Lisa Su Says The "Team Is On It" After Tweet About Open-Source AMD GPU Firmware

[pong]

It's a good start.
The open BIOS has yet to be pervasive (I'm presently suffering through BIOS bugs / limits on consumer AMD X570 systems), so I am
looking forward to the day of open BIOS, fwupd for everything, etc.!

GPUs need to be designed for maintenance / reliability, have SR-IOV support, and non-horrible fans / warranties / fully supported "ECC" etc. (even for the consumer ones) so one can actually take seriously that one can buy a high end consumer GPU today and still have it working reliably in 10 years.

Yeah sure in the mean while I'll probably buy a couple newer generation ones ALSO but if you're selling me effectively "personal supercomputing"
hardware (NN TFLOP/OP/s, N TBy/s RAM BW, ...), something that should be fundamental for all my personal ML, NPU, graphics, simulations, games, HPC, GPGPU, etc. uses then it's not too much to ask for something that works reliably and keeps working. It'll take a few generations of accumulating
still-working GPUs across older / newer systems before my personal compute is where I'd like it to be, and throwing old stuff out / upgrading every year or two is hardly a sane / responsible solution vs scaling by adding and gaining 120% or whatever of last couple-year's model's capability in a next generation one.

And of course SW support -- ROCm for everything, ...

First open source BIOS and now GPU Firmware..

TAKE MY WALLET

[Daktyl198]

I trust AMD for compute as much as I would trust charlie scheen's used condom to be sterile, needless to say, none of the customers i've ever had, or ever will have, have used, or will use AMD in the near future.

Which is entirely an issue with 3rd party apps, and not the problem of the GPU hardware itself. Obviously ROCm also has some issues, but I think the CUDA<>ROCm shim we saw a few weeks ago proves the hardware is generally up to the task.

[zir_blazer]

In the Radeon HD 5000 generation of 14 years ago, there was actually an open source VBIOS, OpenRadeonBIOS. Supposedly it worked on a Radeon 5450, but the project died off and no one was interesed in it.

[Quackdoc]

Which is entirely an issue with 3rd party apps, and not the problem of the GPU hardware itself. Obviously ROCm also has some issues, but I think the CUDA<>ROCm shim we saw a few weeks ago proves the hardware is generally up to the task.

even their first party stuff like their demos have had critical issues even up to like a year ago, AMD just simply can't compute

[Jabberwocky]

Well, NVIDIA is basically printing money for a lot of people. I invested a small amount in NVIDIA a while back. It's up 1200%. I wish I would have invested more. There is at least another quarter (probably several) of AI driven frenzied growth before people need to worry about cashing out.

I put money that I got back from tax into trading crypto. I made a lot of money too. I regret doing that because it bothers me ethically that other people lost so much by being scammed into it.

I would much rather have invested into something meaningful that helped us as people move forward. I feel like deep learning is a similar story all over again.

Crypto could have been so much more and the same goes for machine learning.

PS: I'm not judging people who are investing, you do you. Heck I probably have Nvidia stocks via the plethora of funds that I have invested in.

[Jabberwocky]

Wasn't AMD meant to be looking at opening their PSP code? What ever came of that?

https://www.reddit.com/r/Amd/comment...mment/def5h1b/

I keep looking back at that thread too.

IIRC this is still being worked on. Open-Source Silicon Initialization Library (openSIL) is set to replace AMD Generic Encapsulated Software Architecture (AGESA) firmware.

https://www.phoronix.com/news/AMD-openSIL-Replace-AGESA

Disclaimer: Not sure what the current state of supported CPUs/Motherboards are. As consumers we obviously want this for Desktop CPUs

[libv]

In the Radeon HD 5000 generation of 14 years ago, there was actually an open source VBIOS, OpenRadeonBIOS. Supposedly it worked on a Radeon 5450, but the project died off and no one was interesed in it.

My goal for the RadeonHD driver (the driver that forced ATI open, 2007-2009), was to always be independent of binary blobs for at least display, and basic hardware bringup. We were the driving force for free docs for display and basic hw support. My display driver code was as little dependent on AtomBIOS as possible, allowing us to fix bugs properly and to support hw blocks properly (like the PLL and DVI HPD on r5xx). And by only depending on AtomBIOS's ASICInit() for posting the hardware, i had everything geared towards adding basic board bringup code in C code to the likes of coreboot. My preliminary ogle at ASICInit() had shown that in the worst case scenario, each board would get about 400loc, and that we would quickly find similarities after a few boards.

