I think most of us are using the Nvidia 1070 as a practical reference on ideal bus bandwidth (this Polaris model is also meant to be a 256-bit bus) required rather than matching performance.
Cheers
AMD: Speculation, Rumors, and Discussion (Archive)
- Thread starter iMacmatician
- Start date
- Status
Cheers
Love_In_Rio
Veteran
1060 should be 970 performance at near 100 watts if we take into consideration the perf/watt difference between 1080 and 1070.
D
Deleted member 13524
Guest
All that slide says is that the card has a single 6-pin connector. That only means the PCI-Express' 75W limit isn't enough to power the card, which shouldn't be that much of a surprise for a 5.5 TFLOPs card.
From the single 6-pin, all we know is that the card consumes between 75 and 150W. For example, the GTX 950 needs a 6-pin connector because its TDP is 90W. The GTX960's TDP is 120W with the same 6-pin.
Curiously, the rumored 232mm^2 for Polaris 10 is awfully close to GM206's 227mm^2.
The 290X's TDP is 275W. With the claimed 2.5x better efficiency and similar performance, we would get 275/2.5 = 110W.
Last edited by a moderator:
Thats an internal sale channel slide, not one made for been shown to the public.. (untill they reuse it ofc in the launch material slide )
Infinisearch
Veteran
Any chance that AMD's perf/GFLOP will go up with Polaris? By perf/GFLOP I mean efficiency... as in how maxwell seemed to keep up with AMD with less GFLOP's.*
edit - *theoretical GFLOP's
edit - *theoretical GFLOP's
Last edited:
One would think that AMD has noticed that, and that it's high on their list to improve...
True. As well as cards with a 145W TDP and two 6-pin connectors.
Wall Street Journal has jumped the gun. 
AMD Prices 3-D Tech to Spur Virtual Reality Market
New line of graphics hardware to be priced at $199
Advanced Micro Devices Inc. is angling to lower the cost of virtual reality, targeting the field with a new line of graphics hardware priced at $199—half or less the cost of comparable products.
AMD said the first chips based on its new Polaris design are expected to arrive in graphics cards for personal computers at the end of June. The company aims to help push the starting cost of PCs that can deliver VR experiences as low as...
To Read the Full Story, Subscribe or Sign In
http://www.wsj.com/articles/amd-prices-3-d-cards-to-spur-virtual-reality-market-1464725394
That AMD guy did not lie when he said that Polaris will be the first one who will lower the entry point of PC VR. Judging by the initial reports, GTX 1070 will start at 425€ here in EU, which makes me so fucking sad. :-/ It's much more than what 970 costed when it arrived.
AMD Prices 3-D Tech to Spur Virtual Reality Market
New line of graphics hardware to be priced at $199
Advanced Micro Devices Inc. is angling to lower the cost of virtual reality, targeting the field with a new line of graphics hardware priced at $199—half or less the cost of comparable products.
AMD said the first chips based on its new Polaris design are expected to arrive in graphics cards for personal computers at the end of June. The company aims to help push the starting cost of PCs that can deliver VR experiences as low as...
To Read the Full Story, Subscribe or Sign In
http://www.wsj.com/articles/amd-prices-3-d-cards-to-spur-virtual-reality-market-1464725394
That AMD guy did not lie when he said that Polaris will be the first one who will lower the entry point of PC VR. Judging by the initial reports, GTX 1070 will start at 425€ here in EU, which makes me so fucking sad. :-/ It's much more than what 970 costed when it arrived.
Infinisearch
Veteran
Now that I think about it perf/watt improvments can possibly translate to Perf/GFLOP improvements.
Full article
http://www.neogaf.com/forum/showpost.php?p=205214301&postcount=142
edit - this line is very telling:
"AMD said its new Radeon RX cards, certified for use in VR by HTC and Oculus VR, deliver performance equivalent to that of $500 graphics cards used for VR."
R9 Nano costs $499, and it sits with performance just between 980 and 980Ti. It would be great if AMD can deliver 8GB $199 card that sits in that performance range.
http://www.neogaf.com/forum/showpost.php?p=205214301&postcount=142
edit - this line is very telling:
"AMD said its new Radeon RX cards, certified for use in VR by HTC and Oculus VR, deliver performance equivalent to that of $500 graphics cards used for VR."
R9 Nano costs $499, and it sits with performance just between 980 and 980Ti. It would be great if AMD can deliver 8GB $199 card that sits in that performance range.
Last edited:
I'm not sure how to feel about speculating on something from behind a paywall, but let's say I work only from the number in this thread.
If ~$200 is an anchor point (SKUs under that?) for that level of performance for Polaris, then I'm curious if the choice of Macau for the event might represent a stronger push for more cost-sensitive markets like China.
If ~$200 is an anchor point (SKUs under that?) for that level of performance for Polaris, then I'm curious if the choice of Macau for the event might represent a stronger push for more cost-sensitive markets like China.
].Well, dont forget that there's allways software "efficiency" in this equation. Meaning games optimized or not, drivers etc..
Pretty sure if they got it to nano level performance solidly they could have charged more than $199, or range is starting from $199 and going up sounds more plausible.
Infinisearch
Veteran
Doesn't D3D12 take alot of "software efficiency" out of the equation. As far as I've seen so far NV still matches or exceeds AMD except where Async compute is involved. (AoTS)
Dont really buy on this 199$ .. dont know exactly, what model we are talking about there.
WHo is far different of some situation where you can see the 980TI way ahead ( and in some games, totally abnormally ahead with 50% gain ). How much DX12 games in a panel of 7-10games are used by a review, for "calcul" perf/watts or perf/Gflops ? We know that TR is mostly bugged on DX12, Hitman have his problem too.. and well AOTS is a different manner too.
And i totally agree with you that they was some discrepancy between their theorical max Gflops and the result. I just say, that in some case, drivers, game code, can play a role too.
WHo is far different of some situation where you can see the 980TI way ahead ( and in some games, totally abnormally ahead with 50% gain ). How much DX12 games in a panel of 7-10games are used by a review, for "calcul" perf/watts or perf/Gflops ? We know that TR is mostly bugged on DX12, Hitman have his problem too.. and well AOTS is a different manner too.
And i totally agree with you that they was some discrepancy between their theorical max Gflops and the result. I just say, that in some case, drivers, game code, can play a role too.
Last edited:
Love_In_Rio
Veteran
A rumor(from videocardz.com editor) to help place the RX480 performance: 2 cards in CF faster than a 1080.
If true the 199 dollars price you could have for 400 dollars a system faster than a 1080.
If true the 199 dollars price you could have for 400 dollars a system faster than a 1080.
- Status
