definitely do the comparison with ddr4 and ddr5. most people are still on ddr4

1% and 0.1% lows are absent in ram tests. Thanks Jay :)

Maybe for the next video you could keep one set of fast ram with good timings and just change the Frequency in the BIOS. That way we would know if the differences were down to frequency or timings. You could then compare 3 sticks at the same speed (e.g. 6000MHz) with different timings (CL30, CL 34 and CL 40) and see what difference THAT makes. Thanks for the video

Would have loved to see this for AMD systems as well, X3D and non-X3D to see how much the cache can negate the impact of ram speeds (if any).

i can feel Buildzoid's heavy breathing on this video

I have never thought of RAM as a part of my system that significantly increases FPS. I've always thought of it in terms of gameplay smoothness i.e. 1% lows and 0.1% lows. THAT'S where higher clocked RAM shows it's value.

really? no 1% lows? im baffled you left out THE most imporatnt reason to go with faster memory.

I pretty much always buy higher clocked ram because I typically lower the clock speed and overclock the timings with the extra headroom I get instead, I had very good results with this

Yes Jay... do a DDR4 speed test too and tell me what mouse are you using in the video? I like those headlights...

Upgrade from 4800Mhz to 6400Mhz was a massive upgrade for my performance in actual fps in games.

the 4800 was stable JDEC. Gotta stress test those XMP sticks cause they're probably erroring on stock voltages. Also, the faster RAM is most beneficial for 1% lows.

As you said, this is about gaming performance. I recently switched from running DDR5-5600 to DDR5-8000 and for productivity/general tasks, it does make a noticeable and measurable difference. This is especially down to latency. Also interesting is that the gain I got from frequency (running XMP profile) was doubled when I tuned the timings. So RAM tuning (watch Buildzoid’s videos for this) is actually worth-wile.

Comparing a GPU Ray Tracing benchmark to test Ram Speed differences... LOL

The more and more I learn about PC hardware and gaming, the less and less I trust Jay...

Yes please do the DDR4 video 🙏 Thanx for the video, nice information.

Yes, do a DDR4 vss DDR5 test, possibly fast vs slow. This same test run on an RTX 3070 also could show something (8GB GDDR) A test on Star Citizen could also be very interesting as that game is a HOG for ram and CPUs

There's just no way that 7200 was stable. Potentially not even the 6400 - although maybe that one was just running with terrible auto timings or switching to gear 4 without you noticing. This isn't consistent AT ALL with any previously done DDR4 OR DDR5 testing by other people and you should've really been aware of that before posting false information. At least talk to some other creators if your findings are so far off everyone else's

Jay, you shouldn’t be using AI OC in this scenario. You absolutely cannot guarantee that the OC would be the same as you swap from one ram kit to another. That’s a variable in your benchmarks that needs to be avoided by principle, big or small.

Hey jay, did you get good news or something? You're glowing. Love seeing that.

With this big as screen I would not have been surprised if jay suddenly started talking about the weather.

Man, this was already tested by Hardware Unboxed like a year ago. And their result was that Intel processors are actually more sensitive to the RAM frequency but not much sensitive to the timings BUT Ryzen 7000 processors are not that sensitive to the frequency, as they cannot benefit much from DDR5 faster than 6000MHz but they benefit from low timings and not only from the low Primary timings but especially from the Secondary timings when tuned as low as possible that makes a huge difference in performance!!

so literally no latency tests at all

Hi Jay, love your channel. I'd actually love to see ddr4 test. Mostly game and recently switched from 64GB of 4000 to 128GB of 3600 curious if theres any noticeable changes there

Please do a ddr4 comparison as you mentioned. Thank you for the video!

I am convinced Jay doesn't play video games at home or he has his monitor set to 60hz and didn't realize you can change it in control panel

If you ever had any soft of incling that Jay has no idea what he's just done to himself, just watch this video.

