AMD: Speculation, Rumors, and Discussion (Archive)

Status
Not open for further replies.
I don't understand these strange theories, Robert Hallock posted it straight on Reddit, 62.5 FPS is the average FPS result, and utilization per batch was 51%, 71.9% and 92.3 %, it's dead simple.
62.5 FPS is more than 58.7 FPS > Better Performance
Better efficiency means simply that they achieve more FPS per watt in that test, than a GTX 1080 does

Hallock confirmed on Reddit that 62.5 FPS was the average FPS result, and the average utilization for that result was 1.83x. 1.83x is 91.5% per card.

I don't know about the efficiency stuff, I think people are reading that to mean perf / W.
 
75+30 = 105 :p


think its going to be a bit more than that. The only reason I say this is because otherwise, AMD would have shown something other than 150, cause at 105 watts, the performance this card is going to offer will match the 1070/1080 on the perf/watt metric. And if it did, they would have pointed it out, its an easy positive for marketing to point at.
 
I don't understand these strange theories, Robert Hallock posted it straight on Reddit, 62.5 FPS is the average FPS result, and utilization per batch was 51%, 71.9% and 92.3 %, it's dead simple.
62.5 FPS is more than 58.7 FPS > Better Performance
Better efficiency means simply that they achieve more FPS per watt in that test, than a GTX 1080 does
Dead simple would be giving separate FPS values for each batch.
Deader would be giving FPS values for one card.
I'm still not sure what the numeric relationship is between whatever "utilization" is and the FPS.

FWIW, when I tried to derive what X the utilization percentages would come from to arrive at an average of 62.5, 86 theoretical frames per second. That's roughly 44, 62, 79, if that's what they're calling utilization.

Hallock confirmed on Reddit that 62.5 FPS was the average FPS result, and the average utilization for that result was 1.83x. 1.83x is 91.5% per card.

I don't know about the efficiency stuff, I think people are reading that to mean perf / W.
I think it was 1.83x speedup, which may not be the same thing as utilization. Two cards that are each 51% utilized may see 1.8x speedup over one 51% utilized card, but I don't know anymore after trying to parse AMD PR.
 
I don't understand these strange theories, Robert Hallock posted it straight on Reddit, 62.5 FPS is the average FPS result, and utilization per batch was 51%, 71.9% and 92.3 %, it's dead simple.
62.5 FPS is more than 58.7 FPS > Better Performance
Better efficiency means simply that they achieve more FPS per watt in that test, than a GTX 1080 does
That'd be a max of 95.826* watts per card then. Too good to be true.

*62,5/58,7×180/2
 
Dead simple would be giving separate FPS values for each batch.
Deader would be giving FPS values for one card.
I'm still not sure what the numeric relationship is between whatever "utilization" is and the FPS.

FWIW, when I tried to derive what X the utilization percentages would come from to arrive at an average of 62.5, 86 theoretical frames per second. That's roughly 44, 62, 79, if that's what they're calling utilization.


I think it was 1.83x speedup, which may not be the same thing as utilization. Two cards that are each 51% utilized may see 1.8x speedup over one 51% utilized card, but I don't know anymore after trying to parse AMD PR.

That hadn't occurred to me, you're a much more nuanced thinker than I am. So what you're saying is I'm going to have to go out to the garage and dig up my old linear algebra textbook?

Damnit.

Edit: just so it's handy, the exact wording was, "All together for the entire test, there is 1.83X the performance of a single GPU in what users saw on YouTube."
 
It's piling so many things in the "all together" part that I'm not sure how a specific figure translates into another, or how many non-linear behaviors are creeping in. It might be that the CPU-limited portion has a worse speedup factor, although depending on how everything is put together it cannot be the sole bottleneck since it didn't drag the 1.83 value down further.
 
AMD just trolled the internet with that comparison.

Meanwhile, I'm still waiting to see what the $300 card is.

I wonder if AMD has merely shown the "Pro" and the "XT" (which uses something closer to 150W) will be unveiled on 29th June.
 
Perhaps it's an XT if some big customer like Apple hasn't called dibs. Maybe they're leaving pricing room for 8GB third-party factory-overclocked boards with water blocks?
 
AMD just trolled the internet with that comparison.

Meanwhile, I'm still waiting to see what the $300 card is.

I wonder if AMD has merely shown the "Pro" and the "XT" (which uses something closer to 150W) will be unveiled on 29th June.

I see no purpose of not showing the xt version if there is even one, since the 1070 goes on sale on June 19th, if there was a XT variety that has more performance, I think AMD would want to show it before June 10th just to stop any potential sales of the 1070. Even if it doesn't match the 1070 it would make people think twice to spending 100 bucks more.

We could always see it as a ploy to push nV's hand but again, nV isn't about to release the 1060 till after the June 29th launch I think, at least no rumors of it.
 
AMD just trolled the internet with that comparison.

