Support NeoGAF

[Celcius]

New score for my RTX 3090

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

20,407

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

 

[baphomet]

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]

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.

3DMark - Time Spy Extreme is here - Steam News

Today's big 3DMark update adds a new DirectX 12 benchmark test for 4K gaming PCs. Time Spy Extreme is a benchmark for gamers who want to play at the highest settings, system builders configuring 4K-capable machines, and overclockers pushing the latest hardware for every bit of extra performance...

store.steampowered.com

 

[baphomet]

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.

3DMark - Time Spy Extreme is here - Steam News

Today's big 3DMark update adds a new DirectX 12 benchmark test for 4K gaming PCs. Time Spy Extreme is a benchmark for gamers who want to play at the highest settings, system builders configuring 4K-capable machines, and overclockers pushing the latest hardware for every bit of extra performance...

store.steampowered.com

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.

 

[KyoZz]

can't downlaod free version why?

Just download the 3D Mark demo and launch it, Timespy should be installed by default

 

[Three]

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.

 

[analog_future]

GHG I don’t think your edit is correct. R3TRO is over 40k now and I’m over 39k

 

[baphomet]

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.

 

[GHG]

GHG I don’t think your edit is correct. R3TRO is over 40k now and I’m over 39k

I'm going through the posts one by one, still have quite a few to get through.

 

[Three]

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 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.

 

[Makoto-Yuki]

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 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

I scored 22 991 in Time Spy Extreme

AMD Ryzen 9 9950X3D, NVIDIA GeForce RTX 5090 x 1, 65536 MB, 64-bit Windows 11}

www.3dmark.com

 

[Rossco EZ]

Did another run earlier with a better score.
Rossco EZ - TimeSpy - 20,168

I think that's the best ill get out of my 4070S FE.

 

[GHG]

Table should be all up to date now. If I've made any mistakes or missed anything please let me know.

Been a chaotic weekend though huh? RIP OverHeat 's motherboard.

I appear on that list twice...

 

[Celcius]

Oof, I'm not even top 40
I guess that's what happens when your videocard is two generations old though
RIP

 

[Rossco EZ]

Oof, I'm not even top 40
I guess that's what happens when your videocard is two generations old though
RIP

as long as you can still play the stuff you want at reasonable settings thats all that matters

 

[OverHeat]

Ordered a Asus X870e e gaming wifi mobo should be here Monday or Tuesday.

 

[GHG]

OverHeat right now:

 

[Kilau]

I knew it would be all over for us AMD bros once the i9s showed up.

 

[//DEVIL//]

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.

 

[applesaucejaxon]

applesaucejaxon - Timespy - 18,519

Ehhh... Was expecting worse, so I can't complain.

 

[moogman]

applesaucejaxon - Timespy - 18,519

Ehhh... Was expecting worse, so I can't complain.

I wouldn't be able to live with that graphics score. I'd have to close one app down to get from 19,999 to 20,000

 

[keefged4]

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.

 

[analog_future]

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.

Just depends on a person's budget and what they want to get out of their experience.

4070 is a great card, but many (including myself) don't want to play at 4K DLSS Performance/60fps with no path-tracing or other ultra+ settings. I want to play, at minimum, no worse than 4K DLAA or DLSS Quality/90fps, everything maxed out. And I have the funds to do so, so why not?

Spoiler

Would never pay scalper prices though. Fuck scalpers

 

[notseqi]

Oof, I'm not even top 40
I guess that's what happens when your videocard is two generations old though
RIP

maybe you will find this chart helpful:

 

[DenchDeckard]

Reinstalling 3d mark now. I'll post my results shortly!

Not going to overclock anything. Just run the standard settings.

Well, apart from my memory has xmp enabled.

 

[keefged4]

Just depends on a person's budget and what they want to get out of their experience.

4070 is a great card, but many (including myself) don't want to play at 4K DLSS Performance/60fps with no path-tracing or other ultra+ settings. I want to play, at minimum, no worse than 4K DLAA or DLSS Quality/90fps, everything maxed out. And I have the funds to do so, so why not?

Spoiler

Would never pay scalper prices though. Fuck scalpers

All the power to you man, I hear ya. If I had the funds I'd defo do the same. DLSS4 Performance is wizardry though, I'm able to hit 90fps+ in most games I play. I will most likely mean I'll be sticking with my 4070 for a while.

 

[DenchDeckard]

Ok!

DenchDeckard | TimeSpy | Score: 32 863

GHG please add me to the table me ol' chum.

 

[GHG]

All up to date, interesting to see the amount of distribution here above the 20k score mark.

 

[DenchDeckard]

This was dangerous GHG .....I'm now thinking of pushing my system to the limit to break into the top 10.... RIP

 

[KyoZz]

I'm 16th??

 

[reinking]

maybe you will find this chart helpful:

It will help if I can find a new card at retail prices.

 

[Unknown Soldier]

Oof, I'm not even top 40
I guess that's what happens when your videocard is two generations old though
RIP

I'm #50 with a video card which cost me $1600 about 5 years ago

The march of technology is relentless, and the many people here and on Reddit who scream about Nvidia not improving their products and charging more money are retarded when in reality Nvidia has delivered constant gen-over-gen improvements. At the top end, a 5090 is roughly twice as fast as my 3090 in raster and probably 5 times faster in RT

 

[skneogaf]

It seems I'm in need of an upgrade!

22473 - Timespy

Graphics score of 32301 for a 4090 is meh for thus gpu.

CPU score of 8250 is yikes and expected.

My cpu doesn't affect my 4k max settings gaming if I'm honest, maybe 1 percent lows but I'm always gpu bottlenecked.

Maybe I'll grab a 9800x3d! Which is the best mobo and ram for this cpu, Any suggestions?