[Benchlife] R9 480 (Polaris 10 >100w), R9

[Mopetar]

Both of the larger die products from AMD and NVidia should be more than 30% above the 980 Ti. If they couldn't even do that, they're both massive failures. Even the mid-range die NVidia is expected to release first should be capable of that.

 

[airfathaaaaa]

Both of the larger die products from AMD and NVidia should be more than 30% above the 980 Ti. If they couldn't even do that, they're both massive failures. Even the mid-range die NVidia is expected to release first should be capable of that.

if the mid range is expected to pass 980ti then the high and enthusiast must obliterate the 980ti by 200-300%....
we need to get our feet on the ground and understand what the slides of amd are saying...

polaris line will be from low till maxmium fury line perf

vega will be from fury line and above

 

[alcoholbob]

if the mid range is expected to pass 980ti then the high and enthusiast must obliterate the 980ti by 200-300%....
we need to get our feet on the ground and understand what the slides of amd are saying...

polaris line will be from low till maxmium fury line perf

vega will be from fury line and above

A 3840 core GP100 @ 1.5GHz would be about 70.5% faster than a 980 Ti provided no architectural improvements.

300% improvement would require a 134% improvement in architecture from Maxwell to Pascal. Even 200% improvement would require a 76% improvement in architecture efficiency from Maxwell to Pascal.

Remember Kepler to Maxwell was a 35-40% improvement in cuda core efficiency, and was one of the biggest architectural efficiency jumps ever between generations. Most people are saying Pascal and Maxwell are essentially the same architecture. So thinking you are going to get twice or three times a bigger jump in architectural efficiency from Kepler to Maxwell is probably a pipe dream, especially given how aggressive Nvidia has pushed the clock speed of Big Pascal.

 

[airfathaaaaa]

yeah but because we have an arch that has a lot of new stuff on it saying
" the card will be that fast"
"the card will do so much good"
"the card IS faster"
just kills me fact is we dont know nothing
the only thing we have from both companies so far is the white paper of pascal to give us some speculation

 

[Mopetar]

if the mid range is expected to pass 980ti then the high and enthusiast must obliterate the 980ti by 200-300%....
we need to get our feet on the ground and understand what the slides of amd are saying...

polaris line will be from low till maxmium fury line perf

vega will be from fury line and above

I expect what Nvidia sells as the 1070 to be right around (± 7.5%) the 980 Ti and Polaris 10 to come in somewhere between 390X and Fury X with anything less from either being disappointing.

Vega and big Pascal probably won't be 200% of a 980 Ti, but closer to 75 - 125% is what I'd expect depending on architecture improvements and how big of a die they end up using.

 

[airfathaaaaa]

I expect what Nvidia sells as the 1070 to be right around (± 7.5%) the 980 Ti and Polaris 10 to come in somewhere between 390X and Fury X with anything less from either being disappointing.

Vega and big Pascal probably won't be 200% of a 980 Ti, but closer to 75 - 125% is what I'd expect depending on architecture improvements and how big of a die they end up using.

thats why i added
"we need to get our feet on the ground and understand what the slides of amd are saying..."
people are expecting way too much from those cards those expectations rivals half life 3 lol

 

[parkerface]

I'm all for trying to find the hidden pieces of the puzzle, but it often leads to wishful thinking and hyperbole in a hurry when people have different conclusions and expectations.

I just want the price point on Polaris 10 models and a few leaked benches to confirm its performance so I can make some decisions.. :sneaky:

 

[Det0x]

http://wccftech.com/amd-polaris-10-desktop-polaris-11-notebook-gpu/

The AMD Polaris 10 GPU has a maximum TDP of 175W but cards will actually consume much less than that. The GPU was initially built to support HBM memory but AMD chose to go the GDDR5/X route since it offers a better value currently. We will get to see HBM on AMD GPUs when Vega launches but until then, only Fury series will have HBM support. The Polaris 10 GPU is said to have 3DMark Firestrike Ultra performance around 4000 points which is about what a Radeon R9 Fury X and GeForce GTX 980 Ti score. By 4000 points, we don’t mean exactly 4000 but it’s actually quite a bit less but that’s the number we were told.

