. I am currently in the market to buy a system and honestly do not know how much of an impact 5000mhz CL19 vs 3800 CL14 RAM has an impact on a 3800-3950X and would love to see data that helps me guide which direction to spend my money. Please consider doing a timings video to help some of us with these kinds of questions
Can we just mention how good Jay looks since hes lost weight? Killer bro.
how old is he?
@@mishkatabg7647 like 40
Check out his other channel "Jay Stew Scents", where he pours cans of soup on his old cpu's and describes the smell.
How he loks before lol
Beefy Jay as good as lean Jay . Content 👌
I wanted to see the 1% and 0.1% low result in games with different ram speeds.
That's GN heracy right there
You'll have to look at GamersNexus for that.
The results are basically always the same , larger amounts of ram of course decent speed ram , raises the lows up and keeps them higher.. Do i need 64gb of 3600 ram? Not really... Does it give a slight performance benefit? Yes.
Well there’s Fortnite www.reddit.com/r/FortniteCompetitive/comments/h8bptr/how_ram_frequency_affects_fps_in_cpu_limited/?
yea, because we all know, lowest fps makes the difference >.<
he did the tests, but basically it meant nothing without more accurate information.
I really hope you do a complete Ryzen and memory overclocking tutorial. These new CPU's aren't the same as older ones. It'll be awesome for someone upgrading to this new stuff from hardware from 5 or 6 years ago.
Like i did 4 weeks ago. Went from 8320fx to r5 3600x
tweaking memory and the infinity fabric is the most important part of overclocking and tweaking for better performance for sure on all ryzen cpu's......i gone from 6900 score in ciniR20 to 9800 just tweaking ram to 3800 mhz cl16 timings with the infinity running 1900 mhz...half the speed of ram always for best performance ...remember Weird Ale's song .....its all about the Pentium...lmao......its all about the infinity fabric now ....lmao
Jay is not the guy you want a memory overclocking tutorial from. Such resources already exist though.
Watch gamers nexus video. Jay is good for entertainment but gamers nexus is better for data imo
Or you could, ya know, watch that content from other youtubers who did it already AND are more tech savvy
11:53 for recommendation
Jay: "Gigahertz is everything"
My friend's FX-8350 clocked at 5.5GHz: "Please allow me to introduce myself.."
Only 5.5? The 9350 came 5 ghz stock
@@Tallnerdyguy 5.5 ghz peasant!my friends car sometimes hits 88 mph, and then some serious shit happens!
i have that cpu on my spare pc, it's not very fast
@@Tallnerdyguy The 8350 came 4 GHz stock. 5.5 stable on all cores is pretty decent for a non-professional build. He was 17 at the time and using a Noctua air cooler.
@@IndyMiraaga Depends on silicon lottery. I can't get mine above 4.5. No matter how much I try. FYI I'm talking about the 8350 and I use water. It's literally the rig I'm still running as I type this. Damn thing just won't go beyond 4.5. (Yes I'm poor and can't afford current gen CPUs. I'm trying though. But my build is around 3K and with 2 children, 3K is ALOT for a computer.)
@14:13 "I challenge anyone to say they can see a difference of 142FPS and 150FPS"
What about frame times, 1% lows, and 0.1% lows? That's where the real gain is.
Also, that's crossing back and forth across the 144 sync threshold on a 144 hz monitor, so it WILL be noticable, either stay over 144 or stay under it, but don't float back and forth, causes weird frame time inconsistency.
That's how I noticed my RAM OC didn't stick. MHWorld was stuttering like crazy!
This is what I would like to see is the 1% and 0.1% as that would seem to be where you would really notice the difference as if the RAM can't keep up fast enough with moving files as needed that can introduce stuttering.
I hate these stupid tests, lets test the RAM with CPU and GPU benchmarks :'( RAM tuning makes bigger differece with "lower end" CPUs, for the best experience (low latency) not equal with highest benchmark number, everybody should use 4 stick of dual ranked memory with the tightest timings. It's 16GB for DDR3 and 32GB for DDR4.
First of all they should test 3 type of CPU from AMD, because of the different layout of the chips, 3300X, 3600, 3950X.
Secondly, they sould test multi player games, I would say at least with 20 player on the map. (Only Steve from HW Unboxed do some multi player testing) Just launch Fortnite and you can see the difference in the lobby where 100 player present in a restricted area.