Great video and greetings from Canada. As someone who hasn’t built a computer in long time I have a question for you so I’m not wasting money trying to upgrade an old computer. I only do flight simulation on my old computer (ASUS Prime B350, GTX 1080, Ryzen 7 1300, 32 GB DDR 4 3000 C16, WD Black SN850X 2TB, Corsair 750w power supply, triple 20 inch 1024 monitors). This computer runs FSX maxed out no problem, but I would like to play MSFS 2020 and this computer barely meets the minimum requirements, it will run, but I would like to run at much higher settings. I was thinking Ryzen 7 5700 and RTX 4070 ti or 4070 super, but is this overkill for the current motherboard. Should I be thinking new motherboard, CPU, GPU and better DDR4? Any advice would be greatly appreciated. I was also thinking of switching out the triple display setup for a 32 inch 1440p monitor I already own.

Yes if possible, give us a DDR4 version of this, also it would be highly appreciated if you can include some frametime graphs or any sort of stutter and 1% lows comparison between different kits. What people seem to ignore is that the sheer number of FPS is useless if the game is running like a stuttery mess.

I don't really understand what's going on here. If the idea is to show what's the average use case of people turning on XMP and ignore everything, then sure. Otherwise... weird results. Motherboard is not specified, we don't know if all the configurations have been tested for stability (XMP is not granted to work)

How much difference to the results with a lesser system, something like a ryzen 7600 and a 3060?

I got 15% increase in lows going from 6400 with tight timings to 7200 tight. And many people using stock 5600-6000mhz xmp profiles lol. It is 10-15% average fps and 30-40% in lows difference between 6000 stock and 7200 tight

It's good to see a base test to show the potential diminishing returns with higher clocked memory. I think I would have liked to see multiple runs, and maybe since there are 3 different speeds and also a couple of different memory brands, I think it would have possibly uncovered more potentially interesting results had the tests been done with 3 speeds for each of 2 or 3 different brands (ie 3 speeds of corsair vengeance, 3 speeds of gskill trident, etc). Would have been cool to compare the same speeds from different brands as well as the different speeds.

Jay, if you could explain what CAS latency and timings and such are and what that all means that would be greatly appreciated. Great vid! Do the DDR4 one now, yes please! :)

He says this...but I bet his PC has 7200mhz of ram speed lol

With DDR4, ryzen's infinity fabric was tied to ram speed, so getting ram fast enough for the IO subsystem to be happy was important. This was in large part because jdec for ddr4 was 2133 or 2400, and saw significant performance increases up to about 3000. It saw measurable uplift from 3000 to 3600, but nearly nothing beyond 3600. DDR5 jdec is already so fast you're well onto the tail end of that speed curve. You _might_ need a faster kit if you want to hit jdec speeds on a x4 or x8 ram configuration because you _need_ lots of ram, and want it fast (my ddr4 setup is nominally good to 3800 for 2 sticks, which lets me run 4 sticks at 3600), but that's about the only use case for it.

I needed the questioned answered so badly

Just a few points here: Would be good to include the CAS Latency in the charts and ,preferably to even that, the first word latency. I'm not surprised it's made very little difference in performance changing to faster RAM. I ran a few benchmarks last time I went from 3600CL18 to 3600CL16 (running at 3770CL15 or something similar). It makes zero difference in gaming. Actually few FPS lower in some titles. But the change it made to video and photo editing in Photoshop and Davinci Resolve was easily noticeable and confirmed with a huge difference in benchmark scores. So I made sure I went to 6000CL30 when I went to AM5. Latency over speed but save your money if you're just gaming.

Jay please re do tho video showing 1% lows and ,0.1% lows, this video is not accurate, what about the timings? Jay please do better next time maybe more actual gaming benchmarks show more evidence in your findings

Thanks Jay

This type of video should be based on 1%lows 0%lows

Dude used a GPU benchmark to prove his statement about ram... no a single real life scenario, showed 4 games from which all 4 are absolutely irrelevant..

hi there, first i appreciate your hard work and thanks for this result. (i have gone through several comments and find out none have pointed this thing or i might have missed it) though i think you have done the testing in a bit wrong way because rather playing only one game and not using any other things on background definitely won't bring the proper results. it would be much appreciated if 2x game have been running together to get the result or having some heavy software running on the background like for streaming or video editing. more games like world of warcraft, Path of exiles etc would give more accurate result with having some sort of software running on the background which mention earlier.

