Corsair H170i Elite Capellix XT review
Forspoken: PC performance graphics benchmarks
ASRock Z790 Taichi review
The Callisto Protocol: PC graphics benchmarks
G.Skill TridentZ 5 RGB 6800 MHz CL34 DDR5 review
Be Quiet! Dark Power 13 - 1000W PSU Review
Palit GeForce RTX 4080 GamingPRO OC review
Core i9 13900K DDR5 7200 MHz (+memory scaling) review
Seasonic Prime Titanium TX-1300 (1300W PSU) review
F1 2022: PC graphics performance benchmark review
AMD Greenland Vega10 Silicon To Have 4096 Stream Processors?
Some interesting info surfaced on the web the past day or so. As you guys know, AMD is to release GPUs based on Polaris, however a recent AMD roadmap shows Vega and Navi Architectures. Vega is the successor to Polaris with HBM2 memory to be launched in the 2017 time-frame. An employee on their linked-in page noted Vega to have 4096 shader processors.
It seems that after Polaris in 2017 VEGA will make an appearance, the name is tagged with HBM2 meaning that HBM2 likely will not make it onto Polaris. Vega is the brightest star in the constellation Lyra, the fifth brightest star in the night sky and the second brightest star in the northern celestial hemisphere, after Arcturus. Next in line in the 2018 timeframe we see Navi with the two keywords being scalability and Nextgen memory.
Now the juicy part, The LinkedIn profile of Yu Zheng (who is an R&D manager at AMD) (info was removed btw) shows "shader processor" (stream processor) count of Vega10 to be 4,096, have a peek:
Interestingly enough, that's the same shader processor number as Fiji (Radeon R9 Fury X). Since the info does not state what model GPU (mid-range/high-end/enthusiast) it'll be used for it all remains guess work in terms of performance. We do know that perf per watt will increase significantly over Polaris and that these GPUs will be fitted with HMB2 graphics memory.
Since we are on the topic, "Polaris" architecture based names are oozing out. A high-end chip will be called "Ellesmere" or Polaris10. There will be a mid-range GPU called "Baffin" or Polaris11. And then "Ellesmere" is rumored to get 36 GCN 4.0 compute units, which works out to 2,304 stream processors; and a 256-bit wide bus, indicative for GDDR5/GDDR5X with 8 GB memory amount.
It's going to be an interesting year.
Denial
Senior Member
Posts: 14010
Joined: 2004-05-16
Senior Member
Posts: 14010
Joined: 2004-05-16
#5250958 Posted on: 03/28/2016 07:16 PM
Where are you getting them 30-50% increase?
What about intel 22-14 nm shift. increaces in perf were about 5%. Whats so different about GPU?
On a GPU you can increase core count and expect graphics performance to scare linearly with it. On a CPU increasing core count does little to nothing in most applications. Also this is a larger node jump, from 28 --> 16nm.
Where are you getting them 30-50% increase?
What about intel 22-14 nm shift. increaces in perf were about 5%. Whats so different about GPU?
On a GPU you can increase core count and expect graphics performance to scare linearly with it. On a CPU increasing core count does little to nothing in most applications. Also this is a larger node jump, from 28 --> 16nm.
Fox2232
Senior Member
Posts: 11808
Joined: 2012-07-20
Senior Member
Posts: 11808
Joined: 2012-07-20
#5250959 Posted on: 03/28/2016 07:17 PM
I see what you mean, but it still makes no sense not to account for clocks when talking about relative performance. By that token a 390X is miles faster than a 980, yet it's consistently outperformed by it, especially at a moderate 1500mhz
Clocks matter.
Comparison is performance per transistor per clock, because those are limiting factors of manufacturing technologies.
I have never excluded clock. And I do not intend to. nVidia smartly used lower transistor density on 28nm and reached higher clocks.
But this trick on 14/16nm will help only little. Because once over certain clock, cooling solution will not cope with heat made by GPU. And therefore no 25% higher clock for nVidia (or AMD) on 14/16nm.
It is similar with Apple's A9. People complain that 14nm of exactly same transistor design runs hotter than 16nm. Yeah, why it should not? While both chips have exactly same nand gate sequences connected, to achieve exactly same result upon calling same instructions. Each factory uses different transistor layout and 14nm chip has 8% smaller surface to radiate/transfer heat.
Who's to say that Samsung's 14nm A9 would not actually be as cold as TSMC's 16nm one if they reduced transistor density a bit. We do not have direct comparison.
Where are you getting them 30-50% increase?
What about intel 22-14 nm shift. increaces in perf were about 5%. Whats so different about GPU?
No, clue who you are referring to, Quote button is your friend. But GloFo materials show that in worst case scenario (Maximum clock) they still deliver 27% higher performance per watt than they did on 28nm.
GPUs are in no position to get to those clocks and that means power efficiency is to be expected much higher than this value in range between 1000~1600MHz than before.
