NVIDIA ‘Ampere’ 8nm Graphics Cards

[dsima27]

Why shift the comparing gpu?

Just say this. 3070 quicker than 2080 ti by,x amount and the 3080 is 30 percent quicker than the 2080 ti

2080 Second Level / 3080 Second Level?

Maybe they should have ran some tests comparing the 3090 to the 2080Ti for parity?

 

[jonneymendoza]

2080 Second Level / 3080 Second Level?

Maybe they should have ran some tests comparing the 3090 to the 2080Ti for parity?

Yea maybe so. Is the 3070 twice as fast as the 2070?

Shouldn't the 3090 be above 2x faster than the 2080 ti then? As already the 3080 and possibly the 3070 are twice as fast as there old counter part

 

[Stewart Copson]

I'm scratching my head looking at it, if that fan spins clockwise then its pulling air down into the card, spinning counter clockwise the blades are still not facing the right way to have it be an effective exhaust. Starting to wonder if someone hasn't made a big oversight. Just have to wait on the reviews.

Ok answered in the walkaround video go to 10.45

 

[pBay]

He lost me at bronze....BRONZE?!

 

[TNA]

Not a change that the 3090 will be able to run max settings without DLSS.

Here is one I just did using a single RTX Titan at stock.

You can see the settings I used above, have a go yourself.

Here is what I would do Kaaps, turn off Depth of Field, Motion Blur, Screen Space Contact Shadows. Bet if you try that your FPS will shoot up. To me having things like Depth of Field and Motion Blur off = Max settings, as you get more fps and better image quality (for me)

 

[Gerard]

Ok answered in the walkaround video go to 10.45

https://youtu.be/zxJDdRgs7Lw

Ah now it makes sense, the part after that where he mentioned no screws was also interesting. Might be a right pain to disassemble. Don't think anyone noticed it had no visible screws.

 

[Sargatanas2511]

It is (in pure RT/DLSS computational performance). They never said its was twice as fast IN EVERYTHING.

We all know Nvidia marketing. You only need it to be twice fast in one game with one certain set of settings for them to get away with claiming that.

And that game is Doom Eternal.

But since that game has a bug which cripples the 2080 performance with those settings on and both cards had to run at stock and not their maximum boosts, then its a misleading claim.

Jensen said 2x performance. Not 2x performance in RTX, not 2x performance in DLSS, just a flat 2x performance.

I personally know it's nonsense and will be as you said, but there will be a lot of people who don't know that and his words were pretty definitive.

 

[james.miller]

Could it be to do with the length of the PCB?

It'll be that, yes. It's tiny. Although, they didnt have to do that; nVidia could have easily put the connectors at the end of the card. They just chose not to

 

[Poneros]

https://www.techpowerup.com/review/nvidia-geforce-rtx-3080-unboxing-preview/

 

[tyler_jrb]

Jensen said 2x performance. Not 2x performance in RTX, not 2x performance in DLSS, just a flat 2x performance.

I personally know it's nonsense and will be as you said, but there will be a lot of people who don't know that and his words were pretty definitive.

If you look at the DF performance analysis it is 80-100% faster than RTX 2080 in games without RTX and DLSS and a about the same in games with both.

Obviously it’s questionable on the speed of what both cards were running but I expect the performance difference to be very close to what has been shown.

 

[Greebo]

The basic testing from DF even without DLSS and RTX showed the 3080 at 80-100% the speed of 2080.

2080Ti is only around 30-35% faster than 2080. Maybe 40% at a push in some games at 4k.

So take a game that’s 80% faster than 2080. That could still be 40-50% faster than 2080Ti. Game where it’s 100% could be up to 60/70%.

Its not though.

Lets say your baseline score is 100fps with the 2080 and the 3080 is 80% faster so is 180fps.

And lets go with the 2080ti being 35% faster than the 2080 so would be 135fps.

180-135= 45 fps

45/135 = 33%

So on your figures even when the 3080 is 80% faster than the 2080, its only 33% faster than the 2080ti.

Even in perfect RT?DLSS games where the 3080 is 100% faster than the 2080, the maths work out as follows:

200-135= 65fps

65/135= 48% faster.

So we will probably be looking at 33% faster than a 2080ti and maybe in ideal perfect settings, it might be up to 48% faster.

And that assumes they both clock the same (which isn't looking likely since EVGA has only gone with a 5% overclock on its FTW3 Ultra card which is very very conservative)

So yeah i can easily see the 3080 only being 20-30% faster than a 2080ti in general and up to 40% more in ideal conditions.

 

[tyler_jrb]

https://www.techpowerup.com/review/nvidia-geforce-rtx-3080-unboxing-preview/

Bit strange nvidia say you void warranty if you use a 3rd party cable... how could they ever tell? Haha

 

[Sargatanas2511]

If you look at the DF performance analysis it is 80-100% faster than RTX 2080 in games without RTX and DLSS and a about the same in games with both.

Obviously it’s questionable on the speed of what both cards were running but I expect the performance difference to be very close to what has been shown.

80-100% faster is not "2x performance". Plus these will be cherry picked best case scenario situations.

 

[Heisenberg]

Why did they not put the new power connector in the end of the card, at least it would look a little neater.

The PCB is only just over half the length of the card

 

[Poneros]

Bit strange nvidia say you void warranty if you use a 3rd party cable... how could they ever tell? Haha

Same reason they say things like if you OC then your warranty is void etc. Some people will believe it & that will be a customer they don't have to RMA, or they can use it as an excuse to deny RMA & if the person believes it then - bam, easy $ savings. Or could just be there for legal purposes too.

 

[Kaapstad]

Why did they not use an AIO cooler and put the new connector in the end of the card.

More performance and smaller card, win win.

 

[james.miller]

80-100% faster is not "2x performance". Plus these will be cherry picked best case scenario situations.

UP TO. He said up to 2x.

 

[tyler_jrb]

80-100% faster is not "2x performance". Plus these will be cherry picked best case scenario situations.

It was over several games with and without RTX/DLSS. Wouldn’t say it was cherry picked really.

 

[Sargatanas2511]

It was over several games with and without RTX/DLSS. Wouldn’t say it was cherry picked really.

You are saying Nvidia deliberately didn't choose the best situations they could to market their new products going forward? Are you aware of how marketing works?

They let one single tech tuber verify their claims by ensuring they only use very specific games at very specific settings too, coincidence?

 

[tyler_jrb]

You are saying Nvidia deliberately didn't choose the best situations they could to market their new products going forward? Are you aware of how marketing works?

Considering half the games shown didn’t have RTX or DLSS how is that marketing? Surely if it was nvidia chosen every game would have been a DLSS/RTX title to show the maximum gains.