the problem with this video is that your gpu is being maxed out, so there isn't going to be any noticeable difference in performance. ram speeds entirely affect how the cpu is going to perform
@@crescentmoon256 because its a benchmark???? obviously if you want to test the ram speeds you have to be cpu bottlenecked or this video would actually mean jack shit
@@fhaerhaerhaerheryueryu6470 not true. 6900 xt 16gb card. New gen cards all 12gb+ Only true because 6gb 1660 super. Now if youre playing csgo or valorant... yeah youre probable cpu bottlenecked
@@deion___8354 Most people over limit on their PSU. 500W was plenty for a shitty little system like that. 500 was enough for an rx6600 with a 5600x and overclocked. So.. take your condescending idiocy somewhere else, human worm, or your shitty little genetic tree will get erased from this rock.
The GPU is never the bottleneck in 1080p, cos 1080p resolution is too low to use the GPU. That's why RAM and CPU benchmarks are made in the lowest possible resolution, while GPU tests are made in 1440p and 4k. In 1440p and 4k the CPU and RAM don't matter anymore, cos all they do is calculating and giving zeros and ones. In 1440p and 4k you'll have the same FPS with a 5-10 year old CPU for example.
3600 is the best value currently in terms of price vs performance. I don’t know why but with both Intel and Ryzen setups I get better performance gaming and often in benchmarks with looser timings, sometimes looser than those specified on the sticks.
@@ifarotht5149 not really, ryzen do love fast RAM over intel in DDR4 gen. the 2666 only perform better in some test due to lower primary timings since the RAM that are used with XMP profile are RAM that have loosed Timings. Some games ignore frequency and gain performance a lot by lowering the timings. Some do love high frequency regardless of timings. While some loves high frequency either tight primary and secondary timings. ETC. ETC. The best way is just to learn manual OC to gain free performance and set the frequency to highest possible while tightening primary & secondary timings as low as possible. IF you have more time then you can even go as far doing a GDM OFF 1T/2T which will take more work than just frequency + tight timings OC. I did all of them there's only two setting that are set to auto in my RAM setting. its 1D to 3Days stable with OCCT-RAM-SSE Test, OCCT-RAM-AVX2 Test, OCCT-Large-Extreme- SSE/AVX2/AVX/AVX-512 Test, Prime95-Large-FFT Test, Memtest, Memtest64, TM5 @Antutu Extreme Profile Test. Now as long as i stay with 1080p Monitor performance gain with all type of scenario is noticeable.
It matters if u fight about every fps ... on warzone 3600 cl14 i clocked to 3800 and give me 10fps difference Wow instead of 270 i have around 280 285 and 1procent low better around 3to5fps hahahah
@@ultimate_worm well . in frequency yes . in latency you could go whatever your wallet dream , but is like almost double for every drop in latency . also it depend in your procesor . a 12th that use ddr4 could make good use of 3200mhz and the lowest latency possible , but a i5 10th for example can make the best use of 2666 and the best latency possible . is not a rule and should things like the irq diagram of the motherboard and chipset to see if could be a benefit from higher clock in frequency than the supported by just the cpu
I'm always worried that I should have invested in retiring my old 2017 DDR4 2400 8Gx2 kingston sticks. With DDR5 coming, I'm stretching my thin budget towards there, and your video confirms that I wouldn't got any performance boost on games. Thanks.
for me i will always get ddr4 with 3600 MHz of ram speed. ddr5 is still expensive until today and still many issue with memory bug, cpu bug, motherboard bug for amd socket AM5. not sure with intel but must cost more than amd build
@@ClientsMusic thanks for your input. I've been told is already, it's a latency issue. You also saying this confrms this. Maybe I'll do better if I wait a year.
So what is the real bottleneck? The GPU? I think it's important to max everything else out to see the full potential of the difference in RAM, but it's good to see a more "realistic" test bench, as I would imagine the large majority of gamers have low-mid tier graphics cards. Still, the test could have been done better. All this video has done is highlight that there is no difference among the types of RAM, which really raises the question for viewers - why even bother getting the expensive stuff? Something's missing here, and I suspect it's the test bench set up. Do this again with a maxed out video card for the bigger picture.