If AMD can manage to launch a card at around $329 US with better efficiency and performance close to 980 Ti with twice the VRAM of Fury X, they can have a great product in their hands which many users will be willing to upgrade to. AMD Polaris looks to be an incremental step for team red in the efficiency game, it will be great to see competitive products from AMD in the coming months which will be built for DirectX 12 and VR Gaming.

 

[Saylick]

http://wccftech.com/amd-polaris-10-desktop-polaris-11-notebook-gpu/

The AMD Polaris 10 GPU has a maximum TDP of 175W but cards will actually consume much less than that. The GPU was initially built to support HBM memory but AMD chose to go the GDDR5/X route since it offers a better value currently. We will get to see HBM on AMD GPUs when Vega launches but until then, only Fury series will have HBM support. The Polaris 10 GPU is said to have 3DMark Firestrike Ultra performance around 4000 points which is about what a Radeon R9 Fury X and GeForce GTX 980 Ti score. By 4000 points, we don’t mean exactly 4000 but it’s actually quite a bit less but that’s the number we were told.

If AMD can manage to launch a card at around $329 US with better efficiency and performance close to 980 Ti with twice the VRAM of Fury X, they can have a great product in their hands which many users will be willing to upgrade to. AMD Polaris looks to be an incremental step for team red in the efficiency game, it will be great to see competitive products from AMD in the coming months which will be built for DirectX 12 and VR Gaming.

Don't understand how anyone takes WCCF seriously. First they say one thing, then they say another, add a bunch of qualifiers or exceptions, and hope for the best.

Out of curiosity, has anyone tried trolling WCCF by making up bogus news/specs/leaks? :sneaky:

 

[Mopetar]

Don't understand how anyone takes WCCF seriously. First they say one thing, then they say another, add a bunch of qualifiers or exceptions, and hope for the best.

Out of curiosity, has anyone tried trolling WCCF by making up bogus news/specs/leaks? :sneaky:

Try it out yourself and let us know how it works out.

 

[3DVagabond]

Don't understand how anyone takes WCCF seriously. First they say one thing, then they say another, add a bunch of qualifiers or exceptions, and hope for the best.

Out of curiosity, has anyone tried trolling WCCF by making up bogus news/specs/leaks? :sneaky:

Charlie did it once. Caught quite a few other sites as well claiming they had exclusive info from their sources.

 

[Orvogg]

After reading about this "incremtental step up". Does anyone else get the feeling that Polaris is the safe product? That it is basically Hawaii, a well known chip that has served them well for a long time, for 14nm with some low hanging fruit mixed in. Or if not low hanging, the fruit they had time to validate until the middle of the year.

 

[Adored]

Looks more and more like they've just avoided each other - or more likely AMD avoided Nvidia. They would have known Nvidia's plan (everybody else did) and by concentrating on perf/Watt just slightly below Pascal, they'll snap up massive amounts of mobile share.

Not exactly the exciting outcome of the head-to-head we were hoping for but that's how it looks to me.

 

[Mopetar]

I think it's only the safe product in that it's a smaller die that will yield well enough that AMD can make some money selling it. Otherwise, from what's been show it does have at least some new tech thrown in.

Given that they claimed Vega would have even better performance / watt, it's likely that Polaris doesn't have every new trick added in or that or the difference between HBM2 and GDDR5 memory is significantly larger than I'm thinking.

 

[3DVagabond]

I think it's only the safe product in that it's a smaller die that will yield well enough that AMD can make some money selling it. Otherwise, from what's been show it does have at least some new tech thrown in.

Given that they claimed Vega would have even better performance / watt, it's likely that Polaris doesn't have every new trick added in or that or the difference between HBM2 and GDDR5 memory is significantly larger than I'm thinking.

Looks like a bit more than some new tech thrown in.

 

[Mopetar]

It's a question of efficiency difference between Polaris and Vega though. The slide they showed off had Vega with even better performance per watt than Polaris. I'm not claiming that Polaris isn't a vast efficiency improvement, only that AMD has suggested Vega has even better efficiency and there's a question of where it comes from.

Assuming it's true, it comes down to a few different things:
1) Efficiency scales better with Vega due to fixed front-end so there's more CUs for the same amount of other fixed hardware.
2) HBM2 is a lot more efficient than GDDR5.
3) Vega includes some other efficiency-boosting architecture features that Polaris 10 does not.
4) Polaris having slightly higher clocks which put it at a slight disadvantage in performance per watt.
5) Something else I haven't accounted for here.

