There are a bunch of how-to guides for overclocking memory. Edit: it can be difficult and time consuming, make sure you turn off fast boot so your PC won't skip memory checks in post (you'll likely have to reset cmos if your ram oc isn't stable) Once you think youve got a good oc, run memtest86+ v6 for a few hours to be sure it's not spitting out errors
I'm going to subscribe because I want to see what you can do with the 7800X3D. I like the analysis and it made me wonder whether the game developers are actually using optimized data structures ( maybe for compactness). If improvements to these data structures could be made it would allow for better performance with high cache CPUs.
@@smartmonkey777 The 7800X3D is a single chiplet and will not suffer from the same scheduling issues the 7900X3D and 7950X3D does. It has been shown that windows still does not know how to properly park the non V-Cache die and will share light loads with causing performance dips. Hardware Unboxed did a demo of the 7950X3D with the non V-cache die fully disabled in bios and it had a massive improvement over the other two.
Super interesting results! I'm really glad you took the time to investigate further and figure out why the 1% lows were the way they are. You explained the ram latency bottleneck really well. Awesome video!
You’re goated for this video. I configured the ram you have in this video in my system and I’m seeing drastic improvements. 5900x all Cores -30, I managed to get the ram down to Cl 14-14-14-28 with a 57ns latency!
Nice one. My 5800x3d running on 4000mhz had problems as well in test great higher scores but in games I was lossing 16-18% on my lows FPS vs 3800mhz crazy
Yes! That's a test I have searched and waited for for a long time. The 5800X3D at different RAM speeds and timings. I just got a new 5800X3D with G.Skill 32GB 3600 C14 14 14 36 RAM (got lucky for 150€). I know it's not useful in most cases and games, but already suspected it could help out in SC. And it looks it's doing much more then I even thought. I would suggest to test more common RAM speeds. Most people run at 3200 C14-16 or 3600 C16-18. No one would run at 2666 if they have just a bit of knowledge of PCs. So it would be nice to see the jump from 3200 to 3800.
Those are great timings for 3600.. I know what you mean about the timings.. A lot of people think they are at 16, but the next timing is 18 or 19.. I have my FlareX tuned to 3600 at 16,16,16,36 It's Samsung B die at 3200Mhz and 14,14,14,34 but I had trouble running 3600 with the 14's for timings so I settled for 16's.. I have the 5900X and game at 4K.. But anyway nice price on the memory..
I did 3 Lorville runs at 1080p ultra high/high and got an average FPS of 88.5 and 1% of 33.1 on a 7900X+4080 system using Buildzoid memory timings at 6000MHz. BPO set to “eco mode”, no curve optimizer. Will try next with curve optimizer. I already have a set of stable volumes for all my cores.
Nicely done video with an abundance of information, this must have taken some work. Tuned 3600 b-die at cas 14 with all subtimings optimized and I've seen exactly the same results you are showing.
Great video man i love the benchmarks and how you set the up and had the bonus info at the end thanks it makes a lot of since it was pretty much what i was thinking as well but the fps is going to be crazy!
So I had a Ryzen 3600x and a radeon 5700xt I would get fps on orison around 17 fps or 25 in area 18..and then higher once I get into space. I installed the 5700x3d wich is the same chip with a lower clock speed... My FPS on orison jumped to 45-50 fps and area 18 was above 60. Now with a graphics card upgrade I saw another big boost. Star Citizen absolutely gets gains with V Cache processors.
It's amazing to see what a little tuning can do to performance. I'm impressed with the Ryzen 7 5800X3D and just rebuilt my rig with Lian Li Lan Cool 216/Samsung 980 Pro/Asus Tuf 550 Gaming plus wifi 2, And a Corsair RM850 for future breath ability and upgrades. Thanks Game Day for optimization Guide and comparisons. Peace!
Nice vidéo !! Yes latency of ram is very important for thé 1% low and fluently gameplay. Thats why is décided to pair m'y 5900x with Trident Z Néo 3600 CL 14
Great settings for the 5800x 3D. I run it on b550m MSI board, KomboStrike level3=pbo -30, plus a 100mV offset. 4450 all core cinebench R20. 1.20 Vcore, aircooled. Ram, exact same ram, Patriot Viper, but with 4x8 sticks, 3600, cl16. I used DRam calculator from 1usmus for the settings. stock 1.35v memory previos cpu on this mobo was 5600x with same memory settings/same ram but i don't know if i can push it to 1900 mhz, 1:1 ratio. better ipc and memory controller with the 5800x 3D ? 4 sticks are harder to run than 2. you gain a little bandwidth with 4 sticks, but keeping in mind that it is a 2 memory channel controller so not a real 4 ways memory controller. the stress is higher with 4 sticks. Some users are playing with 100mhz bclk, and cppc preferred cores to get past 4550 mhz boost.......but is it safe longterm ?
Thanks. Would be nice to see a 13900k with decent 7600+ DDR5 sticks and some finetuning. Then 7950x3d and 7800x3d. With a 7950x3d you can easily simulate the 7800x3d too. DDR5 Speed will be heavily limited in comparison to Intel tho.
Hmmm.. AMD is latency/memory sensitive. Huge gains for better RAM. Kind of unclear why the DDR 4 is 'tuned' at CL15 when CL14 should be the target to start. More importantly, why is the Zen 4 Ram timings at cl36 'tuned', when the start should be CL30 baseline and tuned better from that point. Buildzoid has put out way better timings that work for Zen 4 ram chips. Changes in Zen 4 Ram timing fps improvements are absolutely huge (Up to 25% in fps) and only become limited (some games still improve 10-15%) when starting at CL30 and tuning from there.
