• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

Official Neogaf 3DMark Timespy league table - the ULTIMATE epeen showdown

Celcius

°Temp. member
New score for my RTX 3090

+150 core
+500 memory
107% power limit
80% locked fan speeds

20,407

POeK7m8.jpeg


During the previous run I realize I forgot to remove my 80% power limit cap lol
 
Last edited:

baphomet

Member
How do you figure? Are you talking about the 10k CPU difference? My gpu score isn't 10k lower. Speedway it says above average too for 5090s. The GPU score in timespy doesn't scale very well, Timespy Extreme was created for newer GPUs so that it doesn't hit other limits. A near 1/3 clock boost resulted in only 700 more points on 46,000 on old Timespy with a 5090. My GPU is hitting 400fps and showing 80% in those sections. It's clear it's hitting some other limit.

Yes, there are 5090s scoring 10,000 above you in graphics score. If you're hitting 400fps and your GPU is at 80%, then you're likely CPU bound.
 

Three

Member
Yes, there are 5090s scoring 10,000 above you in graphics score.
Who is getting 10k more and at what clock? the Hall of famers? Its possible by addressing the other bottlenecks but not with high temp clocks on the GPU.
If you're hitting 400fps and your GPU is at 80%, then you're likely CPU bound.
Exactly that's the point. The GPU test is being limited by other things that aren't the GPU. It's not the CPU either because that is not hitting 100% either at those moments. It appears to be memory and poor cpu utilisation, something else that Timespy Extreme addresses over Timespy.

 
Last edited:

baphomet

Member
How do you check individual GPU scores of others? Who is getting 10k more and at what clock. Its possible by addressing the other bottlenecks but not with high temp clocks.

Exactly that's the point. The GPU test is being limited by other things that aren't the GPU. It's not the CPU either because that is not hitting 100% either at those moments. It appears to be memory and poor cpu utilisation, something else that Timespy Extreme addresses over Timespy.


Again, it isn't the test.

If neither your CPU or GPU are hitting 100%, then you've got issues elsewhere. These tests are going to hit the bounds of one or the other.

https://www.3dmark.com/spy/53975570
Obviously this is a nitrogen cooled run, but the point is that the test scales with the hardware. The graphics score is at the top under the combined score. 56,106 for the graphics score.
 

Three

Member
Again, it isn't the test.

If neither your CPU or GPU are hitting 100%, then you've got issues elsewhere. These tests are going to hit the bounds of one or the other.

https://www.3dmark.com/spy/53975570
Obviously this is a nitrogen cooled run, but the point is that the test scales with the hardware. The graphics score is at the top under the combined score. 56,106 for the graphics score.
Well obviously there are other bottlenecks to the test but what you said was

"Time Spy will max your GPU, that's the entire point of it. It should stay at 100% and pump out as many frames as it possibly can."

And I said it doesn't hit 100% on the GPU.
The point of the GPU test is to test your GPU and not be limited by other things. Timespy became dated and hit other bottlenecks quickly with newer cards and so they created Timespy Extreme so that the GPU is limited in the test again. You can increase the GPU clock by 1/3 and score only 700 points more in your GPU score. Why? Because you're not limited by your GPU in that GPU test.
 
Last edited:

baphomet

Member
Well obviously there are other bottlenecks to the test but what you said was

"Time Spy will max your GPU, that's the entire point of it. It should stay at 100% and pump out as many frames as it possibly can."
And I said it doesn't hit 100% on the GPU.
The point of the GPU test is to test your GPU and not be limited by other things. Timespy became dated and hit other bottlenecks quickly with newer cards and so they created Timespy Extreme so that the GPU is limited in the test again.

I literally explained all of it to you.

You seemed to think that your GPU was maxing out Time Spy, which it wasn't. Then you also said it isn't maxing out your CPU. Which would mean you've got something going on if a stress test isn't actually stressing your system.
 

Three

Member
I literally explained all of it to you.