@@channelz1xwebr No always. But playing games on lower resolution requires more CPU frame time set up. So if you have a slow CPU, you’re going to have lower frames in that scenario.
I'm currently using 32 gb Cosair Vengance 2666 and from what I see in this video is it's actually better in gaming and just as good as the 4000. I don't think I need to swap it out for now. Great video and thanks.
brother I'm using 1x8GB ddr4 2666mhz Ram. And i ordering another ram with 1x16gb 3200mhz ram. Should i cancel the 3200mhz ram order and Buy 2666mhz ram to pair together. Because if i buy 3200mhz ram Then the other 2666mhz ram is wasted. So I'm thinking to buy one more 2666mhz ram to pair with my old one Ram. is it good ? can someone suggeste me ?
With a gtx1660 I'm not sure you're really stressing things. I have a 9900k and I'm upgrading to a 4070ti from a 1070. I currently have 2666 ddr4, and from what I've seen in some games the extra memory clocks give anywhere from 5% to 40% improvements. It's game dependent. For $70 ill double my ram from 2x8 2666 to 2x16 3600 and push my 9900k a bit further until i upgrade it eventually too.
@@TheGamersRace eSports games needs a strong single core to push higher frames to the maximum. So testing in low settings will max out not just the gpu but cpu as well to push the gpu as much as possible. To put into simple terms, both cpu and gpu matters equally if you want crazy high fps. In this testing, the gpu is holding back the cpu and ram.
For people saying the GPU is a bottleneck: CPU + RAM don't care in graphics. That's why CPU+RAM benchmarks are made in the lowest possible resolution (720p-1080p) - to simulate data dumpster fire. If you switch to good resolutions like 1440p and 4k there's no difference at all, cos CPU+RAM just calculating stuff and handle the data over. You could use a 10 year old CPU+RAM and wouldn't see a difference in 1440p and 4k. It's all in margin of error. A GPU is named graphics processing unit for a purpose... This is perfectly shown here: only GPU matters in games.
A slow cpu can't power a capable gpu. Cpu (central processing unit) controls everything in the computer, including the graphical processing unit. And a bad cpu can't allocate so many resources to control a strong gpu
Thanks, buddy. You made my decision a lot easier now. Didnt expected that "only" 2400 MHz do so well compared to "on paper much faster" 3000 MHz RAM. Seems like timing and CL have very big influence too.
Note: Cache latency is relative to clock speed (cycles of delay), not nanoseconds, and impacts 1% lows, not average fps. 3200mhz is the lowest you should get usually, as this is a deceptive benchmark, the CPU is barely even being used completely invalidating the test. 2400mhz is ancient (budget end of release day, 2014) and will cripple performance when cpu is being pushed.
This is good news When I built my computer years ago, there was no fast memory in stock, and I bought 2 32Gb dims at 2400Mhz Some people have made videos where it seems that the 2400Mhz memory causes a bottleneck, and in a video game a 3200Mhz memory causes growth of more than 10fps I was worried thinking that maybe I should make an improvement
5600x supports up to 3200-3600 Mhz ram speed. that's why there isn't a whole lot of difference between 3200 3600 and 4000. you shouldve used a better cpu and gpu. thanks for the benchmarks nonetheless
I'm using 1x8GB ddr4 2666mhz Ram. And i ordering another ram with 1x16gb 3200mhz ram. Should i cancel the 3200mhz ram order and Buy 2666mhz ram to pair together. Because if i buy 3200mhz ram Then the other 2666mhz ram is wasted. So I'm thinking to buy one more 2666mhz ram to pair with my old one Ram. is it good ? can you suggeste me ?
