Friday, May 13th 2016

More Polaris10 and Polaris11 Specifications Revealed

Industry sources revealed to TechPowerUp some pretty interesting specifications of AMD's two upcoming GPUs based on the 4th generation Graphics CoreNext "Polaris" architecture. The company is preparing a performance-segment GPU and a mainstream one. It turns out, that the performance-segment chip, which the press has been referring to as "Ellesmere," could feature 32 compute units (CUs), and not the previously thought 40.

Assuming that each CU continues to consist of 64 stream processors (SP), you're looking at an SP count of 2,048. What's more, this chip is said to offer a single-precision floating point performance of 5.5 TFLOP/s, as claimed by AMD. To put this into perspective, the company had claimed 5.2 TFLOP/s for the "Hawaii"/"Grenada" based FirePro W9100, which launched earlier this February, and that SKU featured all 2,816 SP present on the chip. So this chip is definitely faster than most "Hawaii" based SKUs.
While "Hawaii" based SKUs feature TDP of no less than 250W, the new chip has a TDP rated no higher than 150W. AMD could pull off a "single 8-pin power connector" feat like NVIDIA, with quite some headroom to spare. The chip features a 256-bit wide GDDR5/GDDR5X memory interface, and 8 GB could be its standard memory amount. The first SKUs based on this chip could feature 7 Gbps GDDR5 memory.

AMD will upgrade the feature-set to include HVEC/H.265 hardware encode/decode acceleration, DisplayPort 1.3, and HDMI 2.0a outputs.

The smaller "Polaris" chip scheduled for 2016, which the press has been referring to as "Baffin," could feature 14 compute units, working out to a stream processor count of 896. It will be a mainstream chip, succeeding the "Tobago" silicon, which drives the current R7 360 series SKUs, although it wouldn't surprise us if it outperformed bigger chips, such as the "Trinidad" based R7 370 series. This chip has its peak single-precision floating-point performance rated at 2.5 TFLOP/s. Its TDP is rated at just 50W, and it is expected to feature a 128-bit wide GDDR5 memory interface, holding 4 GB of memory.
Add your own comment

133 Comments on More Polaris10 and Polaris11 Specifications Revealed

#1
P4-630
The Way It's Meant to be Played
Sorry AMD but not interested in these....
Posted on Reply
#2
Prima.Vera
What an overly boring looking little card :) :) :) Specs are like....lol
Posted on Reply
#3
the54thvoid
btarunr said:
So this chip is definitely faster than most "Hawaii" based SKUs.
How many Hawaii parts were there? I mean, 290 & 290x but that was it was it not? I thought 79xx was Tahitti, and 380x was Tonga refresh of Tahitti. If this comes in at a good price point and matches 290X levels then it could my dad's next card. He's running an older 270X so this would be a great step up.
Posted on Reply
#4
medi01
It turns out, that the performance-segment chip, which the press has been referring to as "Ellesmere," could feature 32 compute units (CUs), and not the previously thought 40.
Yikes.

Let's see how it goes.
I'm definitely not interested in 449$+ 1070 and 699$+ 1080.
Posted on Reply
#5
WhyCry
Interesting, I heard the exact same specs.
@btarunr were you told different GPU codenames as well? ;)
Posted on Reply
#6
RejZoR
From the looks of it, I'll be skipping this generation and just keeping my OC'ed GTX 980. I don't want to hear about damn Hawaii 2 frigging generations later. And while GTX 1000 series do seem to deliver based on PR stuff, question is, do I really need it? Doom 2016 runs maxed out butter smooth at 1080p and quite playable at 4K DSR. Returning me to the part where I'm skipping this entire generation, probably even the next one. Might be returning to the graphics scene when AMD's Vega launches. And NVIDIA's GTX 3000 series...
Posted on Reply
#7
Nihilus
So the Polaris 10 and 11 will be a 290 and 270 with lower power consumption and slower memory.

256 bit GDRR5x on the Polaris 10 will be much slower than the 512 bit GDRR5 of the Hawaii.
As of now, the 256 bit GDDR5x of the 1080 is slower than that of the 384 bit 980ti.
Posted on Reply
#8
btarunr
Editor & Senior Moderator
WhyCry said:
Interesting, I heard the exact same specs.
@btarunr were you told different GPU codenames as well? ;)
Yup, that's exactly why I didn't mention codenames. They sounded too dumb to be true, and could be used to trace back to my Tabasco bottle.
Posted on Reply
#9
john_
Polaris 10 looks like an $199 card with GDDR5 and $249 with GDDR5X and why not, more stream processors. Polaris 11 on the other hand can play ball at around $149. Not bad, but when you create something that the competitor can easily much, you are vulnerable. And while these chips are nice and can be a big success, Nvidia can throw price cuts on 900 series or unveil GP106 sooner than expected and make those new AMD cards less attractive to those looking for a mid range card.

Of course, if AMD puts higher prices, things would NOT be as rosy as I describe them.
Posted on Reply
#10
RejZoR
Makes you wonder, will GCN 4.0 really be that more efficient they can use same GPU layout for the 3rd time?
Posted on Reply
#12
Caring1
Prima.Vera said:
What an overly boring looking little card :) :) :) Specs are like....lol
That's assuming the pictured card is correct for the article.
It shows two six pin power connections.
Posted on Reply
#13
medi01
If I read slides correctly, only 7-8 million PC users got 290+/970+.

So. Yeah, boring, but sweet price.
Posted on Reply
#14
AsRock
TPU addict
So no facts then still ?.

Industry sources revealed to TechPowerUp some pretty interesting specifications of AMD's two upcoming GPUs based on the 4th generation Graphics CoreNext "Polaris" architecture. The company is preparing a performance-segment GPU and a mainstream one. It turns out, that the performance-segment chip, which the press has been referring to as "Ellesmere," could feature 32 compute units (CUs), and not the previously thought 40.

