3090 vs 2080 Ti expected performance difference?

[erek]

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

 

[TheoneandonlyMrK]

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

There's always a span, a from and too.
Nothing ever work's an exact amount better in every application, so that's too loose a question, compared to what too.
How it games?.

 

[X71200]

It's a QHD DX12 test which is irrelevant unless you're rocking that in the games you play with your UHD panel. 3DM has never even been an actually good benchmark. I actually haven't used it in a long time. That RTX bench which portraits humans as braindead zombies in suits in the future wrote the deal for them. I still use Superposition from time to time, although it's not updated or anything.

Try an in-game benchmark if you can.

 

[dgianstefani]

Between 40-50% faster.

 

[X71200]

Except it's less: https://tpucdn.com/review/zotac-geforce-rtx-3090-trinity/images/average-fps_3840-2160.png

The card in that chart has the stock 3090 clocks so it's not an unfair comparison btw.

 

[erek]

I just wonder if Gigabyte is a bad AIB in comparison to even the FE and stock vs stock against the 2080 Ti?

I have the Gigabyte Vision 3090 (basically the same as the Eagle). is +39% basically the lowest 3090-class performance?

 

[X71200]

There's a review for the Eagle OC here, so you can base that. Gigabyte cards are not awful, I however am not a fan of their motherboards a lot.

For the stock 3090, it's no more than a rough %31 difference.

 

[erek]

There's a review for the Eagle OC here, so you can base that. Gigabyte cards are not awful, I however am not a fan of their motherboards a lot.

For the stock 3090, it's no more than a rough %31 difference.

found my limits... might gain another 2.7% by bumping to 10% max power limit or 385W

Gigabyte GeForce RTX 3090 Eagle OC Review

Gigabyte debuts its Eagle brand of graphics cards to the enthusiast segment. Slotted between the WindForce OC and AORUS Gaming series, the RTX 3090 Eagle OC covers all the bases and bling gamers need, at the NVIDIA MSRP price. It also comes with a power connector innovation.

www.techpowerup.com

"Gigabyte only gives an additional 10% manual power limit adjustment range, or up to 385 W, which pretty much maxes out the capabilities of the dual-8-pin power inputs. It seems RTX 3090 really needs three 8-pins. Just like on Turing, NVIDIA's Boost algorithm complicates overclocking because you can no longer dial in specific frequencies. On the RTX 3090, the effect is amplified because the delta between the power limit during normal games and light loads boosting much higher is bigger than on other cards. Still, we managed to achieve 2.7% in real-life performance gains with overclocking."

 

[X71200]

AMD Radeon R9 295X2 8 GB Review

Today, AMD is launching their Radeon R9 295X2, a dual-GPU card based on two fully unlocked, fully clocked Hawaii graphics processors. As a result, the card delivers impressive numbers in 4K and EyeFinity. But with a price of $1500, it is certainly not cheap, no matter how you look at it.

www.techpowerup.com

I don't know how W1zzard came up with that conclusion, but simply putting a 3rd 8-pin will not give you more room for the silicone that already seems to be pushed to its limits. This is likely why none of the Ampere cards can seriously get over some %5-10 OC gains. The card might be limited from the factory to a specific power or clock rating, but adding more cables will help nothing.

 

[erek]

I uploaded my BIOS, it wasn't in the DB: https://www.techpowerup.com/vgabios/226253/226253

AMD Radeon R9 295X2 8 GB Review

Today, AMD is launching their Radeon R9 295X2, a dual-GPU card based on two fully unlocked, fully clocked Hawaii graphics processors. As a result, the card delivers impressive numbers in 4K and EyeFinity. But with a price of $1500, it is certainly not cheap, no matter how you look at it.

www.techpowerup.com

I don't know how W1zzard came up with that conclusion, but simply putting a 3rd 8-pin will not give you more room for the silicone that already seems to be pushed to its limits. This is likely why none of the Ampere cards can seriously get over some %5-10 OC gains. The card might be limited from the factory to a specific power or clock rating, but adding more cables will help nothing.

@X71200
how do i manually increase my Power Limit safely? would setting it to the additional +10% be risky at all?

 

[X71200]

I'd say not unless you're running a very crap configuration such as PSU, mobo with awful PCI-E power traces, etc. EVGA already did a BIOS which is capable of pushing a lot more than some 350W.

However, you might experience instabilities due to the nature of what these cards do at higher clocks.

 

[erek]

I'd say not unless you're running a very crap configuration such as PSU, mobo with awful PCI-E power traces, etc. EVGA already did a BIOS which is capable of pushing a lot more than some 350W.

However, you might experience instabilities due to the nature of what these cards do at higher clocks.

so my gigabyte could draw a max of 390W, but a FE can maybe do 675W because of it's 12-pin connector?