@@tarantulamadness6191 and recently we had an outage where i lost power to the bios and had a decent amount of fps loss without the ram OC, so 100% makes a diff. im using a ryzen cpu too probs thats why
i was afraid bcos my 24 GB ram speed is 2400 mhz, i was planning to buy ram with higher speeds but now i see the truth, thank you for save me money dude
I have 2)AM4 systems. The first is a X370 Motherboard that I always had problems with finding RAM. The highest DDR4 that I had that worked was a set of DDR4-2666-(4x8gb) The Second is a B550, and I have a set of DDR4 3600-(4x8gb) I have re built the PC made up around the X370 Motherboard, but I was not, put out by the limitations of the RAM choice, and I decided it was something I could live with, that is no discernible difference. (except me knowing it was 'only' 2666 vs the 3600 intended) This video verified that, and more. I thought the testing differences would be at least 10 FPS.
It can be seen that a person unfamiliar with the workings of the amd platform made the video, the most important frequencies for the AMD processor were simply not added to this comparison 3800 and 4400, so we can see the minimum differences
my pc is quite old, I use I9 9900k, Rtx 2070 and 32GB ram 3200mhz DDR4, on the lowest settings with Dlss on balanced I get between 60 to 90 fps, it depends on where I am on the map, I have few cases where it dropped below 20fps especially around buildings or inside of them, I think that might be related to shaders. but what makes the game unplayable for me is crashes, I can't complete a single game due to crashes, I understand that the game is still in early access, and it might take a long time to fix all of the performance issues crashes and the poor optimizations.
Your timings on your 3600 give the best latency that would be where youd see anything but very literally 97C on the GPU youre totally bottlenecked by your gpu and you wont see any real difference. Ram speed helps with lost frames if your gpu isnt capped, and helps with cpu intensive games. So you might see a jump on CS:2 or starcraft or arma3 but likely not anywhere else.
In performance matrices i didn't included temperature information, where did u seen that gpu is running at 97°c ? The gpu was running at around 70-75°c. And msi gamingX gpus doesn't drop clock or memory speed even at 85°c bro
Is it true that RAM will have huge impact in the following setup? CPU: Ryzen 5 4500 GPU: RX 6600 motherboard: Gigabyte A520M H 16 GB 2666Mhz I just want to play LoL at stable high FPS (180 or something). But now it decays overtime in game, from average 130 to 110 or something. Some1 mentioned I have a relative weak CPU/RAM combo. Can anyone tell if e.g. 8/16 GB of 3200Mhz RAM will solve this or is my CPU the issue as well?
realistically the ram is more like brain muscle memory, like helping load and render skins in fortnite for example between each game or switching shop each tab in the shops not so much about running around in a game, it stores previous data and loads the data that was already recorded into the ram and it speaks to the cpu back and forth opposed to just going to the hdd but now with m.2 ssd the speed of ram is becoming less required because the speed between nvme and it is so fast
I am having 2 rams using together 1. Corsair 8gb ddr4 ram (2400mhz) 5 year old 2. Corsiar vengewnce rgb 16gb ddr4 (3200mhz) When using both i get speed of 2100mhz So should i remove the 8gb ram and use only 16gb ram this way i can get 2666mhz which is max supported by my motherboard?
Yes both stick's default ram speed is getting set to low to match the speed for both ram, because one of them have lower speed (2400mhz). So u can try 2 things in bios, try to set the mhz value for both of the ram to max mhz supported by ur motherboard, or take out the 8gb stick and set mhz to max supported. And check the performance between both scenarios. Bdw what cpu and motherboard you are ? Model name ?
@@ultimate_worm actually i have a 2019 pc build it had core i5 8th gen and gigabyte b360 gaming hd motherboard which supports 8yh and 9th gen cpu I have only upgraded the gpu 7800 xt and msi 1000w psu
Hello, I bought a Ryzen 5 4500, an ASUS PRIME A520M-K motherboard, an RTX 3050 6GB MSI Gaming X, and there is a VERY CHEAP Corsair Vengeance Lpx 2666mhz 2x8GB offer, would it be a good buy or should I go for 2 3200/3600 modules?
U should go for at least 3200 but stil you can check for vegeance 2666 sticks Overclocking capability. I think vegeance stick can go upto 3000mhz easily and it is safe also and maybe 3200 also possible.