Please do ddr4 and ddr5 ram comparisons... most average consumer are still on ddr4.. Thanks a bunch! will wait for it!

Sorry Jay, but you are wrong, or incomplete at best. You are clearly making some assumptions here. You are aware that not everyone plays FPS games, right? It is anecdotal, but I remember a thread in the ONI (Oxygen not Included) reddit where it was clear that ONI processes its cycles (in-game days) much faster with faster RAM. This is because that is a super memory-heavy process. Early game, no one notices. By the time you are at cycle 1000 and the game needs to process the results of multiple planetoids, the faster RAM will make a difference. As this happens every few minutes, shaving 3-4 seconds off a 20 seconds process makes a huge difference. Having nearly 4000 hours in this game, faster RAM has surely saved me a few hours. :) There are other things factoring in, but here fast RAM makes a difference. Good that you checked this, but you need to consider the use case.

this is such bad information. Ram matters in many games. At 1440p with a 4090/14900k you'll see 20%+ 0.1% lows improvements in many games. No way is jdec faster in any test, unless you're unstable on your faster kits. Did you validate your 7200mhz using kahru/y-cruncher? If the ram is unstable then the conclusion for your video is completely false. It should be something closer to "avoid fast ram if you want plug and play".

I got a great deal on my x590 board bc it was returned. It had the DDR4 manual and I bet they figured it was wrong, but it wasn't it was indeed the DDR5 version. I have 2x16 of 4800. I went from a 10400f to a 12400f but stayed with the 3060 12GB and I get pretty noticable gains in many games, and am glad I made the jump.

I'd love to see a comparison focused on the same speed (MHz) but different timings (CAS, RAS, tRAS, CL, etc).

Please delete this video, it's so full of misinformation. The difference between a stable 7200 MHz and 4800 MHz is like 25% performance.

just like that all the credibility Jay ever had as a tech Tuber is now gone. No 1% lows. no 0.1% lows. (1%+0.1% is the REAL frame-rate of your game, Average FPS is literally meaningless, all gamers knows this which is why they ignore this metric). Oh the best part? he standard benchmarks in SINGLE PLAYER GAMES. He did no runs in any online Multi-player games or E-sports games which are the only games THAT ACTUALLY MATTER. He also uses basic XMP profiles instead of putting in the hours/days of work to due custom voltages and timings for a real overclock and then 24-hour stress test the machine for errors to show us the REAL performance gain from tuning/overclocking which is anywhere from 20-40% or even above 50% in performance boosts depending on the game and server/map your playing on in the 1%/01% lows where it actually matters for competitive E-sports and general premium high refresh rate gaming experience.

What a load of crap. Stopped at Borderlands 3 on 4080 to test modern ram speeds. This guy lost the plot. Yes there is difference between 4800 and high speed tuned memory, especially on am5, sometimes to extend of 10-15%. Just look at actual hardware reviews not some casual pc celebrity. Not to mention how sluggish pc feels at stock ddr 5 at 70-80ns vs 50-65 ns of latency on tuned system. to not see that, You gotta be some dirty casual not any enthusiast.

this is another video why im not subscribed anymore

1k dislikes hmmm I think people figuring out what is going on here. Lame Jay really lame

@JayzTwoCents you really need to test 1% lows and .01% lows in games in all resolutions from 1080p to 4k. i think you will be amazed at the difference in smoothness of gameplay..

Here's a bunch of barcharts that are identical, lmao you convinced me man thank you.

