Not read what was said, but Humbug did this to me recently also. He quoted my post and replied to it as if I said something I never said. It is like at times he reads stuff and see's things there that were not there and winds himself up
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
AMD VEGA confirmed for 2017 H1
- Thread starter Doom112
- Start date
- Status
He's definitely showing us 'hysterically', whatever the case...Not read what was said, but Humbug did this to me recently also. He quoted my post and replied to it as if I said something I never said. It is like at times he reads stuff and see's things there that were not there and winds himself up
Caporegime
It is the leaks, all of which have said 4096 NCUs, that have me bamboozled! Agreed that the compute units are not the only driver for die size, but several Billion transistors is a lot to lose
I've gotta say, I think you're right with >4096 NCUs being the Vega die we saw, or there are some really REALLY dramatic changes to the arch.
Not to be misunderstood here
, i'm not saying any Vega GPU will have 6000 cores, or just more than 4000, i'm saying the idea that for whatever reason they can't, or never would is utter nonsense, of course they can, of course they would if they though it was the right thing to do.That Vega Die is pretty large
Not to be misunderstood here
That Vega Die is pretty large
We have to remember that the same die will also be used in their new Radeon Instinct and Pro line; and unlike NVIDIA they don't have separate cores for top consumer and enterprise similar to GeForce/Quadro and then Tesla.
So it could be that top Vega like top Hawaii is the exact same as Firepro/Instinct just no professional drivers.
So all the compute components will also be there and active.
Caporegime
nVidia's workstation cards are also the same desktop gaming cards, they just have the compute performance gimped.
I doubt it this time round if AMD want to be serious contenders in the HPC and gaming spaces then they need to start seperating the engineering a little. AMD stripped out the FP64 support from the FuryX and Polaris, if they add it back in to Vega gaming chips then performance and power efficiency will be pretty disappointing and a lot f transistors will be needlessly wasted for gaming parts.
nVidia's workstation cards are also the same desktop gaming cards, they just have the compute performance gimped.
Nope, the Tesla line is a bigger die with full compute, and extra cores for deeplearing.
The GP102 used in Titan X/Ti and Quadro all have gimped Double Precision, essentially none really. Just Tesla at P100 have that.
Tesla P40 uses GP104 and P400 uses GP102.
Caporegime
I clearly state "Just Tesla at P100 have that"
Also it seems AMD will be at the NAB Show April 22nd in Las Vegas to show case the latest in Radeon Pro. Should be interesting to see what pops up there.
https://twitter.com/RadeonPro/status/854061908403470336
Soldato
Big news today would not thought to be related but might have an effect is the national election. Thats definitely big enough to move Sterling which could mean a rise in currency value which might help with a lower release price possibly in time with Vega coming out. Lots of maybes
The 580 can work better on intel then the ryzen apparently
The 580 can work better on intel then the ryzen apparently
Display boxes yea - just imagine stacks of boxes in a warehouse for months just flashing away. They have a premium release version for shop front displays and so on but they'd also have a cheaper not special edition version I guessErm..... "The Radeon RX Vega card itself will be coming in a rather slick black, red and silver box, complete with LED lighting on the box itself."
RGB all the things is getting silly
I'm sure they would. Give the latest tech to people paying less doesnt add up exactly. I dont mind if they do that personally, my bar for acceptable performance is lower then most just 1080 or so is fine. With time if its not hype it seems AMD drivers and game devs, etc will develop further gains also, awesome
That is a majority I agree, I think this price level most expect it perfect out of the box and wont buy otherwise. Some will want to break world records with custom cooling and wont even be satisfied with a stock AIO anyway
Last edited:
Humbug said Nvidia's workstation cards(plural) use same die as gaming cards.I clearly state "Just Tesla at P100 have that"
You said no, the 'Tesla line' uses a bigger die with more cores.
Humbug is 2/3's right and you only 1/3 right.
Humbug said Nvidia's workstation cards(plural) use same die as gaming cards.