Hello! Just came to your channel because of TH-cam recommendations and this looks awesome. I still have some questions on how this works though. > Can I tune my own PC components like my Ryzen 5 5500 or is this limited to this CPU version? > I also saw one video from you where you were able to tune Ram (I got 32Gb ram with 3200mhz) so I guess I could also do the same with my GPU? > Would this affect all other games and PC performance or would it be tuned only for Star Citizen? > How do I do all of this? I sadly couldn't find any tutorials on your channel Thanks in advance!
Hey yes all 5000 series can tune their ram. Performance improves in all games some more than others. Star citizen is a great example of a game where it improves extremely well. How to do it all will come later I am just getting started
Ryzen Master - Curve Optimizer will test each core and recommend an undervolt value for a custom curve. Brought my 5800X down from 90* at full load to 83* (which is perfect because the CPU starts thermal throttling at 85*). CPU consistently runs at 4.8-4.9 GHz. I have Corsair 3200 memory (Micron B-Die) that I was able to push up to 3800 with a little looser timings and a bump in voltage. That's a good place to start...
@charlesballiet7074 no he used PBO2 tuner which is a program you run inside windows. When the CPU first launched the PBO and curve optimization in the bios for every board was locked down. PBO2 tuner was a work around for that limitation set by AMD. With the correct bio there is no need to run the program and the neg 30 offset is applied by the motherboard everytime the computer is booted.
@TehnoRey from the quick Google search I did, it looks like that what MSI is doing in their bios. I have a Gigabyte Ultra which was given a beta Bios with the ability to do at the beginning of the month. Every motherboard manfacture will do something different.
Kombo Strike in the msi bios is not the same as the pbo tuning tho it does something for sure but with kombo strike no matter what level, i get worse performance then with pbo2 tuner set to -30 all core
Great video, I'm looking forwards to your Ryzen 7000 content. Did you have any stability issues OC'ing four DIMMs to 3800? I originally ran my memory at 3800 MHz, 1:1 with the memory controller, but had to back off to 3600 because my system became unstable after 30 minutes or so in SC. Running some DDR4 memory stress tests, I noticed that I would get memory errors once my memory hit 55 C. Removing two sticks was also helpful, either because the DIMMs had better airflow or less load on the memory controller.
you should see how the new 7800x3d performs on this game when it comes out fully clocked, I think it would be insane. We're finally getting to a point were a GPU can get to 140-160 fps on max settings at 1080p on star citizen... it would probably take 2 gpu gens for it to be normal to run the game at max settings
Change on bios the memory sync PMIC to by per PMIC instead all PMIC it solve the issue, and sometimes you have to change the voltage VDD and VDDQ 1.35000 on mine I put 1.30000 and worked, take 5000 or put another 5000 on each time and test it works smooth.
@@IrocZIV why should you run 4 dimms at 8000HHz for a system that needs max 32 RAM that you can get from 2 dimms ? and get better stability of overall system because of the mobo!
I need 128GB (or at least 96GB) as I often run short when doing work with only 64GB. But I also don't want my gaming to suffer. I don't need it to run 8000MHz, but being able to get a full 6000MHz out of 4 sticks would be great.@@zontarr22-zon
i loved the video... but pls can you do another where you do the tunning in real time and pls let us know all the tools you used for why the mortals can do the same ty.
Appreciate the work on these tests. Question, would the 'tune' have the same gap with a 4070ti? Looking to upgrade my 1700x ryzen on the cheap. i have the best B die you can get 3200mhz only.
Maximizing ram performance is NOT about ram MHz. It’s about tightening timings……primary…..secondary….and tertiary. Everybody goes about this the wrong way.
Dual channel bandwith is terrible to me compare to my 8yo x99 quad channel setup. have about 90Gb read and 70Gb write over a 2800mhz kit, and similar latency with CL15 timmings. Looks so weird that nothing improve on this point over the years and quad channel not the default for any modern mb xD
Damn. I had no idea. I could have spent $60 more and got that viper 4000 ram for my 5800x3d but I just got ripjaws 3600 cl16 instead because I didn't think there was that much difference. Oh well. Damn.
@@gamedaytoday1 Please correct me if I'm wrong but since Ripjaws V is Hynix DJR you can't tighten the timings much, just increase frequency if you loosen timings, or at least that's the way I understand it. I paid $100 for this 2x16gb dual rank 3600 16-19-19-39 1.35v Hynix DJR kit. Do you think it would be worth the trouble of selling it and getting a viper 4x8 4000 (or 4400) b-die kit for $160? I have a 5800x3d and 3080. Thanks for any comments. I just subscribed, by the way. Good info.
@@pf100andahalf All memory can be tuned and overclocked, your Hynix DJR, should be Hynix D-die which can do can reach 4100Mhz+ but with loser timings, if I where you I'd try 3800Mhz 16-19-19-39 with 1.45 v first before buying a new $160 kit. Just save the $160 towards a future upgrade.
@@gamedaytoday1 3733 at 1.4v is stable with 16-19-19-39 and my latency is improved, so thanks. 3800 wouldn't boot at 1.45v even with looser timings and I think it's a frequency "black hole" with my motherboard. I've heard that some msi am4 motherboards don't like 3800 and mine must be one of them. It might have booted at 3866 but I would've had to loosen timings so I just left it as-is.
@@gamedaytoday1 I bought 2x16gb vesta predator 3600 b-die and tuned it and now I'm ready to go. The smoothness and performance is so much better now, so thanks again for this video.
I have a 5800x3d, ASUS TUF x570 PRO, Kingston Fury Renegade 32GB DDR4-3600 CL16, RTX3080. Trying to figure out the shortest path to better performance. I don't have the stomach for hours and hours spent OC tuning anymore, so do I just chase 1900 FCLK or maybe CL15 on RAM?
