Cheap vs. Premium DDR5: Stock vs. EXPO/XMP Memory

That's interesting cus I was wondering about the same thing too! Idk why, but going from the base speed to EXPO gave me a huge fps boost during benchmarks for Assassins Creeds ,Cyberpunk and Tomb Raider. I use the 7800X3D too. Then again, my 6000 timings are probably very tight, cus I took a long time to find the max before crashing. I have yet to find a game that does not give me a big advantage with the faster speed.

It does make me wonder about the results. Many sites and YouTubers show different results with the same hardware. Some have big fps boost like mine, while others usually have a tiny one (like 1-5% at most) What could be the reason? Its hard to know what to buy if the results for every site and person are all over the place lol. One thing is for sure, if I buy a new platform, I sure won't go to the lowest speeds lol, even if it costs a bit more over the other kits.

Thanks for the tests as usual btw, I love these, especially the 1 from before with the CPUs! Keep up the good work! I was just wondering about this specific topic, cus of my personal experience ;p

Is this correct?

"performance of 7700X was improved by ??3x?? more using the same DDR5-6000 memory with a 12% improvement to the average frame rate."

" For those who consider themselves PC enthusiasts but don't understand why reviewers test at this resolution, please check our our explainers here. These articles explain the basics of CPU benchmarking and why it is done in a certain way. Testing at 1440p or 4K would focus on the RTX 4090's performance, making the data less relevant for the subject at hand."

That's a rather unsubtle shade. It's a bit disappointing to see this kind of patronizing attitude in the article proper, not in the comments as it usually is. But ok, I will respond in kind, so...

...for those who consider themselves tech journalists, but don't understand why many people complain about the methodology used, here's a little explainer.

As it happens, I'm pretty sure most of us understand very well the reasoning behind your choice of testing methods. It's not exactly rocket science, so there's no need to wheel out this "they just don't get it" trope every time these tests happen.

What we can't fathom though is why do you think this methodology helps anybody but a handful of people interested in hardware theory? Because that's precisely what testing only with 4090@1080p is good for: theory. It is at best a waste of time for the most of us, at worst it can be misleading for people whose buying choices are influenced by these articles. Your byline asks "Should Gamers Buy OC RAM?" and its a great question, unfortunately your article completely fails to show whether they should or shouldn't, since it's using only one, quite improbable to boot, hardware use case.

Your results might or might not apply to the wider spectrum of setups but quite probably it might also be a complete wash (given the well known distortion the extreme combination such as 4090/1080 produces) - we just don't know that. And unless these kinda tests will start including more realistic setups (even if the results they produce are considered "boring" by the "real enthusiasts" ) then we will continue to be disappointed by these articles.

Nictron said:

Is this correct?

"performance of 7700X was improved by ??3x?? more using the same DDR5-6000 memory with a 12% improvement to the average frame rate."

Click to expand...

Yes, because the improvement was 4% for the 7800X3D. 12% is 3x 4%

passingposeidon said:

Yes, because the improvement was 4% for the 7800X3D. 12% is 3x 4%

Click to expand...

Thanks, missed that.

10% improvement is negligible .

zx8086 said:

" For those who consider themselves PC enthusiasts but don't understand why reviewers test at this resolution, please check our our explainers here. These articles explain the basics of CPU benchmarking and why it is done in a certain way. Testing at 1440p or 4K would focus on the RTX 4090's performance, making the data less relevant for the subject at hand."

That's a rather unsubtle shade. It's a bit disappointing to see this kind of patronizing attitude in the article proper, not in the comments as it usually is. But ok, I will respond in kind, so...

...for those who consider themselves tech journalists, but don't understand why many people complain about the methodology used, here's a little explainer.

As it happens, I'm pretty sure most of us understand very well the reasoning behind your choice of testing methods. It's not exactly rocket science, so there's no need to wheel out this "they just don't get it" trope every time these tests happen.

What we can't fathom though is why do you think this methodology helps anybody but a handful of people interested in hardware theory? Because that's precisely what testing only with 4090@1080p is good for: theory. It is at best a waste of time for the most of us, at worst it can be misleading for people whose buying choices are influenced by these articles. Your byline asks "Should Gamers Buy OC RAM?" and its a great question, unfortunately your article completely fails to show whether they should or shouldn't, since it's using only one, quite improbable to boot, hardware use case.

Your results might or might not apply to the wider spectrum of setups but quite probably it might also be a complete wash (given the well known distortion the extreme combination such as 4090/1080 produces) - we just don't know that. And unless these kinda tests will start including more realistic setups (even if the results they produce are considered "boring" by the "real enthusiasts" ) then we will continue to be disappointed by these articles.

Click to expand...

I saw that too and just shook my head because the tester didn't even go into any kind of depth on the IC manufacturer or timings. These memory kits are most likely using the same exact memory chips. Timings can have a greater effect than transfer rates. You need to shop the IC manufacturer not just the memory manufacturers and speeds alone.

Other places test at all 3 resolutions to show you there is zero difference at 4k between DDR5-4800 and DDR5-8000. Articles that don't test realistic scenarios and don't even go into any depth, are just ad clicks.

What is someone who works in 4k supposed to take out of these tests? Skip them I guess.

Starfals , let s assume memory contributes 25% (1) to the overall 3D performance , and also compare 4800MT/s DDR5 to 6000MT/s DDR5 . We ll get 25%(1) x 25% ( mem dif) = roughly 6.25%

zx8086 said:

What we can't fathom though is why do you think this methodology helps anybody but a handful of people interested in hardware theory? Because that's precisely what testing only with 4090@1080p is good for: theory. It is at best a waste of time for the most of us, at worst it can be misleading for people whose buying choices are influenced by these articles.

Click to expand...

this is a scientific way of testing the impact of cpu/memory on gaming. I don't think you were elected to represent any group of readers, so I don't get why you keep referring to yourself as a collective. If you find the methodology does not suit your interest, you are free to not go through the article, so no waste of time. It is clear that running a game in a more balanced configuration you will see a smaller difference between memory configurations running current games, so what would pe the point of testing it?