You said no, the 'Tesla line' uses a bigger die with more cores.
Humbug is 2/3's right and you only 1/3 right.
P100 is a 610mm2 die is with the current top tesla card sporting 3584, up to 3840, while GP102 is 471mm2.
Tesla is not a workstation card, it has No display outputs. It's primarily an accelerator, and interfacing.
Quadro's ( workstation cards ) use the same cores as GeForce, hence gimped Double precision.
Last edited:
Permabanned
I clearly state "Just Tesla at P100 have that"
Also it seems AMD will be at the NAB Show April 22nd in Las Vegas to show case the latest in Radeon Pro. Should be interesting to see what pops up there.
https://twitter.com/RadeonPro/status/854061908403470336
That looks like it'll probably be to do with HP & Dell desktops coming with Ryzen and Polaris.
I mean, you're still wrong here, though. Quadro P5000 and P6000 again, respectively use the same GP104 and GP102 dies as the gaming cards.
Like, you're not disagreeing with humbug's original comment which was:
"nVidia's workstation cards are also the same desktop gaming cards, they just have the compute performance gimped."
I dont understand what you're arguing against!
Bah, anyways...
Caporegime
Humbug said Nvidia's workstation cards(plural) use same die as gaming cards.
You said no, the 'Tesla line' uses a bigger die with more cores.
Humbug is 2/3's right and you only 1/3 right.
lol... that did make me giggle
Associate
- Joined
- 29 Jun 2016
- Posts
- 541
Not to be misunderstood here
That Vega Die is pretty large
Fair enough. Will be really interesting to see where all those transistors have gone to!
Looking at the 580 launch, two things feel likely:
- The Polaris range of cards are here to stay
- Vega must be significantly more powerful than Polaris to avoid large price cuts to the 570/80, particularly the high spec cards.
Last edited:
Humbug said Nvidia's workstation cards(plural) use same die as gaming cards.
You said no, the 'Tesla line' uses a bigger die with more cores.
Humbug is 2/3's right and you only 1/3 right.
Together, they are 3/3 right!
Permabanned
Big news today would not thought to be related but might have an effect is the national election. Thats definitely big enough to move Sterling which could mean a rise in currency value which might help with a lower release price possibly in time with Vega coming out. Lots of maybes
The 580 can work better on intel then the ryzen apparently
Display boxes yea - just imagine stacks of boxes in a warehouse for months just flashing away. They have a premium release version for shop front displays and so on but they'd also have a cheaper not special edition version I guess
I'm sure they would. Give the latest tech to people paying less doesnt add up exactly. I dont mind if they do that personally, my bar for acceptable performance is lower then most just 1080 or so is fine. With time if its not hype it seems AMD drivers and game devs, etc will develop further gains also, awesome
That is a majority I agree, I think this price level most expect it perfect out of the box and wont buy otherwise. Some will want to break world records with custom cooling and wont even be satisfied with a stock AIO anyway
Come on that's gotta be a graph error on Linus's side, especially seeing as the graph after that shows where they tend to sit for the game regardless of cpu.
I never once stated the Quadro cards used different cores?
The way I read that he's saying just the gaming cards have gimped compute, when in fact also the Quadro cards do.
The Tesla 100 is the first NVIDIA card since Kepler to feature full Compute performance and FP64, as even the Quadro's don't have it.
Where as the original thread of ours was about AMD's older cards from Hawaii and prior never had that issue; the top gaming cards had the same compute as top FirePro.
So I surmised that it could be the reason Vega is so darn big considering it's core count; which could mean that they went back to that design of huge chip ( 500-540mm2 ), and using it for top consumer as well. Especially since HBM2 has ECC as a spec feature as well.
Instead of like NVIDIA which has separated those lines now, DP and FP64 only available on G/P100 610mm2, and everything in the Consumer and Workstation line "gimped"; and they only recently even announced the Quadro GP100 which will use the same core as Tesla P100.
Last edited:
- Status