EDIT: here is an older benchmark on a Ryzen 1600 CPU ua-cam.com/video/JtVRs_Q1ngQ/v-deo.html
@@tcclaviger Why would frame time inconsistency be problematic when the fps fluctuates between 130-150? Present an argument please, because you sound like you just heard and read from here and there that 'frame time inconsistency is baaad', ignoring the actual reason that causes it to be a bad thing.
Having the Windows 95 logo at first I thought I might have clicked on a LGR episode instead.
Windows 95 is coming up on its 25 year anniversary. ;-)
I think it because Bill Gates said something about no one needing more than 192somethingbyte for ram
@@dexterous2513 He never said that though. Those were fake news by authors putting words in his mouth.
@@vidyamancer7135 oh
I still love 98 and XP
"It's only wednesday and it's been a long week"
I felt that
Afaik, timings gives a pretty big boost. You can't just leave it at cl18 in an overclocking video.
Yup, checkout Hardware Numb3rs channel for some interesting results when optimising timings on a Zen 2 setup.
For games xmp settings are still most stable ... I tried and still is the best for my system atleast
He said it's cl15 at 3800. That means at 2133 the auto would be cl12
@ZPinch I tried ram calculator. It doesn't have my Patriot 3200MHz ram in it
@@snipernote I'm using XMP on my MSI X470 Gaming plus and its perfectly stable with gaming and everything else
Now that’s some funny stuff right there, been building intel for years. Just finished my first ryzen build maybe 3 hours ago. An hour later this comes out, thanks Jay!!!
imagine microcenter actually having full shelves like that, post covid :P
And yet Microcenter is still short on power supplies.
@@venividivici4253 Everyone is short on power supplies. :(
@@Sandriell Agreed :( MC just restocked on a 750 watt and I rushed to purchase one. Now they out of stock again xD
We aren't post covid yet.
@@venividivici4253 I bought mine at best buy. Both places were the same price.
9:40 I own both a 3950X system and a 3990X system. In Ryzen Master, you can set clocks by core, by CCX, by CCD, or all core, as well as turn entire CCDs off. The issue (in my admittedly somewhat limited testing) is that you cannot set the voltage per CCD-at least, not through Ryzen Master. So all that fancy undervolting we might be able to do (for example, my 3990X is stable with up to a .075V offset and the clocks otherwise left to auto, and I can run my 3950X at 1.10625 and still hit 4150MHz all-core) is wasted when we have to give the okay to 1.3V+ just so a single CCD can hit that higher clock. It would be really nice if Ryzen Master let us set per-core or per-CCD voltage, that way we could set our programs' affinities so we have a low heat, decent clock on most of the CPU, but then have our 4-8 cores set up to be a little higher clocked for gaming. More granularity is always better, if the user takes the time to understand what it is they might be tweaking of course!
As far as the ideal RAM clock speed, Steve made a really detailed video a few weeks or months back (what even is time?) that explained how the uclock, mclock, and fclock like to be in ratios with each other, and pushing fclock past 1800 whacks out the uclock to be 2:1 instead of 1:1 (basically cuts its clock in half), which robs you of...I want to say 9ns. But of course depending on your kit, if you just push even harder on your mclock and tighten the timings even more, you can still overcome this drawback. For normal people who want to spend their time using their computer instead of tweaking it (both of which endeavors are valid and good fun), 3200C14 or 3600C16 are some pretty attainable sweet spots currently.
My question is why do some of yall stuff run better on lower power when I have gone the opposite direction qnd see my stuff running world's smoother at a higher than stock power limit
Why is this
@@jaygee3032 because you don't know what you're doing. And every chip is different. Duh
Watched the vid, instructions unclear. Got stuck staring at your beautiful hairline.
Ed, go to sleep you re drunk
I will send my setup soon bro
Instructions unclear. Finger stuck in chassis.
Send help
UA-camr with millions of subscriber didn't get even 100 likes what?
@WolraadWoltemade 1652 psst, it wasnt the "finger"...
I love how you kept the mistakes in. Lol. Really makes your content more enjoyable, and shows that you're just as human as the rest of us. :)
I know in Ryzen Master I can individually set core speeds for my 2950x
Same for my 3600x
same with my 2600x
Same for my 3600
same
Same for a 3900x
A slight FPS boost is always nice, but the big difference I've seen/noticed when increasing RAM and FCLK was with **FRAME PACING** (frame times)
Meaning, it elliminated most of the stutters and inconsitencies that I got when my 3800x ran with default RAM and InfFab speeds.