Assuming that each CU continues to consist of 64 stream processors (SP), you're looking at an SP count of 2,048. What's more, this chip is said to offer a single-precision floating point performance of 5.5 TFLOP/s, as claimed by AMD. To put this into perspective, the company had claimed 5.2 TFLOP/s for the "Hawaii"/"Grenada" based FirePro W9100, which launched earlier this February, and that SKU featured all 2,816 SP present on the chip. So this chip is definitely faster than most "Hawaii" based SKUs.
Posted on Reply
#15
atomicus
Well this makes it abundantly clear that AMD really HAD to leak that possible VEGA release in October given they new these specs were pathetic in light of Nvidia's offering. AMD are at serious risk of fading away in to nothingness if they don't deliver something almighty with VEGA and get the jump on Nvidia, but Nvidia doubtless have the 1080Ti tucked under their sleeve which they'll whip out the moment AMD try to surge ahead. Poor AMD, I feel sorry for them, but ultimately no one wins here, least of all the consumer. Nvidia know they only need to do just enough and they're laughing all the way to the bank as their recent financial results demonstrate.
Posted on Reply
#16
ERazer
damn thats just pathetic
Posted on Reply
#17
Nihilus
384 bit with 6 gb on the polaris 10 would have been much more attractive on the polaris 10. For the 1440p gaming this card is designed for, 6 gb would have been plenty.
Posted on Reply
#18
OneCool
Wow!! None of that sounds worth a damn. I may have to go green now
Posted on Reply
#19
medi01
atomicus said:
Well this makes it abundantly clear that AMD really HAD to leak that possible VEGA release in October given they new these specs were pathetic in light of Nvidia's offering. AMD are at serious risk of fading away in to nothingness if they don't deliver something almighty with VEGA and get the jump on Nvidia, but Nvidia doubtless have the 1080Ti tucked under their sleeve which they'll whip out the moment AMD try to surge ahead. Poor AMD, I feel sorry for them, but ultimately no one wins here, least of all the consumer. Nvidia know they only need to do just enough and they're laughing all the way to the bank as their recent financial results demonstrate.
Jeez, people make it sound as if AMD's 480 series was supposed to compete with 1080. Let me remind ya:

We are here:
370 - 950 (? does it even exist?)
380 - 960
390(x) - 970/980
Fury Nano - 980
Fury X - 980Ti/TX

We are moving here:
470 - ?
480 (5.5Tf) - ?
? - 1070 (6.5Tf) <= 449$, no thanks
? - 1080 (9.5Tf)
? - ?



1080 is in "too expensive" area, expensive, and doesn't offer too much an upgrade for existing 980Ti users (comparable price) with good overclocks.
So those guys are better off waiting

The only "mass market" chip that threatens AMD is 1070. But at 449$, uh, oh, who would buy that?

It is just about PR, in practice, we will likely get cards targeted at compeltely different parts of the market.
Posted on Reply
#20
ShurikN
If the specs are true, then i am more than interested in Polaris 10.
The problem is the pricing. If it is true that 10 will cost $350, then i can just get a 390x or smth... this needs to be $300 max. And with GDDR5X. If not it will fail.
Posted on Reply
#21
okidna
Nihilus said:
So the Polaris 10 and 11 will be a 290 and 270 with lower power consumption and slower memory.

256 bit GDRR5x on the Polaris 10 will be much slower than the 512 bit GDRR5 of the Hawaii.
As of now, the 256 bit GDDR5x of the 1080 is slower than that of the 384 bit 980ti.
Polaris 10 should be better than 290 (4,8 TFLOPS FP32/single precision), IMO it will be closer to 290X (5,6 TFLOPS). That's not bad at all considering the rumored TDP and if it priced accordingly (I'm thinking $199-249), will be a good card for 1080p and 1440p.

The more concerning part is the 2nd part of your post, memory bandwidth. Let's hope AMD create a more efficient compression engine for Polaris.
Posted on Reply
#22
heydan83
Why people think that this GPUs are for competing with the 1080? these GPUs are for low/mid market, things that nvidia hasn't reveal for this generation....
Posted on Reply
#23
Frick
Fishfaced Nincompoop
medi01 said:
Jeez, people make it sound as if AMD's 480 series was supposed to compete with 1080. Let me remind ya:

We are here:
370 - 950 (? does it even exist?)
380 - 960
390(x) - 970/980
Fury Nano - 980
Fury X - 980Ti/TX

We are moving here:
470 - ?
480 (5.5Tf) - ?
? - 1070 (6.5Tf) <= 449$, no thanks
? - 1080 (9.5Tf)
? - ?



1080 is in "too expensive" area, expensive, and doesn't offer too much an upgrade for existing 980Ti users (comparable price) with good overclocks.
So those guys are better off waiting

The only "mass market" chip that threatens AMD is 1070. But at 449$, uh, oh, who would buy that?

It is just about PR, in practice, we will likely get cards targeted at compeltely different parts of the market.
Aye this. Given a good price they'll be succesfull.
Posted on Reply
#24
Steevo
Its the same formula they had going into a few other chips, small die mainstream chips first to get the process down before launching higher performance larger die chips. Considering how successful a chip that runs everything at 1080P with max settings would be with a super cheap price? Kinda like the 750Ti on crack.


Lets just hope they overclock well, and AMD gets out Vega in October.
Posted on Reply
#25
WhyCry
btarunr said:
Yup, that's exactly why I didn't mention codenames. They sounded too dumb to be true, and could be used to trace back to my Tabasco bottle.
Yes, that's actually why I didn't post this leak. However those specs were for mobile chips as I was told.
Posted on Reply
Add your own comment