Answered about 5 distinct different questions I get on the daily in the PC groups I teach people in. Also, did you watch the fights tonight!? Thanks again for making high quality fantastic content.
Another great analysis Matt! I'm on AM4 with an R7-5800X3D on an ASRock X570 Pro4 motherboard and 64GB of Team Vulcan DDR4-3600 so I have no intention whatsoever of upgrading in the near future. My PC is already stacked.
@@AvroBellow I know this is an old comment but I have a similar, slightly less beefy setup- AM4, R7 3700x with 32gb DDR4-3600. I found this video while searching around about possible upgrades, still not sure about it. Do you have a specific need for 64gb or just for extra headroom?
Wstched this video late but it would have been interesting to have the GOAT x3D gaming CPUs in. When I first built with AM4 memory tuning and over clocking was a hot topic, while now cheaper slower memory is often delivering the same performance.
I was planning to do an SSD comparison between Gen 3, Gen 4 and Gen 5 ... I could add a modern high speed HDD to the mix ... Gen 5 SSD's are great but most motherboards can only support them if you reduce the lanes going to the GPU, which really sucks.
@@blackbirdpctech this is new to me, i didnt know that there were different generations to SSDs. This is interesting, thanks ill look into it :) P.s I am rocking a 4TB HDD so I am deffo going to get an SSD when I build my new setup hopefully
Matt I saw that you have a open bench table too. I got one too and I even managed to travel with my whole setup with it. Took it through TSA. Loved the video mate. Cheers
Thank you! I already did a video comparing ddr5-6000 vs ddr5-8000 for the 7800x3d (th-cam.com/video/p21bsVYAgvA/w-d-xo.html) … unfortunately I can’t do a ddr4 vs dd5 test for AMD without changing the cpu's, that’s why I did the testing for this video with a 14th gen Intel cpu.
With AMD you can just test different Speeds... Only Intel provides support for two RAM Generations on the same CPU Socket, so Intel is the only CPU to test the different RAM Generations lol
@@FunFoxHDlol no shit this comment was 6 months ago but im pretty sure what I was referring to was how does fast ram affect the x3d chips. Not how does ddr4 affect them because that would be impossible….
@@blackbirdpctechyes the fast ddr5 vs slow ddr5 with the x3d chips is what I was referring to thanks u. This was 6 months ago but I have since learned that faster ram doesn’t seem to be as relevant for x3d chips.
@@danebeee1 oh... Yeah, I have an 7800X3D myself and can confirm it's nearly identical in games... Also for some reason my RAM is unstable over 5600MHz, gladly I have an X3D CPU
i can't confirm if it's good for gaming but i am on 3600 CL 14 and for work it's noticeable computing/vid editing/ photo editing that is my workflow and i begin to create some 3d nonsense sometimes despite the fact my kit looks amazingly
Thank you for saving me £300 on an unnecessary upgrade for my 12700k. Too many reviews only focus on low res game testing for the bigger, attention grabbing differences (HUB). I expected small gains for DDR5 at 4k but the negative results were surprising. I wonder if the UE5 engine games will be similarly indifferent at 4k
It's always interesting to look at the data versus listening to opinions ... I discovered the same thing when I tested Gen 4 M.2's against Gen 5 M.2's ... there is no difference in gaming, so don't waste your money.
Interesting detail I found with Ry 5 5600x was an energy penalty going above DDR4-3200 that created a significant hotspot as the IOD memory controller was not entering an energy saving state. This meant I lost any benefit of pushing DDR4 to 3400MT/s and reducing heat was better. The subsequent Zen3 stepping had metal layer fixes from after Q3/21 which helped the 5800x3D be efficient, was mostly ignored but helped the low end. With a small TDP 65W/88W PPT the watts saved help the per core power budget. I saw an interesting test result where disabling 2c/4t on a Ry 5 5600x actually delivered small fps gains albeit in older titles. Greater use of SMT would mitigate memory stalls and per core power and cache was boosted by 50%
Kinda like cars...the last 5% in performance doubles your budget. And yet, us 'enthusiasts' bite like rabid dogs, every-single-time. hehe Another good vid, brotha!
The crazy thing is I know what the logical choice is but I can’t help myself … it’s a disease … wait until you see my next video … I’m building my ultimate gaming pc … feels good to build again 😉
I would've liked to seen the temps while gaming as well. You didnt mention if you overclocked the cpu so and during the productivity areas where you showed the temp a few times, it was close to a 10° difference. Thats a lot of room on the table to oc the cpu and if so, would the numbers then be closer or better? Of course assuming you only oc the ddr4/cpu setup to match the temps of the ddr5/cpu setup.
It's always a trade-off when deciding on how much data to show ... if I add temps to the gaming charts I worry that it will impact their readability ... will see if I can figure out a good way of doing it in the future. With respect to CPU overclocking, I did explain in the video that I enforced all Intel CPU limits in bios for both motherboards. That said the temp difference was, as you point out, relatively large. I think most of that difference is due to the memory controller on the CPU needing to work a lot harder with DDR5. It would be interesting to see what happens when you normalize temps, great suggestion.
Im pretty sure im just going to stick with ddr4 128gb 3200mhz until ddr6 comes out, then switch to ddr5, and so on... Im building a workstation on a budget so performance gains at my price/gb level are probably negligible, im okay with staying a gen behind
That's a good perspective to have ... I wouldn't even worry about RAM, I would just focus on GPU and CPU and get whatever RAM you need ... for a workstation more RAM will always be better than faster RAM.
Y-Cruncher VST/VTT will be best to test stability, Karhu will be confusingly stable for 8hrs then err out around 9-10hrs..lol. Y-cruncher will drive that error out in just a few minutes of running.
There is a meaningful difference in 1% lows, which is something most people will notice, but overall I don’t think it’s worth upgrading just for faster ddr5 memory.
I just killed 128gb ram ddr4 3600mhz ram set to 3400mhz. Overclocking kills the ram. Also in winter if you have a cold room... circuits eventually crack on the solders... so either way it'll die eventually.
I think it make sense. Because if you're editing especially videos or heavy apps DRR5 will be great uprade. But when it comes to gaming, I heard majority of the games are not optimize on new hardware like RAM speed.
@@blackbirdpctech I think back to how long ddr3 was relevant, and around this time into it's life it was still good for another half a decade. Same was true for DDR2 as well. It takes a very long time for the latest DDR to become standard.
That’s true and a good point however the pace of technology is increasing, so I don’t think those timeframes will remain the same. With Intel now moving to a DDR5 only platform I think the writing is on the wall … my guess would be 2-4 years, but it’s just a guess.
@@blackbirdpctech Trust in Intel is very shaken right now though with its botched 13th and 14th gen microcode. Gonna be a slow adoption rate to their new platform. Probably many are switching to AMD.
Wait until you see me video that I’m working on for next week … this makes Intel look even worse … not sure who is making decisions there but they definitely need to be fired.
the test could be lot more interesting if was tested with ddr4 4000mhz cl 16 gear mode 1 , 1 to 1 ... the best ddr4 vs best ddr5 ... idont think the test was fair but thats only my opinion ...thank you for the video man .... lot of work to make this video ... dam man ... thx again have a good one !
Appreciate your comment. I specifically selected these kits (3600 and 7200), I used the exact same cpu for both and I used Z690 motherboards, so it was as fair as I could make it. The problem with 8000 MT/s ddr5 memory is that it requires a 2-dimm motherboard and some luck with your cpu silicon (specifically the memory controller) to get stable … not something most people will be able to achieve. Hence I selected DDR5-7200 CL34 as a speed/latency combination that most people should be able to get stable. Same goes for DDR4-4000 ... I actually tried to get that stable on my Z690 motherboard and I was having issues ... I've found that anything above DDR4-3600 requires tweaking to get stable ... again, not something most people will be willing/able to do. So I settled on DDR4-3600 CL14 as a high performance option that most people should be able to get stable. Thanks for watching!
Good vid, i watch frame chasers myself. If youve never watched him u might wanna check him out. Id be curious how a tuned set of ddr4 currently compares. Minimum 4000ddr4+ to be exact. I seen you used tomb raider which scales w everything. I run my 13900k 4070ti w ddr4 4133 cl17. I learned to tune myself after watching many vids. If youve tested the 4000+mhz speeds lmk your results. How much harder is it to tune ddr5 vs ddr4. Another voltage i adjust as well is cpu core voltage aside from sa and vddq
Thanks … I met Jufus at CES earlier this year and we had dinner a few times … he can come across as a bit arrogant but he’s a nice guy and he really does know his stuff, especially wrt memory optimization. I have a very high speed ddr4 kit (4600) but I couldn’t get it stable on the z690 motherboard I was using. I have a z790 ddr4 board so I was planning to give it another shot on that, just haven’t had time. Someone else I watch for memory optimization insight is buildzoid … his videos are pretty boring but they have a lot of great insight … you just have to find it. DDR5 is getting much easier to tune compared with when the 12th gen chips were first released … so I think that will continue as IMC’s get better. Will let you know when I re-test.
