Man I feel like a student who actually enjoys listening to his teacher , you really explain everything so good that it is actually interesting to listen and learn, glad I found this channel!
OG White Apex. yes The 24gb SK Hynix M-Dies are easier to push it seems. I am using the Team Group Xtreem kit. I guess it's their binned memory IC ram kits. @@blackbirdpctech
1: Great test, great video.. thanks! 2: When deciding which RAM to go for - I do not for a second consider its effect on FPS. The only thing I'm interested in - is loading times. _(...and by extension, how large % of frames are below 30 FPS.... or below 15 FPS... or worse, as freeze-frames are instant "quit-moments" for me.)_
For me this was a win for 6000 Mhz kit lmao the price/performance just isn't there with 8000 - Much more expensive kit and motherboard for what? 5% of difference? Yeah no
I’ve never been able to get my Team Group DDR5 ram to actually run at its rated 7200 speed on my 14700K system. It is 2 x 16GB single rank. The ram runs fine at 5200 and 6000.
What motherboard are you using? I have a test bench with a high end Asus Z690 extreme with a 14700k and I have the same problem … I need to bin the speed down to get it stable.
the issue is speed vs cl once you go to high on the cl its pretty much a downfall. You have to find the happy medium between speed and cl, a balancing act. Then there are the sub timings that will also play a role. Like take G.Skill Trident Z5 ARGB 7200mhz cl34 vs Corsair Dominator Titanium/Platinum's both kits are cl34 but Corsairs more refined sub timings gives it more speed than the G.Skill. Then there is the game it will affect harder and at what resolution you still want it to be gpu bound so your 1% lows are higher this will come into play with Cyberpunk 2077 at 3440x1440 DLAA+Path Tracing+Frame Generation with everything at Psycho/Max if you get the right board, the right ram, the right card you are looking at 60-80 fps with a frame time of 12-18, if you didnt get it right your frames will drop below 60 into the 40-50 range and frame time will also shoot up to over 20. Sadly with DDR4 4000mhz cl14 and 16 that you cannot get at least at a reasonable price.. I have those kids luckily and they are on par at the sweet spot running in gear 1 as 7200mhz cl34. Its pretty wild when I upgraded was like ohhh, I also have some higher cl 18 ddr5 kits ranging from 3600-4000mhz that also could not run in gear 1 talk about butchering your 1% lows by a good 15-30 fps.
Aorus master + 13900k here.....I wish I had seen this video yesterday, before buying two ddr5 8000 cl38 kits (one 2x16gb, one 2x24gb). So no chance to make one of them work at 8000 mt/s I guess 😂😂 So far I'm running stable 64gb(2x32) corsair 6600 cl32 ram kit, but I wanted to test higher frequency, but now I understand it's not worth it
Don’t feel too bad, you should be able to get 7600 or 7400 to work and the performance difference at those speeds is nothing meaningful. I noticed that a lot of UA-camrs in this space really have no idea what they are doing, so they unintentionally give bad advice. Hopefully this channel can help fix that.
Yes, nowhere else I had found the clear info that only few top end dual slot motherboards could MAYBE if you're lucky accept 8000 Mt/s modules. Besides Aorus master is declared to support ddr5 up to 8000. Anyway, If you say that performance increment at 7600 would not be meaningful, I can save my time (and money) trying to get there. I would rather try 2x48 gb kits like corsair 6000 to 6600, that could work in my case also being an improvement
@@frankthepriest1220 what I meant was that the performance difference from 7600 to 8000 is not meaningful. You might want to try the 8000 kit out but with the speed set to 7600 if you are comfortable doing that in Bios before returning it ... you might get lucky with silicon quality and have it run stable. A 96GB kit from Corsair would be great ... I am running a 96GB 6400 corsair kit on my Intel Z690 testbench and it runs great. For some reason the newer 24GB/48GB DIMM's appear to run more stable ... something I will have to dig in to more. The fastest kit that I would tell most people to get if they want to run XMP is 7200 ... that should run stable with 2-DIMM's on a 4-DIMM motherboard.