All of this was countered by the likes of John Bridgman (former ATI employee), Alex Deucher (then fresh "AMD" employee drinking the ATI coolaid), Dave Airlie (red hat), Adam Jackson (red hat), Matthew Garrett (then red hat), and Daniel Stone (then nokia). The former did so because he was protecting ATIs way of working, the latter bunch did so because RadeonHD did not allow them to take the spotlight. This gaggle forked RadeonHD code, replaced bits of it with calls into AtomBIOS, and then used any means possible to hinder the development of an actual C based driver (even actively labelling C code as "legacy", culminating in vandalizing the RadeonHD git repository).

The second RadeonHD died (when AMD got hit by the credit crunch, and stopped the SuSE partnership, which also coincided with Novell throwing out 20% of developers at SuSE in Nuremberg) is also the very second when free docs stopped. Showing just how wrong the RadeonHD haters were all along.

In my credit crunch forced sabbatical, i finished the VIA unichrome work i had started late 2006 (before i got "sidetracked" by RadeonHD), and produced the first fully native display driver for Coreboot. I had been on this path since i started prying apart the nasty VIA code in 2003. While any VIA hardware support has long been removed from coreboot, native code exists in coreboot for intel graphics.

So an open source video BIOS was always the goal if the RadeonHD driver had survived one of the nastiest and most underhanded stories of open source graphics.

[pWe00Iri3e7Z9lHOX2Qx]

My goal for the RadeonHD driver (the driver that forced ATI open, 2007-2009), was to always be independent of binary blobs for at least display, and basic hardware bringup. We were the driving force for free docs for display and basic hw support. My display driver code was as little dependent on AtomBIOS as possible, allowing us to fix bugs properly and to support hw blocks properly (like the PLL and DVI HPD on r5xx). And by only depending on AtomBIOS's ASICInit() for posting the hardware, i had everything geared towards adding basic board bringup code in C code to the likes of coreboot. My preliminary ogle at ASICInit() had shown that in the worst case scenario, each board would get about 400loc, and that we would quickly find similarities after a few boards.

All of this was countered by the likes of John Bridgman (former ATI employee), Alex Deucher (then fresh "AMD" employee drinking the ATI coolaid), Dave Airlie (red hat), Adam Jackson (red hat), Matthew Garrett (then red hat), and Daniel Stone (then nokia). The former did so because he was protecting ATIs way of working, the latter bunch did so because RadeonHD did not allow them to take the spotlight. This gaggle forked RadeonHD code, replaced bits of it with calls into AtomBIOS, and then used any means possible to hinder the development of an actual C based driver (even actively labelling C code as "legacy", culminating in vandalizing the RadeonHD git repository).

The second RadeonHD died (when AMD got hit by the credit crunch, and stopped the SuSE partnership, which also coincided with Novell throwing out 20% of developers at SuSE in Nuremberg) is also the very second when free docs stopped. Showing just how wrong the RadeonHD haters were all along.

In my credit crunch forced sabbatical, i finished the VIA unichrome work i had started late 2006 (before i got "sidetracked" by RadeonHD), and produced the first fully native display driver for Coreboot. I had been on this path since i started prying apart the nasty VIA code in 2003. While any VIA hardware support has long been removed from coreboot, native code exists in coreboot for intel graphics.

So an open source video BIOS was always the goal if the RadeonHD driver had survived one of the nastiest and most underhanded stories of open source graphics.

I just wanted to say thanks for sharing this. Insider nuggets like this are often the most interesting responses on Phoronix for me.

[Panix]

It's still hard to take AMD seriously on any comments related to GPU compute. There was that 'big things are coming soon for ROCm we pinky swear!' message. George's message is accurate but applies at a much higher level. 'FIX YOUR SHIT! ' applies to their entire strategy around GPU compute.

AMD market cap: $340B
NVIDIA market cap: $2.2T

Basic logic says that as the x86 CPU leader (at least in terms of current tech/arch/performance/efficiency) AND one of the two big GPU vendors, those numbers above should be reversed. And that $2.2T market cap has basically nothing to do with desktop or gaming. GPU compute rules the day. AMD's constant fail here is hard to understand. NVIDIA already wrote the playbook. The ROI is obvious. So indeed, AMD, fix your shit.

Lots of ppl here have been saying to AMD, 'fix your ****' and the fanboys on here then come crying and whining - and defending AMD. Pretty pathetic when you see ppl saying the same thing in the PC world and even the CEO feels like a response is necessary. Just PR stuff. AMD hasn't put any investment, resources or attention into GPU Compute - they are too late, it seems. The only thing they care about now is AI and if it's related - they might do something there.

[ssokolow]

AMD hasn't put any investment, resources or attention into GPU Compute - they are too late, it seems. The only thing they care about now is AI and if it's related - they might do something there.

Does A.I. Art count? Stable Diffusion is one of the more leisurely things that motivated me to buy an nVidia card because I could trust it would Just Work™, while the AMD GPU in my Ryzen 5 only shows up in the Vulkan port of the Waifu2x upscaler that I also have installed.