@@blackbirdpctech ok cool, i have watched quite a few buildzoid videos for sure. not many guys do mem tuning. i cant get mine up to 4600 stable on my mobo. i do have a top tier bdie kit though. im running a z690 msi mobo currently myself. ill keep an eye out for your memory vid for sure. i like jufes myself personally. its hard to find honest guys in this stuff. you cant believe the paid mouthpieces or the fanboys ya know. good talkin to you though and take care.
As I mentioned in the video, I purposely tested both kits at default XMP settings. I did this to avoid silicon lottery related benefits … that is the only way to ensure the performance shown will be achievable by the majority of viewers.
Thank you for making this awesome video, i am on the brink if i should update from a z690 Mobo to z790 (owning 14700k) in order to upgrade to 48gb 7200 CL34 from 32gb 3800 CL16 and not really sure it is worth it. I see definitely some gains in the 1% lows, Cyberpunk being quite significant even. I wondered if you maybe read this comment @blackbirdpctech and are willing to give some input on the thought that DDR5 gains on 4k but using DLSS and Frame Gen, when those technologies are used wouldn't it benefit from DDR5 speeds more substantially because of "loosening" the GPU limit on 4k? I play a lot of games using DLSS and Frame gen and wondered if there might be a niche more impacted by higher memory speeds. Not many benchmarks include frame gen or dlss i've noticed. Hope this question makes sense, i would really wish to know from someone with insights (You already mentioned the interesting find on the latency issue during high gpu load, where lower latency DDR4 seems to be performing better). Thank you for making this video.
Great question ... so I think you are asking if the use of frame gen and DLSS will reduce the load on the GPU, especially at resolutions and game settings where it's the bottleneck. And the answer almost certainly is yes. It would definitely be interesting to dig into that, but if you offload the GPU and place more load on the CPU then your RAM speed/latency will become much more important, as shown by the 1080p/low benchmarks. When you use RT in games, such as in F1, it does place more load on the CPU even though the GPU remains heavily loaded. I will have to dig in to this a little more.
@@blackbirdpctech Thank you for the quick answer, yes that is what i'm asking, it would be awesome to get some insights on that. Thanks again, so far i haven't found any benchmarks looking specifically into this topic 🙏
I think the only time you'll ever need Uclk higher than Fclk is when you have dual CCD's like the Ryzen 9 series CPUs as they can access memory at different times. That also may benefit from the higher bandwidth of the Mclk 8,000mhz. The setting would really weird. If running 8,000mhz ram on a 9950x you'd have max your Uclk to 3,000to 3,300mhz. Then Fclk could be 2,000 to 2,200mhz. Which means not ever in sync at all. The sync seems to mainly benefit single ccd designs are they're is only one ccd accessing the memory control. I do wonder though if that higher speed ram helps when using Multi-GPUs setups.
Take a look at my recent RAM video for AMD AM5 Ryzen CPUs: th-cam.com/video/JuUhnQaGG_I/w-d-xo.html You can't independently set the UCLK from the MCLK, it's set automatically when you select 1:1 or 2:1 mode. In 2:1 mode the UCLK for DDR5-8000 is 2000 ... you can't set it to 3000. You can however change your FCLK, but speeds above about 2200 tend to be unstable.
@@blackbirdpctech I have to look through my comments, but I know I was talking to someone/Video uploader on here who did exactly that 3:2 ration with the uclk somehow for 8000 ram 2133mhz infinity fabric
I just put an MSI Z690 Unify motherboard in my PC for the i5 12600k CPU I got, And the difference to my old Gigabyte H610i is like night and day, The H610i is 32bg of DDR4, And the Z690 is 16gb of DDR5, And I can say that just the 16gb of DDR5 at 5200MHz is more snappier than the 32gb of DDR4 at 3200MHz, I never thought a decent motherboard would make such a noticeable difference, I do Sim Racing in VR, And AMS 2 do look better, So there is something about the DDR5 ram, That improves your PC, I am glad now I did get the MSI Z690 unify motherboard, It had made the i5 12600k better for the VR sim racing, So yes if anyone is thinking of getting a motherboard with DDR5, Just do it if you can, It is a move I don't regret, And the PCIe Gen 5 on the MSI Z690 I think also makes a noticeable difference, Plus the MSI Z690 got an AI tool that works out the best settings for you, And the settings on the MSI Z690 is almost endless lol.... But yeah I am glad I made the move to a DDR5 motherboard, That way I am also getting all my moneys worth out of the i5 12600k, And I am sure the I5 12600k would be more stable when in DDR5, I wouldn't go back to DDR4 now I am on DDR5, So if that answers people questions, lol, If you can that is, If money is tight then yeah wait, Another two reason's I got the MSI Z690, As it was on sale on Amazon, For 160 pounds from 230 pounds, And I plan on upgrading the 12gb GPU to a 16bg 40 series, Just waiting for the prices to come down, lol, But for now the 12gb 3060 on the MSI Z690 with the i5 12600k is amazing on AMS 2 in VR, But yeah if you can afford the extra cost, Then yeah you will notice the difference to DDR4 and DDR5, Or at least I notice my Pc is more snappier, And with the MSI Z690 there is a lot of options to play with, That can be used to improve your system over all, And in the long term, So getting a Z690 motherboard with DDR5 is only going to help the CPU and GPU in the long term, I would imagine, I could be wrong, I am no expert, lol, But it just makes sense that DDR5 would be better for a CPU that is designed for DDR5, And again I am sure Intel and AMD is designing the CPU with DDR5 being more stable so to speak, More stable something is then the faster that something can handle yeah, lol, it just makes more sense in theory that DDR5 will be faster and better for gaming in 2024-25, I can even see Z690 motherboards dropping in price to get people on to DDR5, Again I would imagine the I5, I7 and i9 would work better or be more stable in the DDR5..... I am glad that I made the DDR5 move now, And if you got an i9 12900k then yeah upgrading to the DDR5 will give you better and punchier gaming, Colors stand out more in VR as well with the MSI Z690 compared to the Gigabyte H610i I was using last week, The VR sim racing is better lets just say, lol...... Again if you can afford the move to DDR5 then getting a Z690 motherboard would be the way to go, As I can tell you it has made a difference to me, it is actually pretty nuts when I think about it, lol, The same CPU and GPU, But, Looks better and is better for the sim racing experience in VR than when I had the Gigabyte DDR4 3200MHz, So yeah for me amazing, If you can afford a Z690 and the DDR5 ram just do it, you won't regret it if you is a gamer.....
Appreciate your comment and it's great that you are much happier with your system. For gamers the objective should be to build a PC that offers a great gaming experience. That said it's unlikely that DDR5 will be more stable than DDR4, in fact DDR4 should be more stable because it's a much more mature standard and it puts a lower load on CPU memory controllers. So my guess is that there was another issue with your old system that was causing stability problems. The good news is that you solved it with your upgrade, which is awesome.
@@blackbirdpctech I have also learned how to put a PC together, lol,. If I was a PC company, I would do a PC build option that would be Father and Child project PC, lol, Nor jokes aside it is a good thing to know, On how to put a PC together, And the Z690 is also pretty straight forward to build on, I got an MSI MAG 320r forge Airflow gaming case not long ago, So it seemed fitting I put an MSI motherboard in an MSI case, lol, See I did think that the DDR5 ram would be more stable on a CPU that is DDR5, That is the great thing about the PC world is it is always a learning curve, There is always something new to learn, And the combinations of CPU and GPU within different motherboards is endless, I actually do wonder how many different CPU and GPU combinations there is, lol, I do find since I put in the MSI Z690 motherboard, The VR on AMS 2 seems to be more punchier in the graphics if that makes sense, Almost as if the colors are brighter if that makes sense, lol, I have got older vids of Race Room, I just re-downloaded Race Room as I thought Race Room wasn't VR, lol, So now I can see when I do go on Race Room how much of a difference there is, lol...... But yeah I did honestly think that the DDR5 Ram would be more stable with more wiggle room having higher frequencies, Than the DDR4 being lower, Lower is more controllable even if you is a PC then, lol.... Again always something to learn in the PC world, lol.......