The ration of MCLK to FCLK speed is/should actually always be 1:1 as DDR (double data rate) 3600Mhz means that it actually runs at 1800Mhz.
The reason why I mention this is because of overclocking the FCLK.
The maximum InfFab speed is 1867Mhz (DDR3733), past that it won't run 1:1 syncronous mode which will **hurt** performance.
I like Steve's videos for the granular data and info, I like Linus' videos for the wild tests and random content, I like Jay's videos for a mix of both worlds.
I like Kyles Videos for Kyle
@@Davorii Some people like Kyle's videos for Lyle
I don't like Kyle or Lyle.....
Just to clarify the bit about the FCLK (Infinity Fabric clock) at 8:26 and 12:04 - you do NOT want to run FCLK at half the speed of the memory, you want to run it at the same speed as the memory (1:1 ratio). Keep in mind the effective clock speed of DDR memory will always be double the memory clock speed shown in BIOS because it is double data rate. Just bringing this up so people won't go and set their FCLK at 1:2 by accident for 3600 speed or lower. So, like Jay says at 11:53, this is why the general consensus is that 3600 speed memory is the sweet spot for Ryzen. Most (if not all) Ryzen CPUs will do 1800 MHz FCLK for a 1:1 ratio with DDR4-3600 memory. As to 14:29 - you absolutely can run 4200 memory if you want on Ryzen without any additional effort (assuming XMP/DOCP for your memory kit works at that speed on your motherboard without additional tweaking) but you would probably be looking at setting FCLK to 1050 MHz in 1:2 ratio and lose some or all of the performance difference. And it costs a lot more, so probably not worth it.
Hey Jay love your vids!
Would have been good to see a comparison of performance with different ram timings as well, to see if there's more gains to be had from low latency ram, or just outright ram clock speed, and where the best balance of those two is.
Best case scenario after fiddling with Ryzen for the first time was lock my cores to 4.3 and set faster RAM to 3600 with tighter timings and run 1:1 with IF. PBO was always inconsistent even under water for my 3700x. I find it comparable to my 7800x locked at 4.8GHz in gaming while also having 2 extra cores.
The f clock automatically sets itself at one half the memory speed at up to 3600 MHz on Ryzen 3000 processors. The f clock goes up to 2000 MHz on 5000 series processors so 4000 MHz is the sweet spot. You can overclock the f clock on Ryzen 3000 series to 2000 MHz and the 5000 series to 2200 MHz but that involves undervolting the cpu and overvolting the I/o die. Way too complicated for most people.
Would love to see how well tuned b-die kit does vs typical docp settings
5-6 % overall performance improvement on tightly timed bdie, but 0% in some tasks and more than 10% in others, it's all over the place.
The IF matters more than me freq.
14-14-14-28-42 3733 vs 3200 DOCP CL14.
It made a huge difference for me in game FPS at least. Borderlands 3 ran about 20% faster.
Makes huge diff to minimums in some games. Like 20%+ in some. Averages in those same games only go up maybe 5% tho so people think it's not a big deal. Running samsung bdie 3600c16 at c14 with pretty much every timing tuned.
Hynix c die is just as good. I got 3200mhz cl16 hynix c die and it ocd to 3800 cl16. I saw my fps in modern warfare go from 130fps to 250fps was kind of insane and i didnt expect gains like that.
@@yoltsbp cjr is junk compared to samsung bdie it scales well frequency wise when you throw voltage at it but timing wise its mediocre at best. That being said samsung isn't worth the premium. I own hynix djr 3600c16 kit and samsung bdie 3600c16 kit.
Used your Microcenter link. Purchaced a laptop and stand. Liked the video. Commented. Been subbed for a bit. I think life's good right now :)
Really appreciate the large monitor. I can actually read benchmark number.
February 22, 2020 i run my Ryzen 7 3700x Plus Msi x570 atx + Rtx 2060 8gb Super & 32gb 3000 Ram with Xmp On = Rock -N- Roll 👾 Zero issues 6 months later 🤘🏽 Thanx for Posting 🇵🇹🇵🇹🇵🇹🇵🇹
Phil, love the glueing of Jay's face animation. Gold.