@@blackbirdpctech @blackbirdpctech ok sorry I got it, you meant 7600/8000 is almost the same. So you tested 96gb too, interesting. As I said, now I run stable 64gb Corsair 6600 cl32, straight xmp profile no tweaking.. you would go with 1) 96gb at 6400/6600 or 2) 48gb at 7200 (hopefully 7600 but I'm not that confident). I mean, my goal is to have better overall performance, not pfs
@@frankthepriest1220honestly I would stick with what you currently have working … if you need more ram for professional reasons then go for the 96 kit, but it won’t make any difference in games.
I've never found RAM speed to be a big deal. The difference when using RAM rated faster might be measurable, but in most applications is not noticeable. AM4 was a PITA regarding RAM speed at first, and I suspected this could be the case with AM5 as well. That's why I went with LGA1700. My 13700K runs on a Taichi Z690 board with 2x32GB DDR5 5600 CL 36, chosen because at the time DDR5 prices were still quite high. Based on my experience with DDR3 and DDR4 kits on various platforms, I doubt if going to 6000 CL30 would be meaningful.
I just got ddr5-8000 running on my 7800x3d … I’m going to see if there is any performance difference versus 6000 for AM5 … I’m really curious if it scales the same way that Intel does.
Well try Pubg at comp settings, ddr5 5600Mhz runs like trash. Now try 7600Mhz cl34 tight timings and watch 0.1% lows. It all depends on the game you play. If you dont have a 240/360Hz monitor you can even see the effects in practice. Install Riva tuner statistic and watch 0.1%Fps lows. So many ppl talk about things they cant even test themselves and therefor theres tons of wrong info in the gaming world right now.
@@blackbirdpctech Thats already tested by Buildzoid and the infinity fabric limitations made it so that it had no had no effect. Try running Pubg on that chip, ppl now claim that Pubg finally runs well on x3d cpus. Which I highly dounbt since the Pubg devs are the most lazy devs I have ever come across. But you get a tons of VIEWS if its true since theres a ton of Pubg players in EU and Asia. Pubg have traditionally liked low latency ram systems so if they got the cache working it might be time for to go AMD.
@@impuls60 You're right about the problem of people echoing what they've read or seen on UA-cam as if they had done it themselves. But what a given person reports based on his or her own actual observation is not wrong information; it's that person's experience. I don't play high-FPS shooter or battle royale games, but I have read that fast RAM can make a difference in such games. If that's your observation in PUBG, then you absolutely should run faster RAM for playing PUBG.
That’s why I tested at multiple resolutions. The challenge when benchmarking is to decide on a philosophy … should I test how someone who purchases 8000 MT/s memory will likely play or test at settings no one will play at but will show max performance of the memory? I chose the first option because I think it is more representative of what most people will do.
@@blackbirdpctech I understand that, but you should had put some edge cases like CS2 bench and Shadow of Tombraider at lowest possible settings to show how memory can scale in the future and also e-sports
It's a been a bliss learning from memory overclockers on frequency limits, IMC lottery and manual timings tuning. Performance in games seems to be scale up with tight (lower) primary/secondary/tertiary timings, but of course within small margins of CPU limited senarios. I see no reason for regular DIY PC users to go for more than 6400 for 12th gen and more than 6800 for 13th gen.
I think that’s a very reasonable recommendation … I like watching memory overclockers too … I particularly like buildzoid … he understands memory parameters better than anyone else I’ve seen and he explains what each parameter does.
Do not buy 8000Mhz for gaming, it does not scale. Buy any kit of ram that can do 6400C30 and tune all the timings manually. It takes a while to learn but you can really pull performance out of cheaper RAM kits when you learn to tune them. I got my Samsung B-die 6000C36 kit to 6800C34 with massive timing reductions in the secondaries and tertiaries, 50.4ns in AIDA64
I assume that you are using an Intel based system. I follow buildzoid … he provides great advice for memory overclocking. The new AMD agesa update provides two paths for AM5 … latency reduction or bandwidth increase … I am going to test both with the same kit to see which one is better.
Thanks, I purposely ran the benchmarks at XMP with stock settings. This system used to be my primary gaming rig, so I would tighten the timings, kill the e-cores and oc the cpu … I’m pretty sure I was able to get just below 50ns (something like 49.7).