That's a great attitude ... I am still learning too, if you stop learning you you stop growing. That's one of the reasons I engage in my comment section, I learn a lot from my subscribers.
@@blackbirdpctech Fair play, When someone takes a few out to reply to comments, Then I subscribe, It is when a You Tuber creator doesn't reply I think why should I subscribe, But I love the attitude you have for this PC stuff, It shows dedication, For when someone is dedicated to a certain subject, Then 9 out of 10 they make top mentors, One mistake I made growing up is finding the wrong mentor's, But then I grew up in the 90s, And it could be said the 90s was a nuttier time to be alive, Hence the term Noughties, lol... Nor jokes aside life is a constant learning curve, The reason someone becomes wise is from learning over a certain amount of years yeah, And with the PC stuff, We are always going to use computers now, The reason I love VR so much is coz VR can be used as a tool to learn all sorts, Want to learn how to put a car engine together, Get Car Mechanic simulator, Want to learn the insides of a human, There is even things you can do in VR that involves medical operations, lol, VR is even helping people who suffered a stroke, Getting motor functions back using the VR, Oculus does it, So even Meta can see how much practical learning and education a VR headset can bring to your brain, Even poor kids in Africa can have a classroom within a VR headset, I did read there is things being set up so old VR head sets can be programmed with software, To actually bring VR to poor villages, How real it actually is, But isn't that a great idea, Giving poor village children the chance to learn with VR, They can charge the headsets using Solar power as well apparently, Was in the article about using VR as a tool for poor kids to turn a hut into a classroom, I hope it is real and it do happen, Like I said every kid deserves some sort of education, So Amen for Stand alone VR, Plus VR is a brilliant wat to do sim racing, lol.... I will check out some more of your vids over the weekend.....
good ideea for a video, but i think that the 4k results are useless in this comparition simply becouse the bottleneck was on the gpu and the diferences between the ddr4 and ddr5 memory kits were 0. Though the 1080p test at low settings indeed is relevant. Not a big diference, i was expecting more. It seems that the latency and the timings are important, not just how many megahertz.
Thanks, the reason that I decided to keep the 4k results in the benchmarks is to show people that are gaming at 4k that it's basically irrelevant what memory kit you game with ... I think it's always good to have the data even if the data simply supports what you thought would happen. Thanks for watching!
@@blackbirdpctech agree i'm one of those gamers that use Blender and game at 4K and was thinking of upgrading but every video I see there is literally not much difference besides maybe 5-10% at best
Bro, I have a 12700KF , ROG STRIX Z690-A GAMING WIFI D4 and GSILL TRIDENT Z 3000 14-14-14-34. I bought a 4070ti last year september. Do you think i need to upgrade motherboard and Ram for a DDR5 system now or should i wait and buy the board and ram later. Its because i will buy a 5070 ti level card when it releases.. By the time 5070ti releases new cpu and board will release. thats why i am asking.. Its is really bad to keep a 4070ti on a ddr4 3000mhz system ? I just use pc for gaming and media.
I would wait and upgrade your system when your new GPU becomes available ... you may wish to consider the new AMD CPU's when they come out. One thing you could do while waiting is pickup a cheap DDR4-3600 CL16 kit but it's not going to significantly increase your FPS.
I got a Ryzen 5 3600 its am4 with 2070 super. I want to buy a 7900 XT. should a get a new am5 motherboard + cpu + ddr5 combo? or just buy Ryzen 7 5700X3D 8-Core am4 and ddr4 ? mostly for 1440. Thanks.
Good question ... you might find this video helpful: th-cam.com/video/JIjD1mZUGz0/w-d-xo.html In it I upgrade a system with a 3600X/3060 to a 5700X3D/4060Ti. I would get the 7900 XT and a 5700X3D ... the next time you upgrade there will be no choice but to upgrade the entire platform.
Would really like to have seen you test with much tighter control over the variables. Things I see that could cause differences in performance, false positives, or otherwise make the data misleading. Different capacity memory, different main storage drive, different windows install, different model and spec of mainboard. This kind of test really needs to be done with as many variables controlled for as possible. Ideally I would think using the same brand memory and capacity, use the same cpu and cpu cooler, use as close as possible to the same motherboard and ideally you would find the exact same model motherboard with the same chipset and the only difference being ddr4/ddr5. Rant over I will watch the rest of the video now.
For a perfect test I agree however most of these variables have no meaningful impact on gaming performance. The two components that impact gaming performance the most are the CPU and GPU, and as I mentioned in the video I used the exact same CPU and GPU for each system. I also used Z690 motherboards and made sure that the Bios for each board was setup to enforce all Intel CPU limits. Components like SSDs do not impact gaming performance even when comparing Gen5 versus Gen4 drives (see results from my recent video). For memory the brand is not really relevant, it's the memory die manufacturer that is relevant and even then there are differences between DDR5 and DDR4 that make using the same die manufacturer pointless (i.e. Samsung B-die is best for DDR4 while Hynix A-die is best for DDR5). Also, memory capacity is only relevant when you run out of memory ... for these tests no games required anything close to 32GB of memory, so the additional 16GB of memory for the 48GB kit had no impact.
Did you check the troubleshooting guide in the video? It shows likely sources of your issue so you can hopefully find the problem and resolve it rapidly. For what you describe you should look under the >10sec (turns on, but no display) row.
My DDR4 motherboard was a Z690 board, so I wanted to be consistent with the chipset when comparing performance to help eliminate as much variability in the results as possible. Yes, both Z690 boards required a bios update to recognize the 14900K.
if i was about to buy a new pc with a budget of 1000-1000 €. I would rather go with a ddr4 system at 3600mhz with a 4070 than with a ddr5 system with a 4060 and have fun with it for 7 years, right? I dont see the value of ddr5 yet.
Absolutely, you will get a much larger performance boost going with a better GPU virtually every time ... what is the CPU you have and/or are considering buying?
charts alone done tell the whole story. been on 6400 ddr5 from 3600 and there was differences in certain games. ie getting that extra hit from a win or lose situations. INMO now its more of a choice you want to stay up to date upgrade if u dont care dont.
Appreciate your comment. I think the charts do a pretty good job of telling the story, the part that is tough to show via charts is how an increase in 1% low "feels" in games ... that might actually be an interesting topic for an upcoming video. The main point I wanted to get across for people still on systems with DDR4 is that you should wait for next gen chips/motherboards to be released before upgrading. Thanks for watching!
I would love to see 4 sticks of DDR4 4000mhz vs the DDR5 7200 because DDR4 is more stable and runs perfectly fine with 4 sticks where DDR5 if you use 4 sticks and you set the XMP on you will just keep crashing
DDR4 at 4000 with 4 dimms will still be a challenge to get stable. As DDR5 matures it’s getting easier, for example it’s relatively easy to get 4 dimms of 6000 stable now.
@@blackbirdpctech Strange I actually got rid of my maximus hero z790 with Cl32 6400 mhz SWAPPED it out for a Z790 strix ddr4 gaming feels so much better no more dsync in shooter games
@@blackbirdpctech maybe I started with the 12900k and a z690 maximus gskill 6000 then 13900KF Z790 Maximus 6400 Gskull. could be just me but Iv always felt that DDR4 is better in games even though DDR5 has fps.
With respect to stability you are spot on, but as DDR5 matures and the platforms that use them mature, it will get much better. Remember DDR4 has been around for a while and is a very mature solution at this point.
I typically don’t benchmark games that do not provide consistent results between runs … that said I wouldn’t expect the relative performance differences in these games to vary too much from the averages shown in the video.
That's a great question and I just finished testing and optimizing the 265K, so I can say confidently that the Z890 motherboards have much better memory support than Z790, and it's not even close. In addition, the memory controller on Core Ultra Series 2 chips appears to be much better as well. So even without using CUDIMMS, you should be able to easily get DDR5-8000 stable on a 4-dimm Z890 motherboard. It's shocking to me that no one else is talking about this because it's a significant improvement.
@@blackbirdpctech Which brings up the question, which set to buy? I've only seen up to 8000 MT on an amazon search. Most people say I should get a lower tier, but I need something to help me make the best decision. What about timing and latency, are they still valid. Would you do another video?!?
@@darkman237 I just completed one for AM5, so I should do another one for Intel Arrow Lake. Intel APU's scale really well with memory, so I would recommend getting a fast kit.
Isn't it a bit of a scam on part of the cpu/memory/mother board manufacturers that they advertise based on the memory speed with expo/xmp while on the same time don't guarantee these speeds/timings are achievable/stable. Honestly reviewers should just benchmark with jdec settings until the manufacturers start to guarantee that these speeds/timings are actually achievable.