9jlwvuqzhjg
Now, those are power consumption of entire systems. Let's presume that AMD has as low CPU overhead as nVidia and therefore CPU eats same watts as CPU in nVidia's (while in reality CPU has to work bit more to feed AMD card).
Pick your CPU+MB+HDD+FANs+... consumption 25 / 35 / 45 / 55?
Entire nV system: 144W => 119W / 109W / 99W / 89W GPU?
Entire AMD system: 85W => 60W / 50W / 40W / 30W GPU?
I think you get idea of 28 vs 14/16nm expected power efficiency as it looks like it is at lease twice as good. With possibly 3 times as good?
I see what you mean, but it still makes no sense not to account for clocks when talking about relative performance. By that token a 390X is miles faster than a 980, yet it's consistently outperformed by it, especially at a moderate 1500mhz
Clocks matter.
Comparison is performance per transistor per clock, because those are limiting factors of manufacturing technologies.
I have never excluded clock. And I do not intend to. nVidia smartly used lower transistor density on 28nm and reached higher clocks.
But this trick on 14/16nm will help only little. Because once over certain clock, cooling solution will not cope with heat made by GPU. And therefore no 25% higher clock for nVidia (or AMD) on 14/16nm.
It is similar with Apple's A9. People complain that 14nm of exactly same transistor design runs hotter than 16nm. Yeah, why it should not? While both chips have exactly same nand gate sequences connected, to achieve exactly same result upon calling same instructions. Each factory uses different transistor layout and 14nm chip has 8% smaller surface to radiate/transfer heat.
Who's to say that Samsung's 14nm A9 would not actually be as cold as TSMC's 16nm one if they reduced transistor density a bit. We do not have direct comparison.
Where are you getting them 30-50% increase?
What about intel 22-14 nm shift. increaces in perf were about 5%. Whats so different about GPU?
No, clue who you are referring to, Quote button is your friend. But GloFo materials show that in worst case scenario (Maximum clock) they still deliver 27% higher performance per watt than they did on 28nm.
GPUs are in no position to get to those clocks and that means power efficiency is to be expected much higher than this value in range between 1000~1600MHz than before.
9jlwvuqzhjg
Now, those are power consumption of entire systems. Let's presume that AMD has as low CPU overhead as nVidia and therefore CPU eats same watts as CPU in nVidia's (while in reality CPU has to work bit more to feed AMD card).
Pick your CPU+MB+HDD+FANs+... consumption 25 / 35 / 45 / 55?
Entire nV system: 144W => 119W / 109W / 99W / 89W GPU?
Entire AMD system: 85W => 60W / 50W / 40W / 30W GPU?
I think you get idea of 28 vs 14/16nm expected power efficiency as it looks like it is at lease twice as good. With possibly 3 times as good?
Denial
Senior Member
Posts: 14010
Joined: 2004-05-16
Senior Member
Posts: 14010
Joined: 2004-05-16
#5250964 Posted on: 03/28/2016 07:27 PM
8.9B Transistors on a 596mm2 die for Fury X
8.0B Transistors on a 601mm2 die for Titan X.
Idk, I get where everyone is coming from in general -- but I don't think heat density will be an issue. There is quite a large gap when going from WC to Air cooling on a GPU. My 980 went from like 78c loaded to 35c loaded. I'm sure they can improve air coolers further to match the heat dissipation necessary. And if not, just go with water cooling. Gamers will get over it. Fury X works fine with it, I haven't heard of any major issues and it was the first iteration.
Biggest problem and prevention of larger GPU's is going to be manufacturing costs. Both companies are going to want profitable chips and they are going to want it in similar price ranges to what we see now. I think it's pretty clear that we see out of the demos of Polaris, is that we are going to end up with Fury X/980Ti levels of performance at ~120w/300mm2 die chips. Hopefully architecture/small clock increases/more bandwidth can help the newer chips edge out of the older ones.
I personally don't even care if something bigger is coming. Tired of playing Division at Medium settings on my 980. I'm upgrading to whatever is the fastest single card that's out by the end of August. If Nvidia can't get Pascal out by then I'm not even considering them.
If better stuff comes out next year that's way faster I'll just upgrade again.
8.9B Transistors on a 596mm2 die for Fury X
8.0B Transistors on a 601mm2 die for Titan X.
Idk, I get where everyone is coming from in general -- but I don't think heat density will be an issue. There is quite a large gap when going from WC to Air cooling on a GPU. My 980 went from like 78c loaded to 35c loaded. I'm sure they can improve air coolers further to match the heat dissipation necessary. And if not, just go with water cooling. Gamers will get over it. Fury X works fine with it, I haven't heard of any major issues and it was the first iteration.
Biggest problem and prevention of larger GPU's is going to be manufacturing costs. Both companies are going to want profitable chips and they are going to want it in similar price ranges to what we see now. I think it's pretty clear that we see out of the demos of Polaris, is that we are going to end up with Fury X/980Ti levels of performance at ~120w/300mm2 die chips. Hopefully architecture/small clock increases/more bandwidth can help the newer chips edge out of the older ones.