Recently swapped my 5800X our for the 5800X3D. Before tuning, the 5800X3D was about 75C. After tuning, it's around 70C or less. In game frame performance seems to be about the same, feels a tad more stable. Resolution is 5120x1440. GPU: 3090 FE Mem: Corsair Vengeance 32GB, 3600 MHz, CL 16 SSD: WD Black 2TB SN850 NVMe For performance in game, running around Orison, it typically doesn't go lower than 33fps. At Lorville worst it gets is down to about 22fps when flying to the hangars, usually around 40-50 when running around in the city but dips to 28 when I look outside a window. (Clouds are high but not max). This about as good as it gets or would more tuning help?
if i put in pbo neg -3 all core and then cpu max boost +200 then my cpu is sticking at 3500mhz ! if i dont put the cpu max boost on +200 then my pc boost to 4,450 and 2 core sometimes 4,550mhz How is that ?? wy is that? wy is cpu max boost not working on my x570s aorus master ?? is that because we on gigabyte can't boost the cpu ??? then its stuck at 3500mhz ?
@@gamedaytoday1 Got a nice curve on the cpu, boost all core´s 4975Mhz some times 4995Mhz. Ram 3800mhz cl14 with tight sub-timings. to bad, my 5900x doesen´t like IF2000.. 1900Mhz without a problem.
so i saw on another video that on the 7950x3d (i think) one of the ccd's can be disabled (the non 3d one), is this possible on the 5800x3d too and would this also yield performance gains as demonstrated on the 7950x3d?
I am guessing that 4,45GHz stable was the highest clock did you try 4,55 Ghz ? And have you been able to run ddr4 at 4000Mhz and how was it, worst then 3800 if so?
Yes I tried both 4.55Ghz and DDR 4000mhz which was not stable for this CPU. I saw someone get lucky and he could do 4100Mhz but he had a better motherboard which may have helped.
You can download it form here: drive.google.com/file/d/1OswZcZ72jhm_Neek9c7PV-aRhM1EuOrX/view Follow these instructions: github.com/PrimeO7/How-to-undervolt-AMD-RYZEN-5800X3D-Guide-with-PBO2-Tuner/blob/main/README.md#2-why-are-we-doing-it-what-are-the-benefits
@@gamedaytoday1 when you do and tune the memory I would like to see that :) ive got Gskill 6000hz CL30 64GB to go with 7950x3d in the hope the other 8 cores help with background tasks like Tobii 5, Discord and other windows crap
Help out, why my bclk is 99.8, i already tried everything, b450 tuf last bios, spreadspectrum disabled already...with 5600 i got 99.9, now with x3d it decreased...using pbo2tuner etc
@@gamedaytoday1 its 9mhz (bus speed/bclk 99.8x44.5= 4441.0mhz total I am an experience ln2 overclocker and i was looking for someone with knowleage to help out out of the ordinary, this bus speed changed when i changed cpu from the 5600
Great video, however this game in itself is very poorly optimized. Considering how long this game has been in beta/early access, and that the newest hardware still can't push 120+fps @ 1080p on average is proof enough of that. Also, the same kind of tuning doesn't scale as significantly in other titles, probably for the exact same reason.
I need to oc my memory from ddr4 3200 to something higher and I did a +30 CO all cores and got 4.549 to 4.899 and I turned on the OC option im bios without touching anything else. I would like to see about tuning the CPU like the max boost mhz like +200 and mess with the other fabric settings. But just clicking the OC on and -30 voltage all core have me lower temps and high clocks. Temps are 30-60C
This only matters at 1080p, any resolution higher and you gain nothing. I havent gamed on 1080p in many years. Time to get a better monitor if you have a system this powerful.
Way to completely miss the point of a CPU benchmark, brother. SC is GPU bound in most locations, you're right, but there are plenty of spaces where the CPU performance becomes the bigger issue. Which, you know, is exactly why he's testing it in heavy CPU loaded situations at low res to max out the CPU instead of the GPU.
@@jackinthebox301 well unless you have an old 8 core cpu you will never be cpu bound in star citizen. Once I bumped up to 12+ cores my usage stays around 50% or less on all cores. With occasional spikes above. I saw higher usage while using Nvidia cards but on AMD systems, my cpu is not stressed at all. So this video is only good for people with old computers? Either way, if someone is planning to buy a high end system like the Ryzen 5000 series, you are better off going with the 5900x, which I bought. More cores for productivity and streaming capabilities AND it is much cheaper than the 5800x3d. Unless of course, you plan to use your new beastly PC with an old 1080p monitor....
@@Toutvids Well, there isn't a game on the planet that can use even 8 cores fully, so bumping up to 12 isn't going to change anything in games. I can't find it now, but I once watched a Star Citizen CPU benchmark video that showed the 5800x3d was the best even at high res. Obviously not nearly as dramatically as 1080p, but there is still a notable percentage to be gained by throwing more cache at the problem. That was before the 13900ks, though, so the point is moot until we get the Zen 4 x3d CPU's benchmarked.
Dont know what software you used. dont know how you ocerclocked or changed the values. You flew through it explaining next to nothing. Pretty much just wasted time to anyone trying to learn something.
if RSI just updatet the game so it runs on non tunet pc as normal people own then it will be grast lucky for me my computer is 5 days old an i have the best of the best serusly is sad they dont fix the game or alfa alittle better for other users
3800 is too high for the 5800x3d. 3600 is a better memory speed for this CPU. I can see that AMD gave you a high binned 5800x3d CPU. most retail CPU’s will not hit 4.55 GHz. The overwhelming majority will fall within the 4.45 GHz range after tuning. Herein lies the main problem with the majority of TH-cam reviews. The free components they receive from manufacturers are all hand picked skews……….meaning they are just better overall than a retail component available to you and I. We can get lucky and get one of these retail, but in the last 5 years of gaming PC building, I have received 2 (two)…… and I’ve built over 250 rigs over that time. Manufacturers benefit from sending hand picked skews to reviewers. Why wouldn’t they send their best silicon to a reviewer. So………bottom line………take these reviews and overclocking videos with a “grain of salt”. Especially from reviewers that receive components from the manufacturer for review. The most honest reviews are from TH-camrs like myself who buy all their components retail …………. and out of pocket. Those would be more real world and align more closely to your PC hardware components. As you can clearly see, these reviewers also use top tier hardware ………… high end motherboards ………. high end graphics card ………… quality ram ……….. fast Nvme storage …………. high end cooling ………. high airflow cases…………. etc, etc. Most users will NOT have this higher end hardware in their PC’s. So don’t expect your results to match those of reviewers. Every component within a PC plays a role in the results achieved. Consider any review as results that can be achieved with the right hardware………..not as results you can achieve with your particular hardware. BTW………those AMD numbers showing the 7800x3d uplift over the 5800x3d are just not possible when comparing specs between the 2. At best you might see a 10% uplift max.
