Overclock.net - An Overclocking Community - Reply to Topic

Thread: [Digital Foundry] Memory Overclocking and How it Affects FPS in 8 Different Games Reply to Thread
Title:
Message:

Register Now

In order to be able to post messages on the Overclock.net - An Overclocking Community forums, you must first register.
Please enter your desired user name, your email address and other required details in the form below.
User Name:
If you do not want to register, fill this field only and the name will be used as user name for your post.
Password
Please enter a password for your user account. Note that passwords are case-sensitive.
Password:
Confirm Password:
Email Address
Please enter a valid email address for yourself.
Email Address:

Log-in


  Additional Options
Miscellaneous Options

  Topic Review (Newest First)
01-21-2016 04:10 AM
TheBloodEagle Are there similar benches for content creation? I'd love to know more about RAM sped in regard to Adobe Premiere, 3D modeling, large databases, etc.
01-20-2016 01:35 PM
Tivan
Quote:
Originally Posted by dubldwn View Post

Yeah that's what I'm looking at. Anand did a test that's a little old now that showed margin of error differences.

http://anandtech.com/show/8959/ddr4-haswell-e-scaling-review-2133-to-3200-with-gskill-corsair-adata-and-crucial/6

MSI GTX 770 Lightning+maximum settings in GPU heavy games

Doesn't seem like they stressed the CPU much.
01-20-2016 12:08 PM
dubldwn
Quote:
Originally Posted by raclimja View Post

Is there any memory overclocking test on X99 Platform to see how it affects FPS in games?

I am wondering as X99 CPU's have much high cache and Quad Channel Memory instead of Dual Channel Memory.

Yeah that's what I'm looking at. Anand did a test that's a little old now that showed margin of error differences.

http://anandtech.com/show/8959/ddr4-haswell-e-scaling-review-2133-to-3200-with-gskill-corsair-adata-and-crucial/6
01-20-2016 08:41 AM
Silent Scone
Quote:
Originally Posted by Cyro999 View Post

Sure, but most of these games also respond to memory performance differences last gen. And the gen before. And the gen before that - and that's just talking major gens, not die shrinks.

It's not a straight fire question. Different games and different engines push these subsystems further. Others don't have quite as big a bottleneck, so gains will never be entirely linear. You can all discuss it to you're blue in the face, but the reality of it is that it's application dependant. There are however quite obviously gains to be had.
01-20-2016 07:49 AM
Cyro999
Quote:
These new platforms are very different.

Sure, but most of these games also respond to memory performance differences last gen. And the gen before. And the gen before that - and that's just talking major gens, not die shrinks.
01-20-2016 07:38 AM
Silent Scone His problem is explained in his own post, he said many years ago. These new platforms are very different.
01-20-2016 07:28 AM
Cyro999
Quote:
I'm having hard time believing those kinds of FPS gains due to memory speed in games. Memory speed being a serious influence on performance stopped many platforms ago

What's your source on that other than feeling? Reporting has been a bit biased against memory gains because people typically don't test in scenarios friendly to them.

If you have a game that's always quite limited by the GPU, performance differences between even different CPU's are hidden. RAM is usually smaller. If you benchmark RAM performance with some games that are traditionally CPU limited, you've been able to see gains easily for a long time.

It's very hard to benchmark MMO's to show a 5-10% performance gap. For other genres though, there are games like Starcraft 2 which are not highly GPU bound which are easy to benchmark (completely deterministic so you can get 5 results within 0.1% of eachother).

SC2 and some other games like it have shown significant gains from memory for A LONG TIME; First gen i7, second gen, third, fourth, sixth (probably 5'th too) have shown gains equivelant or better to gaining 200mhz on the CPU core if you go from the cheapest RAM to the little-more-expensive-but-still-really-cheap RAM. For example with Haswell, that was 2133c9 instead of 1600c9. With Skylake, it's 3000c15 instead of 2133c15.

It's not much performance but it's notable. Enough to make upgrading RAM better perf/$ and better absolute performance than doing something like running a high end cooler instead of low-midrange for those games. Many users on OCN have posted solid scientific tests backing RAM performance having this impact on games. You have these videos from Digitalfoundry which seem solid too and say the same thing that i've seen everywhere since i really got into tech 5 years ago - zero gains in some games, notable gains in others. I can't speak much for memory performance and gaming in the days of core 2, phenom I and previous.
01-20-2016 05:29 AM
Silent Scone
Quote:
Originally Posted by Danzle View Post

If i understand correctly one would have to find the sweet spot of timing/frequency to get the highest performance increase in games.

So DDR4-3000, CL15-15-15-35 = DDR4-3733, CL17-17-17-37 ?~

3466 is probably the optimal frequency on SKL
01-20-2016 03:23 AM
MoGTy
Quote:
Originally Posted by STEvil View Post

Dont bother paying attention to windows.
In general it is, if you dont work on your timings finding what is optimal for your setup. For example Winbond BH-5 back in the day could run 260-300mhz at CAS2 and keep TRCD and TRP low (usually 2, some kits could run TRP at 1). CH-5 chips required TRCD of 3 usually, except some that were found in the "re-run" sets and they didnt respond to voltage as well either... so basically it was best to run 2-3-2-5/6 for CH-5 and 2-2-2-5/6 for BH-5/6 then push the mhz as high as you could. But if you didnt mind losing a few mhz top end you could manage better bandwidth by adjusting TRCD-W and TRCD-R separately and pushing TRP to 1, if your board supported it.

I remember running my old BH-6 at over 276mhz 1.5-2-1-1-5 (CAS-TRCD-w-TRCD-r-TRP-TRAS). It took 4.0v to get there but the bandwidth and latency were on par with BH-5 kits pushing nearer 290mhz.

good old days..

edit - may have TRCD-R and TRCD-W reversed in the "proper" order, been way long since memory timings were something a lot of people took time to work with.

And here I am, lowering timings like the good old days. redface.gif

My kit just doesn't like higher frequencies tongue.gif
01-20-2016 02:00 AM
STEvil
Quote:
Originally Posted by bl4ckdot View Post

Hi,
Quick question. I bought a shiny new kit of 2400 CL11 DDR3 to upgrade my 1600 CL9 "value" RAM. I set everything in the BIOS, however CPU-Z and Windows are telling me different thing. Windows sees the memory speed at 1600 when CPU-Z tells me 2400. I would believe CPU-Z more, but how to be sure ?


Dont bother paying attention to windows.
Quote:
Originally Posted by MoGTy View Post

Indeed. It's no exact science, however where did you read frequency has become more important overall?

In general it is, if you dont work on your timings finding what is optimal for your setup. For example Winbond BH-5 back in the day could run 260-300mhz at CAS2 and keep TRCD and TRP low (usually 2, some kits could run TRP at 1). CH-5 chips required TRCD of 3 usually, except some that were found in the "re-run" sets and they didnt respond to voltage as well either... so basically it was best to run 2-3-2-5/6 for CH-5 and 2-2-2-5/6 for BH-5/6 then push the mhz as high as you could. But if you didnt mind losing a few mhz top end you could manage better bandwidth by adjusting TRCD-W and TRCD-R separately and pushing TRP to 1, if your board supported it.

I remember running my old BH-6 at over 276mhz 1.5-2-1-1-5 (CAS-TRCD-w-TRCD-r-TRP-TRAS). It took 4.0v to get there but the bandwidth and latency were on par with BH-5 kits pushing nearer 290mhz.

good old days..

edit - may have TRCD-R and TRCD-W reversed in the "proper" order, been way long since memory timings were something a lot of people took time to work with.
This thread has more than 10 replies. Click here to review the whole thread.

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off