3090 vs 2080 Ti expected performance difference?

[mrthanhnguyen]

Want a faster 3090? Put a waterblock on it and shunt mod

I scored 15 179 in Port Royal

Intel Core i9-10900K Processor, NVIDIA GeForce RTX 3090 x 1, 32768 MB, 64-bit Windows 10}

www.3dmark.com

I scored 21 586 in Time Spy

Intel Core i9-10900K Processor, NVIDIA GeForce RTX 3090 x 1, 32768 MB, 64-bit Windows 10}

www.3dmark.com

 

[erek]

That's how I run all my cards never had any issues but I upgrade pretty frequently.


I don't run my fans at 100% though I just set up a custom profile to a level I'm comfortable with noise wise

Typically 40db ish

Decent?

Result

www.3dmark.com

 

[oxrufiioxo]

Decent?

Result

www.3dmark.com

Now you're in the ballpark. Although gaming performance is what matters not synthetics. I would enjoy my card and not be overly worried about a couple percentage points in synthetics.

 

[X71200]

Want a faster car? Take out your hood plastic, put a bigger turbo, toy with the ECU and there you have a faster car. Does it matter when it is already fast, and is it worth it to botch up your card with some crap soldered by yourself to the back of your GPU? Reliability, warranty, hello? Shunt mods are pointless unless you're extreme overclocking. Good luck with selling the card.

I think you really need to figure out some stuff. Are you idling at 45C with your GPU fans at full blast? Why are you directing a HOUSE fan at your case? You are doing it wrong. None of that is necessary.

You need to fix your airflow, and if needed be, adjust fan directions of your computer fans - so you get a better, balanced airflow. Your card also has a temp limit next to the power limit and a performance curve accordingly. You could increase that if you're not getting there, but seems to me like you may have been overheating.

 

[erek]

Want a faster car? Take out your hood plastic, put a bigger turbo, toy with the ECU and there you have a faster car. Does it matter when it is already fast, and is it worth it to botch up your card with some crap soldered by yourself to the back of your GPU? Reliability, warranty, hello? Shunt mods are pointless unless you're extreme overclocking. Good luck with selling the card.

I think you really need to figure out some stuff. Are you idling at 45C with your GPU fans at full blast? Why are you directing a HOUSE fan at your case? You are doing it wrong. None of that is necessary.

You need to fix your airflow, and if needed be, adjust fan directions of your computer fans - so you get a better, balanced airflow. Your card also has a temp limit next to the power limit and a performance curve accordingly. You could increase that if you're not getting there, but seems to me like you may have been overheating.

having issues with stability in GTA V... boosts are hitting 1950+

i'm back to stock power limits too, so 100%

after some time in GTA V (only, so far)

monitor goes blank, gpu fans ramp up to 100%

have to hard kill the power and reboot it

 

[X71200]

I'd suggest reducing clock settings and capping your framerate, in that case.

 

[erek]

I'd suggest reducing clock settings and capping your framerate, in that case.

i set 60 fps (vsync'ed) also clock settings and everything were reduced to totally out of the box...


it seems to be primarily limited to GTA V, haven't been able to reproduce it yet even in Watch Dogs: Legion

 

[X71200]

If you're using mods or such things in GTA V, that might be why. I'd also suggest using FS as V-sync is going to introduce lag.

 

[oxrufiioxo]

Always verify more than one demanding games does it for sure as it could just be a bug with a specific game.

 

[erek]

If you're using mods or such things in GTA V, that might be why. I'd also suggest using FS as V-sync is going to introduce lag.

Always verify more than one demanding games does it for sure as it could just be a bug with a specific game.

no mods being used,

does Watch Dogs: Legion count?

or Mafia III? so far those haven't crashed on me. I also ran the Red Dead Redemption 2 benchmark three or so times

could it be somehow my 10+ year old Corsair AX1200?

 

[X71200]

I thought you had the i version, in fact I even totally forgot that PSU existed. Yeah, that's the old AX1200 which competed with the Huntkey unit (cough, my nickname).

It was also a very good Flextronics unit. These PSUs are said to be single rail but in Corsair Link for the digital ones, there are OCP triggers for the cables. In that case, your PSU would shut your system down.