Appreciate your comment. I recently watched a video explaining JEDEC memory standards ... apparently when they set the standard they have to guarantee that it will work in every possible scenario, which would include super hot environments with minimal airflow, server rooms, edge devices, industrial applications, etc. which is why the actual speeds that they guarantee are relatively low. That said, what I think is a scam is when motherboard manufacturers market their boards for high speed memory when there is basically no chance of achieving that speed. So when all of the big manufacturers (Asus, MSI, Gigabyte, etc.) marketed their Z790 4-dimm motherboards at speeds of 8000 and above this was BS ... there is absolutely no way those boards can run those speeds stable. The problem is people buy the boards, build their systems, install the RAM and just lower the speed to get it stable because it's too much trouble to RMA the board. The manufacturers know this ... that's why they will keep doing it. I highlighted this terrible business practice in one of my earlier memory videos. What we need are the large influencers to call it out, people like Linus and Jay ... the problem is they have no idea what they are doing most of the time ... Jay couldn't even get his AM5 system to run EXPO at 6000.
@@blackbirdpctech Exactly. They should market their products at the speeds that are achievable in normal conditions (no golden sample cpus, memory with specific dies (that aren't even clearly stated on the packaging and can change from one batch to another of the same memory), only two ram slots occupied, ...). If the cpu/memory/motherboard is advertised as supporting speed x you should be able to just buy it, plug it in and have it work. If it doesn't it should be considered faulty and replaced by the store you bought it at or a full refund issued.
Agreed ... at a minimum they should provide conditions under which the advertised speed is likely ... so they could say the speed is 8000 however it requires a 2-dimm motherboard and your probability of reaching that speed is 50% but 7600 with the same kit is 90% ... perhaps even develop a standard that states the advertised speed is guaranteed to function on 85% of motherboards, or something like that. What they are doing now is wrong.
Got my Sub for Bruce :D Just kidding. Awesome comparison. Will stick to my 3600 DDR4 for a while i guess :D But tbh the comparison isn´t really fair. 3600 is the sweetspot of DDR4 but 7200 ist ultra high end for DDR5 already. 6000 is the sweetspot for DDR5 and should be the comp here.
Great add ... I've played around with lots of different memory testing tools and the reason I recommend Karhu is that it's super simple to use and finds errors quickly ... the only downside is that it's not free.
I tested with a 4090 so I don’t understand your statement … there is no doubt that ddr5 improves your 1% lows but you only see a difference in gaming at lower resolutions when the cpu is more heavily loaded. So this means the gpu model has no impact, which makes sense … you will only see an impact when your system loads your cpu more than your gpu.
I call BS on the Avatar 4k result - your take is that at 4k the game becomes more latency dependant and thus the ddr4 kit wins - however, the 1% is still higher on the ddr5 kit? How do you explain that?
What I said in the video was that at 4K the DDR4-3600 kit beat the DDR5-7200 kit in at least half the titles, which would indicate that in these titles where the GPU is highly loaded, latency becomes more important to extracting maximum performance. The 1% low average for the DDR5 kit is higher than the DDR4 kit, but not by a meaningful amount (
I tested the 7800X3D with DDR5-6000 and DDR5-8000 memory and there is a performance benefit, particularly in the 1% lows however the behavior is different when compared with Intel systems due to the 3D V-cache and infinity fabric: th-cam.com/video/p21bsVYAgvA/w-d-xo.html Also, it's impossible to properly test DDR4 vs DDR5 for AMD systems because AM4 only supports DDR4 and AM5 only supports DDR5, so there is no way to use the same CPU for testing ... that's why I tested on LGA1700 with a 14900K. By the way, these new Bios settings that Asus just rolled out will negatively impact the performance of the 14900K ... so I'm thinking of doing a comparison with the 7800X3D ... will be interesting to see how they compare now that the 14900K doesn't have an infinite power limit 😉
@blackbirdpctech I also have done tests with my 7800X3D with various different ram, and the benefits are low on faster Ram, the real benefit for higher 1% lows on X3D is lower Clock Timing and more ram getting 30 to 50% higher 1% lows, the best is 2 dims at 32GB DDR5 for a total of 64GB at 6000MT CL 30 and 64GB of 5600MT CL28 which is what I am rocking right now, it also improves better stability, less stuttering and more stable frame rates.
I was watching buildzoid and he was playing around with timings on a 7900X (non x3d) … I am interested to see if I can get 6000 stable at CL28 on the 7800 x3d … would be interesting to see the performance impact.
@blackbirdpctech Do the test so you can see show the benefits and differences, G-Skill Ripjaw Sells on newegg the 32 and 64GB 5600MT CL28 on Newegg. I am not a content creator, but it would be cool if more youtubers like yourself show the difference with more ram and lower Timings compared to faster Ram but higher Timings.
The amount of RAM has absolutely no impact on performance unless you don’t have enough. None of the benchmarks I ran required anything close to 32GB of memory.
Awesome video. Thank you for providing quality content as always!
Thank you for watching, appreciate your support!
Such an underrated TH-cam channel! Great video as always… that troubleshooting guide is awesome!
Thank you, glad you like it!
Answered about 5 distinct different questions I get on the daily in the PC groups I teach people in. Also, did you watch the fights tonight!? Thanks again for making high quality fantastic content.
Thanks, really appreciate that. Unfortunately I wasn’t able to watch … just caught up on the results now … looks like I missed a great event!
Which are this 5 questions and can you explain please?
I'm staying on am4 at least until am6 comes out , then i'll be switching to am5 , another great vid!
Thanks … that’s exactly what I would do at this point … there may also be an updated AM5+ socket as well.
Thanks bro you had the only good DDR4 vs DDR5 test I managed to find, keep it up
Thank you !!
Another great analysis Matt!
I'm on AM4 with an R7-5800X3D on an ASRock X570 Pro4 motherboard and 64GB of Team Vulcan DDR4-3600 so I have no intention whatsoever of upgrading in the near future. My PC is already stacked.
Exactly, that was the point … it’s easy to get wrapped up in the hype but the data doesn’t lie … it’s not a meaningful difference yet.
@@AvroBellow I know this is an old comment but I have a similar, slightly less beefy setup- AM4, R7 3700x with 32gb DDR4-3600. I found this video while searching around about possible upgrades, still not sure about it. Do you have a specific need for 64gb or just for extra headroom?
@@cds124ful probably extra head room ^^ for gaming 32gb ram is enough (I do dual screen gaming and I am doing fine^^)
Wstched this video late but it would have been interesting to have the GOAT x3D gaming CPUs in. When I first built with AM4 memory tuning and over clocking was a hot topic, while now cheaper slower memory is often delivering the same performance.
The really big difference will be between DDR5 vs DD6 and that will be the day when it will be worth updating from DDR4.
I think as DDR5 matures, and the memory controllers on newer chips mature, we will see bigger gains.
Amazing video!! 👏🏼
Thank you!!
@@blackbirdpctech You should make a SSD vs HDD speed comparison!!
I was planning to do an SSD comparison between Gen 3, Gen 4 and Gen 5 ... I could add a modern high speed HDD to the mix ... Gen 5 SSD's are great but most motherboards can only support them if you reduce the lanes going to the GPU, which really sucks.
@@blackbirdpctech this is new to me, i didnt know that there were different generations to SSDs. This is interesting, thanks ill look into it :)
P.s I am rocking a 4TB HDD so I am deffo going to get an SSD when I build my new setup hopefully
Yes, pretty large speed differences between generations … you will see a huge uplift in performance going from a hdd to ssd.
Matt I saw that you have a open bench table too. I got one too and I even managed to travel with my whole setup with it. Took it through TSA. Loved the video mate. Cheers
Yeah, they are fantastic bench tables … really makes testing components easy. Thanks for watching!
Would love to see this test done with the 7800x3d, well done man 👊👊
Thank you! I already did a video comparing ddr5-6000 vs ddr5-8000 for the 7800x3d (th-cam.com/video/p21bsVYAgvA/w-d-xo.html) … unfortunately I can’t do a ddr4 vs dd5 test for AMD without changing the cpu's, that’s why I did the testing for this video with a 14th gen Intel cpu.
With AMD you can just test different Speeds... Only Intel provides support for two RAM Generations on the same CPU Socket, so Intel is the only CPU to test the different RAM Generations lol
@@FunFoxHDlol no shit this comment was 6 months ago but im pretty sure what I was referring to was how does fast ram affect the x3d chips. Not how does ddr4 affect them because that would be impossible….