Meanwhile, I'm still waiting to see what the $300 card is.

I wonder if AMD has merely shown the "Pro" and the "XT" (which uses something closer to 150W) will be unveiled on 29th June.

They certainly have people excited that two of their new card might be about as good as one of Nvidia's new cards. It's an unorthodox strategy, but I think it's going well so far.
 
I see no purpose of not showing the xt version if there is even one, since the 1070 goes on sale on June 19th, if there was a XT variety that has more performance, I think AMD would want to show it before June 10th just to stop any potential sales of the 1070. Even if it doesn't match the 1070 it would make people think twice to spending 100 bucks more.

We could always see it as a ploy to push nV's hand but again, nV isn't about to release the 1060 till after the June 29th launch I think, at least no rumors of it.

Yes but it could be a selling strategy, get the "big" card reveal like a cheaper alternative to the 1070 and make the "300 vs 400 same performance". Im not saying this is the case just that it could be.

Perhaps it's an XT if some big customer like Apple hasn't called dibs. Maybe they're leaving pricing room for 8GB third-party factory-overclocked boards with water blocks?

Why would you even get a water block on a card that consume less than 150W? Until the chip can hold 2v I dont see the need for it...

http://www.digitimes.com/news/a20160530PD201.html

Not official but NX may use VR and it seens like PS and XB will also use it.

Could AMD design Polaris to specifically server as a chip for this 3 consoles? Because to me it seems that it fits all of their needs.
 
Last edited:
Yes, but both Rise of TR and Witcher3 games are nvidia gamework title. Despite that it use technology from AMD in the case of TR.

Honestly, i have the feeling they wanted to show as little possible of "performance" of thoses gpu's.... nearly no benchmark have been shown outside this small comparaison on AoTS. Like the 150W "Power" who is way enough vague. Even the >5Tflops instead of give a right number.

Several reason maybe behind it. 1060Ti who should launch soon. or they have yet not completely fix the clock rate.

The last time i have seen this from AMD, it was on the FuryX or Nano launch ( not remember ), and the gpu was finally coming with a bit more Tflops than previous initial presentation and higher core speed. ( should have been the Nano )
Well just because they have Nvidia Gameworks should not exclude them as AMD went out of their way to compare HW, and in a way this is where AMD needs to compete - as Carsten rightly mentioned AoTS has a limited narrative because it already plays to the strengths of AMD GCN architecture.

The benefit of both those games is that they have aspects worth showing, if AMD does compete.
They may had missed a trick not showing those games if the 480 does have great performance, but that would take away a bit of shine from the 2 cheap cards perform as well as a very expensive one narrative, which is what they wanted to push I think.
Not disagreeing just their narrative could had been a bit broader in terms of performance comparison because in theory the improvements to GCN should also reflect in both those games while also giving them an edge in terms of fidelity in one with setting maxed on Pure Hair (RoTR), while the other is known to be a headache in terms of tessellation for both GPU companies, so both good when promoting your product.
On the plus side from an AMD marketing perspective, a lot the talk-disussions since the event is again around DX12 and looking at AoTS, so maybe this is a success.

Cheers
 
Why would you even get a water block on a card that consume less than 150W? Until the chip can hold 2v I dont see the need for it...
I was putting forward the scenario that AMD is leaving room for premium third-party products that would be allowed to exceed the base limits as part of the price range for Polaris. Sometimes the things they add to those don't necessarily make the most sense, although thermal density or noise concerns might be lessened by such a decision.
 
Yes but it could be a selling strategy, get the "big" card reveal like a cheaper alternative to the 1070 and make the "300 vs 400 same performance". Im not saying this is the case just that it could be.


That will only make sense if they reveal their bigger card before the availability of the 1070 ;)
 
Personally I think the RX 480 is a VR marketing PR release. If the Polaris 10 core actually is 232 mm2 there is no way that this is the top dog. No way!
I expect a Polaris 10 with at least 2 560 sp with higher clock speed and higher memory bandwidth. Maybe that is with Gddr5x or a 384 bus. Maybe I'm wrong but this feels like a carbon copy of the Tonga-release done again.
 
That will only make sense if they reveal their bigger card before the availability of the 1070 ;)
Not really. If you do that by the time you place your card into availability Nvidia could just adjust their price and deny that "same performance less price" but if you reveal your card after 1070 availability then you can make compare price vs performance in "today market"
 
Not really. If you do that by the time you place your card into availability Nvidia could just adjust their price and deny that "same performance less price" but if you reveal your card after 1070 availability then you can make compare price vs performance in "today market"

Honestly, i think the bigger sku are all based on Vega, not Polaris.. WHo is somewhat funny because im pretty sure a sku " in between" ( like is GP104 ) could be made, maybe witth 190-200W and 1600-1700mhz clock.
 
Last edited:
Status
Not open for further replies.
Back
Top