2:08 - "The capacity of RAM should have no impact on your gaming whatsoever, unless your graphics card's RAM is so low, in terms of GDDR RAM available to the graphics chip, that it's doing swap, all the time, with the system RAM." Pardon me. I'm not a computer expert at all. However, I HAVE been a computer gamer for over 35 years. That statement is TOTAL BS. As a current example, the game "Star Citizen" (SC) is currently in "open alpha" (open to players, while being developed). They recommend a minimum of 16 gigs of RAM to run the game, however we players know that 32 gigs is a lot better, as it makes the game run much more smoothly. HOWEVER, I specifically put in 64 gigs into my latest machine, because the difference is NOTICEABLE, even though the game never uses all the RAM on 32 gig systems. It can get VERY VERY close, sometimes, however. Now, SC is not optimized, at this point. However, optimization will only LESSEN the difference. It will never eliminate it. There will ALWAYS be a difference. 7:24 - A "my raid" of things??? WTaF? The word is "myriad" (MEER-ee-ad). Please don't try to use words you don't know. +facepalm+

Unsubscribed. Misinformation.

as always makes tons of sense!

Love these types of videos! I always budget for max performance per dollar and this analysis made it easy for me. Thanks!

disappointing video. I'll probably refrain from watching you in the future for any benchmarks because there are so many things wrong with this video that lose you your credibility

Why didn't you include the 1% & 0.1% lows? Huge L take Jay. that's not how you test the influence of RAM Speeds on Gaming Performance. you're literally missing the most impactful test results.

Jay/ I have a good question. Is it worth it to upgrade to DDR5 from DDR4? Don't know if you have done a video about that yet or not. But I think that would be a good topic for another video if you haven't. Or expand on the one you already did.

A DDR4 vs DDR5 comparison would give me a good idea of the value I would get out of upgrading from AM4 to AM5 - from a RAM perspective. I expect the test would be done on an Intel system that supports both otherwise the CPU upgrade would also skew the results. However, seeing a full AM4 vs AM5 comparison using the same GPU would be helpful to me as well. (RTX 4080S owner)

Worst UKpostsr up there with Linus... this is just a rich TV channel showing off hardware. Take this dislike. Frame chasers is a better channel

Did not compare load times, modding and mod loading, performance when multitasking, ram load and unloading speed. Amazing benchmark about ram, congrats!

it is very interesting the tests, but it is also worth noting that the motherboards, OS, applications and games are not really ready for these speeds, because with 8600 I have seen improvements depending on which game is running.

Yes, pls to a ddr4 vs ddr5 comparison! Thank you!

thanks for your awareness

This is generally true, except for 1 very specific use case. iGPU's making use of the system memory. That's the only one I could think of where faster memory could make sense, at least up to a point before it just get's way too expensive still.

I remember needing to read reviews for MOTHERBOARDS in order to get the best overclockable motherboard available. in the mid 2000s it was a big deal trying to get the most performance

Well... I feel great about my Corsair DDR5 5600 32GB kit I bought 2~yrs ago when I built my 12700K / Z690 pc. Video card is a 3070FE because video card prices were very messed up at the time and got the FE @ BB for MSRP which was much cheaper than other options at the time.

Thanks man.

loved the opening dad jokes. thank you jay

Good info. Question: What about timings? Is a slightly better timing at same clock a notable performance improvement?

Came here through the avoid on buying for pc video glad to see the ram video got made

Normally, I install RAM at the rated speed of the memory controller on the CPU. On the other-hand, installing faster RAM primarily benefits those that are either running onboard graphics (like those of the G series Ryzen CPUs or Intel’s UHD/Iris enabled CPUs) or for those committed to overclocking.

Hey, where's the mention of single versus dual rank memory? What about Hynix A versus Hynix M? And what about primary, secondary, and tertiary timings? Let's not forget about channel configuration and its impact on performance. Oh, and motherboards play a crucial role too. We can't ignore bus latency and first word latency either. And did I mention that faster RAM can fix texture chokes in 1 percent lows, not just average or maximum FPS? We need to be more clear and thorough in our writing to truly inform builders. Isn't that your motto? Also, I'm guessing your DDR5 OC experience is limited, and I can't blame you as it's quite the dive, but you certainly are not making valid comparisons or observations here. I have noted up to 30 percent increases in 1 percent lows over multiple brands, timings, and configurations. That is huge for dips.

I was using Corsair dominator 6200 and went to Corsair dominator titanium 8000 and I’ve seen a small gain in fps in gaming and my pc is noticeably faster at loading windows or whatever program I load.

