• 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.

Poll: ** The AMD VEGA Thread **

On or off the hype train?

  • (off) Train has derailed

    Votes: 207 39.2%

  • (on) Overcrowding, standing room only

    Votes: 100 18.9%

  • (never ever got on) Chinese escalator

    Votes: 221 41.9%
  • Total voters
    528
Status
Not open for further replies.
awaybreaktoday said:
What has the AMD VEGA FRONTIER EDITION been designed to do, what is it primary function?
Click to expand...

Deep Learning, content creation, and for data scientists.

In deep learning tests it's 39-50% faster than Tesla P100.

g0mST2kLR5_3ZrVDskxCfw.png
 
N19h7m4r3 said:
Hold up, doesn't the Titan Xp use GDDR5X? That site lists GDDR5, and Half Precision at 190 GFLOPs, the Titan Xp should be doing Half Precision at 12 TFLOPS, same as Single Precision.

Looks like they either can't get their info correct, or it's mostly all made up.

As I pointed out before, the Radeon Pro WX 7100 ( 5.7 TFLOPs ) is matching the Vega FE results they've posted. Hmm.
Click to expand...


I mentioned that a few pages back. They list the double precision of the Titan with the half-precision of the Vega, complete fail.

Then they have bench-marked professional software with the gaming drivers on Tiatn instead of using a quadro with the quadro drivers, so those numbers are meaningless. Perhaps the Vega numbers are also with gaming drivers which then makes the entire table a joke.



The only worrying thing is the 300w/375w. Could also be made up nonsense but could come form AMD specs. AMD did have a slide with <300w which makes the 300w seem possible. And then there is the power difference between the air cooled and liquid cooled, which means they wont be running at the same clocks. We have seen leaked benchmarks of Vega running at 1100-1200MHz, but also 1500-1600MHz. Good chance the 1600MHz pulls 375w which would mean the gaming card would have to be clocked a lot lower. Of course with the BS numbers on this table It is probably not worth pondering too long over.
 
zornyan said:
neither of them run 'professional' drivers, thats the quadro type of market, hence why the p4000 is quicker in professional workflow (as the drivers are optimised for it)

technically you can run quadro drivers on the titan with a bios flash. I'm not sure how amdwill handle the drivers for this card, but currently as it is, it's not running their pro software.



well it performs nothing like a quadro in professional level, it's not priced like a professional card.

it's price and performance are just like a titan...
Click to expand...

Well it's supposed to have pro drivers....

N19h7m4r3 said:
Directly from Raja from the reddit AMA.
https://www.reddit.com/r/Amd/comments/6bklro/we_are_radeon_technologies_group_at_amd_and_were/

Looks like it can use the Gaming drivers, and the Pro ones.
Click to expand...

The Frontier Edition was designed for a variety of use-cases like Machine Learning, real-time visualization, and game design. Can you play games on Frontier Edition? Yes, absolutely. It supports the RX driver and will deliver smooth 4K gaming. But because it is optimized for professional use cases (and priced accordingly), if gaming is your primary reason for buying a GPU, I’d suggest waiting just a little while longer for the lower-priced, gaming-optimized Radeon RX Vega
Click to expand...
 
D.P. said:
I mentioned that a few pages back. They list the double precision of the Titan with the half-precision of the Vega, complete fail.

Then they have bench-marked professional software with the gaming drivers on Tiatn instead of using a quadro with the quadro drivers, so those numbers are meaningless. Perhaps the Vega numbers are also with gaming drivers which then makes the entire table a joke.
Click to expand...

They also tested Vega Frontier with gaming drivers, it's also been shown it matches up with the Radeon pro WX7100 which is a £660 Polaris card.

Also the place that listed those specs didn't get them from AMD. The retail company messed up the Titan Xp specs.
 
Wasn't it previously calculated at 1600Mhz for 12.5Tflops? This would show it to be above that for the blower at 13.1TFlops. The AIO, by pulling more power than this, should be above that though I am skeptical of the whole thing.
 
D.P. said:
Then they have bench-marked professional software with the gaming drivers on Tiatn instead of using a quadro with the quadro drivers, so those numbers are meaningless. Perhaps the Vega numbers are also with gaming drivers which then makes the entire table a joke.
Click to expand...

It looks like they were....

five8five said:
The drivers listed as being used were AMD graphics driver 17.20 & NVIDIA graphics driver 382.05. Does that mean they were the gaming drivers?
Click to expand...
 
Sargatanas2511 said:
Wasn't it previously calculated at 1600Mhz for 12.5Tflops? This would show it to be above that for the blower at 13.1TFlops. The AIO, by pulling more power than this, should be above that though I am skeptical of the whole thing.
Click to expand...
The table could be faked but some of it is genuine. It "looks" like an official document.
 
five8five said:
It looks like they were....
Click to expand...

They were also using gaming drivers for Vega, unless you truly honestly believe a 13.1TFLOP vega card can only be 5% faster than a 5.7 TFLOP Polaris card if both are on Radeon Pro drivers.

Vega FE
Catia = 135.75
CREO = 83.94
SolidWorks = 114.88
Cinebench OpenGL = 183.28

Radeon Pro WX 7100 ( £660)
Catia = 104.34
CREO = 82.08
SolidWorks = 111.55
Cinebench OpenGL = 197.11

Titan Xp
Catia = 107.29
CREO = 65.2
SolidWorks = 67.75
Cinebench OpenGL =169.72


5LuslYx.png
 
N19h7m4r3 said:
They also tested Vega Frontier with gaming drivers, it's also been shown it matches up with the Radeon pro WX7100 which is a £660 Polaris card.

Also the place that listed those specs didn't get them from AMD. The retail company messed up the Titan Xp specs.
Click to expand...
It's strange that it matches up with the blog so could've been taken from an official document.
 
N19h7m4r3 said:
Deep Learning, content creation, and for data scientists.

In deep learning tests it's 39-50% faster than Tesla P100.
Click to expand...

Those numbers are really pointless because they weren't even using the CUDA version of DeepBench, using the CUDA version the P100 is almost certainly faster. Bottom line is best case scenario for Vega would be about 15-20% improvement based on the half-precision performance difference but in reality Nvidia' CUDA architecture and software system ensure a very high throughput that can more than overcome that small difference.
 
D.P. said:
Those numbers are really pointless because they weren't even using the CUDA version of DeepBench, using the CUDA version the P100 is almost certainly faster. Bottom line is best case scenario for Vega would be about 15-20% improvement based on the half-precision performance difference but in reality Nvidia' CUDA architecture and software system ensure a very high throughput that can more than overcome that small difference.
Click to expand...

Why are you so 100% certain of all of this based on charts you said yourself were unreliable? Why would you expect AMD to bench using CUDA as well?
 
five8five said:
Well it's supposed to have pro drivers....
Click to expand...

to be honest that doesn't say pro drivers, that says optimised, as in the 16gb memory is actually more optimised for deep learning for example because you need all the vram you can get.

much like how Intel doesn't advertise the titan as a gaming card and advertise is as a deep learning card that 'can' game.
 
Status
Not open for further replies.
Back
Top Bottom