@@toonnut1I sold the system to a friend just after making this video … my oc settings should still be saved in bios (I doubt he would update the bios), so I should be able to get them … will post on my website (not enough content for a video)
DDR5 6000 and Expo setting for me. Just flip the switch and it works. No days or months messing around. I don't have time nor the patience to be chasing problems.
I plan to look into the impact of memory speed on a 7800X3D now that the new AGESA update was released. That said it appears that AMD 7000 series chips don’t really benefit with memory speeds above 6000. The 3D V-cache certainly helps to offload the need for faster memory.
If the only reason you want to upgrade is to get better memory performance then i would say wait a generation … however if you are buying new or upgrading the motherboard then definitely get ddr5, I simply wouldn’t recommend getting the high speed memory kits.
It really depends on the game.. The game that benefits the most from good ram is star citizen currently. i shit you not but there's almost a 20 fps difference between ddr4 and ddr5 in the big city landing spots so from around mid 30's to high 50's on the most demanding areas
The latency or CL that you run will not impact the life of the memory, so it's safe to run at CL 30. However, in order to get it to run stable at that latency you may have to increase the voltages and that will impact the memory module. If the voltage is too high you could indeed damage your memory, you will increase the temp and it will have a negative impact on your RAM life. If you don't have much experience tuning RAM in BIOS then I think the best way to proceed is to find a tighter timed kit from the same manufacturer, say a CL30 6000 kit, and try to run those timings at 6400 and see if it's stable. You might get lucky.
@@TheMichaelparisunless you change the voltage in Bios manually you will not have an issue. When you increase the speed or tighten the timings it is typically necessary to increase the voltage to achieve memory stability. A great resource to learn about memory timings is Buildzoid. That said, you will likely not gain much performance by tweaking these settings so if it's not something you really have time to test then I wouldn't bother ... just buy the kit you want and run it at XMP or EXPO settings.
i ran a occt at cl 30 and there was a lot of errors. i changed it to cl32 and it passed the stress test. I used memory try it on a msi motherboard too scared to change everything manually lol @@blackbirdpctech
NO MOBO RAM is al about the manufacturer. I ahve tried other brands like kingston and so on , But corsair has never let me down. The rest are absolute shit. I been building and gaming since the 80`s.
I've had good experience with G.Skill and T-Force, but I agree that Corsair is also good. All of these RAM modules get their die's from the same manufacturers, all they really do is put unique head spreaders on and bin/test them. So for DDR4 Samsung B-die was the best and for DDR5 Hynix A-Die is the best. The CPU memory controller will have the largest impact on what RAM speed you can get stable, followed by the motherboard layout.
just a comment, i agree that some ddr4 particularly good b-die such as 3600c14 is competitive vs ddr5. That said for those with intel 13 and now 14, it is possible to go an extra step , and run ddr4 4000+ at c15/16 in gear1 mode ( 14700s and 14900 can go at least to 4200 and depending on imc lottery, between 4266 to 4400c16 is achievable), and that tends to perform even better than 3600c14 , and by extension is even more competitive with ddr5. the drawback is that it takes some time to get the vccsa and vddq voltages right and ddr4 4400mhz quits are not cheap
Great comment and insight. I only have one ddr4 board that will support 14th gen with a bios update … it’s an Asrock z690 extreme … will be interesting to run some tests. Will see if I can find some higher speed ddr4 ram (fastest I have is 4000).
@@blackbirdpctech you should be able to oc a 4000 kit to 4400, if your 14 gen Imc is able to handle 4400g1 is another story (most are not) but at least 4200 should be achievable. obviously it is easier and much faster if you can have something like g.kills f4-400016d-32gvka (16-16-16-36) or f4-440017d-32gvk (17-18-18-38) as the default xmp profiles of both are already relatively tight (by xmp standards) and any time investment to make a video will be shorter and mostly tied with finding stable vccsa and vddq values. My experience (need to point I do not have a 4090) is that 4300c16g1 tends to outperform dd5 up to 6000 or so in most cases, will trade blows with ddr5 7200 at resolutions above 1440 but will generally lose to ddr8000 . The saving grace is that it is considerable less expensive to get a run of mill ddr4 z690 mb and a good ddr4 kit ( total cost around 400) then it is to get one of the few z790 boards capable of really running 8000, that are all on the 500+ dollars price category plus the 8000 kit, that is probably 200+/ And even if one is willing to fork the dollars, there is no assurance that speeds above 7600 will work.