 

[X71200]

If you're using two 8-pins to a 12-pin, how would that be possible? Think of it. You're not going to pull anywhere close to 700W while you're gaming with framesync locking your FPS to 60, and if you consistently do try to pull 700W from your PSU, you might not even be able to do it. I know I had my 1500i shut down couple times with some random cable configs and a V64 LC OC'ed. The OCP might trigger and the PSU would shut down. The OCP should be at 40a on that unit on every cable, so you might have to make sure you're running two different cables from different spots. I don't know.

More pins do not always equal to more power.

See those grounds? It's actually only 6 pins that provide power, aka double of an 8-pin. This however does not mean it's not going to work. The issue is not with these cables, but with the core of the GPU itself not being able to overclock a lot. Probably because Nvidia already pushed them to their limits.

 

[erek]

If you're using two 8-pins to a 12-pin, how would that be possible? Think of it. You're not going to pull anywhere close to 700W while you're gaming with framesync locking your FPS to 60, and if you consistently do try to pull 700W from your PSU, you might not even be able to do it. I know I had my 1500i shut down couple times with some random cable configs and a V64 LC OC'ed. The OCP might trigger and the PSU would shut down. The OCP should be at 40a on that unit on every cable, so you might have to make sure you're running two different cables from different spots. I don't know.

More pins do not always equal to more power.

See those grounds? It's actually only 6 pins that provide power, aka double of an 8-pin. This however does not mean it's not going to work. The issue is not with these cables, but with the core of the GPU itself not being able to overclock a lot. Probably because Nvidia already pushed them to their limits.

just compared the BIOS defined limits...

Mine:
Board power limit Target: 370.0 W Limit: 390.0 W

FE:
Board power limit Target: 350.0 W Limit: 400.0 W

... theoertically in stock form my card ought to be faster or not? or at least potentially faster

 

[oxrufiioxo]

The Strix that pulls nearly 480W overclocked is about 5% better than your card in synthetics which means it'll at most be 2-3% better in actual games. The FE also draws around 400w so other than it having a better cooler its probably not gonna be all that different than yours.

Most of this is gonna come down to silicon lottery as even if you tested 10 different gigabyte vision 3090s they could deviate by 2-3% in performance.


Also depends on what 2080 ti you're comparing it to in that synthetic mines scores over 16000

 

[X71200]

It's probably silicon lottery really, it might be that way, or the other way around. One thing might be sure though and that is stock clock speeds. You might be boosting higher, but that doesn't mean some random reference 3090 won't be able to beat yours. You might beat a FE, but then again, another FE might beat yours... really though around %5 OC gains, it's mostly pulling hairs.

 

[erek]

The Strix that pulls nearly 480W overclocked is about 5% better than your card in synthetics which means it'll at most be 2-3% better in actual games. The FE also draws around 400w so other than it having a better cooler its probably not gonna be all that different than yours.

Most of this is gonna come down to silicon lottery as even if you tested 10 different gigabyte vision 3090s they could deviate by 2-3% in performance.


Also depends on what 2080 ti you're comparing it to in that synthetics mines scores over 16000

2080 Ti FE

 

[oxrufiioxo]

2080 Ti FE

Just saying a score under 14000 for a 2080 ti is pretty low.

 

[erek]

Just saying a score under 14000 for a 2080 ti is pretty low.

it was all stock and so forth, no OCs, nothing just a completely stock situation FE... like every single profile totally default

 

[X71200]

The 2080 Ti had more overclocking headroom, but the FE was kinda eh.

3DM heavily depends on other things that include drivers and CPU. You could get some 20+ core CPU with a rather puny card beat a 6-8 core with a decent GPU.

 

[oxrufiioxo]

it was all stock and so forth, no OCs, nothing just a completely stock situation FE... like every single profile totally default

You seem to be about 1000 points off with your 3090 but that can literally be down to your ambient temps and not being in an open test bench which most reviews are done on.

The vision could also have a crap cooler i haven't seen any reviews yet on it


Your average temp was 82C I think the FE sticks in the low 70c high 60c range depending on ambient probably explaining why it seems to be better.

 

[erek]

You seem to be about 1000 points off with your 3090 but that can literally be down to your ambient temps and not being in an open test bench which most reviews are done on.

The vision could also have a crap cooler i haven't seen any reviews yet on it


Your average temp was 82C I think the FE sticks in the low 70c high 60c range depending on ambient probably explaining why it seems to be better.

opened up the case and put a box fan at max right directly on the side:

https://www.3dmark.com/3dm/52528915?

 

[oxrufiioxo]

Getting closer 19000 Graphics score is closer to the typically 20000 scores I see from like the FE/Strix etc

You also seem to have dropped your average temps by 20c


Not enough people actually own these cards to get an idea of how much variance there is gpu to gpu reviewers could have gotten above average silicon

 

[erek]

Getting closer 19000 Graphics score is closer to the typically 20000 scores I see from like the FE/Strix etc

You also seem to have dropped your average temps by 20c


Not enough people actually own these cards to get an idea of how much variance there is gpu to gpu reviewers could have gotten above average silicon

Board power limit Target: 370.0 W Limit: 390.0 W Adj. Range: -73%, +5%


safe?

 

[oxrufiioxo]

safe?

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