I dont even know what you're talking about or even afford anything above your workbench, i still enjoy listening to your videos. Good stuff!
Jayz: hundred perceet..
Also jayz: hundred perspent...
Ill never forget the time my girlfriends friend was setting up a cellphone for us I said "please dont hit my credit a million times", her response was "dont worry, Im a ferpessional"
this guy is crazy he is really underrated even tho he has 2.5m subs but he inspired me to build a pc and learn about how pc's work, Thank You!
I need a Micro Center in FL!
I need it in New Jersey
Jk I have it in NJ
@@dexterous2513 Been meaning to go but they lack some item's I need so I'm supplementing it with newegg.com
I need one in Sydney Australia matching USD Prices in AUD
Need one in the UK, closest thing here to microcenter only sells a couple computers and mostly sells toasters and fridges
3800 C14 Fclock 1900 is very much doable on some CPUs. Most can do 3733/1866. Custom timings also make a big deal on 1% lows even more than outright FPS.
That background really took me back, I was so confused for like 10 seconds trying to wrap my head around how you got the testing suite to work on that, then I saw the taskbar lmao, great video!
A RAM timing video would be much appreciated, I've tried messing with it on my mobo and results have been mixed...
When I started doing memory overclocking on my b350 board, if I locked it up with bad settings I had to remove the memory completely, clear the CMOS, boot it without the memory in the slots, then turn it off and reinstall the memory, then it would boot with default memory settings.
Like how you summed up the ending. Can you maybe make a series where you go over all of the words/terminology(specs), so that a noob can maybe understand what your saying. Love the channel.
I've spent an average of 35 percent of my income at Microcenter since 2014
Only 35? You have will power!
Dude that’s housing level spending. 😮
Never stepped foot in microcenter, and never ordered from them online either.
Wh-How? You make minimum wage and spend $8k a year?
Damn son!! 35% would be right around 18000 of my annual income. I take it you're not only buying PC components?
Please do a video about timings on ryzen as well. Becasue some people say you can get hella better performance with that.
Oh I forgot. Great video Jay. :)
Did he do the video if he did can u please tag me in it lol.. I am also hearing the same thing on timings.
Timings don't matter
AHOC fans: What
As some one who watches AHOC as well I have to agree, plus Cinebench is not memory sensitive and is a good example of why memory speed can and should be tuned speed wise more for your workloads as you can save money on your RAM that way. Plus 3200 is actually a JDEC spec so you can find kits that do it out of the box with out setting XMP at least when dealing with Ryzen not sure on Intel as they have not been where I've been paying attention as I'm a core heavy user. Just ask my poor Ryzen 5 3600 that has seen more full core loading than most of them will see :D
AOC fans: RACIST!
@@justjoeblow420 i dont think a single kit sold so far has a 3200 JEDEC profile yet. Kits only recently started shipping with 2400 and 2666 JEDEC profiles...
Anything above that is 99% XMP profiles.
@@TheHavocInferno I've definitely seen one, not for sale, but they exist. I think it was a kit of ECC server memory.
TheHavocInferno there are plenty of sticks that you can buy, and I mean PLENTY that you can buy with JEDEC 3200. I would know it too given I literally get an email from JEDEC every morning about random ram stuff. The majority of JEDEC 3200 sales come from OEM sticks. Actually, you can buy some JEDEC 3200 B die from Samsung right now.
Jesus...this video in particular is clear as mud. So many starts and stammers and new directions. Get it together, dude!
Love this 6:37
When you try so hard to say something. It happens to all of us
Oh hey someone actually recommended what I did. Got 3600MHz CL16 RAM for $70 for my 3600.
Me too Buddy
thats what everyone knows, jays late to this party by months...
I wouldn't say 3600 is a 'sweet spot' at all. It is for Ryzen 3000 if you don't manually OC your FCLK cuz it maxes out at 1800mhz by default.
That said 3600 cl16 is a great kit of memory for out of the box XMP/DOCP. There's a lot of 3600 CL 18 kits out there which is technically slower than 3200 CL15
Put 3200mhz in my x570 with ryzen 3600. XMP was loose cl16. I manually changed it to 3600mhz tight cl18. Boost was easily measurable across the board. Highly recommend 3600mhz
Amd ryzen master, you can specifically target cores to overclock, I oc 2 of my cores to 4.4 and the other 4 at 4.3 on my 3600. It's fun to mess with too.