@@Solrac-SiulI just picked up a g.skill 4600 kit … I have a few 14th gen chips to play around with … hopefully I get lucky and get a golden sample … would make a cool video to compare ultra high speed ddr4 with ddr5-8000+ (whatever I can get stable). Thanks for your input, really appreciate it.
@@blackbirdpctech that will work, you will probably need to actually downclock , the good part is you will not need to worry much about ram voltage. I will be checking out to see that eventual video. I know the future is ddr5. it is just that right now.... not only we have to contend with the IMC on the chip but also motherboards, as manufacturers are coming out with claims of 8000+ support, when many times they cannot even run 8000 at default xmp stable...
@@Solrac-Siulyes, they all have started doing it with the refreshed Z790 boards … adding memory to their QVL that has no chance of running stable at XMP settings. It’s marketing BS as I mentioned in my video. They really need to be held accountable for blatantly lying.
Great question. AMD 7000 series chips behave very differently with DDR5 memory than do Intel chips. The X3D chips are also much less reliant on RAM due to their large L3 cache. When AM5 and 7000 series chips were first released it was tough to get anything past 6000 MT/s stable, with 6400 the typical upper limit. With the relatively recent AGESA update that has now changed. From what I've seen online however there does not appear to be any benefit from matching fast RAM with 7000 series chips ... I plan to do this testing for myself because I'm really curious to see the results. All of that said my recommendation is to get a DDR5-6000 EXPO kit with tight timings at CL30 from a brand like G.Skill, Corsair, Kingston, etc.
Yeah alrighty thanks mate surprised you don’t have more subs but with the quality of your content won’t be surprised to see you past 100k in a few months
@@kieranlyall6140thank you … I just launched my channel 2 weeks ago so it’s growing rapidly, but it will take some time to hit those big numbers … really appreciate your support, means a lot !!
Not sure I really understand what you are saying. If you are testing a CPU or memory then testing at 4K doesn't make any sense because it will be GPU limited, so that's why we test at 1080p even with a 4090. For my upcoming GPU testing I just tested at 4K for similar reasons.
@@blackbirdpctech yes thats what i mean, your video include test at 4k and there's no improvement in gpu limited sceneario. While other youtuber mostly just listing at 1080p and the audience hoping buying higher end cpu and ram get more improvement, but that's actually not the case when you really push your gpu
Another great piece of software that I use to test memory stability is Karhu … it’s super fast and I highly recommend it, but it’s not free.
I use karhu myself. Great memtest.
Karhu running at the same time as Modern Warfare or Pubg is a great way of testing ram stability.
Man I feel like a student who actually enjoys listening to his teacher , you really explain everything so good that it is actually interesting to listen and learn, glad I found this channel!
I really appreciate that, thank you for watching and being a valued part of our community!
OMG love the Bruce Buffer cameo at minute 6:40 😆
😂
Running DDR5 8400 CL38 in a 2x24gb config on my Apex with 14900K.
Awesome … it appears that the 24gb modules are easier to get stable … are you using an apex or apex encore?
OG White Apex. yes The 24gb SK Hynix M-Dies are easier to push it seems. I am using the Team Group Xtreem kit. I guess it's their binned memory IC ram kits. @@blackbirdpctech
@@TalonsTechyeah, I like the original apex a little more … although the memory fan is pretty cool on the encore.
Holy shit, more people need to see this.
Thank you !!
1: Great test, great video.. thanks!
2: When deciding which RAM to go for - I do not for a second consider its effect on FPS. The only thing I'm interested in - is loading times. _(...and by extension, how large % of frames are below 30 FPS.... or below 15 FPS... or worse, as freeze-frames are instant "quit-moments" for me.)_
thx very good test
Glad you liked it!
For me this was a win for 6000 Mhz kit lmao the price/performance just isn't there with 8000 - Much more expensive kit and motherboard for what? 5% of difference? Yeah no
Agreed, it’s easy to get carried away with wanting the best but it typically doesn’t make sense given the cost and actual performance increase.