@@blackbirdpctechyes the fast ddr5 vs slow ddr5 with the x3d chips is what I was referring to thanks u. This was 6 months ago but I have since learned that faster ram doesn’t seem to be as relevant for x3d chips.
@@danebeee1 oh... Yeah, I have an 7800X3D myself and can confirm it's nearly identical in games... Also for some reason my RAM is unstable over 5600MHz, gladly I have an X3D CPU
Spectacular video
Thanks, glad you liked it!
i can't confirm if it's good for gaming but i am on 3600 CL 14 and for work it's noticeable computing/vid editing/ photo editing that is my workflow and i begin to create some 3d nonsense sometimes despite the fact my kit looks amazingly
Thank you for saving me £300 on an unnecessary upgrade for my 12700k. Too many reviews only focus on low res game testing for the bigger, attention grabbing differences (HUB). I expected small gains for DDR5 at 4k but the negative results were surprising. I wonder if the UE5 engine games will be similarly indifferent at 4k
It's always interesting to look at the data versus listening to opinions ... I discovered the same thing when I tested Gen 4 M.2's against Gen 5 M.2's ... there is no difference in gaming, so don't waste your money.
Interesting detail I found with Ry 5 5600x was an energy penalty going above DDR4-3200 that created a significant hotspot as the IOD memory controller was not entering an energy saving state. This meant I lost any benefit of pushing DDR4 to 3400MT/s and reducing heat was better.
The subsequent Zen3 stepping had metal layer fixes from after Q3/21 which helped the 5800x3D be efficient, was mostly ignored but helped the low end.
With a small TDP 65W/88W PPT the watts saved help the per core power budget. I saw an interesting test result where disabling 2c/4t on a Ry 5 5600x actually delivered small fps gains albeit in older titles.
Greater use of SMT would mitigate memory stalls and per core power and cache was boosted by 50%
Kinda like cars...the last 5% in performance doubles your budget. And yet, us 'enthusiasts' bite like rabid dogs, every-single-time. hehe Another good vid, brotha!
The crazy thing is I know what the logical choice is but I can’t help myself … it’s a disease … wait until you see my next video … I’m building my ultimate gaming pc … feels good to build again 😉
Hard not to go for the best of the best instead of the best value!
That's what makes us PC enthusiasts 😉
@@blackbirdpctech Let's goooooo!!! These are the most fun projects. Budgets be damned!!!
fantastic video!
Thank you!
I would've liked to seen the temps while gaming as well. You didnt mention if you overclocked the cpu so and during the productivity areas where you showed the temp a few times, it was close to a 10° difference. Thats a lot of room on the table to oc the cpu and if so, would the numbers then be closer or better? Of course assuming you only oc the ddr4/cpu setup to match the temps of the ddr5/cpu setup.
It's always a trade-off when deciding on how much data to show ... if I add temps to the gaming charts I worry that it will impact their readability ... will see if I can figure out a good way of doing it in the future. With respect to CPU overclocking, I did explain in the video that I enforced all Intel CPU limits in bios for both motherboards. That said the temp difference was, as you point out, relatively large. I think most of that difference is due to the memory controller on the CPU needing to work a lot harder with DDR5. It would be interesting to see what happens when you normalize temps, great suggestion.
Im pretty sure im just going to stick with ddr4 128gb 3200mhz until ddr6 comes out, then switch to ddr5, and so on... Im building a workstation on a budget so performance gains at my price/gb level are probably negligible, im okay with staying a gen behind
That's a good perspective to have ... I wouldn't even worry about RAM, I would just focus on GPU and CPU and get whatever RAM you need ... for a workstation more RAM will always be better than faster RAM.
Excelent job, man. I'm stick with ddr4 for a while.
Thank you!
Man. I love your content!
Really appreciate that!
Y-Cruncher VST/VTT will be best to test stability, Karhu will be confusingly stable for 8hrs then err out around 9-10hrs..lol.
Y-cruncher will drive that error out in just a few minutes of running.
Thanks, will check it out … I’ve found Karhu to be pretty good for finding errors rapidly … if it lasts 8hrs then I take that as a win 😉
Have been using Y-Cruncher its the worst possible case workload you can hammer on your RAMS.
Good stuff Bro...keep it up
Thank you!
you sir got a subcriber , just returned my DDR5 board
Welcome to the Blackbird PC Tech community!
Rocket Scientist eh
Not exactly brain surgery is it.
Ha ha … that actually made me laugh 😂
So... It does not matter. DDR4 still is good to go. Okay.
There is a meaningful difference in 1% lows, which is something most people will notice, but overall I don’t think it’s worth upgrading just for faster ddr5 memory.
I just killed 128gb ram ddr4 3600mhz ram set to 3400mhz.
Overclocking kills the ram.
Also in winter if you have a cold room... circuits eventually crack on the solders... so either way it'll die eventually.
@@dylandesmondthat’s extremely unlikely … you didn’t even run them at the rated speed. What voltage did you run them at?
If you can get your DDR4 to 4000mhz at cl15/16 it’s the same as DDR5 7200
@@alexburke4569 I have not had much luck getting ddr4 to run stable at those speeds and timings, even with kits that are sold at those speeds.
I think it make sense. Because if you're editing especially videos or heavy apps DRR5 will be great uprade. But when it comes to gaming, I heard majority of the games are not optimize on new hardware like RAM speed.
Nice to see someone finally not overselling DDR5 to less tech savvy people. DDR5 won't be worth it for another 5 - 8 years people, save your wallet.
Glad you liked it, 5-8 years might be a bit too long, but the value just isn't there for most people at the moment.
@@blackbirdpctech I think back to how long ddr3 was relevant, and around this time into it's life it was still good for another half a decade. Same was true for DDR2 as well. It takes a very long time for the latest DDR to become standard.
That’s true and a good point however the pace of technology is increasing, so I don’t think those timeframes will remain the same. With Intel now moving to a DDR5 only platform I think the writing is on the wall … my guess would be 2-4 years, but it’s just a guess.
@@blackbirdpctech Trust in Intel is very shaken right now though with its botched 13th and 14th gen microcode. Gonna be a slow adoption rate to their new platform. Probably many are switching to AMD.
Wait until you see me video that I’m working on for next week … this makes Intel look even worse … not sure who is making decisions there but they definitely need to be fired.
the test could be lot more interesting if was tested with ddr4 4000mhz cl 16 gear mode 1 , 1 to 1 ... the best ddr4 vs best ddr5 ... idont think the test was fair but thats only my opinion ...thank you for the video man .... lot of work to make this video ... dam man ... thx again have a good one !
Appreciate your comment. I specifically selected these kits (3600 and 7200), I used the exact same cpu for both and I used Z690 motherboards, so it was as fair as I could make it. The problem with 8000 MT/s ddr5 memory is that it requires a 2-dimm motherboard and some luck with your cpu silicon (specifically the memory controller) to get stable … not something most people will be able to achieve. Hence I selected DDR5-7200 CL34 as a speed/latency combination that most people should be able to get stable. Same goes for DDR4-4000 ... I actually tried to get that stable on my Z690 motherboard and I was having issues ... I've found that anything above DDR4-3600 requires tweaking to get stable ... again, not something most people will be willing/able to do. So I settled on DDR4-3600 CL14 as a high performance option that most people should be able to get stable. Thanks for watching!
Good vid, i watch frame chasers myself. If youve never watched him u might wanna check him out. Id be curious how a tuned set of ddr4 currently compares. Minimum 4000ddr4+ to be exact. I seen you used tomb raider which scales w everything. I run my 13900k 4070ti w ddr4 4133 cl17. I learned to tune myself after watching many vids. If youve tested the 4000+mhz speeds lmk your results. How much harder is it to tune ddr5 vs ddr4. Another voltage i adjust as well is cpu core voltage aside from sa and vddq
Thanks … I met Jufus at CES earlier this year and we had dinner a few times … he can come across as a bit arrogant but he’s a nice guy and he really does know his stuff, especially wrt memory optimization. I have a very high speed ddr4 kit (4600) but I couldn’t get it stable on the z690 motherboard I was using. I have a z790 ddr4 board so I was planning to give it another shot on that, just haven’t had time. Someone else I watch for memory optimization insight is buildzoid … his videos are pretty boring but they have a lot of great insight … you just have to find it. DDR5 is getting much easier to tune compared with when the 12th gen chips were first released … so I think that will continue as IMC’s get better. Will let you know when I re-test.