What voltage do you set?
Anywhere around 1.4375 to 1.4875
But the main oc that I set is a 4.375 overclock on all cores and I have the volts set to 1.4375
It was able to run cinebench r20 3 times in a row with no problems
That's weird, I am running a 4.3 Ghz all core oc using ryzen master at 1.225V , anything above 1.3V makes my processor heat up like crazy in prime 95. Have you ever run prime95 for stress testing or just cinebench?
We need a micro center in FLORIDA. We only have Best buy so needless to say they would have no competition.
The way he said Perspent so confidently makes me die from laughter
Perspent is the percentage performance you gain for the percentage money spent.
Spend $500 and get 500 performance points
Spend $1000 and get 1100 performance points
This gives it a +10 Perspent value.
Magnus Nilsson I feel enlightened
i do believe that ryzen master gives you the option to per core settings.
yep
this
Great video Jay. I have been having issues with my ram for a while where I couldn’t get it to run at 3200mhz even with docp on and all the turbo boost and core enhancement, etc. Thanks to this video I now know to go in and set the Fclock which I shall try later when I’m home from work. Thanks again for the great content you’re putting out.
This is great, but it's fairly well known that 3600MHz is the sweet spot for Zen 2. How about a comparison between kits? CL18, CL16, CL14 etc.
thanks jay this helps me a lot with my parts selection for my ryzen build im switching over to ryzen from intel for my next build
come to the dark side my friend !
@@camerongrieve5521 ikr im late to the party
I switched back to Intel and I feel good about myself. Ryzen was not for me.
Veni Vidi Vici ryzen isn’t for me but yet I keep buying it. Not sure tbh
@Tinkering With Tech I’ve ran 4 different Ryzen cpus since zen1 launched and never had a single issue. It’s cute when people with bad analytical skills make broad generalizations about computer parts they just don’t know how to use.
Been running a Ryzen 3600 X with fclock @ 1800 MHz on 3600 ddr4 Viper cl14 & core enhancement Since January 2020, a week after I built it. Good to know it already doing the best.
0:44 "Was" known? It still "is" known!!!
Usually with ad spaces I'm like oh cool I'll avoid that company because if they need ads they probably suck but microcenter is my favorite place. It's better than buying online and if I have to buy online it's nice to know that I'm supporting a cool store and business.
Also nice vid Jay lol
Please do a timing video :)
A video highlighting the DRAM Calculator for Ryzen would be really cool Jay. Easily got my 2666 set up to 3000 in a few minutes and its super simple.
So is the Dram Calculator pretty accurate?
I learned so much about ram speeds this was excellent
Fabric clock isn't half of memory speed when it's half of *reported* RAM frequency. DDR means Double Data Rate, which means RAM talks twice per clock. That is shown as double actual frequency. Fabric clock is exactly RAM freq when it's "half", probably also why it gives the best performance because they are in sync.
I want you to do a montage of the guy in the background laughing. Im sorry I dont know his name, im REAL bad with names. But he is hilarious.
Yes memory timings video please!! I've been into building computers since 2013 and I've overclocked everything I can in all those computers except memory
0:08 well, microcenter doesn't have nothing right now. They have 0 motherboards, 0 decent budget ryzen cpu's (3200g, 1600af, even 2600).
It depends on where you are. Mayfield Heights, OH has a BUNCH of motherboards, not sure about the budget CPUs, though.
The main issue I'm seeing GLOBALLY would be power supplies. I can't find a single decent 650W+ Gold that isn't price gouged to fucking smithereens.
1600af is discontinued, 3200g is a quad core with apu, 2600 is meh. Get a 3300x.
Ryzen 3 3200G is probably the most overrated CPU I have seen. It costs as much as Ryzen 5 1600, but is MUCH worse. Not only in game performance, 1600 is 6c 12t CPU which gives you better framerate on lower usage.