@@blackbirdpctech I could very well upgrade my GPU a tier or two higher and get much bigger performance uplift than going this way
@@MartinMlBanicabsolutely, especially if you are running at higher resolutions.
I’ve never been able to get my Team Group DDR5 ram to actually run at its rated 7200 speed on my 14700K system. It is 2 x 16GB single rank. The ram runs fine at 5200 and 6000.
What motherboard are you using? I have a test bench with a high end Asus Z690 extreme with a 14700k and I have the same problem … I need to bin the speed down to get it stable.
@@blackbirdpctechAn Asus Rog Strix Z790-E WiFi II . It’s a fairly new mobo that was only release in the last couple of months.
@@anorax001does it have a mc sp calculation? Also an SP calc? Asus boards tend to have those … helpful to determine silicon quality.
@@blackbirdpctech That is a good question. I did try to look for that yesterday but could not find it in the BIOS.
@@anorax001the MC one is under the AI option … will find specifics for you and post a screenshot.
the issue is speed vs cl once you go to high on the cl its pretty much a downfall. You have to find the happy medium between speed and cl, a balancing act. Then there are the sub timings that will also play a role. Like take G.Skill Trident Z5 ARGB 7200mhz cl34 vs Corsair Dominator Titanium/Platinum's both kits are cl34 but Corsairs more refined sub timings gives it more speed than the G.Skill. Then there is the game it will affect harder and at what resolution you still want it to be gpu bound so your 1% lows are higher this will come into play with Cyberpunk 2077 at 3440x1440 DLAA+Path Tracing+Frame Generation with everything at Psycho/Max if you get the right board, the right ram, the right card you are looking at 60-80 fps with a frame time of 12-18, if you didnt get it right your frames will drop below 60 into the 40-50 range and frame time will also shoot up to over 20.
Sadly with DDR4 4000mhz cl14 and 16 that you cannot get at least at a reasonable price.. I have those kids luckily and they are on par at the sweet spot running in gear 1 as 7200mhz cl34. Its pretty wild when I upgraded was like ohhh, I also have some higher cl 18 ddr5 kits ranging from 3600-4000mhz that also could not run in gear 1 talk about butchering your 1% lows by a good 15-30 fps.
2 more to 100 subs. Lets Go!!!!🤩
Love the energy ... I'm going to do a giveaway when I hit 1,000 😉
Something cool to share with you … can you email me at mjtarascio@blackbirdpc.tech
Aorus master + 13900k here.....I wish I had seen this video yesterday, before buying two ddr5 8000 cl38 kits (one 2x16gb, one 2x24gb). So no chance to make one of them work at 8000 mt/s I guess 😂😂
So far I'm running stable 64gb(2x32) corsair 6600 cl32 ram kit, but I wanted to test higher frequency, but now I understand it's not worth it
Don’t feel too bad, you should be able to get 7600 or 7400 to work and the performance difference at those speeds is nothing meaningful. I noticed that a lot of UA-camrs in this space really have no idea what they are doing, so they unintentionally give bad advice. Hopefully this channel can help fix that.
Yes, nowhere else I had found the clear info that only few top end dual slot motherboards could MAYBE if you're lucky accept 8000 Mt/s modules. Besides Aorus master is declared to support ddr5 up to 8000.
Anyway, If you say that performance increment at 7600 would not be meaningful, I can save my time (and money) trying to get there.
I would rather try 2x48 gb kits like corsair 6000 to 6600, that could work in my case also being an improvement
@@frankthepriest1220 what I meant was that the performance difference from 7600 to 8000 is not meaningful. You might want to try the 8000 kit out but with the speed set to 7600 if you are comfortable doing that in Bios before returning it ... you might get lucky with silicon quality and have it run stable. A 96GB kit from Corsair would be great ... I am running a 96GB 6400 corsair kit on my Intel Z690 testbench and it runs great. For some reason the newer 24GB/48GB DIMM's appear to run more stable ... something I will have to dig in to more. The fastest kit that I would tell most people to get if they want to run XMP is 7200 ... that should run stable with 2-DIMM's on a 4-DIMM motherboard.
@@blackbirdpctech @blackbirdpctech ok sorry I got it, you meant 7600/8000 is almost the same.