Those PSUs use Jap Nippon caps, and you weren't loading it too much. Although if it's really over 10 years old, you might want to retire it to a different PC or something just for the sake of that.

 

[erek]

I thought you had the i version, in fact I even totally forgot that PSU existed. Yeah, that's the old AX1200 which competed with the Huntkey unit (cough, my nickname).

It was also a very good Flextronics unit. These PSUs are said to be single rail but in Corsair Link for the digital ones, there are OCP triggers for the cables. In that case, your PSU would shut your system down.

Those PSUs use Jap Nippon caps, and you weren't loading it too much. Although if it's really over 10 years old, you might want to retire it to a different PC or something just for the sake of that.

I tried -10% power limit and the same results in gta v

 

[X71200]

The problem with GTA V probably lays somewhere else as online lobbies are a cheating festival, might be a bunch of things in that situation.

 

[erek]

The problem with GTA V probably lays somewhere else as online lobbies are a cheating festival, might be a bunch of things in that situation.

able to re-produce in Single Player :-\

 

[X71200]

I haven't played SP of GTA V much, more of MP. Not sure what's wrong there, could again be some other things. I'd try validating game files, if not, maybe reinstalling.

 

[Zyll Goliat]

is a +39% performance increase in Timespy a good expected result for a 3090 over a 2080 Ti or is it supposed to be closer to 49%?

Result

www.3dmark.com

OP seems like yours GPU scores are a bit lower on both cards then it's supposed to be here on picture bellow you can see average/normal/stock-non OC GPU scores for 3090 and 2080TI and on the bottom of the picture is yours provided results...

Not sure what but possibly something is choking your cards........

 

[erek]

I haven't played SP of GTA V much, more of MP. Not sure what's wrong there, could again be some other things. I'd try validating game files, if not, maybe reinstalling.

GTA V Single Player testing...

1080@30: Good
1080@60: Good

4K30: Good

4K30+Ultra: Good

4K (no FPS limit): crash

 

[X71200]

It might be worth locking your FPS to your refresh rate in Nvidia control panel next to FS, that might be it.

 

[erek]

It might be worth locking your FPS to your refresh rate in Nvidia control panel next to FS, that might be it.

isn't the crashing in general a concern though

 

[X71200]

I wouldn't know, these chips might be pushed real hard as far as their silicone goes. Sometimes you get a bad GPU that isn't even stable at stock clocks, happens. Could be other stuff though.

 

[Zyll Goliat]

OP I was also checking the "usual" CPU score for 3950X in Time Spy and it seems that your score is also possibly a bit lower as I keep seeing that majority of scores are somewhere in between 15.000-16.000...you score was 13.750 possibly but I can't be sure your CPU is not working in full potential and because of that choking a GPU.....maybe you need memory tweaking or something else maybe you have some open programs in background that "eating"your performance tho' you have a lot of threads&power in that CPU so....anyway how's your cinebench 20 results?Pls. check that and see if that is in line with others people results with the same CPU...GL

 

[X71200]

I'd second that, 3DM also has that combined bench which affects the end result a lot. You get very low frames there when both the CPU and GPU are loaded to their peak. Cinebench is a favorable benchmark for the CPU.

 

[erek]

what's the most demanding title out now to stress test

 

[Rei]

what's the most demanding title out now to stress test

Doom Eternal, Crysis Remastered & Unigine Superposition are the one I could think of at the top of my head right now.

 

[X71200]

Doom Eternal is not all that demanding unless you manage some form of exquisite way of cranking settings tbh. It's in the nature of the ID Tech engine, the game can be ran on a potato yet perform well while looking good.

Crysis remastered, I don't know. It's probably demanding but not just because of the upgraded textures, but also because of the broken nature of the game itself not scaling all that properly / game related bugs. Personally not willing to shell out cash for that, especially on the Epic launcher.

Superposition is good, it can be demanding but it's an older benchmark after all. You could try Flight Sim but it's just overly hard to run at your desired settings. It is reluctant to stability from my observation. I had some crashes in Ghostrunner too, which is also demanding when you crank RTX. I heard the RTX implementation there is eh, though.