I would say the best budget CPUs are Ryzen 5 1600, 1600af, 2600, Ryzen 3 3100x. Eventually I3 9100F
@@MrDutch1e 3300x is basically non existent right now, can't get em anywhere! I had this planned on a custom build and had to bite the bullet on 3600xt because it was all they had in newer amd cpus with relative or better performance, although of course, way less price/performance value for me sadly which is probably why it was available! 😂
I would like to see a new test with the new Ryzen 5000. The fps improvement for faster RAM should be higher, since the clock speed of zen 3 is higher as well I think? Also need a new gaming rig and kind of confused on which RAM to by. But speed should matter since with their new gpu in combination to access the faster DDR6 graphic memory they also boost performances, or am I mistaken?
I like the fart sound at the end. Nice touch.
It’s fitting. Jay is prone to making dad jokes, and dads fart a lot. As a dad, I know this to be true.
@Narendra Lau WTF LOL
You're 100 perspent my favourite channel 🤣🤣
Percent*
Stupid auto correct 😡😤😠😤😡
Jkjk ik he said perspent
Hello Jay, oc by ccx , normally 4 options check with hwinfo64 what cores go faster 4.7ghz etc. On bios you will oc by ccx and not all core. Ccx0 44.75 ccx1 44.50 etc just an example. Voltage don't go over 1.38v be careful with the LLCs so it won't go over 1.4v. my 3900x like this does 7850 on cinebenq with a x370 gaming Pro carbon. Hope it helps
Phil's laugh cracks me up everytime. This is the funniest tech channel imo
Yeah I can agree here
Complete opposite for me. Not entertaining and annoying as its becoming more prevalent. As well as focussing on Jay's stutters over words...
Definitely is
More videos with Phil!
Indeed.
So I was able to overclock my ram from 3200 to 3600 at CL14 with tight timings using the DRAM calculator that's been floating around for a while. It was a significant performance boost and it's literally never crashed or locked up in about a year's time now. The question I have is how do I know that I'm not getting errors? Not every error causes a crash, but I am assuming it could cause it to have to recalculate something if fails a check or something.
Jay struggling just to say PERCENT is the best thing ever, and Phil's laugh makes it even funnier 😂
Does fake laughs make me cringe!
Jayz, what's the relationship and impact of CL against a motherboard's QVL. Eg. A 3600MHz RAM with lower CL but not on the motherboard's QVL vs a 3600MHz RAM with higher CL but is on the motherboard's QVL. Does it even matter for a RAM to be on the QVL? Can I take a non-QVL RAM with lower CL (eg. CL16) and use "DRAM Calculator for Ryzen" to try and tighten the timings, or is it still faster to use a QVL-certified RAM that's maybe CL18 or CL19?
I swear I'm having deja vu. Haven't we done this exact video at least once in the past??
I think Linus and GN have done this before, I had the same feeling of Deja' Vu.
No not the specifics of this one but yes, a very similar one was uploaded before
Looking back at his uploads what I think it might be is the fact that he's uploaded like 3-4 videos in the past year with "ram speed effecting gaming" in his titles.. I guess they weren't specifically about ryzen but I'm pretty sure his test bench is ryzen now anyways lol
Feels like I've seen at least two before this one this last year.
@1Me I didn't know jay charged for anything or that I was buying anything.
If only I understood a few of the words you are saying. I am almost 60, don't know much, and trying to learn this way, is just over my head.
this kinda guy sucks now. gamersnexus would be a better teacher if you’re just starting
@@MacBoiiPr0 thank you. Just beyond my level of understanding which I'm sure some will make fun of. But thank YOU for being so kind.
thank you for mentioning this I checked my ram speed and it was 2666 instead of the 3200 it should've been
I took some cheap corsair dom 3200mhz 16-18-18 up to 3600mhz 18-19-19 and my score went up a lot . My fps in games also noticeably up. I feel frequency/infinity fabric boost is more noticeable than when I did 3200mhz with tighter timings. X570 3600 cpu
People: Arguing about RAM speed performance
Me, an intellectual: Download more memory to get the extra performance
fell off my dinosaur last time i heard this joke
ofc
I downloaded so much memory COX throttled my neighborhood
how do i download internet, so i can download ram?? Help
@@r21174 you know you have to download the computer then the monitor then the wifi so you can download more ram 😀
06:40 that right there is the reason i love this channel so much XD
"AMD Ryzen Master" let's you set speeds for individual cores.
