• 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.

Now's your time AMD

D1craig said:
so this is another thing AMD did first and nvidia painting as their invention?

whats the info on radeonrays? when was it introduced and has it been used?
Click to expand...

RadeonRays 1.0 back in 2013 with Hawaii.

RadeonRays 2.0 start of 2018.

Both versions support GCN & CUDA. And yes that includes Pascal, Maxwel etc.
AMD has published a very detail documentation how to do Ray Tracing with CUDA.
And of course they will amend the API to support Turing and it's new architecture.

It doesn't require dedicated hardware on the chip to do the job.
Now what is the performance, we know from another post yesterday, a small gaming company by an ex Unreal employee, said that Vega can do 4.88Grays, at 1 ray per pixel at 4K 60fps on real time rendering (no rasterization) graphics. He also said that is comparable performance to Titan Xp, with the latter having bit more edge.

And here is the issue with the "Nvidia" Gigarays as unit of measure which is not a good indicator of performance.

What is 10 gigarays?
2 rays per pixel at 4K 60fps?
or 200 rays per pixel at 1080p 24fps like the Nvidia videos (Star Wars)?

we posted the tweet and video above in this discussion.

Also bear in mind, AMD is supporting Sound Rays also since 2013, with the new updated version right now.
Nvidia has nothing

Btw here is how you can start with the API :p

https://32ipi028l5q82yhj72224m8j-wp...2016/08/169798-A_AMD_RadeonRays_Intro_FNL.pdf

open source baby :D
 
Last edited:
melmac said:
Might just the 12nm APUs? They are due first aren't they?
Click to expand...

Might be,but GN said they saw some evidence for AMD ordering faster GDDR5 a few months ago. TBH,they already developed a 2560 shader Vega/Polaris part for the XBox One X,so surely then can make an updated Polaris replacement by now on GF 12NM??

If not Polaris will be over 3 years old by the time it is hopefully replaced by Navi!!
 
LePhuronn said:
Cut down Vega 10 to replace Polaris? 7nm shrink of Vega 10?
Click to expand...

*shrug*
Contrary to Nvidia, AMD GPU department seems tight lipped since Raja left and Lisa Su took over, downplaying even their own business growth estimations, even when business analysts saying that is more than 6 months ahead of Nvidia on the GPU servers (datacenters etc). *shrug*
Look at the Vega 20 for example. She presented the chip in May, and the only thing we know is an almost 21tflop 5nm Vega tiny behemoth already getting sampled. It's drivers are in Linux since early July. Yet nobody knows anything about it. Nada.

The only thing we know, is the AMD software director saying that something big coming by end of the year, and we should be excited. Yet that could be just the HDMI 2.1 driver update on all RX cards, promised in January. *shrug*
 
I used to only buy Nvidia GPU's (current have an AMD though). But what I do love about AMD is their adoption of open standards. They always have multiple displayport, they use freesync and they release open source drivers.
 
Hades said:
I used to only buy Nvidia GPU's (current have an AMD though). But what I do love about AMD is their adoption of open standards. They always have multiple displayport, they use freesync and they release open source drivers.
Click to expand...

To try to gain traction against the "Giant" but yeah ideal world and all that, sadly it's commercial....
 
Hades said:
I used to only buy Nvidia GPU's (current have an AMD though). But what I do love about AMD is their adoption of open standards. They always have multiple displayport, they use freesync and they release open source drivers.
Click to expand...

If you go to the AMD GPUOpen site, you will be impressed of the content AMD puts in public domain.
Also don't forget AMD promised full HDMI 2.1 support on all RX cards by driver update :)
 
Panos said:
If you go to the AMD GPUOpen site, you will be impressed of the content AMD puts in public domain.
Also don't forget AMD promised full HDMI 2.1 support on all RX cards by driver update :)
Click to expand...
Didn't know about that site. Thanks.
 
Cant wait for AMD's :
Trailers
Posters
HYPE Train

Just to underdeliver like they did EVERY SIGNE TIME pass Hawaii gpu....
 
Panos said:
*shrug*
Contrary to Nvidia, AMD GPU department seems tight lipped since Raja left and Lisa Su took over, downplaying even their own business growth estimations, even when business analysts saying that is more than 6 months ahead of Nvidia on the GPU servers (datacenters etc). *shrug*
Look at the Vega 20 for example. She presented the chip in May, and the only thing we know is an almost 21tflop 5nm Vega tiny behemoth already getting sampled. It's drivers are in Linux since early July. Yet nobody knows anything about it. Nada.

The only thing we know, is the AMD software director saying that something big coming by end of the year, and we should be excited. Yet that could be just the HDMI 2.1 driver update on all RX cards, promised in January. *shrug*
Click to expand...

I think they can see their stock prices taking off like a rocket into outer space when you look at the 1 year chart.
https://www.google.com/search?q=NAS...=1908&bih=913#scso=_1h2DW6GhFYmk_QbU7rv4Bw1:0
 
Zeed said:
Cant wait for AMD's :
Trailers
Posters
HYPE Train

Just to underdeliver like they did EVERY SIGNE TIME pass Hawaii gpu....
Click to expand...

Ryzen says hello. Nvidia's latest launch could be right up there with AMD's past few if these performance figures don't turn head. Unless you wanna look at some shadows and drop to 1080p just for some fps.
 
Hades said:
Didn't know about that site. Thanks.
Click to expand...

A lot of white papers there with code how to hook on the APIs. Including how to interface with RadeonRays 2.0, and even how to make it work on CUDA cores.
Also look at the compute section. Tensorflows for Image and voice recognition etc.
 
Panos said:
*shrug*
Contrary to Nvidia, AMD GPU department seems tight lipped since Raja left and Lisa Su took over, downplaying even their own business growth estimations, even when business analysts saying that is more than 6 months ahead of Nvidia on the GPU servers (datacenters etc). *shrug*
Look at the Vega 20 for example. She presented the chip in May, and the only thing we know is an almost 21tflop 5nm Vega tiny behemoth already getting sampled. It's drivers are in Linux since early July. Yet nobody knows anything about it. Nada.

The only thing we know, is the AMD software director saying that something big coming by end of the year, and we should be excited. Yet that could be just the HDMI 2.1 driver update on all RX cards, promised in January. *shrug*
Click to expand...

The absolute lack of news is a big shift - development times should indicate something coming out soon and we have seen Vega in a heap of APU varieties but nothing about a new GPU. I wonder if there was actually a bigger problem that Raja knew was coming which is why he jumped.

Is there a big difference in price between a 775m2 12nm chip and a 360m2 7nm chip? I get that the 12nm is a much more mature process, but it is still a very large piece of silicon Vs a chip half the size on a process that should mature rapidly over the next 12 months.

I want there to be a skunkworks Vega 20 with 8GB of HBM2 that they have been developing on the D/L.
 
You must log in or register to reply here.
Back
Top Bottom