God I hate these kind of tests where people have a weak setup and try to show a difference in fps when their running too demanding games where the gpu is maxed out. IF YOUR GPU BOUND THE BENCHMARK IS POINTLESS
Restart/power on your laptop. After pressing power button immediately start tapping delete button on keyboard. In few seconds your laptop will enter into bios. In overclocking section of bios you will find out a option called xmp/docp profile, Click on that and you will found 1 or 2 profile with different speeds just select any of them. And then you can save & reboot your laptop
Best is to pick 4000Mhz it should be also samsung b-die its more cheaper now as of today since the newer gen focuses on DDR5 now. Then you can just manually adjust it to 3600/3800Mhz and copy the stock XMP primary timing, DRAM voltage of Samsung B-die kits in the market. And when you have time try to learning how to manually overclock your DDR4 RAM there are now tons of very accurate guides how to overclock a DDR4 RAM specially with samsung B-die since its really popular RAM die-type in DDR4 gen. Follow the guide and properly stress test your RAM with multiple RAM stability app if it pass all those (the longer the better basic rule of 24hrs testing per app to avoid stability issue). The benefit of getting a 4000Mhz++ Frequency w/ Samsung B-die chip is due to the XMP profile of at least 4000Mhz you have no problem hitting a 4000Mhz 1:1 ratio with just enabling XMP with IF OC using a ryzen 5000 series (only if your IMC can do a 4000Mhz) you can even push it higher if it does. If it doesn't then you can lower it from 4000-3966-3900 etc. until stable since its originally a 4000Mhz vs a 3600Mhz the 3600Mhz might only reach 3733/3800/3866Mhz even if your IF/IMC can reach 4000Mhz or above.. due to your RAM being lowered binned. There's a reason why they are 3600Mhz out of the box they are lower binned variant of the same model... Plus the higher binned 4000Mhz can also do tighter timing the a 3600Mhz, higher binned 4000Mhz is also easier to OC than 3600Mhz.
Nice video but unfortunately not really testing ram speed impact as it's significant in CPU bound scenarios ... Upgrade gpu and you will see the difference
Yes u r later on i realised that because of weak gpu the difference wasn't that noticeable. Will fix this in next video. Thankyou for watching and also for suggestion ♥️
No anything above processor's specified ram frequency is Overclocking, 5600x's specified memory clock is 3200mhz but if we use anything this then it will be Overclocking
Agreed in Intel it really doesn't matter.. even manually tuning it doesn't make much of difference. But with ryzen it really does affect CPU overall performance since IMC = IF is within the CPU not in the MOBO like intel (this is only with DDR4).
Running a Ryzen 9 5900x with G Skill Rip Jaw on a 10 layer PCB cl14 3600mhz 4x8gb overclocked to cl14 3800mhz with the tightest timings at 1.50 volts and a 120mm fan pointing right at the ram modules they idle at 23c and when gaming or running benchmarks never go above 27c saw a major FPS increase in Aida64 Extreme the latencey is 52.3ns the OS just seems snappy love Samsung B-die.
This test is technically useless. Those frames are so close on all systems because the 1660 is rendering at its max Performance here. This way it wont show us any improvement because CPU+RAM still have headroom even on the lowest RAM Speed. RDR2, WD2 or Hogwarts Legacy, Jurassic Park Evolution 2 are highly CPU intensive Titles. Short example here: I have a Ryzen 7 1700 at 3,9ghz paired with 32GB of DDR4 Ram. I had them running on 2666 (bcs that CPU wouldnt allow me to set it any higher). I got a 3060 recently and really see a CPU bottleneck (not much, but it is there). Now I gave RAM OC another shot. For some reason (made no BIOS updates recently bcs its up to date since last year march) my RAM now allows me to go to 3000mhz. What did it change? Hogwarts Legacy really tumbled down in Hogsmeade. I get stable 60fps anywhere with 1440p/Ultra(custom) settings. But in Hogsmeade it dropped to 45 in many instances. With the RAM on 3000 now it stays around 55 which is a lot better to bear and way closer (no real hiccup) to the 60fps usually
It also solved a problem I had in Anno 1800. When I had huge Cities the game had big stutters. With those increased Ram Speeds they are still there but way less noticeable
Yes you're right. For new gen games gtx 1660 is very low-mid range gpu. That's why difference isn't noticeable. Thankyou for the suggestion brother. Will try with better gpu in future video 🤝
This is useless, one isn't that much faster than the other because you're using the same kit. You need to have higher quality ram kits to show the delta between low-mid-high-tier rams. 4000 19-20-20-40 is laughable and is about the same as the 3600 when you account in the timing diff. For reference, 4000 15-15-15-30 is okay for a high tier DDR4 ram (b-die). Also, your processor is shite for ram bottleneck testing and so is your gpu. So essentially everything is wrong in this test. Live and learn.
