The AMD Execution Thread [2007 - 2017]

Status
Not open for further replies.
Any ideas if the founders section for HSA has a typo in it, or did something happen since March?
http://www.hsafoundation.com/members/

I think the A section is a little more sparse than it once was.

The site structure has changed somewhat, but indeed AMD was but is now not part of the Members chart.

d1ek2gkn.h4z.png

flfyilng.hvv.png
 
If it's not some kind of error, there are some other items of note like the ARM's refusal to implement HSAIL in its latest cache-coherent Bifrost GPU architecture, Qualcomm's switch to the new ARM license for new cores, and ARM's significant influence as IP provider to at least half of the foundation.

HSAIL has so far been mostly known in conjunction with AMD's hardware and AMD's efforts.
Anandtech's article on ARM's new GPU architecture has ARM rather curtly dismissing HSAIL in favor of OpenCL.

edit: Actually, the Anandtech section is a bit more ominous, although not a direct quote and might have other context. It addresses ARM's concerns as never "totally buying into" HSAIL. The never is not in quotes, but might have more finality if representative.
 
Last edited:
If it's not some kind of error, there are some other items of note like the ARM's refusal to implement HSAIL in its latest cache-coherent Bifrost GPU architecture, Qualcomm's switch to the new ARM license for new cores, and ARM's significant influence as IP provider to at least half of the foundation.
This is surprising to say the least. Looking into it...

HSAIL has so far been mostly known in conjunction with AMD's hardware and AMD's efforts.
Anandtech's article on ARM's new GPU architecture has ARM rather curtly dismissing HSAIL in favor of OpenCL.

edit: Actually, the Anandtech section is a bit more ominous, although not a direct quote and might have other context. It addresses ARM's concerns as never "totally buying into" HSAIL. The never is not in quotes, but might have more finality if representative.
The specific quote in my notebook is “We never totally bought into [HSAIL]”. It was in response to a question about HSAIL support.
 
This is surprising to say the least. Looking into it...

The specific quote in my notebook is “We never totally bought into [HSAIL]”. It was in response to a question about HSAIL support.

That quote sounds like a Freudian slip about HSAIL's prospects.


edit: corrected punctuation
 
Last edited:
Somewhat humorously, Lawrence Livermore National Laboratory and Argonne National Laboratory have both undergone mitosis, at least with the browsers I've seen it from. The academics section's logo pics seem a little inconsistent as well.
Those sections do appear to have some legitimately new entries since the last snapshot, though some are illegible or unreadable to me.

It'll hopefully look better when another blurb concerning HSA comes up and provides a reminder that this site exists.
 
Somewhat humorously, Lawrence Livermore National Laboratory and Argonne National Laboratory have both undergone mitosis, at least with the browsers I've seen it from. The academics section's logo pics seem a little inconsistent as well.
Those sections do appear to have some legitimately new entries since the last snapshot, though some are illegible or unreadable to me.

It'll hopefully look better when another blurb concerning HSA comes up and provides a reminder that this site exists.

If you speak about the quality of the logo's there.. Its the problem when thoses "company" university have not send vectorized logo, or dont have a graphic charts where take them with good quality, look like the graphic designer of the site have not search to do anything for it. ( just placing them as they are )
 
That's my assumption for the academics section. The duplicated national laboratory logos have what appear to be a properly scaled logo that is then cropped incorrectly, or a blurry logo that fits the dimensions of the picture.
 
And just to follow up on the above, I reached out to ARM about HSA after the 1.1 announcement and got some further information from them on where they stand with HSA. In short they're not going to use HSAIL (even with 1.1) but like ROCm they are embracing HSA for the hardware aspects of the specification and the advantages of a well-defined, common platform for how heterogeneous compute should work. G71 is basically an HSA 1.1 GPU in that regard.

With yesterday's announcement of the HSA 1.1 specification, I went back to ARM to ask them whether the new specification impacts the company’s heterogeneous compute plans at all, especially given that their architecture doesn’t support the 1.0 standard. As it turns out, ARM is going a route very similar to AMD’s ROCm platform: while the company isn’t utilizing the HSAIL – and thus in the strictest sense isn’t a complete HSA platform – they are using the HSA standard in the development of their hardware.

At a hardware level, the HSA specification standardizes a number of aspects of the hardware for common interoperability and easier programming purposes, including signals, queues, floating point number handling, and other, low-level minutiae about how heterogeneous execution should work. This is separate from the HSAIL, which is more concerned with the software aspects of heterogeneous programming, and though helpful, is not necessary for heterogeneous compute. As a result while Mali-G71 is technically not an HSA platform, in practice it is HSA hardware, using the HSA specification as a means to offer a common and well understood execution model for heterogeneous compute. So ARM is very much on-board with HSA – and is essentially supplying one of the first non-AMD HSA 1.1 hardware designs – even if they’re not using HSAIL itself.
 
Am I off-base in seeing perverse incentive right now, and perhaps permanently to do what ARM and AMD have done?
It sounds like finalizers are more of an impediment for time to market, and if you are trying to further your own platform, there's less of a reason to expend resources and time on giving workloads the ability readily use someone else's platform.

Like you mentioned in the earlier article, a company like Mediatek might have an interest in allowing objects to freely move between ARM, Qualcomm, IMG, and AMD(?) hardware.
Qualcomm might want to allow shifting between its Hexagon, GPU, and CPU resources, although how many workflows are going to straddle that many domains?

ARM and AMD wanted to take standard and mature languages and front ends that are supported and better-understood by everyone, and then feed them into the specific platforms they know intimately, and do so in a timely fashion while under competitive pressure. They can do more to tweak or add value at either end, and the credit and benefit goes to them.
There's no 1.0 representation and tool chain in the middle, nor an avenue via the IL or finalizers to allow other vendors to benefit from their value-add on either side.

The tune might change if the question was whether Future AMD or Future ARM want to keep better continuity with their existing ecosystems or the additional complexity they must manage alone, but that's a Future Them problem.

Not using the vendor-agnostic underpinnings of HSA sort of leaves the "don't make dumb hardware" portion of HSA, which is nice.
 
I would that that the assumption that this is for competitive reasons is off base. The impression I get is that this has to do more with what software paradigms make sense; both ARM and AMD are pursuing OpenCL 2.0/C++ style paradigms, which would be fully portable among vendors.
 
Current Professional GPU market share report:

http://jonpeddie.com/publications/workstation_report

Nvidia holds a 4-1 edge over AMD.

In the market for discrete professional GPUs, largely paralleling the workstation market it serves, the market for discrete showed no significant change in the battle between rivals Nvidia and AMD. Nvidia maintained its dominant share (79.7% of units) while AMD picked up the remaining (20.3%).
 
So, why is the stock gaining?
Buyout rumors/insider trading shit, or some solid actual good news coming out of the company for once? :p
 
The most recent event from a quick search is an investment firm upgrading AMD to a buy with a doubled price target.
They seem positive on AMD's prospects with Zen, VR, and 4k. The latter two seem to be reinforced by AMD's apparent semicustom wins in the form of upcoming consoles.
Some of AMD's selling off of itself also apparently makes them feel better about its immediate cash position and cash burn.
 
Status
Not open for further replies.
Back
Top