why do all you ppl use gpus that 99% of gamers don't own???? i understand that your not trying to bottleneck but come on now a 4090 overclocked loll such useful info
you can't tune 5800x3d for fps games. its just shit anyone reading this. trust me don't buy it . 1% lows or even 0.1% this is % of a total. it does not include 0.01% and on 5800 it drops all time. every fucking second it drops to 1 - 5 fps. i would choose 8700k over 5800x3d. and fps games will run 10000% smoother. i feel being scammed and tortured and laughed at. i've waisted 3 years of my life trying to play pubg well. don't do my mistakes :((( PC happy 13600 owner with 0.1% at 160fps instead of 1fps with amd.
@@gamedaytoday1 Used Ryzen ram calc and time too proof on 3700x x470. Just transfered all primary and secondary timings to new chip/board. I could not belive it myself.
Absolute folly purchasing CPU's and GPU's with Star Ciizen in mind. Ther must be so many people who bought !080ti's in the hope of haviung the very best GPU for Star Citiziens release. My advice,wait until the games actually released int 2050 before buying the latest GPU or CPU..
![เปิด อาณาจักร หมื่นล้าน "สาระตั้ม" l [Nickynachat]](http://i.ytimg.com/vi/3XE0nPF4YkQ/mqdefault.jpg)
![[LIVE] BABYMONSTER(베이비몬스터) - DRIP | 두시탈출 컬투쇼](http://i.ytimg.com/vi/UG05pxLGwzc/mqdefault.jpg)
Would love to see a how-to memory tuned 5800x3d guide. Then we won't complain about optimization 😉
Yess plz
tuning ram to the maximum causes physical pain.
@@mryellow6918absolutly fact
There are a bunch of how-to guides for overclocking memory.
Edit: it can be difficult and time consuming, make sure you turn off fast boot so your PC won't skip memory checks in post (you'll likely have to reset cmos if your ram oc isn't stable)
Once you think youve got a good oc, run memtest86+ v6 for a few hours to be sure it's not spitting out errors
load docp then lower trc
I'm going to subscribe because I want to see what you can do with the 7800X3D. I like the analysis and it made me wonder whether the game developers are actually using optimized data structures ( maybe for compactness). If improvements to these data structures could be made it would allow for better performance with high cache CPUs.
From what i saw from GN is the 7950x3d actually performs worse in most scenarios because of the cache only interfacing with half the cores.
Star Citizen uses the Entity Component System (ECS) to optimize data structures, but I assume they still have debugging enabled for it.
@@smartmonkey777 The 7800X3D is a single chiplet and will not suffer from the same scheduling issues the 7900X3D and 7950X3D does. It has been shown that windows still does not know how to properly park the non V-Cache die and will share light loads with causing performance dips. Hardware Unboxed did a demo of the 7950X3D with the non V-cache die fully disabled in bios and it had a massive improvement over the other two.
@@TdrSld Yes, i wrote my statement carefully, if this ended up being the case with the 7800x3d.
Your benching has an issue. You used a different Bios for the third and final test, look here 3:30. It may have made a difference.
Super interesting results! I'm really glad you took the time to investigate further and figure out why the 1% lows were the way they are. You explained the ram latency bottleneck really well. Awesome video!
You’re goated for this video. I configured the ram you have in this video in my system and I’m seeing drastic improvements. 5900x all Cores -30, I managed to get the ram down to Cl 14-14-14-28 with a 57ns latency!
Glad to hear it
Nice one. My 5800x3d running on 4000mhz had problems as well in test great higher scores but in games I was lossing 16-18% on my lows FPS vs 3800mhz crazy
3800 on memory and 1900 fclk for 5800X3D is really, really good.
Yes! That's a test I have searched and waited for for a long time. The 5800X3D at different RAM speeds and timings.
I just got a new 5800X3D with G.Skill 32GB 3600 C14 14 14 36 RAM (got lucky for 150€). I know it's not useful in most cases and games, but already suspected it could help out in SC. And it looks it's doing much more then I even thought.
I would suggest to test more common RAM speeds. Most people run at 3200 C14-16 or 3600 C16-18. No one would run at 2666 if they have just a bit of knowledge of PCs. So it would be nice to see the jump from 3200 to 3800.
Those are great timings for 3600.. I know what you mean about the timings.. A lot of people think they are at 16, but the next timing is 18 or 19.. I have my FlareX tuned to 3600 at 16,16,16,36 It's Samsung B die at 3200Mhz and 14,14,14,34 but I had trouble running 3600 with the 14's for timings so I settled for 16's.. I have the 5900X and game at 4K.. But anyway nice price on the memory..
I did 3 Lorville runs at 1080p ultra high/high and got an average FPS of 88.5 and 1% of 33.1 on a 7900X+4080 system using Buildzoid memory timings at 6000MHz. BPO set to “eco mode”, no curve optimizer. Will try next with curve optimizer. I already have a set of stable volumes for all my cores.
Good job man! Been looking for a vid like this.
Nicely done video with an abundance of information, this must have taken some work. Tuned 3600 b-die at cas 14 with all subtimings optimized and I've seen exactly the same results you are showing.