I’m planning on building my first pc. So far my parts list is the R5 3600, rtx2060, b450 tomahawk max, 16gb 3600 g skill ram, 250gb m.2 nvme ssd, and 1 tb Seagate barracuda hard drive. What are the best choices for a case and powersupply for my build, since I don’t know where to choose from in terms of best cooling and such. I’m mainly looking for help on what case since ik psu r in low stock and once they get back up I can find one
What about 1% lows. In my experience, that's where the faster RAM makes a difference.
It's the exact opposite. Faster RAM raises your avg fps because it removes some throughput bottlenecking from the top % of your fps where your CPU load is minimal. For example looking mostly at a skybox in a game will see your fps dramatically increase as the CPU is allowed to run free, and with faster RAM you will get higher fps there. However, when then CPU itself is under a heavy load (as Jay showed with Cinebench) the effect of RAM speed becomes negligible as the bottleneck moves to compute instead of throughput. That is to say - if you only want a stable fps (e.g. you're capping it somehow for G-sync or Freesync), RAM speed has very little effect on your experience.
I think you just did an underpants gnome reference at the end there....I approve \o/
It’s 3 am I’m dying but I still love ur work
Same he makes me moisty
You dying? Call an ambulance dude... seen too much on the internet lately..
Same here
@@albertsang5428 u want me to do u
U Wana join in to do the do
Awesome video, great info as always. I am running to my PC to make the changes!
I thought the requirements for ryzen is really fast memory to maximize its performance
Ryzen 3000 series processors get noticeable gains for ram speeds all the way up to 3600mhz. Past that, however, the gains and much more negligible.
@@squigglefitness should I bother trying to OC my 3600mhz kit then?
Not fast memory but low latency memory. Memory at 3600 with FCLK at 1800 is the sweet spot, but Ryzen wants low timings to improve performance, unlike Intel. This video was kinda pointless imho, it told nothing about how to improve Ryzen gaming performance. It just said what we all know already and kinda misled people into thinking that memory timings are inconsequential for Ryzen as they are for Intel.
@@marsovac what sticks would you buy for a 3700x I honestly watch stuff like this and still kinda don't get the info I'd like for gaming purposes
@@Trizjoe low timings 3600mhz kit is the sweet spot for ryzen
Love all the vids, keep it up brother!
Thank you for stating what every other youtuber said 6 months ago
😂
On my 3700X processor, Cinebench R20 scores were improved about 15% when I went from 2133 to 3000.
0:04 uh oh, I think his computer got a flagrant system error
Computer over? Virus = very yes? That's not a good prize!
Came into the comments to see if anyone else got the reference. Am not disappointed.
Alright...so what about a 16GB kit of DDR4 3600 CL14. More specifically talking about the Teamgroup T-Force Xtreem ARGB kit. Would the faster speed and really low latency be a huge/noticeable boost for Zen 2?
Illidan: "You are not prepared!"
Jay:"quid pro quo!"
Illidan: ?¿???¿?¿?¿?
illidan: fear the burning legion"
jay: burning hot? let me put an AIO on it. that´ll fix it"
Jay FCLK it's actually on a 1:1 ratio if you put it at "half" ram speed, base ram speed it's exactly the half of what you see (DDR = Double Data Rate), so that's actually 1:1, and that's why you have the best results with such settings.
And no, CL timings DOES COUNT. At least in gaming. I have my ram at 3733 with a mix of cl14/15 with very tight subtimings, and difference with cl16 and relaxed sub timings is it there.
The only downside of oc ram too much, it's you will get really higher temps on the cpu, because you just increase the cpu ipc by having way faster memories, but temps really spikes up.
If I put my ram at 3200mhz cl14 (stock xmp/docp profile) I have something like 10 Celsius LESS on my cpu, and spikes (mainly in idle) are way less prominent.
I tried also with 3733mhz but cl16 timings, and temps were always lower than with very tight timings.
My 2 cents? If you have a good cpu cooler, leave zen 2 do what he does (boost, or turn on pbo with some UV, but be careful, or you will get core stretching), and play a LOT with rams, it's where you will get the major boost, especially in gaming.
me: goes to parts picker.com and watches the cost of 3600mhz ram double for the next 6 months.