@@blackbirdpctech ok cool, i have watched quite a few buildzoid videos for sure. not many guys do mem tuning. i cant get mine up to 4600 stable on my mobo. i do have a top tier bdie kit though. im running a z690 msi mobo currently myself. ill keep an eye out for your memory vid for sure. i like jufes myself personally. its hard to find honest guys in this stuff. you cant believe the paid mouthpieces or the fanboys ya know. good talkin to you though and take care.
That ddr4 is b-die though, you could get very close to the 7200 if you tuned it. Running b-die at 60ns on intel is borderline criminal.
As I mentioned in the video, I purposely tested both kits at default XMP settings. I did this to avoid silicon lottery related benefits … that is the only way to ensure the performance shown will be achievable by the majority of viewers.
Thank you for making this awesome video, i am on the brink if i should update from a z690 Mobo to z790 (owning 14700k) in order to upgrade to 48gb 7200 CL34 from 32gb 3800 CL16 and not really sure it is worth it. I see definitely some gains in the 1% lows, Cyberpunk being quite significant even. I wondered if you maybe read this comment @blackbirdpctech and are willing to give some input on the thought that DDR5 gains on 4k but using DLSS and Frame Gen, when those technologies are used wouldn't it benefit from DDR5 speeds more substantially because of "loosening" the GPU limit on 4k? I play a lot of games using DLSS and Frame gen and wondered if there might be a niche more impacted by higher memory speeds. Not many benchmarks include frame gen or dlss i've noticed.
Hope this question makes sense, i would really wish to know from someone with insights (You already mentioned the interesting find on the latency issue during high gpu load, where lower latency DDR4 seems to be performing better).
Thank you for making this video.
Great question ... so I think you are asking if the use of frame gen and DLSS will reduce the load on the GPU, especially at resolutions and game settings where it's the bottleneck. And the answer almost certainly is yes. It would definitely be interesting to dig into that, but if you offload the GPU and place more load on the CPU then your RAM speed/latency will become much more important, as shown by the 1080p/low benchmarks. When you use RT in games, such as in F1, it does place more load on the CPU even though the GPU remains heavily loaded. I will have to dig in to this a little more.
@@blackbirdpctech Thank you for the quick answer, yes that is what i'm asking, it would be awesome to get some insights on that. Thanks again, so far i haven't found any benchmarks looking specifically into this topic 🙏
@@blackbirdpctech Any luck looking into this? :D
Im running 4133mhz cl15 gear 1 ddr4, thinking about upgrading but not sure to be honest. Would like to see a benchmark between high OC ddr4 vs ddr5
I will have some benchmarks in my video next week that show the performance uplift of running DDR4-3600 vs DDR4-4000 for an AMD AM4 system.
@@blackbirdpctech sounds nice, what timings? Is it also cl16 for the 4000mhz?
@@jewelersofwhale DDR4-4000 CL18-22-22-42 however the challenge for AM4 is running an FCLK of 2000MHz.
親切で真面目な情報を教える姿勢が大好きです。とても良い動画！👍
Thank you !!
Ram speed is very important if ure running integrated graphics.
That makes sense.
Thank you , Дуже цікаво тебе дивитися
Thanks for watching!
I think the only time you'll ever need Uclk higher than Fclk is when you have dual CCD's like the Ryzen 9 series CPUs as they can access memory at different times. That also may benefit from the higher bandwidth of the Mclk 8,000mhz. The setting would really weird.
If running 8,000mhz ram on a 9950x you'd have max your Uclk to 3,000to 3,300mhz. Then Fclk could be 2,000 to 2,200mhz. Which means not ever in sync at all. The sync seems to mainly benefit single ccd designs are they're is only one ccd accessing the memory control.
I do wonder though if that higher speed ram helps when using Multi-GPUs setups.
Take a look at my recent RAM video for AMD AM5 Ryzen CPUs:
th-cam.com/video/JuUhnQaGG_I/w-d-xo.html
You can't independently set the UCLK from the MCLK, it's set automatically when you select 1:1 or 2:1 mode. In 2:1 mode the UCLK for DDR5-8000 is 2000 ... you can't set it to 3000. You can however change your FCLK, but speeds above about 2200 tend to be unstable.
@@blackbirdpctech I have to look through my comments, but I know I was talking to someone/Video uploader on here who did exactly that 3:2 ration with the uclk somehow for 8000 ram 2133mhz infinity fabric
I just put an MSI Z690 Unify motherboard in my PC for the i5 12600k CPU I got, And the difference to my old Gigabyte H610i is like night and day, The H610i is 32bg of DDR4, And the Z690 is 16gb of DDR5, And I can say that just the 16gb of DDR5 at 5200MHz is more snappier than the 32gb of DDR4 at 3200MHz, I never thought a decent motherboard would make such a noticeable difference, I do Sim Racing in VR, And AMS 2 do look better, So there is something about the DDR5 ram, That improves your PC, I am glad now I did get the MSI Z690 unify motherboard, It had made the i5 12600k better for the VR sim racing, So yes if anyone is thinking of getting a motherboard with DDR5, Just do it if you can, It is a move I don't regret, And the PCIe Gen 5 on the MSI Z690 I think also makes a noticeable difference, Plus the MSI Z690 got an AI tool that works out the best settings for you, And the settings on the MSI Z690 is almost endless lol....
But yeah I am glad I made the move to a DDR5 motherboard, That way I am also getting all my moneys worth out of the i5 12600k, And I am sure the I5 12600k would be more stable when in DDR5, I wouldn't go back to DDR4 now I am on DDR5, So if that answers people questions, lol, If you can that is, If money is tight then yeah wait, Another two reason's I got the MSI Z690, As it was on sale on Amazon, For 160 pounds from 230 pounds, And I plan on upgrading the 12gb GPU to a 16bg 40 series, Just waiting for the prices to come down, lol, But for now the 12gb 3060 on the MSI Z690 with the i5 12600k is amazing on AMS 2 in VR, But yeah if you can afford the extra cost, Then yeah you will notice the difference to DDR4 and DDR5, Or at least I notice my Pc is more snappier, And with the MSI Z690 there is a lot of options to play with, That can be used to improve your system over all, And in the long term, So getting a Z690 motherboard with DDR5 is only going to help the CPU and GPU in the long term, I would imagine, I could be wrong, I am no expert, lol, But it just makes sense that DDR5 would be better for a CPU that is designed for DDR5, And again I am sure Intel and AMD is designing the CPU with DDR5 being more stable so to speak, More stable something is then the faster that something can handle yeah, lol, it just makes more sense in theory that DDR5 will be faster and better for gaming in 2024-25, I can even see Z690 motherboards dropping in price to get people on to DDR5, Again I would imagine the I5, I7 and i9 would work better or be more stable in the DDR5.....
I am glad that I made the DDR5 move now, And if you got an i9 12900k then yeah upgrading to the DDR5 will give you better and punchier gaming, Colors stand out more in VR as well with the MSI Z690 compared to the Gigabyte H610i I was using last week, The VR sim racing is better lets just say, lol......
Again if you can afford the move to DDR5 then getting a Z690 motherboard would be the way to go, As I can tell you it has made a difference to me, it is actually pretty nuts when I think about it, lol, The same CPU and GPU, But, Looks better and is better for the sim racing experience in VR than when I had the Gigabyte DDR4 3200MHz, So yeah for me amazing, If you can afford a Z690 and the DDR5 ram just do it, you won't regret it if you is a gamer.....
Appreciate your comment and it's great that you are much happier with your system. For gamers the objective should be to build a PC that offers a great gaming experience. That said it's unlikely that DDR5 will be more stable than DDR4, in fact DDR4 should be more stable because it's a much more mature standard and it puts a lower load on CPU memory controllers. So my guess is that there was another issue with your old system that was causing stability problems. The good news is that you solved it with your upgrade, which is awesome.
@@blackbirdpctech I have also learned how to put a PC together, lol,. If I was a PC company, I would do a PC build option that would be Father and Child project PC, lol, Nor jokes aside it is a good thing to know, On how to put a PC together, And the Z690 is also pretty straight forward to build on, I got an MSI MAG 320r forge Airflow gaming case not long ago, So it seemed fitting I put an MSI motherboard in an MSI case, lol, See I did think that the DDR5 ram would be more stable on a CPU that is DDR5, That is the great thing about the PC world is it is always a learning curve, There is always something new to learn, And the combinations of CPU and GPU within different motherboards is endless, I actually do wonder how many different CPU and GPU combinations there is, lol, I do find since I put in the MSI Z690 motherboard, The VR on AMS 2 seems to be more punchier in the graphics if that makes sense, Almost as if the colors are brighter if that makes sense, lol, I have got older vids of Race Room, I just re-downloaded Race Room as I thought Race Room wasn't VR, lol, So now I can see when I do go on Race Room how much of a difference there is, lol......