Indeed this took a lot of time thanks for the appreciation
Very good analysis. I'm impressed by the quality of you work. Keep up the good work!
Great video man i love the benchmarks and how you set the up and had the bonus info at the end thanks it makes a lot of since it was pretty much what i was thinking as well but the fps is going to be crazy!
So I had a Ryzen 3600x and a radeon 5700xt I would get fps on orison around 17 fps or 25 in area 18..and then higher once I get into space. I installed the 5700x3d wich is the same chip with a lower clock speed... My FPS on orison jumped to 45-50 fps and area 18 was above 60. Now with a graphics card upgrade I saw another big boost. Star Citizen absolutely gets gains with V Cache processors.
wow thank you for this informative breakdown
Excellent video, thanks !
It's amazing to see what a little tuning can do to performance. I'm impressed with the Ryzen 7 5800X3D and just rebuilt my rig with Lian Li Lan Cool 216/Samsung 980 Pro/Asus Tuf 550 Gaming plus wifi 2, And a Corsair RM850 for future breath ability and upgrades. Thanks Game Day for optimization Guide and comparisons. Peace!
Great information, thank you sir. See you in the Verse
Nice vidéo !! Yes latency of ram is very important for thé 1% low and fluently gameplay. Thats why is décided to pair m'y 5900x with Trident Z Néo 3600 CL 14
Try clocking your ram to 3800Mhz CL14 with 1:1 FCLK and Memclock should give another 15%+ FPS boost
You are the best tester and optimizer so far.
The only thing unoptimized is the brain.
Great settings for the 5800x 3D. I run it on b550m MSI board, KomboStrike level3=pbo -30, plus a 100mV offset.
4450 all core cinebench R20. 1.20 Vcore, aircooled.
Ram, exact same ram, Patriot Viper, but with 4x8 sticks, 3600, cl16.
I used DRam calculator from 1usmus for the settings. stock 1.35v memory
previos cpu on this mobo was 5600x with same memory settings/same ram but i don't know if i can push it to 1900 mhz, 1:1 ratio.
better ipc and memory controller with the 5800x 3D ?
4 sticks are harder to run than 2. you gain a little bandwidth with 4 sticks, but keeping in mind that it is a 2 memory channel controller so not a real 4 ways memory controller.
the stress is higher with 4 sticks.
Some users are playing with 100mhz bclk, and cppc preferred cores to get past 4550 mhz boost.......but is it safe longterm ?
Yes its safe
good informative video!
Great video.
would be cool to see 1440p. great effort put in and nice graphics/data
Check out my 4090 review where I run through 1440p and 4K
now i need to see this, but for a 5950x
Thanks. Would be nice to see a 13900k with decent 7600+ DDR5 sticks and some finetuning. Then 7950x3d and 7800x3d. With a 7950x3d you can easily simulate the 7800x3d too. DDR5 Speed will be heavily limited in comparison to Intel tho.
could you do a tune / explanation video for the AMD 7800x3D?
Thank you for the knowledge transfer!
Excellent video. Broke down the 1% lows issue really well and provided great analysis. Will definitely subscribe :)
Hmmm.. AMD is latency/memory sensitive. Huge gains for better RAM. Kind of unclear why the DDR 4 is 'tuned' at CL15 when CL14 should be the target to start. More importantly, why is the Zen 4 Ram timings at cl36 'tuned', when the start should be CL30 baseline and tuned better from that point. Buildzoid has put out way better timings that work for Zen 4 ram chips. Changes in Zen 4 Ram timing fps improvements are absolutely huge (Up to 25% in fps) and only become limited (some games still improve 10-15%) when starting at CL30 and tuning from there.
well with gdm on cl15 isnt possible at all memory controller overrites it...
Hello! Just came to your channel because of TH-cam recommendations and this looks awesome. I still have some questions on how this works though.
> Can I tune my own PC components like my Ryzen 5 5500 or is this limited to this CPU version?
> I also saw one video from you where you were able to tune Ram (I got 32Gb ram with 3200mhz) so I guess I could also do the same with my GPU?
> Would this affect all other games and PC performance or would it be tuned only for Star Citizen?
> How do I do all of this? I sadly couldn't find any tutorials on your channel
Thanks in advance!
Try typing ryzen 5000 tuning into the youtube search box.
Hey yes all 5000 series can tune their ram. Performance improves in all games some more than others. Star citizen is a great example of a game where it improves extremely well. How to do it all will come later I am just getting started
@@gamedaytoday1 looking forward for your tutorial!
Ryzen Master - Curve Optimizer will test each core and recommend an undervolt value for a custom curve. Brought my 5800X down from 90* at full load to 83* (which is perfect because the CPU starts thermal throttling at 85*). CPU consistently runs at 4.8-4.9 GHz. I have Corsair 3200 memory (Micron B-Die) that I was able to push up to 3800 with a little looser timings and a bump in voltage. That's a good place to start...
How to tune a Ryzen ? How to use PBO tuning ?
with Ryzen master, my 5950X don't appreciate the tuning. Probably a bad tune in my BIOS...
Just a heads up alot of Motherboard manfactures are unlocking PBO tuning for undervovlting the 5800x3d thru the bios, this will require a Bio update.
isent that what he did @ 2:37?
i know that MSI have kombo strike for 5800x3d, which lvl 3 = curve -30 all cores, but which manufacturer have the same? i know only msi
@charlesballiet7074 no he used PBO2 tuner which is a program you run inside windows. When the CPU first launched the PBO and curve optimization in the bios for every board was locked down. PBO2 tuner was a work around for that limitation set by AMD. With the correct bio there is no need to run the program and the neg 30 offset is applied by the motherboard everytime the computer is booted.