You seemed to think that your GPU was maxing out Time Spy, which it wasn't. Then you also said it isn't maxing out your CPU. Which would mean you've got something going on if a stress test isn't actually stressing your system.

I'm not sure what we're discussing then. It seems like we're saying the same thing? I said the benchmark isn't maxing out the GPU not that the GPU is maxing out some kind of fps hard limit on timespy:

What settings do you mean? it's a benchmark issue I believe. Timespy is just kind of dated and they released Timespy Extreme in 2017 to fix some of the issues in the older Timespy benchmark. On timespy's percentage graph you often get sections where it's using less than 80% of GPU and CPU in the GPU test. On speedway it maxes out at 100% constantly for GPU.

Time Spy will max your GPU, that's the entire point of it. It should stay at 100% and pump out as many frames as it possibly can. If Your CPU is hitting 100% trying to push out 400+ fps then your GPU may still have some headroom.

If you were maxing out the test there wouldn't be other 5090's scoring ~10,000 higher than you in graphics. Of course those are probably liquid nitrogen cooled and boosted super high, but they're not even maxing out the test.

I said it's not maxing the GPU or CPU on newer cards. It's not nitrogen cooled overclocked GPUs since its GPU test doesn't scale well with GPU potential. I said clocking a 5090 30% higher gets you less than 2% increase in GPU score because the benchmark has issues scaling and they addressed some of its issues in Timespy Extreme. Only newer GPU benchmarks like Speedway 100% the GPU. Makoto-Yuki Makoto-Yuki has the very highest end CPU yet his 5090 doesn't hit 100% either. Timespy itself has problems utilising the GPU and CPU properly, issues they tried to address in Extreme.
 
Last edited:

Makoto-Yuki

Gold Member
I'm not sure what we're discussing then. It seems like we're saying the same thing? I said the benchmark isn't maxing out the GPU not that the GPU is maxing out some kind of fps hard limit on timespy:





I said it's not maxing the GPU or CPU on newer cards. It's not nitrogen cooled overclocked GPUs since its GPU test doesn't scale well with GPU potential. I said clocking a 5090 30% higher gets you less than 2% increase in GPU score because the benchmark has issues scaling and they addressed some of its issues in Timespy Extreme. Only newer GPU benchmarks like Speedway 100% the GPU. Makoto-Yuki Makoto-Yuki has the very highest end CPU yet his 5090 doesn't hit 100% either. Timespy itself has problems utilising the GPU and CPU properly, issues they tried to address in Extreme.
should we have done the 4K time spy instead?

other benchmarks have no problem hitting 100% usage but time spy will go to as low as ~79%. my RAM is 6000 CL32. i've turned all virtualisation off and disabled v/g-sync on software and hardware level. nothing i've tried makes my GPU hit 100% in the basic time spy. i've done countless clean installs with DDU across the latest official WHQL driver and the latest hotfix driver.

steel nomad hits 100% all the way through but i don't know if that is a GPU only test.

EDIT: done time spy 4k. stayed 99-100% gpu usage. if we want to see who has the most powerful PC then it makes more sense to run benches with at least 4K imo 🤷‍♂️

 
Last edited:

Celcius

°Temp. member
Oof, I'm not even top 40
I guess that's what happens when your videocard is two generations old though
RIP
 
Last edited:

//DEVIL//

Member
I just remembered that my test was done with silent bios on not performance. I know the performance Bios on Astral has a higher clock. but I am too lazy to do the test again. happy with my current score.

Also. the new Bios update for Astral ( silent bios update ) is great. i don't hear the card at all now. Debaur still not happy but I am lol.
 
Last edited:

keefged4

Member
Oof, I'm not even top 40
I guess that's what happens when your videocard is two generations old though
RIP
Given the prices of the latest cards, do you really need a higher score? I am near the bottom of the rankings and can play everything I want to at 4k using DLSS, sometimes not, with framerates well above 60 at all times. So it doesn't bother me, I can only imagine the outlay involved for some of the larger epeen scores.
 
Top Bottom