Can you hear that? that is the screeching from the Frame Chaser fan boys haha Thanks Jay

Tighter secondary timings perfom far better with ddr5 as opposed to tighter primaries with ddr4.

Hi Jay. You are are my daily cup of coffee. I'm using a new mug that looks great on my gamer desk. If you haven't made a video concerning QVL, please make one for us PC builders . I have two INTEL Z790 MBs. One is an ASUS, and the other MSI. When I checked the QVL list for DDR5 memory, both MBs were approved to use the same m mory I purchased, same provider same spects, same size, and timing. both MBs have the latest BIOS and firmware updates. The memory sticks would work on one MB but not the other. I thought this was strange and decided to see what you would say. My bottom line would have to be how accurate are QVLs?

Yes please on the ddr4 tests!

Here we see a wild Jay reinventing the price per gigabyte unit.

It is impossible to tell me there's no difference. YOUR RAM IS UNSTABLE AT 7200mhz. There should at least be a 20 - 25% FPS difference on average and 1% lows. Stick to basic reviews rather.

I had to disable XMP profile to get most direct X12 ultimate application to work one of them being fortnite for my son I have a 13900 K in a 4090 with an Asus z790 Maximus hero motherboard with 32 gigabytes of ddr5 6800 megahertz turned XMP off for good

Would love to see a video on ryzen X3D vs Non X3D. Also timings vs speed. Also testing 2 sticks vs 4 sticks on am5

This is a super useful video, advice the community can use to keep dollars in their pocket.

clear .. effective .... nice video

My issue with gaming benchmarks is that you could consider them synthetic as well. What I mean by that is usually the machine under test is only running the game, and very little else. When I game, I have discord running, a web browser open, sometimes I play music from the PC and run other apps in the background. It would be interesting to have a workload that has multiple processes running during the benchmark. Then you could look at testing with 8, 16 and 32G and compare whether having more memory is better than having fast memory. So, is a 4800 32G PC equal to or slower than a 7200, 16G PC? I've never thought that RAM had a huge impact on games, as long as you have enough.

I gained 20fps in tarkov at ultra wide (2x2k) going from ddr5 6000 to 8000. CPU bound games it can impact. also cleaned up my micro stutters

Yes, I know I want to see DDR4 vs DDR5, or how different DDR4 specs can affect gaming

I would definitely appreciate a DDR4 video. Chances are, my next ground up PC build will be DDR6.

Would love to see a speed vs. timing comparison at DDR4. AM4 seems very memory sensitive. Lots of people are probably consider upgrading AM4 platforms (or older Intel) from 8 or even 16GB for a bit more life. I need to re-build a home server soon and will probably go AM4 with it again just based on performance to value.

Yes. I will definitely appreciate video with ddr4

I remember when DDR5 first came out and it was like $10 a gig, and I got a good deal on DDR4 at $90 for 32GB, now you can get DDR5 for the old DDR4 price, and DDR4 sometimes is half of that.

I would love to see the ddr4 run. that would be very intersting, especially for those of us who built a pc 3 years ago and might see an uplift from faster spd ram

Would be interesting to see if these results remain consistent between AMD and Intel, especially given that AM4 chips were generally much more effected by RAM speeds than DDR4 Intel chips were.

I love your tech explanations like this. Could you take a look at PCI lane usage? Specifically using a gen 5 m.2 stealing lanes from the main gpu slot. Thanks

I had to get 64GB of DDR4 for work reasons and I got a deal on some "slower" Hynix RAM than the Samsung B-die with the quick timing I had been using. If there was drop in gaming performance, I honestly couldn't detect it. I think RAM speed is one of the most overblown aspects of builds some folks spend way too much time obsessing over, especially those who aren't trying to get a record on Time Spy or whatever.

Would be great to see on an AMD system. For the editing stuff, Puget Systems shows the 14900k ahead, but they use 5600mhz cl40. Whereas Gamers Nexus shows the 7950x/3d ahead, but they use 6000mhz cl30

I just specifically specced my new mid-budget system to get decent DDR5 speed, based on YOUR recommendation.