@TehnoRey from the quick Google search I did, it looks like that what MSI is doing in their bios. I have a Gigabyte Ultra which was given a beta Bios with the ability to do at the beginning of the month. Every motherboard manfacture will do something different.
Kombo Strike in the msi bios is not the same as the pbo tuning tho
it does something for sure but with kombo strike no matter what level, i get worse performance then with pbo2 tuner set to -30 all core
Great video, I'm looking forwards to your Ryzen 7000 content. Did you have any stability issues OC'ing four DIMMs to 3800? I originally ran my memory at 3800 MHz, 1:1 with the memory controller, but had to back off to 3600 because my system became unstable after 30 minutes or so in SC. Running some DDR4 memory stress tests, I noticed that I would get memory errors once my memory hit 55 C. Removing two sticks was also helpful, either because the DIMMs had better airflow or less load on the memory controller.
I did an hour memory stress test to confirm stability, no errors
Slap on a fan on top if you get errors from overheating
you should see how the new 7800x3d performs on this game when it comes out fully clocked, I think it would be insane.
We're finally getting to a point were a GPU can get to 140-160 fps on max settings at 1080p on star citizen... it would probably take 2 gpu gens for it to be normal to run the game at max settings
you know, that your RAM memory clockspeed will be significantly reduced if you use 4 ram slots rather then 2?
Any guidance on how to replicate your tunning?
The biggest issue I see with the 7000x3D is that AM5 doesn't do well with RAM above 6000Mhz, so squeezing more out could be difficult.
Change on bios the memory sync PMIC to by per PMIC instead all PMIC it solve the issue, and sometimes you have to change the voltage VDD and VDDQ 1.35000 on mine I put 1.30000 and worked, take 5000 or put another 5000 on each time and test it works smooth.
Actually there is a BIOS update that can run DDR 5 with 8000MHz now as Buildzoid showed us.
@@zontarr22-zon I saw that. Does that also help run 4 dimms at a higher speed?
@@IrocZIV why should you run 4 dimms at 8000HHz for a system that needs max 32 RAM that you can get from 2 dimms ? and get better stability of overall system because of the mobo!
I need 128GB (or at least 96GB) as I often run short when doing work with only 64GB. But I also don't want my gaming to suffer. I don't need it to run 8000MHz, but being able to get a full 6000MHz out of 4 sticks would be great.@@zontarr22-zon
i loved the video... but pls can you do another where you do the tunning in real time and pls let us know all the tools you used for why the mortals can do the same ty.
Appreciate the work on these tests. Question, would the 'tune' have the same gap with a 4070ti? Looking to upgrade my 1700x ryzen on the cheap. i have the best B die you can get 3200mhz only.
Maximizing ram performance is NOT about ram MHz. It’s about tightening timings……primary…..secondary….and tertiary. Everybody goes about this the wrong way.
Can`t wait new X3D chips!! So hyped! 🚀
from what i saw from gamer nexus prepare for disappointment
Dual channel bandwith is terrible to me compare to my 8yo x99 quad channel setup. have about 90Gb read and 70Gb write over a 2800mhz kit, and similar latency with CL15 timmings.
Looks so weird that nothing improve on this point over the years and quad channel not the default for any modern mb xD
great video! got a RTX 2070 and AMD RYZEN 7 5800x3D. You think a better GPU will make a huge difference?
Thanks :)
Try Dual Rank DDR4 Tuned sticks (2x16gb) you see 😈
How the Frick you adjust MCLK and UCLK. Just leave that part out im sure it is well know info everywhere on internet
Damn. I had no idea. I could have spent $60 more and got that viper 4000 ram for my 5800x3d but I just got ripjaws 3600 cl16 instead because I didn't think there was that much difference. Oh well. Damn.
You can still tune your 3600 cl16 sticks and get a frame boost
@@gamedaytoday1 Please correct me if I'm wrong but since Ripjaws V is Hynix DJR you can't tighten the timings much, just increase frequency if you loosen timings, or at least that's the way I understand it. I paid $100 for this 2x16gb dual rank 3600 16-19-19-39 1.35v Hynix DJR kit. Do you think it would be worth the trouble of selling it and getting a viper 4x8 4000 (or 4400) b-die kit for $160? I have a 5800x3d and 3080. Thanks for any comments. I just subscribed, by the way. Good info.
@@pf100andahalf All memory can be tuned and overclocked, your Hynix DJR, should be Hynix D-die which can do can reach 4100Mhz+ but with loser timings, if I where you I'd try 3800Mhz 16-19-19-39 with 1.45 v first before buying a new $160 kit. Just save the $160 towards a future upgrade.
@@gamedaytoday1 3733 at 1.4v is stable with 16-19-19-39 and my latency is improved, so thanks.
3800 wouldn't boot at 1.45v even with looser timings and I think it's a frequency "black hole" with my motherboard.
I've heard that some msi am4 motherboards don't like 3800 and mine must be one of them.
It might have booted at 3866 but I would've had to loosen timings so I just left it as-is.
@@gamedaytoday1 I bought 2x16gb vesta predator 3600 b-die and tuned it and now I'm ready to go. The smoothness and performance is so much better now, so thanks again for this video.
Nice oc results 🤩 What ram did u use?
amzn.to/3EEAudN
@@gamedaytoday1 that’s a CPU…
Hey GD, so is your plan to pick up the 7800x3d?
Yes
Impressive
I would love to do that on my ryzen 5800x3D , I have 3200mhz ram cl16 but 64gb ram
Great Video thank you but was this 3.17.4 or was it 3.18?
3.17.4, will test 3.18 when it is on live servers
Command Rate 2? You're leaving ~ 5% performance on the table (depending on the game, even more).
you didnt go over on how to actually do this
What voltage are you running for the RAM?
I have a 5800x3d, ASUS TUF x570 PRO, Kingston Fury Renegade 32GB DDR4-3600 CL16, RTX3080. Trying to figure out the shortest path to better performance. I don't have the stomach for hours and hours spent OC tuning anymore, so do I just chase 1900 FCLK or maybe CL15 on RAM?