I personally don't even care if something bigger is coming. Tired of playing Division at Medium settings on my 980. I'm upgrading to whatever is the fastest single card that's out by the end of August. If Nvidia can't get Pascal out by then I'm not even considering them.
If better stuff comes out next year that's way faster I'll just upgrade again.
Ieldra
Senior Member
Posts: 3490
Joined: 2007-01-27
Senior Member
Posts: 3490
Joined: 2007-01-27
#5250966 Posted on: 03/28/2016 07:42 PM
8.9B Transistors on a 596mm2 die for Fury X
8.0B Transistors on a 601mm2 die for Titan X.
Idk, I get where everyone is coming from in general -- but I don't think heat density will be an issue. There is quite a large gap when going from WC to Air cooling on a GPU. My 980 went from like 78c loaded to 35c loaded. I'm sure they can improve air coolers further to match the heat dissipation necessary. And if not, just go with water cooling. Gamers will get over it. Fury X works fine with it, I haven't heard of any major issues and it was the first iteration.
Biggest problem and prevention of larger GPU's is going to be manufacturing costs. Both companies are going to want profitable chips and they are going to want it in similar price ranges to what we see now. I think it's pretty clear that we see out of the demos of Polaris, is that we are going to end up with Fury X/980Ti levels of performance at ~120w/300mm2 die chips. Hopefully architecture/small clock increases/more bandwidth can help the newer chips edge out of the older ones.
I personally don't even care if something bigger is coming. Tired of playing Division at Medium settings on my 980. I'm upgrading to whatever is the fastest single card that's out by the end of August. If Nvidia can't get Pascal out by then I'm not even considering them.
If better stuff comes out next year that's way faster I'll just upgrade again.
I'm also open to selling my 980Ti for gp104 if it's worth it, problem is I'm bound to nvidia because of Cuda and their libraries, so AMD have to work a small miracle to win me over, unless I separate my gaming and workstation and just buy a card for each. Dat cash.
As for the heat density, I just recognize it's an issue ; just look at skylake. They soldered the heatspreader on because delidding it isn't the greatest option; the die is so small the surface area available for thermal transfer is just insufficient to cool it effectively, obviously this doesn't affect gpu's quite as severely because they're much bigger by nature. I think the cooling will be fine, and I also expect double everything. Double transistor density, and double efficiency in terms of compute throughput - my only worry came from you actually, that transistor cost has remained the same, if not slightly more.
IF Polaris 10 and gp104 will only MATCH fury X and Ti then I'll wait for Vega and big pascal, or maybe volta
8.9B Transistors on a 596mm2 die for Fury X
8.0B Transistors on a 601mm2 die for Titan X.
Idk, I get where everyone is coming from in general -- but I don't think heat density will be an issue. There is quite a large gap when going from WC to Air cooling on a GPU. My 980 went from like 78c loaded to 35c loaded. I'm sure they can improve air coolers further to match the heat dissipation necessary. And if not, just go with water cooling. Gamers will get over it. Fury X works fine with it, I haven't heard of any major issues and it was the first iteration.
Biggest problem and prevention of larger GPU's is going to be manufacturing costs. Both companies are going to want profitable chips and they are going to want it in similar price ranges to what we see now. I think it's pretty clear that we see out of the demos of Polaris, is that we are going to end up with Fury X/980Ti levels of performance at ~120w/300mm2 die chips. Hopefully architecture/small clock increases/more bandwidth can help the newer chips edge out of the older ones.
I personally don't even care if something bigger is coming. Tired of playing Division at Medium settings on my 980. I'm upgrading to whatever is the fastest single card that's out by the end of August. If Nvidia can't get Pascal out by then I'm not even considering them.
If better stuff comes out next year that's way faster I'll just upgrade again.
I'm also open to selling my 980Ti for gp104 if it's worth it, problem is I'm bound to nvidia because of Cuda and their libraries, so AMD have to work a small miracle to win me over, unless I separate my gaming and workstation and just buy a card for each. Dat cash.
As for the heat density, I just recognize it's an issue ; just look at skylake. They soldered the heatspreader on because delidding it isn't the greatest option; the die is so small the surface area available for thermal transfer is just insufficient to cool it effectively, obviously this doesn't affect gpu's quite as severely because they're much bigger by nature. I think the cooling will be fine, and I also expect double everything. Double transistor density, and double efficiency in terms of compute throughput - my only worry came from you actually, that transistor cost has remained the same, if not slightly more.
IF Polaris 10 and gp104 will only MATCH fury X and Ti then I'll wait for Vega and big pascal, or maybe volta
Click here to post a comment for this news story on the message forum.
Member
Posts: 35
Joined: 2015-09-26
Where are you getting them 30-50% increase?
What about intel 22-14 nm shift. increaces in perf were about 5%. Whats so different about GPU?