So you tested 96gb too, interesting.
As I said, now I run stable 64gb Corsair 6600 cl32, straight xmp profile no tweaking.. you would go with 1) 96gb at 6400/6600 or 2) 48gb at 7200 (hopefully 7600 but I'm not that confident). I mean, my goal is to have better overall performance, not pfs
@@frankthepriest1220honestly I would stick with what you currently have working … if you need more ram for professional reasons then go for the 96 kit, but it won’t make any difference in games.
I've never found RAM speed to be a big deal. The difference when using RAM rated faster might be measurable, but in most applications is not noticeable. AM4 was a PITA regarding RAM speed at first, and I suspected this could be the case with AM5 as well. That's why I went with LGA1700. My 13700K runs on a Taichi Z690 board with 2x32GB DDR5 5600 CL 36, chosen because at the time DDR5 prices were still quite high. Based on my experience with DDR3 and DDR4 kits on various platforms, I doubt if going to 6000 CL30 would be meaningful.
I just got ddr5-8000 running on my 7800x3d … I’m going to see if there is any performance difference versus 6000 for AM5 … I’m really curious if it scales the same way that Intel does.
Well try Pubg at comp settings, ddr5 5600Mhz runs like trash. Now try 7600Mhz cl34 tight timings and watch 0.1% lows. It all depends on the game you play. If you dont have a 240/360Hz monitor you can even see the effects in practice. Install Riva tuner statistic and watch 0.1%Fps lows. So many ppl talk about things they cant even test themselves and therefor theres tons of wrong info in the gaming world right now.
@@blackbirdpctech Thats already tested by Buildzoid and the infinity fabric limitations made it so that it had no had no effect. Try running Pubg on that chip, ppl now claim that Pubg finally runs well on x3d cpus. Which I highly dounbt since the Pubg devs are the most lazy devs I have ever come across. But you get a tons of VIEWS if its true since theres a ton of Pubg players in EU and Asia. Pubg have traditionally liked low latency ram systems so if they got the cache working it might be time for to go AMD.
@@impuls60 You're right about the problem of people echoing what they've read or seen on UA-cam as if they had done it themselves. But what a given person reports based on his or her own actual observation is not wrong information; it's that person's experience. I don't play high-FPS shooter or battle royale games, but I have read that fast RAM can make a difference in such games. If that's your observation in PUBG, then you absolutely should run faster RAM for playing PUBG.
You using Max/Ultra settings where the GPU is the bottleneck and the main memory subsystem inst pushed to the limits.
That’s why I tested at multiple resolutions. The challenge when benchmarking is to decide on a philosophy … should I test how someone who purchases 8000 MT/s memory will likely play or test at settings no one will play at but will show max performance of the memory? I chose the first option because I think it is more representative of what most people will do.
@@blackbirdpctech I understand that, but you should had put some edge cases like CS2 bench and Shadow of Tombraider at lowest possible settings to show how memory can scale in the future and also e-sports
@@Akgis32good suggestion, thanks.
It's a been a bliss learning from memory overclockers on frequency limits, IMC lottery and manual timings tuning.
Performance in games seems to be scale up with tight (lower) primary/secondary/tertiary timings, but of course within small margins of CPU limited senarios.
I see no reason for regular DIY PC users to go for more than 6400 for 12th gen and more than 6800 for 13th gen.
I think that’s a very reasonable recommendation … I like watching memory overclockers too … I particularly like buildzoid … he understands memory parameters better than anyone else I’ve seen and he explains what each parameter does.
Do not buy 8000Mhz for gaming, it does not scale. Buy any kit of ram that can do 6400C30 and tune all the timings manually. It takes a while to learn but you can really pull performance out of cheaper RAM kits when you learn to tune them. I got my Samsung B-die 6000C36 kit to 6800C34 with massive timing reductions in the secondaries and tertiaries, 50.4ns in AIDA64
Keep in mind.....I forgot to say this before, my DIMM's are single rank 16Gb, if you have dual rank or using 4 sticks it's likely to be harder to tune
I assume that you are using an Intel based system. I follow buildzoid … he provides great advice for memory overclocking. The new AMD agesa update provides two paths for AM5 … latency reduction or bandwidth increase … I am going to test both with the same kit to see which one is better.