Yes chase higher ram overclock and tighter timings + oc the GPU and windows optimisation
2 sticks of RAM is always better & more stable than 4.
Depends on which IC but yes 2 sticks with the same IC are better than 4
How to tune 5800x3d?
I’m using this CPU with 4090 running 12288x2160. Using stock setting running average 110fps.
Have you tried turning off the SMT on the 5800X3D to see if there is any difference?
SMT on is better since patch 3.17
@@gamedaytoday1 LOL ZEN 4 3D is real good, are you going to test it tuned?
@@gamedaytoday1 I wonder how the 7950x3d will handle sc
Can you test Kombo Strike 3 please
Only getting about 45fps in Lorville with 5800x3D, 7900 XTX, 32gb3600mhzCL15. Any ideas?
wish mines was stable at 1900 cause my ram c an easily do it i got bdie ram was able to get it to 3900 cl14 but the the fclk didnt like it
Spec sheet says it's an Intel config.
Recently swapped my 5800X our for the 5800X3D. Before tuning, the 5800X3D was about 75C. After tuning, it's around 70C or less. In game frame performance seems to be about the same, feels a tad more stable.
Resolution is 5120x1440.
GPU: 3090 FE
Mem: Corsair Vengeance 32GB, 3600 MHz, CL 16
SSD: WD Black 2TB SN850 NVMe
For performance in game, running around Orison, it typically doesn't go lower than 33fps. At Lorville worst it gets is down to about 22fps when flying to the hangars, usually around 40-50 when running around in the city but dips to 28 when I look outside a window. (Clouds are high but not max).
This about as good as it gets or would more tuning help?
Tuning your memory and infinity fabric is the next step that can significantly boost your performance
My ram XMP is 3600 CL14-15-15-35.
What should my timings be for tuning at 3800?
if i put in pbo neg -3 all core and then cpu max boost +200 then my cpu is sticking at 3500mhz ! if i dont put the cpu max boost on +200 then my pc boost to 4,450 and 2 core sometimes 4,550mhz How is that ?? wy is that? wy is cpu max boost not working on my x570s aorus master ?? is that because we on gigabyte can't boost the cpu ??? then its stuck at 3500mhz ?
Would it be worth to downgrade from a 5900x to the 5800x3d for only gaming @ 3440x1440???
First tune your memory and overclock the 5900x as far as you can and see if you like the performance improvement before considering changing it.
@@gamedaytoday1 Got a nice curve on the cpu, boost all core´s 4975Mhz some times 4995Mhz. Ram 3800mhz cl14 with tight sub-timings. to bad, my 5900x doesen´t like IF2000.. 1900Mhz without a problem.
so i saw on another video that on the 7950x3d (i think) one of the ccd's can be disabled (the non 3d one), is this possible on the 5800x3d too and would this also yield performance gains as demonstrated on the 7950x3d?
The 5800X3D only has 1 CCD so it is not possible.
Dude show your bios settings
I am guessing that 4,45GHz stable was the highest clock did you try 4,55 Ghz ? And have you been able to run ddr4 at 4000Mhz and how was it, worst then 3800 if so?
Yes I tried both 4.55Ghz and DDR 4000mhz which was not stable for this CPU. I saw someone get lucky and he could do 4100Mhz but he had a better motherboard which may have helped.
Thanks great to one that :)
first i was, cool he got very high, then i checked it was 1080p...
1080p is cpu bottlenecked
Big mistake! U running the rams in 2T Command instead of 1T.
ok so you are saying that if we buy Crucial Ballistix MAX 16GB 4400MHz it should be as fast as 13900???
5800X3D can't reach to 4400Mhz with 1:1 ratio. If your are luck you can get 4000-4100Mhz which ad an additional 10-15% better performance
Do you have a link to the PBO tuner?
You can download it form here: drive.google.com/file/d/1OswZcZ72jhm_Neek9c7PV-aRhM1EuOrX/view
Follow these instructions:
github.com/PrimeO7/How-to-undervolt-AMD-RYZEN-5800X3D-Guide-with-PBO2-Tuner/blob/main/README.md#2-why-are-we-doing-it-what-are-the-benefits
That non tuned 5800X3D is what RAM frequency?
3600Mhz CL16
will you be doing the 7950x3d as well?
Yes
@@gamedaytoday1 when you do and tune the memory I would like to see that :) ive got Gskill 6000hz CL30 64GB to go with 7950x3d in the hope the other 8 cores help with background tasks like Tobii 5, Discord and other windows crap
Help out, why my bclk is 99.8, i already tried everything, b450 tuf last bios, spreadspectrum disabled already...with 5600 i got 99.9, now with x3d it decreased...using pbo2tuner etc
It’s ok if the BCLK is at 99.9 it does not matter.
@@gamedaytoday1 but why did it change, its 9mhz missing everywhere and an enthusiastic owner ;)
@@pav1u Based on what you said you are missing 0.1Mhz not 9Mhz.
@@gamedaytoday1 its 9mhz (bus speed/bclk 99.8x44.5= 4441.0mhz total
I am an experience ln2 overclocker and i was looking for someone with knowleage to help out out of the ordinary, this bus speed changed when i changed cpu from the 5600
Where is Intel ddr5?
Great video, however this game in itself is very poorly optimized. Considering how long this game has been in beta/early access, and that the newest hardware still can't push 120+fps @ 1080p on average is proof enough of that. Also, the same kind of tuning doesn't scale as significantly in other titles, probably for the exact same reason.
which build was this on though? 3.17 or 3.18?
3.17
@@gamedaytoday1 planning on doing one with 3.18?
what did u use to tune your ram ?
I used the BIOS
@@gamedaytoday1 thanks but i mean to get them timings u got whats the best way for getting useable timings for the ram
So do you really consider a CPU bound game optimized?