But yeah I did honestly think that the DDR5 Ram would be more stable with more wiggle room having higher frequencies, Than the DDR4 being lower, Lower is more controllable even if you is a PC then, lol....
Again always something to learn in the PC world, lol.......
That's a great attitude ... I am still learning too, if you stop learning you you stop growing. That's one of the reasons I engage in my comment section, I learn a lot from my subscribers.
@@blackbirdpctech Fair play, When someone takes a few out to reply to comments, Then I subscribe, It is when a You Tuber creator doesn't reply I think why should I subscribe, But I love the attitude you have for this PC stuff, It shows dedication, For when someone is dedicated to a certain subject, Then 9 out of 10 they make top mentors, One mistake I made growing up is finding the wrong mentor's, But then I grew up in the 90s, And it could be said the 90s was a nuttier time to be alive, Hence the term Noughties, lol...
Nor jokes aside life is a constant learning curve, The reason someone becomes wise is from learning over a certain amount of years yeah, And with the PC stuff, We are always going to use computers now, The reason I love VR so much is coz VR can be used as a tool to learn all sorts, Want to learn how to put a car engine together, Get Car Mechanic simulator, Want to learn the insides of a human, There is even things you can do in VR that involves medical operations, lol, VR is even helping people who suffered a stroke, Getting motor functions back using the VR, Oculus does it, So even Meta can see how much practical learning and education a VR headset can bring to your brain, Even poor kids in Africa can have a classroom within a VR headset, I did read there is things being set up so old VR head sets can be programmed with software, To actually bring VR to poor villages, How real it actually is, But isn't that a great idea, Giving poor village children the chance to learn with VR, They can charge the headsets using Solar power as well apparently, Was in the article about using VR as a tool for poor kids to turn a hut into a classroom, I hope it is real and it do happen, Like I said every kid deserves some sort of education, So Amen for Stand alone VR, Plus VR is a brilliant wat to do sim racing, lol....
I will check out some more of your vids over the weekend.....
I love your attitude and I agree that VR/AR can be super helpful as an education tool.
good ideea for a video, but i think that the 4k results are useless in this comparition simply becouse the bottleneck was on the gpu and the diferences between the ddr4 and ddr5 memory kits were 0. Though the 1080p test at low settings indeed is relevant. Not a big diference, i was expecting more. It seems that the latency and the timings are important, not just how many megahertz.
Thanks, the reason that I decided to keep the 4k results in the benchmarks is to show people that are gaming at 4k that it's basically irrelevant what memory kit you game with ... I think it's always good to have the data even if the data simply supports what you thought would happen. Thanks for watching!
@@blackbirdpctech agree i'm one of those gamers that use Blender and game at 4K and was thinking of upgrading but every video I see there is literally not much difference besides maybe 5-10% at best
Bro, I have a 12700KF , ROG STRIX Z690-A GAMING WIFI D4 and GSILL TRIDENT Z 3000 14-14-14-34. I bought a 4070ti last year september. Do you think i need to upgrade motherboard and Ram for a DDR5 system now or should i wait and buy the board and ram later. Its because i will buy a 5070 ti level card when it releases.. By the time 5070ti releases new cpu and board will release. thats why i am asking.. Its is really bad to keep a 4070ti on a ddr4 3000mhz system ? I just use pc for gaming and media.
I would wait and upgrade your system when your new GPU becomes available ... you may wish to consider the new AMD CPU's when they come out. One thing you could do while waiting is pickup a cheap DDR4-3600 CL16 kit but it's not going to significantly increase your FPS.
I got a Ryzen 5 3600 its am4 with 2070 super. I want to buy a 7900 XT. should a get a new am5 motherboard + cpu + ddr5 combo? or just buy Ryzen 7 5700X3D 8-Core am4 and ddr4 ? mostly for 1440. Thanks.
Good question ... you might find this video helpful: th-cam.com/video/JIjD1mZUGz0/w-d-xo.html
In it I upgrade a system with a 3600X/3060 to a 5700X3D/4060Ti. I would get the 7900 XT and a 5700X3D ... the next time you upgrade there will be no choice but to upgrade the entire platform.
I have tight options, pls help.
I want to build new PC, which one is better 12400f ddr4 32gb and rtx 4060, or 7500f ddr5 32gb and rx 6600?
The RTX 4060 will give you more performance. Wrt the memory, you would have to provide additional information (speed, latency).
@@blackbirdpctech standard memory, DDR4 32GB 3200mhz cl 16, and DDR5 32GB cl 36 6000mhz
Would really like to have seen you test with much tighter control over the variables. Things I see that could cause differences in performance, false positives, or otherwise make the data misleading. Different capacity memory, different main storage drive, different windows install, different model and spec of mainboard.
This kind of test really needs to be done with as many variables controlled for as possible. Ideally I would think using the same brand memory and capacity, use the same cpu and cpu cooler, use as close as possible to the same motherboard and ideally you would find the exact same model motherboard with the same chipset and the only difference being ddr4/ddr5.
Rant over I will watch the rest of the video now.
For a perfect test I agree however most of these variables have no meaningful impact on gaming performance. The two components that impact gaming performance the most are the CPU and GPU, and as I mentioned in the video I used the exact same CPU and GPU for each system. I also used Z690 motherboards and made sure that the Bios for each board was setup to enforce all Intel CPU limits. Components like SSDs do not impact gaming performance even when comparing Gen5 versus Gen4 drives (see results from my recent video). For memory the brand is not really relevant, it's the memory die manufacturer that is relevant and even then there are differences between DDR5 and DDR4 that make using the same die manufacturer pointless (i.e. Samsung B-die is best for DDR4 while Hynix A-die is best for DDR5). Also, memory capacity is only relevant when you run out of memory ... for these tests no games required anything close to 32GB of memory, so the additional 16GB of memory for the 48GB kit had no impact.
I have a rog zephyrs 4070. I tried to replace ram. The keyboard rgb and power light turn on. But just a black screen. How to fix?
Did you check the troubleshooting guide in the video? It shows likely sources of your issue so you can hopefully find the problem and resolve it rapidly. For what you describe you should look under the >10sec (turns on, but no display) row.
Why not 790? 690 need bios update for 14gen yeah?
Interesting 4k results.
My DDR4 motherboard was a Z690 board, so I wanted to be consistent with the chipset when comparing performance to help eliminate as much variability in the results as possible. Yes, both Z690 boards required a bios update to recognize the 14900K.
DDR 4 to DDR 5 , would be stupid to change mobo for a overall 13% increase 😂😂😂😂
DDR 6 - okay then 🎉🎉
It’s easy to get sucked in by the hype but when you look at the actual data it usually tells a different story.
So glad i picked up a z790 strix d4 on clearance. 12700k and 48gb ddr4 definitely still doing it's job
Really looking like a smart move given the issues with 13/14th gen CPU’s 😉
if i was about to buy a new pc with a budget of 1000-1000 €. I would rather go with a ddr4 system at 3600mhz with a 4070 than with a ddr5 system with a 4060 and have fun with it for 7 years, right? I dont see the value of ddr5 yet.
Absolutely, you will get a much larger performance boost going with a better GPU virtually every time ... what is the CPU you have and/or are considering buying?
@@blackbirdpctech thanks for the answer. The CPU planned is the AMD Ryzen 5 5500, 6x 3600 MHz.
Have you considered a 5700X3D?
@@blackbirdpctech wow, thats like 30 fps more in all games. It will make the system 130 € more expensive. I have to consider it.
@@gagagag7051 definitely worth considering, it's a great chip.
comment section is the conclusion of this video
My COD Warzone running 20000mb on ddr5 6000mhz, 4k 65" lg oled tv.
Wow great job!
Yeah I guess my 11900k and 3090 are fine where they are for now
I definitely wouldn’t sweat that setup, enjoy!
charts alone done tell the whole story. been on 6400 ddr5 from 3600 and there was differences in certain games. ie getting that extra hit from a win or lose situations. INMO now its more of a choice you want to stay up to date upgrade if u dont care dont.
Appreciate your comment. I think the charts do a pretty good job of telling the story, the part that is tough to show via charts is how an increase in 1% low "feels" in games ... that might actually be an interesting topic for an upcoming video. The main point I wanted to get across for people still on systems with DDR4 is that you should wait for next gen chips/motherboards to be released before upgrading. Thanks for watching!