I'm running 8400 36-49-49-96 the 1% lows is where faster ram shines. Less frame dips
Awesome, what motherboard are you using?
59.7ns seems very high for 8k ram speeds in aida64 im sure you could get that down. I really liked this video thanks!
Thanks, I purposely ran the benchmarks at XMP with stock settings. This system used to be my primary gaming rig, so I would tighten the timings, kill the e-cores and oc the cpu … I’m pretty sure I was able to get just below 50ns (something like 49.7).
@@blackbirdpctech excellent can you upload a video of your bios settings when fully overclocked and tightened memory thanks
@@toonnut1I sold the system to a friend just after making this video … my oc settings should still be saved in bios (I doubt he would update the bios), so I should be able to get them … will post on my website (not enough content for a video)
DDR5 6000 and Expo setting for me. Just flip the switch and it works. No days or months messing around. I don't have time nor the patience to be chasing problems.
I plan to look into the impact of memory speed on a 7800X3D now that the new AGESA update was released. That said it appears that AMD 7000 series chips don’t really benefit with memory speeds above 6000. The 3D V-cache certainly helps to offload the need for faster memory.
@@blackbirdpctech That has been said to be the sweet spot and that is what I am sticking with. I just wanna have fun.
@@Methodical2great attitude !!
I guess wait another year or so until ddr5 get some experience. Then it gets better
If the only reason you want to upgrade is to get better memory performance then i would say wait a generation … however if you are buying new or upgrading the motherboard then definitely get ddr5, I simply wouldn’t recommend getting the high speed memory kits.
It really depends on the game.. The game that benefits the most from good ram is star citizen currently. i shit you not but there's almost a 20 fps difference between ddr4 and ddr5 in the big city landing spots so from around mid 30's to high 50's on the most demanding areas
Interesting … I will check it out … thanks for sharing.
is it safe to run the ram at cl 30 because my kit comes at cl 40 6400 mhz and is stable and gets no crash ? thank you
The latency or CL that you run will not impact the life of the memory, so it's safe to run at CL 30. However, in order to get it to run stable at that latency you may have to increase the voltages and that will impact the memory module. If the voltage is too high you could indeed damage your memory, you will increase the temp and it will have a negative impact on your RAM life. If you don't have much experience tuning RAM in BIOS then I think the best way to proceed is to find a tighter timed kit from the same manufacturer, say a CL30 6000 kit, and try to run those timings at 6400 and see if it's stable. You might get lucky.
thanks for the reply after doing a passmark memory test the temperature was at 38C and 1.365V. The module is rated at 1.4V@@blackbirdpctech
@@TheMichaelparisunless you change the voltage in Bios manually you will not have an issue. When you increase the speed or tighten the timings it is typically necessary to increase the voltage to achieve memory stability. A great resource to learn about memory timings is Buildzoid. That said, you will likely not gain much performance by tweaking these settings so if it's not something you really have time to test then I wouldn't bother ... just buy the kit you want and run it at XMP or EXPO settings.
i ran a occt at cl 30 and there was a lot of errors. i changed it to cl32 and it passed the stress test. I used memory try it on a msi motherboard too scared to change everything manually lol @@blackbirdpctech
@@TheMichaelparisawesome … if you want to be sure try running karhu
NO MOBO RAM is al about the manufacturer.
I ahve tried other brands like kingston and so on , But corsair has never let me down.
The rest are absolute shit.
I been building and gaming since the 80`s.
I've had good experience with G.Skill and T-Force, but I agree that Corsair is also good. All of these RAM modules get their die's from the same manufacturers, all they really do is put unique head spreaders on and bin/test them. So for DDR4 Samsung B-die was the best and for DDR5 Hynix A-Die is the best. The CPU memory controller will have the largest impact on what RAM speed you can get stable, followed by the motherboard layout.
Would be great to see the same with AMD 7900x cpu!