Why on earht would someone do a comparison about ram speeds, when the system is hardbottlenecked by the gpu. Ram Speeds only start to matter once we approach +100-200 fps... sorry for all the hard work, but this was absolutely pointless...
Yes i know later on i realised that my gpu was not that powerful enough to produce high frame rates & gpu was the bottleneck. And yes we can see difference with high end GPUs. So some how this video point for those viewers who have low-mid range gpu old systems and thinking to buy high speed ram.
Man your GPU cant show differences between better and bad ram. And moreover, your timings, how you can see you have the best performance in 3600 C17. So you should stay in that timing. It will not answer the best mhz in games because of timings. You can make comparison between the real thing, as B-die, E-die, C-die and others. That will give you really big difference. You can buy 3200, and clock it for 4400 and more with b-die by changing timings and voltage. In e-die you can just clock and change timings, voltage till 1.375. and with hynix, good luck!
i have Sapphire Nitro+RX6800 16GB normal and Pocessor intel Core i9-7900X and Motherboard MSiPro X299 LGA2066 on Quad Channel Kingston Fury Renegade Black 2666MHz CL-13 but work's 3200MHz it XMP very nice for me but i can use 4200MHz my Motherboard its the max i can put on it
So the conclusion is GPU spesification more important that RAM with high Mhz right? For the addition, I like how the editing in this video & Thanks you to made this video!
Later on after making this video i realised that, If u have average gpu. Low-mid range gpu(in my case my gtx1660) u wont notice any huge difference in different speed ram. It only starts to matter when your gpu & cpu both are powerful. Specially gpu
anyone looking at this video in 2024 this is a bad bench, firstly i doubt this person changed their FCLK to match the ram speeds, and secondly then you're REALLY gpu bottled necked like in this video it will matter less, but if you're at lower gpu usage and maybe -low 90s in gpu usage it will help, i can vouch ram makes a huge difference
Brother I have Ryzen 5 3400g with 2666mhz ram(1x8) and I am going to upgrade it should I get a same stick or should I buy 2 separate (8x2) rams. Please reply
I'm using 1x8GB ddr4 2666mhz Ram. And i ordering another ram with 1x16gb 3200mhz ram. Should i cancel the 3200mhz ram order and Buy 2666mhz ram to pair together. Because if i buy 3200mhz ram Then the other 2666mhz ram is wasted. So I'm thinking to buy one more 2666mhz ram to pair with my old one Ram. is it good ? can someone suggeste me ?
@@skypka_telefoniv Брат, спасибо за твое предложение. Я в замешательстве, что мне делать. Потому что я уже использую оперативную память 2666 МГц. И с амазона заказать еще один с другой частотой (3200 МГц). Сейчас подумываю отменить заказ на оперативку (3200 МГц). и купите еще одну оперативную память 2666 МГц, чтобы использовать ее со старой как двухканальную.
pointless test, you should compare all frequencies at the same delay. For example, 3200Mhz CL16 = 3600 CL18. At the same delay, we can check the pure impact of frequency on performance.
People or reviews said ram speed really affectd AMD cpu speed a lot, so actually this is not true? I am using 2133 ram and actually waste of money and time to upgrade?
It also depends on game & gpu. But in this condition i am using a average gpu that most of people are lower end or average gaming pc users and with type of gpu they wont notice much until they are going really high end in every components