It seems likely that we see a little bit of performance gains for all of those reasons.

 

[airfathaaaaa]

or (very unlikely) all of the new technologies we are hearing about is for vega only and polaris is just a literally shrink of gcn 1.2/1.3

 

[Mopetar]

They would have known Nvidia's plan (everybody else did) and by concentrating on perf/Watt just slightly below Pascal, they'll snap up massive amounts of mobile share.

They might also have an even bigger advantage if Zen turns out to be good, or at least not as worthless as Bulldozer. They've been pushing their dGPU + APU approach for a while and they may finally have enough parity with their competition for them to excel in that market segment.

 

[Mopetar]

or (very unlikely) all of the new technologies we are hearing about is for vega only and polaris is just a literally shrink of gcn 1.2/1.3

The image 3DVagabond linked states new 4th generation GCN, so it's not a shrink. The results of Hitman demo with Polaris 10 also wouldn't be possible if it were just a shrink either. There's at least a little bit of special sauce in Polaris.

 

[airfathaaaaa]

They might also have an even bigger advantage if Zen turns out to be good, or at least not as worthless as Bulldozer. They've been pushing their dGPU + APU approach for a while and they may finally have enough parity with their competition for them to excel in that market segment.

cant be worse than faildozer it was the bottom lol

 

[Saylick]

The image 3DVagabond linked states new 4th generation GCN, so it's not a shrink. The results of Hitman demo with Polaris 10 also wouldn't be possible if it were just a shrink either. There's at least a little bit of special sauce in Polaris.

I'm thinking Polaris incorporates most, if not all, of the architectural improvements RTG mentioned in that slide. I'm thinking Vega is an upscaled, HBM2'd Polaris with slightly lower clocks.

 

[3DVagabond]

It's a question of efficiency difference between Polaris and Vega though. The slide they showed off had Vega with even better performance per watt than Polaris. I'm not claiming that Polaris isn't a vast efficiency improvement, only that AMD has suggested Vega has even better efficiency and there's a question of where it comes from.

Assuming it's true, it comes down to a few different things:
1) Efficiency scales better with Vega due to fixed front-end so there's more CUs for the same amount of other fixed hardware.
2) HBM2 is a lot more efficient than GDDR5.
3) Vega includes some other efficiency-boosting architecture features that Polaris 10 does not.
4) Polaris having slightly higher clocks which put it at a slight disadvantage in performance per watt.
5) Something else I haven't accounted for here.

It seems likely that we see a little bit of performance gains for all of those reasons.

Pitcairn was a lot more efficient than Tahiti. I think it was mostly mem controller between those two. Not sure what the differences are between Polaris and Vega, but it doesn't mean it's because Polaris is "mostly" a shrunk Hawaii. Tahiti was a completely new chip, but they found better efficiency still with Pitcairn.

 

[3DVagabond]

cant be worse than faildozer it was the bottom lol

Bulldozer was unbelievably bad. They literally would have been better off shrinking what they already had and sent Bulldozer off to the key chain supplier.

While I don't think Zen is going to match/beat Intel's latest and greatest. I'm hoping that it at can match Haswell-E. Offer that at the right pricing and they'll sell some CPU's.

 

[Silverforce11]

Looks more and more like they've just avoided each other - or more likely AMD avoided Nvidia. They would have known Nvidia's plan (everybody else did) and by concentrating on perf/Watt just slightly below Pascal, they'll snap up massive amounts of mobile share.

Not exactly the exciting outcome of the head-to-head we were hoping for but that's how it looks to me.

Not exactly since their big chips with a higher performance ceiling requires HBM2 and that's just not ready for prime-time until Q4 2016 or Q1 2017.

To say AMD is avoiding it would mean they don't have VEGA at all, but that's just not right since they push the roadmap with Vega where it is on purpose. So folks knows something big and fast is coming after Polaris with HBM2.

 

[maddie]

Not exactly since their big chips with a higher performance ceiling requires HBM2 and that's just not ready for prime-time until Q4 2016 or Q1 2017.

To say AMD is avoiding it would mean they don't have VEGA at all, but that's just not right since they push the roadmap with Vega where it is on purpose. So folks knows something big and fast is coming after Polaris with HBM2.

If Polaris is very good, price, power use and performance, do you see many top end buyers waiting for Vega?