Mystery Radeon engineering sample appears on OpenVR benchmark to best Nvidia's RTX 20

We will see bigger Navi cards but i can say with absolute certainty that they will not have a single card beat a single 2080ti.

Match a 2080 Super at best. Though probably still behind power consumption.
 
Well since amd are referring to one internal card as nvidia killer I'd at least expect it to be on par with a 2080 ti but I'm waiting for both company's to show their cards so I know what my next upgrade is but first comes the saving process :(
 
I'm not sure this is really it, but matching an RTX2080Ti would be quite easy for AMD now, as demonstrated by the fact these results are more or less exactly what you'd expect from a 64CU, 384-bit Navi card.

An RTX2080Ti is 30-40% faster than a 5700XT on average.
If you made a 64CU version of the 5700XT that's 60% more cores.
If you gave it 16Gbps chips and a 384-bit bus width (Same as RTX2080Ti) that's 70% more theoretical bandwidth than 5700XT.

So even counting for practical losses and lower clock speeds a 50% gain over the 5700XT is pretty doable. Die size should be around an RTX2070 in that scenario, which is still a good chunk larger than a Radeon VII but not exactly out of this world.
 
Last edited:
I would be content with RTX 2080 Super performance for £425/500 € and 25 % discount on existing RX 5000 series, just like the strategy they took with AMD Ryzen. Anything better than that would be great.
 
I'm not sure this is really it, but matching an RTX2080Ti would be quite easy for AMD now, as demonstrated by the fact these results are more or less exactly what you'd expect from a 64CU, 384-bit Navi card.

An RTX2080Ti is 30-40% faster than a 5700XT on average.
If you made a 64CU version of the 5700XT that's 60% more cores.
If you gave it 16Gbps chips and a 384-bit bus width (Same as RTX2080Ti) that's 70% more theoretical bandwidth than 5700XT.

So even counting for practical losses and lower clock speeds a 50% gain over the 5700XT is pretty doable. Die size should be around an RTX2070 in that scenario, which is still a good chunk larger than a Radeon VII but not exactly out of this world.

I'll believe it when I see it. Time and time again I have seen these quotes. Who remembers the titan killer aka Fury X.. yeah well we remember it for the wrong reasons.

They will just not be able to compete with the Nvidia flagship. Even if in some miracle, they are on par with the 2080ti. Nvidia will just release the 2080ti Super. That will have the world flocking for the next biggest thing meaning AMD just helped Nvidia profits soar once again.
 
I think it's worth considering NVidia are currently competing with an 18 month old architecture and 3 year old node that was a close derivative of a 6 year old node. Turing will likely be their previous generation by the end of the year and the least we should expect is for AMD to be able to match their by then older Turing arch after this time and with a leading edge node.

Ofc NVidia may have been put in a tight situation with Samsung's time to market on 7nm if they committed to one foundry for Samsung's lead on EUV as rumoured but I'm sure they'll have a response as soon as they can build up some supply.
 
Last edited:
I'll believe it when I see it. Time and time again I have seen these quotes. Who remembers the titan killer aka Fury X.. yeah well we remember it for the wrong reasons.

They will just not be able to compete with the Nvidia flagship. Even if in some miracle, they are on par with the 2080ti. Nvidia will just release the 2080ti Super. That will have the world flocking for the next biggest thing meaning AMD just helped Nvidia profits soar once again.

In fairness, the Fury X was far better value than the Titan X. The issue was that Nvidia released a $650 variant of it that performed better because it had factory overclocks. If the Fury X was supposed to compete with the Titan X, it did incredibly well at that.

I think it's more likely that Nvidia will release Ampere just a couple of months after Navi 21/23 launches than refresh the 2080Ti. There's definitely performance left behind with the 2080Ti, but not 17%. Obviously that's if the next generation of Navi cards are 17% faster.

Personally, I think it's a sure thing for AMD to surpass a 2080Ti in terms of performance per dollar. The issue I see is, will it come out just two months before Ampere, will it consume far more power than an RTX 3080, and will it still be prohibitively expensive with short supplies. For me, these are very likely.
 
Once Nvidia releases a 7nm GPU. AMD is screwed.

At best AMD wins for a few months. Then Nvidia will come back and obliterate them. If AMD can barely match a 12nm design on an architecture that's over 12 months old(once AMD launch), all Nvidia would need to do is release Turing on 12nm and boom they are once again ahead.
 
Quick fact: if you overclock a Radeon VII and cool it properly, it will trade blows with a 2080, and beat it in some titles. The Radeon VII was a cut-down Instinct workstation card. It only stands to reason that if the VII could compete with the 2080 in some areas (albeit water cooled and OCed to the moon), and that card is a cut-down version of crap they just had laying around that didn't make the cut for Instinct, then they could conceivably have the ability to release a 2080TI competitor.
 
Quick fact: if you overclock a Radeon VII and cool it properly, it will trade blows with a 2080, and beat it in some titles. The Radeon VII was a cut-down Instinct workstation card. It only stands to reason that if the VII could compete with the 2080 in some areas (albeit water cooled and OCed to the moon), and that card is a cut-down version of crap they just had laying around that didn't make the cut for Instinct, then they could conceivably have the ability to release a 2080TI competitor.

And there is the problem sadly. AMD will always dominate in the price vs performance field. But as you say, cool it properly "a la" buy a waterblock, means your product now costs $100 more. It then loses that price vs performance to quite an extent.

Furthermore, OC the hell out of it bumps your electric bill (if you pay it at all) and yearly costs loses even more of that value you chose AMD for.
 
I'm not sure this is really it, but matching an RTX2080Ti would be quite easy for AMD now, as demonstrated by the fact these results are more or less exactly what you'd expect from a 64CU, 384-bit Navi card.

An RTX2080Ti is 30-40% faster than a 5700XT on average.
If you made a 64CU version of the 5700XT that's 60% more cores.
If you gave it 16Gbps chips and a 384-bit bus width (Same as RTX2080Ti) that's 70% more theoretical bandwidth than 5700XT.

So even counting for practical losses and lower clock speeds a 50% gain over the 5700XT is pretty doable. Die size should be around an RTX2070 in that scenario, which is still a good chunk larger than a Radeon VII but not exactly out of this world.

Seems reasonable.
Hope it happens.
 
I don't see a card faster than a 2080ti as a problem or the fact Nvidia will hit back with something faster.

That sort of performance at a cheaper price will do many things for us.

Maybe it will be hot and loud (ok so definitely) but if it weren't for AMD I wouldn't have a 2070 super now.
 
So long as they are not looking to charge the ludicrous prices Nvidia are, they are going to gain market share, Ray traced or not.
Bring out something that offers 2080ti performance (non raytraced) at around £700 and they will be the heroes. Its still between £1100 and £1300 for a 2080ti and just to add, knowing AMD, they won't put there raytracing algorithms in a walled garden for those willing to pay for it, they'll offer it out for free. I mean why not when your tech is powering all the next gen consoles
 
DigitalFoundry calculated about 405mm^2 for the Xbox Series X die shots released by Phil Spencer by the way.

This confirms AMD can produce 7nm chips at the size required for a 64CU Navi model/"2080Ti killer", for reference Radeon VII was 331mm and 5700XT was 251mm.
 
Back
Top