• 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.

Its not GTX, its RTX 2080, nVidia Turing Raytracing card.

you'll see the cards at gamescon :)

NV called CM/MM for AIB last week to arrange it all - prob ok to post this up now since cat is truly out the bag.

Hoping they are allowed to display their own Custom AIB which would be nice instead of waiting for media press shots to be shown or leaked out
 
bru said:
Well the Tensor cores on the new Quadro, alone are 5 times more powerful per card than an MI25 instinct (125TFlops fp16 vs 24.6 fp16) so the server garden would look about the same if he had of used a set of AMD servers.
But of course he used the Skylake example because that is what is currently used by most of the film CGI producers to actually do the rendering.
Click to expand...

Tensor cores are not the same as stream processors or "CUDA cores" though, they are specialised in matrix operations. The Cuda core FP16 capability of the newest Quadro RTX 8000 is 32 TFLOPs, the MI25 does 24.5 TFLOPs FP16 and the newest Instinct Vega 20 will do around 40 TFLOPs FP16.

So really the RTX 8000 is around 25% more powerful at FP16 than the MI25, at a much higher price. Of course if you can utilise the specialised tensor core hardware in the RTX 8000 in your application, you can potentially get much more performance out of it but it's not for everyone, especially with the high price. Vega 20 will be a big competitor with the massive TFLOPs and the much much smaller die size, making it a lot cheaper to produce.
 
Killer7xx said:
Tensor cores are not the same as stream processors or "CUDA cores" though, they are specialised in matrix operations. The Cuda core FP16 capability of the newest Quadro RTX 8000 is 32 TFLOPs, the MI25 does 24.5 TFLOPs FP16 and the newest Instinct Vega 20 will do around 40 TFLOPs FP16.

So really the RTX 8000 is around 25% more powerful at FP16 than the MI25, at a much higher price. Of course if you can utilise the specialised tensor core hardware in the RTX 8000 in your application, you can potentially get much more performance out of it but it's not for everyone, especially with the high price. Vega 20 will be a big competitor with the massive TFLOPs and the much much smaller die size, making it a lot cheaper to produce.
Click to expand...

And run cooler as it seems AMD decided to go for power consumption drop, instead of all out "more power" keeping the power consumption at same levels.

Also we haven't seen the dual Vega behemoth, which price is cheaper than the top RTX 8000.
 
Btw. If you haven't caught up the news, according to Chaos group the Star Wars ray tracing demo video, only had real time ray tracing rendering on the $10000 quadro.
The rest of the graphics was pre-baked on the video, not real time rendered.
 
The new Quadro’s are very bling bling packaging, I’d prefer a more aesthetically pleasing example of material reflectance and physical properties.
However, a K7 Slot A Athlon let me tick raytracing on back in the day, it had a Geforce 256, now those bounces of light can be done on the GPU.
Bravo.
 
Last edited:
Like most posters on the forums I like the excitement of a new generation of Gpu`s but I always give the first Gen a miss and go for the Ti versions more cost effective in the long run.
 
Dicehunter said:
Curious to see how much Asus' top end 2080 will be.
Click to expand...

Ob6qLcG.jpg
 
If the RTX2080 performance is as the AdoredTV video says it is, and I expect it will be as he’s pretty on the mark with his predictions, it’ll save me a lot of hassle as I’ll happily cling onto my 1080ti!!
 
X-Mikuru_Beam-X said:
If the RTX2080 performance is as the AdoredTV video says it is, and I expect it will be as he’s pretty on the mark with his predictions, it’ll save me a lot of hassle as I’ll happily cling onto my 1080ti!!
Click to expand...
It was a prediction based on some random person emailing him though. So taking it with a pinch of salt. I hope it is completely wrong, otherwise the gains are very poor imo.
 
TNA said:
It was a prediction based on some random person emailing him though. So taking it with a pinch of salt. I hope it is completely wrong, otherwise the gains are very poor imo.
Click to expand...
True, but I wouldn’t be at all surprised if it was bang on the money - we’ll find out Monday!

We all know Nvidia love selling mid (lower mid now?) tier parts as high end, what better way to shift the focus off the lacklustre performance improvements than rebranding with all this Ray-traycing business? They’ll be so busy wowing the crowds with a technology that for quite some time won’t benifit them in an significant way, everyone will forget they’re being shafted :D:D:D
 
You must log in or register to reply here.
Back
Top Bottom