I found at Amazon that they got 2x8 3600 for 80 bucks Corsair something
I am still seeing a lot of 3200mhz 8x2 sticks on sale for around $60 -- G-Skill, Crucial, etc. Just sign up for newsletters from etailers and you will be notified when parts go on sale. I doubt DDR4 will see a big price increase when they need to clear stock for DDR5 in 2021.
linus did same video awhile back and nothing happened with prices so doubt they will now unless stock runs out.. and seeing as all the MOBOs are gone unless u buy the highend ones they may be next on the list. I did however upgrade from 16g 2933 tridents to 32g of 3600 after they went on sale
You can just buy 3200 mhz CL 16 DRAM and overclock it to 3600 with minimal effort like mentioned.
I don't know about you but ordering multiples of twin packet sticks (ie x2 packs of two sticks/pkt for 16GB DDR4 @3600MHz) saves a coupla hundred, compared to 4-pack 'matched' ram. Its much like Crossfire and SLI to me. The bet I make is to stack twin stick packs in the A1 B1 +A2 B2 dimm slot order so the electronic abstraction sees each matched ram pair in their own channel (if I have that right).
**Can't wait to finish my build: now 64GB of 3600MHz ddr4 richer, just arrived. G.Skill Ripjaws V Black, in a Black Corsair Crystal 280X mATX case with its glass top pane sitting on stacked nylon washers and the grommets upside down for extra standoff height, main glass side panel grommets reversed to set panel flush against case, front panel dust filter removed. I have an MSI Radeon RX 5700XT Mech otw (this model is only 232mm long). Almost everything going on inside my case is black, except the brushed aluminium heatsinks of the Asrock X570M Pro4 mobo, and its white printed stuff, and the Ryzen 9 IHS. Not a fan of pretty lights. I might custom build a puppet theatre stage curtain-esque drape to the glass side panel .. I'll think it over.
I love how, after so many years of being succesfull, you still take the time to explain stuff most of your viewers know. (Like how DDR4 out of the box runs, free benchmarking programs, difference between Intel CPU's) I know you know the majority of your viewers know, but some don't (me :-) )
OS : Windows 10
Wallpaper : Windows 95
upgrade windows 95 to windows 10 that's the shit dude
R5 3600X here with DDR4 running at 3733Mhz and Fabric Clock set to 1866Mhz (OC).
Most Zen 2 chips will be able to overclock the Fabric Clock to this and if you plan to upgrade to upcoming Zen 3 (aka Ryzen 4000) it makes sense to invest in higher frequency RAM.
Also don't forget to tweak your subtimings too using Ryzen DRAM Calculator as it makes a noticeable difference.
Absolute first?
Edit: Only 12 seconds after the upload, I think it's just my second time being first in a UA-cam comment
last
No
Only the sith deal in absolutes
I have 32oomhz memory but I bought the ASUS Prime X370 when I got my Ryzen 7 all those years ago only for the board to claim it runs 3200mhz but becomes unstable over 2600mhz. I will NEVER buy any asus product again. Nor will I buy nvidia for lying when i bought my gtx970. PC companies are the worst.
"3,5 is slow these days"
Me, with my "slow" 1700 at 3,8 GHz: 😮
me with my 1700 at 3.7Ghz 😠
3.5 is my max ☹
Microcenter - what PC World in the UK should of been.
tl:dr 10:42
Why does no one increase the priority to high on the Cinebench 20 process while running a benchmark?
Constant and pretty much max scores every time.
Backgrund tasks don´t bother me anymore!
Do I miss something here?
The JEDEC DDR4 SPD specification allows speeds other than 2133 (both slower and faster); so non-overclock 2400, 2666, etc is quite possible. The standard includes examples up to 3200. It's just a combination of manufacturing and marketing that has made most DDR4 modules ship with SPD configured as 2133 (and XMP for the real rated speed after binning).
Isn't there a ram divider that kicks in after certain ram speeds that essentially reset memory performance? You need to test 3200, ~3400, 3600, ~3700, ram speeds as well, and check when the dividers kick in. There are certain ram speeds that will peak performance at a particular divider, but then drop after a new divider kicks in (ie. 3600 is peak of one divider, but a new divider kicks in at 3700 and performance drops down a bit).
I was disappointed that this video didn't address the concerns of timings. There is a substantial lack of information on this but the few channels and sites that test it seem to show significant difference in performance in a variety of CPUs. I am currently in the market to buy a system and honestly do not know how much of an impact 5000mhz CL19 vs 3800 CL14 RAM has an impact on a 3800-3950X and would love to see data that helps me guide which direction to spend my money. Please consider doing a timings video to help some of us with these kinds of questions