I was planning to do it on a 7800X3D … will see at what speed I can get the memory stable.
just a comment, i agree that some ddr4 particularly good b-die such as 3600c14 is competitive vs ddr5. That said for those with intel 13 and now 14, it is possible to go an extra step , and run ddr4 4000+ at c15/16 in gear1 mode ( 14700s and 14900 can go at least to 4200 and depending on imc lottery, between 4266 to 4400c16 is achievable), and that tends to perform even better than 3600c14 , and by extension is even more competitive with ddr5. the drawback is that it takes some time to get the vccsa and vddq voltages right and ddr4 4400mhz quits are not cheap
Great comment and insight. I only have one ddr4 board that will support 14th gen with a bios update … it’s an Asrock z690 extreme … will be interesting to run some tests. Will see if I can find some higher speed ddr4 ram (fastest I have is 4000).
@@blackbirdpctech you should be able to oc a 4000 kit to 4400, if your 14 gen Imc is able to handle 4400g1 is another story (most are not) but at least 4200 should be achievable.
obviously it is easier and much faster if you can have something like g.kills f4-400016d-32gvka (16-16-16-36) or f4-440017d-32gvk (17-18-18-38) as the default xmp profiles of both are already relatively tight (by xmp standards) and any time investment to make a video will be shorter and mostly tied with finding stable vccsa and vddq values.
My experience (need to point I do not have a 4090) is that 4300c16g1 tends to outperform dd5 up to 6000 or so in most cases, will trade blows with ddr5 7200 at resolutions above 1440 but will generally lose to ddr8000 .
The saving grace is that it is considerable less expensive to get a run of mill ddr4 z690 mb and a good ddr4 kit ( total cost around 400) then it is to get one of the few z790 boards capable of really running 8000, that are all on the 500+ dollars price category plus the 8000 kit, that is probably 200+/
And even if one is willing to fork the dollars, there is no assurance that speeds above 7600 will work.
@@Solrac-SiulI just picked up a g.skill 4600 kit … I have a few 14th gen chips to play around with … hopefully I get lucky and get a golden sample … would make a cool video to compare ultra high speed ddr4 with ddr5-8000+ (whatever I can get stable). Thanks for your input, really appreciate it.
@@blackbirdpctech that will work, you will probably need to actually downclock , the good part is you will not need to worry much about ram voltage. I will be checking out to see that eventual video.
I know the future is ddr5. it is just that right now.... not only we have to contend with the IMC on the chip but also motherboards, as manufacturers are coming out with claims of 8000+ support, when many times they cannot even run 8000 at default xmp stable...
@@Solrac-Siulyes, they all have started doing it with the refreshed Z790 boards … adding memory to their QVL that has no chance of running stable at XMP settings. It’s marketing BS as I mentioned in my video. They really need to be held accountable for blatantly lying.
What ram would you recommend for a 7800x3d mate don’t mind spending a bit extra
Great question. AMD 7000 series chips behave very differently with DDR5 memory than do Intel chips. The X3D chips are also much less reliant on RAM due to their large L3 cache. When AM5 and 7000 series chips were first released it was tough to get anything past 6000 MT/s stable, with 6400 the typical upper limit. With the relatively recent AGESA update that has now changed. From what I've seen online however there does not appear to be any benefit from matching fast RAM with 7000 series chips ... I plan to do this testing for myself because I'm really curious to see the results. All of that said my recommendation is to get a DDR5-6000 EXPO kit with tight timings at CL30 from a brand like G.Skill, Corsair, Kingston, etc.
Yeah alrighty thanks mate surprised you don’t have more subs but with the quality of your content won’t be surprised to see you past 100k in a few months
@@kieranlyall6140thank you … I just launched my channel 2 weeks ago so it’s growing rapidly, but it will take some time to hit those big numbers … really appreciate your support, means a lot !!
Yes we need more gpu bound test, like who the hell buy rtx 4090 just to play at 1080p lol
Not sure I really understand what you are saying. If you are testing a CPU or memory then testing at 4K doesn't make any sense because it will be GPU limited, so that's why we test at 1080p even with a 4090. For my upcoming GPU testing I just tested at 4K for similar reasons.
@@blackbirdpctech yes thats what i mean, your video include test at 4k and there's no improvement in gpu limited sceneario. While other youtuber mostly just listing at 1080p and the audience hoping buying higher end cpu and ram get more improvement, but that's actually not the case when you really push your gpu
🥰