Interesting...
I need to oc my memory from ddr4 3200 to something higher and I did a +30 CO all cores and got 4.549 to 4.899 and I turned on the OC option im bios without touching anything else. I would like to see about tuning the CPU like the max boost mhz like +200 and mess with the other fabric settings. But just clicking the OC on and -30 voltage all core have me lower temps and high clocks. Temps are 30-60C
This only matters at 1080p, any resolution higher and you gain nothing. I havent gamed on 1080p in many years. Time to get a better monitor if you have a system this powerful.
Way to completely miss the point of a CPU benchmark, brother. SC is GPU bound in most locations, you're right, but there are plenty of spaces where the CPU performance becomes the bigger issue. Which, you know, is exactly why he's testing it in heavy CPU loaded situations at low res to max out the CPU instead of the GPU.
@@jackinthebox301 well unless you have an old 8 core cpu you will never be cpu bound in star citizen. Once I bumped up to 12+ cores my usage stays around 50% or less on all cores. With occasional spikes above. I saw higher usage while using Nvidia cards but on AMD systems, my cpu is not stressed at all.
So this video is only good for people with old computers?
Either way, if someone is planning to buy a high end system like the Ryzen 5000 series, you are better off going with the 5900x, which I bought. More cores for productivity and streaming capabilities AND it is much cheaper than the 5800x3d. Unless of course, you plan to use your new beastly PC with an old 1080p monitor....
@@Toutvids Well, there isn't a game on the planet that can use even 8 cores fully, so bumping up to 12 isn't going to change anything in games.
I can't find it now, but I once watched a Star Citizen CPU benchmark video that showed the 5800x3d was the best even at high res. Obviously not nearly as dramatically as 1080p, but there is still a notable percentage to be gained by throwing more cache at the problem. That was before the 13900ks, though, so the point is moot until we get the Zen 4 x3d CPU's benchmarked.
Dont know what software you used. dont know how you ocerclocked or changed the values. You flew through it explaining next to nothing. Pretty much just wasted time to anyone trying to learn something.
hmmmm and i mean that
run 4k and then test
if RSI just updatet the game so it runs on non tunet pc as normal people own then it will be grast lucky for me my computer is 5 days old an i have the best of the best serusly is sad they dont fix the game or alfa alittle better for other users
SC is almost completely unoptimized, it is in alpha...throwing the latest cpu/gpu at it is completely useless
is your whole system upside down? lmaooo
It is inverted
Star Citizen is not a game!it will never be in our lifetime!
3800 is too high for the 5800x3d. 3600 is a better memory speed for this CPU. I can see that AMD gave you a high binned 5800x3d CPU. most retail CPU’s will not hit 4.55 GHz. The overwhelming majority will fall within the 4.45 GHz range after tuning. Herein lies the main problem with the majority of TH-cam reviews. The free components they receive from manufacturers are all hand picked skews……….meaning they are just better overall than a retail component available to you and I. We can get lucky and get one of these retail, but in the last 5 years of gaming PC building, I have received 2 (two)…… and I’ve built over 250 rigs over that time. Manufacturers benefit from sending hand picked skews to reviewers. Why wouldn’t they send their best silicon to a reviewer. So………bottom line………take these reviews and overclocking videos with a “grain of salt”. Especially from reviewers that receive components from the manufacturer for review. The most honest reviews are from TH-camrs like myself who buy all their components retail …………. and out of pocket. Those would be more real world and align more closely to your PC hardware components. As you can clearly see, these reviewers also use top tier hardware ………… high end motherboards ………. high end graphics card ………… quality ram ……….. fast Nvme storage …………. high end cooling ………. high airflow cases…………. etc, etc. Most users will NOT have this higher end hardware in their PC’s. So don’t expect your results to match those of reviewers. Every component within a PC plays a role in the results achieved. Consider any review as results that can be achieved with the right hardware………..not as results you can achieve with your particular hardware. BTW………those AMD numbers showing the 7800x3d uplift over the 5800x3d are just not possible when comparing specs between the 2. At best you might see a 10% uplift max.
That 4090 is too much GPU for a 5800x3d.
why do all you ppl use gpus that 99% of gamers don't own???? i understand that your not trying to bottleneck but come on now a 4090 overclocked loll such useful info
Because if you're gpu bound you wouldn't gleam anything from this. Watch hardware unboxed's video on it.
you can't tune 5800x3d for fps games. its just shit
anyone reading this. trust me don't buy it .
1% lows or even 0.1% this is % of a total. it does not include 0.01% and on 5800 it drops all time. every fucking second it drops to 1 - 5 fps.
i would choose 8700k over 5800x3d. and fps games will run 10000% smoother.
i feel being scammed and tortured and laughed at. i've waisted 3 years of my life trying to play pubg well.
don't do my mistakes :(((
PC happy 13600 owner with 0.1% at 160fps instead of 1fps with amd.
Had a 3700x on a x470 chipset and w b-die got 64ns. Now on a 5800x3d and x570 same ram getting 34ns. 3200/cl14@ 3733/cl16
34ns amazing, first time I have heard some get it this low, any tips?
@@gamedaytoday1 Used Ryzen ram calc and time too proof on 3700x x470. Just transfered all primary and secondary timings to new chip/board. I could not belive it myself.
@@gamedaytoday1 Check you bus email. Sent some ss
All this video showed me was not to buy AMD what a bag of shit my 10th Gen is getting between 40 to 70 FPs @ 4k with my 2080Ti 🤣🤣 "COOL STORY THO"
Absolute folly purchasing CPU's and GPU's with Star Ciizen in mind.
Ther must be so many people who bought !080ti's in the hope of haviung the very best GPU for Star Citiziens release.
My advice,wait until the games actually released int 2050 before buying the latest GPU or CPU..