AMD's Longtime Open-Source Linux Graphics Driver Advocate Retires

[Termy]

Thanks for all the hard work and enjoy your well deserved retirement!

[coder]

thats the problem of the forum in general people do not even try to be brilliant some just want to be destructive here.

Yes, you're mostly right and I believe your suggestions are made with good intentions.

I just think it's funny that you seem to believe everyone in a big company is like a happy family, where anyone can just phone or message anyone else with a great idea. That's not how even much smaller businesses work. If there's a truly important message to pass from one part to the other, it can be done. However, people normally exist in silos and Bridgman's was just a small corner of their GPU business. Not only that, but his office was ex-ATI, located in a different actual country (i.e. Canada) than where the CPU work happens!

[bridgman]

Thanks for all the kind words. It's been a lot of fun, and I had the privilege of working with a lot of great people both inside and outside of AMD. We have been getting more key people involved with open source graphics for over a decade now - twriter not only took on management of the open source graphics team but (along with agd5f) has been a very active evangelist internally. Even better, we have active support from multiple layers of executives above us.

Most of my recent work has been related to early SW development & testing for compute parts - starting with Vega10 but really ramping up with Vega20 and beyond. Last project was MI300, which was interesting not just because of its size and complexity but because we had both APU and dGPU configurations using the same silicon.

Anyways, I'll still be around here and hopefully will have time to get back to being more hands-on with the hardware and software. Thanks again.

[Lanz]

@bridgman has done so much for open source. Truly a great man who made a lasting contribution to freedom. Thank you!

[DarkFoss]

bridgman Congratulations on your well earned retirement! Thank you for everything you've done for AMD and the open source initiative. I'm glad to see your intending to stick around the forums! I have always enjoyed your posts and look forwards to your occasional updates.

[qarium]

Thanks for all the kind words. It's been a lot of fun, and I had the privilege of working with a lot of great people both inside and outside of AMD. We have been getting more key people involved with open source graphics for over a decade now - twriter not only took on management of the open source graphics team but (along with agd5f) has been a very active evangelist internally. Even better, we have active support from multiple layers of executives above us.
Most of my recent work has been related to early SW development & testing for compute parts - starting with Vega10 but really ramping up with Vega20 and beyond. Last project was MI300, which was interesting not just because of its size and complexity but because we had both APU and dGPU configurations using the same silicon.
Anyways, I'll still be around here and hopefully will have time to get back to being more hands-on with the hardware and software. Thanks again.

i know you are in pension now... i am in privat in this forum to.

i have a question for you can you give me some explanation what could be the reason for forum posts like this one:

I have msi b650m as motherboard and ryzen 7 7700 cpu. All the resets and hangs I have observed was already reported. For instance https://gitlab.freedesktop.org/drm/amd/-/issues/2156 this and this https://gitlab.freedesktop.org/drm/amd/-/issues/3284 when you are using anything that utilize vaapi. But this is not a regular thing. I got 2-3 of these over 2-3 weeks of usage.
I had regular freezes in ac:valhalla https://gitlab.freedesktop.org/mesa/mesa/-/issues/5701 and some regular ring gfx timeout https://gitlab.freedesktop.org/drm/amd/-/issues/1974 in all kinds of different games. ac:valhalla resets were more frequent.
I had problems in wolfensteins games and watch dogs 2 about textures but those are unrelated to stability of drivers.
These hangs are totally random. You can play something for a week or 2 without any crashes and get a reset randomly.
One other thing that bothers me is that in windows my gpu consumes 45 watts when decoding a video. But on linux same settings and configuration it consumes 75-95 watts. Which is talked in here https://gitlab.freedesktop.org/drm/amd/-/issues/3195. Again unrelated to stability but bothering still.

do you think this will become better with this news: " AMD Working To Release MES Documentation & Source Code"

i hope you have fun write in the phoronix.com in privat to...

[Space Beer]

Congratulations Mr. bridgman, and thank you for your work

[bridgman]

I just think it's funny that you seem to believe everyone in a big company is like a happy family, where anyone can just phone or message anyone else with a great idea. That's not how even much smaller businesses work. If there's a truly important message to pass from one part to the other, it can be done. However, people normally exist in silos and Bridgman's was just a small corner of their GPU business. Not only that, but his office was ex-ATI, located in a different actual country (i.e. Canada) than where the CPU work happens!

Sorry, just noticed this. One of the things I really enjoyed about AMD was that we enjoyed more of this "anyone can just phone or message anyone" freedom than I have seen or heard about at any comparably sized company. During the initial Zen launch (remember the segfault bug) I was pulled into the CPU discussions quite early and immediately found myself with full access to the top designers.

It works at two levels though... one is the true "anyone can phone anyone" and there's a good chance the person on the other end will listen to you even if they have no idea who you are, but they are more likely to investigate what you said than just do what you recommend.

The other is more of a peer-level "web of trust" where people in widely different areas of the company know and trust each other, and will listen to someone "far away" as much as they would listen to their boss or co-worker. In my opinion that web of trust is what allows an engineering company to function as it grows.

AMD does a lot to build and maintain that web of trust via technical forums where the primary focus is cross-pollination and relationship-building between otherwise widely separated teams. The last "Fellows Forum" I attended in Monterrey had over 1000 of the most senior technical people in the company put together, with most of the time spent doing presentations about what *our* group did to all of the *other* group's technical leaders.

One could argue that the MI300 is a direct result of these cross-org forums - basically a huge data fabric project but externally more about CPUs and GPUs. It was the first time that we (GPU) worked full-time with the CPU data fabric team and it was a bit eye-opening just how good they were.

[coder]

Sorry, just noticed this. One of the things I really enjoyed about AMD was that we enjoyed more of this "anyone can just phone or message anyone" freedom than I have seen or heard about at any comparably sized company.

Even so, you don't just phone up the product manager for ThreadRipper with a random idea for how they should spec their product lineup. Even if you did, you wouldn't reasonably expect that person to give much weight to what you said, because you're telling them how to do their job without even knowing anything about how or why the structured the product lineup in the way they did.

Just to put this back in the context of the kinds of requests qarium was making.

[bridgman]

Exactly... if it isn't someone you know then your idea has to be good enough to sell on its own merits almost immediately, ie "oh I hadn't thought about that". That doesn't happen very often although probably more than you might expect when it's engineer-to-engineer.

On the other hand if it is someone who knows you then your idea is going to get taken pretty seriously... and one of the things I enjoyed about AMD was that web of relationships across teams and orgs that you would normally expect to be silo'ed away from each other, and the fact that they go out of their way to build and maintain that web.

ThreadRipper was an interesting example because it was not part of the original product plan and happened in part because "someone phoned someone with an idea", although I don't remember the organizational distance.