I would love to see 4 sticks of DDR4 4000mhz vs the DDR5 7200 because DDR4 is more stable and runs perfectly fine with 4 sticks where DDR5 if you use 4 sticks and you set the XMP on you will just keep crashing
DDR4 at 4000 with 4 dimms will still be a challenge to get stable. As DDR5 matures it’s getting easier, for example it’s relatively easy to get 4 dimms of 6000 stable now.
@@blackbirdpctech Strange I actually got rid of my maximus hero z790 with Cl32 6400 mhz SWAPPED it out for a Z790 strix ddr4 gaming feels so much better no more dsync in shooter games
You may have been unlucky with the silicon lottery for your IMC
@@blackbirdpctech maybe I started with the 12900k and a z690 maximus gskill 6000 then 13900KF Z790 Maximus 6400 Gskull. could be just me but Iv always felt that DDR4 is better in games even though DDR5 has fps.
With respect to stability you are spot on, but as DDR5 matures and the platforms that use them mature, it will get much better. Remember DDR4 has been around for a while and is a very mature solution at this point.
should have tested esports titles like pubg ,cs2 ,fortnite , rainbow 6 ,warzone 2,
I typically don’t benchmark games that do not provide consistent results between runs … that said I wouldn’t expect the relative performance differences in these games to vary too much from the averages shown in the video.
So what about z890 MBs?
That's a great question and I just finished testing and optimizing the 265K, so I can say confidently that the Z890 motherboards have much better memory support than Z790, and it's not even close. In addition, the memory controller on Core Ultra Series 2 chips appears to be much better as well. So even without using CUDIMMS, you should be able to easily get DDR5-8000 stable on a 4-dimm Z890 motherboard. It's shocking to me that no one else is talking about this because it's a significant improvement.
@@blackbirdpctech Which brings up the question, which set to buy? I've only seen up to 8000 MT on an amazon search. Most people say I should get a lower tier, but I need something to help me make the best decision. What about timing and latency, are they still valid. Would you do another video?!?
@@darkman237 I just completed one for AM5, so I should do another one for Intel Arrow Lake. Intel APU's scale really well with memory, so I would recommend getting a fast kit.
@@blackbirdpctech Thank you! I'll be looking for it.
nice video
Thanks!
Isn't it a bit of a scam on part of the cpu/memory/mother board manufacturers that they advertise based on the memory speed with expo/xmp while on the same time don't guarantee these speeds/timings are achievable/stable. Honestly reviewers should just benchmark with jdec settings until the manufacturers start to guarantee that these speeds/timings are actually achievable.
Appreciate your comment. I recently watched a video explaining JEDEC memory standards ... apparently when they set the standard they have to guarantee that it will work in every possible scenario, which would include super hot environments with minimal airflow, server rooms, edge devices, industrial applications, etc. which is why the actual speeds that they guarantee are relatively low. That said, what I think is a scam is when motherboard manufacturers market their boards for high speed memory when there is basically no chance of achieving that speed. So when all of the big manufacturers (Asus, MSI, Gigabyte, etc.) marketed their Z790 4-dimm motherboards at speeds of 8000 and above this was BS ... there is absolutely no way those boards can run those speeds stable. The problem is people buy the boards, build their systems, install the RAM and just lower the speed to get it stable because it's too much trouble to RMA the board. The manufacturers know this ... that's why they will keep doing it. I highlighted this terrible business practice in one of my earlier memory videos. What we need are the large influencers to call it out, people like Linus and Jay ... the problem is they have no idea what they are doing most of the time ... Jay couldn't even get his AM5 system to run EXPO at 6000.
@@blackbirdpctech Exactly. They should market their products at the speeds that are achievable in normal conditions (no golden sample cpus, memory with specific dies (that aren't even clearly stated on the packaging and can change from one batch to another of the same memory), only two ram slots occupied, ...). If the cpu/memory/motherboard is advertised as supporting speed x you should be able to just buy it, plug it in and have it work. If it doesn't it should be considered faulty and replaced by the store you bought it at or a full refund issued.
Agreed ... at a minimum they should provide conditions under which the advertised speed is likely ... so they could say the speed is 8000 however it requires a 2-dimm motherboard and your probability of reaching that speed is 50% but 7600 with the same kit is 90% ... perhaps even develop a standard that states the advertised speed is guaranteed to function on 85% of motherboards, or something like that. What they are doing now is wrong.
👍 !!! 👍 !!! 👍 !!!
Thanks for watching!!
Got my Sub for Bruce :D Just kidding. Awesome comparison. Will stick to my 3600 DDR4 for a while i guess :D
But tbh the comparison isn´t really fair.
3600 is the sweetspot of DDR4 but 7200 ist ultra high end for DDR5 already. 6000 is the sweetspot for DDR5 and should be the comp here.
Thank you and welcome to the Blackbird PC Tech community!
@@blackbirdpctech 7200 is not ultra high end :) That would be like 8400 cl36. 7200cl34 kits are sweetspots for DDR5 now.
@@jiri281 agreed
Wait how did Bruce Buffer end up on here smh lol
I reached out to his team and he was awesome about it … extremely nice guy!
Ddr5 is super meh atm 😂 probably in one or two years with something like 8000mts CL32 it will be a more worthy upgrade.
There is performance to be gained, but there is just not a lot of value ... it's definitely not something I would upgrade an entire platform for.
Testmem5 is free and finds errors very fast too
Great add ... I've played around with lots of different memory testing tools and the reason I recommend Karhu is that it's super simple to use and finds errors quickly ... the only downside is that it's not free.
@@blackbirdpctech yeah this one has similar interface to Karhu
until you hit the 4090 no it doesn't matter
I’m not sure if I understand your statement … if anything a 4090 will rely less on RAM because it has a large amount of onboard VRAM.
@@blackbirdpctech ddr4 you lose 15% its to slow for the 4090
I tested with a 4090 so I don’t understand your statement … there is no doubt that ddr5 improves your 1% lows but you only see a difference in gaming at lower resolutions when the cpu is more heavily loaded. So this means the gpu model has no impact, which makes sense … you will only see an impact when your system loads your cpu more than your gpu.
I call BS on the Avatar 4k result - your take is that at 4k the game becomes more latency dependant and thus the ddr4 kit wins - however, the 1% is still higher on the ddr5 kit? How do you explain that?
What I said in the video was that at 4K the DDR4-3600 kit beat the DDR5-7200 kit in at least half the titles, which would indicate that in these titles where the GPU is highly loaded, latency becomes more important to extracting maximum performance. The 1% low average for the DDR5 kit is higher than the DDR4 kit, but not by a meaningful amount (
And then we have the 7800X3D that does Not care about Ram speed, completely destroys the 14900K and only draws 50 watts 😎😎💪💪
I tested the 7800X3D with DDR5-6000 and DDR5-8000 memory and there is a performance benefit, particularly in the 1% lows however the behavior is different when compared with Intel systems due to the 3D V-cache and infinity fabric:
th-cam.com/video/p21bsVYAgvA/w-d-xo.html
Also, it's impossible to properly test DDR4 vs DDR5 for AMD systems because AM4 only supports DDR4 and AM5 only supports DDR5, so there is no way to use the same CPU for testing ... that's why I tested on LGA1700 with a 14900K.
By the way, these new Bios settings that Asus just rolled out will negatively impact the performance of the 14900K ... so I'm thinking of doing a comparison with the 7800X3D ... will be interesting to see how they compare now that the 14900K doesn't have an infinite power limit 😉
@blackbirdpctech I also have done tests with my 7800X3D with various different ram, and the benefits are low on faster Ram, the real benefit for higher 1% lows on X3D is lower Clock Timing and more ram getting 30 to 50% higher 1% lows, the best is 2 dims at 32GB DDR5 for a total of 64GB at 6000MT CL 30 and 64GB of 5600MT CL28 which is what I am rocking right now, it also improves better stability, less stuttering and more stable frame rates.
I was watching buildzoid and he was playing around with timings on a 7900X (non x3d) … I am interested to see if I can get 6000 stable at CL28 on the 7800 x3d … would be interesting to see the performance impact.
@blackbirdpctech Do the test so you can see show the benefits and differences, G-Skill Ripjaw Sells on newegg the 32 and 64GB 5600MT CL28 on Newegg. I am not a content creator, but it would be cool if more youtubers like yourself show the difference with more ram and lower Timings compared to faster Ram but higher Timings.
this is 32GB DDR4 against 48GB DDR5 not really apples to apples?
The amount of RAM has absolutely no impact on performance unless you don’t have enough. None of the benchmarks I ran required anything close to 32GB of memory.