Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

I was at AMD in the mid-late 2000s helping design CPU/APU/GPUs

twitter.com

Post reply on HN
Print view View original

131–139 of 139 posts

Re: I was at AMD in the mid-late 2000s helping design CPU/APU/GPUs

#131

Post by einsteinx2 »

JonChesterfield wrote: post #114
Apreche wrote: post #80

I predicted years ago they would make a CPU and you would be able to buy an All-NVidia PC. I think the reason that hasn't happened is because of the failed purchase of ARM. And looking at the market dominance of NVidia, it seems they were right to block that acquisition.

There was an arm CPU from nvidia branded "Tegra" from circa 2010. I remember a laptop (or maybe a mini pc style thing) based on that and power consumption being shameful but my recollections are hazy. Vaguely interesting side note, yandex found that from poor search terms very easily and google abjectly failed to. I hope google are tracking how frequently people use their engine to find yandex, while remembering bing…

The Nintendo Switch uses a Tegra X1 chip.

Re: I was at AMD in the mid-late 2000s helping design CPU/APU/GPUs

#132

Post by p_l »

shmerl wrote: post #120
wmf wrote: post #118

Earlier quoted context omitted.

AFAIK a lot of Hollywood visual effects are done on Linux + Nvidia so they probably support that market.

Not really familiar with it, but hopefully they can get unstuck from Nvidia, especially on Linux. Only very recently things started improving it seems and not even with Nvidia's effort but outside community working on nova + nvk.

nvidia linux drivers efforts were pretty much always driven by that market, it's not getting unstuck because it's supported well.

Hell, nvidia drivers might be often complained about, but for years I would take nvidia because the crappiness was manageable and close to nothing if you were in the target market (desktop workstations running X11 on only nvidia GPUs? The only issue was if you were running super latest kernel).

Re: I was at AMD in the mid-late 2000s helping design CPU/APU/GPUs

#133

Post by p_l »

AlexandrB wrote: post #42
luyu_wu wrote: post #38

Earlier quoted context omitted.

The obvious issue with both your points is that NVidia's competitors did do as such. AMD has had workable Linux drivers for many years now and there were numerous alternatives to CUDA pushed.

A common talking point is that CUDA is a formidable moat for Nvidia, but - as someone who has never done AI dev - I'm curious to understand what makes CUDA so sticky. From an outsider perspective it looks like a re-run of DirectX vs. everything else but AI is not like gaming and end users often don't have to run the model themselves. So it seems like the network effects should be less than that for a graphics APIs.

The real moat of CUDA is that CUDA... works. Simply works out of the box, even on cheapest GPUs. Unless you want some specific high end stuff, everything will work on the cheapest GPU of given generation, with the same base tooling.

And because of that, their OpenCL implementation also works better than others. So there's more tooling not just from nvidia using it, because it. just. works.

Compare this with AMD, whose latest framework is a total mess of "will it work on this GPU?", sometimes needing custom wrangling to enable, etc. etc. and it's effectively supported only on the most expensive compute-only cards.

Re: I was at AMD in the mid-late 2000s helping design CPU/APU/GPUs

#134

Post by shmerl »

p_l wrote: post #132
shmerl wrote: post #120

Earlier quoted context omitted.

Not really familiar with it, but hopefully they can get unstuck from Nvidia, especially on Linux. Only very recently things started improving it seems and not even with Nvidia's effort but outside community working on nova + nvk.

nvidia linux drivers efforts were pretty much always driven by that market, it's not getting unstuck because it's supported well . Hell, nvidia drivers might be often complained about, but for years I would take nvidia because the crappiness was manageable and close to nothing if you were in the target market (desktop workstations running X11 on only nvidia GPUs? The only issue was if you were running super latest ke…

Half baked support that fixes major issues at the rate of once per decade can hardly be called well. I would stay away from such garbage when possible.

Now that tools like Blender and the like are increasingly picking up Vulkan support, there is no reason for the above to use Nvidia anymore.

Re: I was at AMD in the mid-late 2000s helping design CPU/APU/GPUs

#136

Post by p_l »

shmerl wrote: post #134
p_l wrote: post #132

Earlier quoted context omitted.

nvidia linux drivers efforts were pretty much always driven by that market, it's not getting unstuck because it's supported well . Hell, nvidia drivers might be often complained about, but for years I would take nvidia because the crappiness was manageable and close to nothing if you were in the target market (desktop workstations running X11 on only nvidia GPUs? The only issue was if you were running super latest ke…

Half baked support that fixes major issues at the rate of once per decade can hardly be called well. I would stay away from such garbage when possible. Now that tools like Blender and the like are increasingly picking up Vulkan support, there is no reason for the above to use Nvidia anymore.

For the targeted use case, their drivers tended to work just fine.

It was when you went outside said use case that things started getting worse, and you had to wait long time for fixes. Sometimes it was because the changes in XFree/X.Org were effectively fixated on how some other vendors did things (cough intel cough), or involved things that effectively nobody wanted to spend engineering to fix properly (like rebuilding rendering path to be able to handle hybrid graphics properly when hybrid graphics came into world years after critical set of X.Org devs decided to stop any real development into X.Org...).

Vulkan Compute also is nowhere close to feature parity with CUDA, so not sure it would be picked up instead.

Re: I was at AMD in the mid-late 2000s helping design CPU/APU/GPUs

#137

Post by shmerl »

p_l wrote: post #136
shmerl wrote: post #134

Earlier quoted context omitted.

Half baked support that fixes major issues at the rate of once per decade can hardly be called well. I would stay away from such garbage when possible. Now that tools like Blender and the like are increasingly picking up Vulkan support, there is no reason for the above to use Nvidia anymore.

For the targeted use case, their drivers tended to work just fine. It was when you went outside said use case that things started getting worse, and you had to wait long time for fixes. Sometimes it was because the changes in XFree/X.Org were effectively fixated on how some other vendors did things ( cough intel cough ), or involved things that effectively nobody wanted to spend engineering to fix properly (like rebu…

Well, Wayland support was a sore point for Nvidia for a very long time. And it's the prime example of why they don't see Linux as their normal support target. Otherwise they would have worked with upstream a long time ago, instead of just trying to start now.

Re: I was at AMD in the mid-late 2000s helping design CPU/APU/GPUs

#138

Post by p_l »

shmerl wrote: post #137
p_l wrote: post #136

Earlier quoted context omitted.

For the targeted use case, their drivers tended to work just fine. It was when you went outside said use case that things started getting worse, and you had to wait long time for fixes. Sometimes it was because the changes in XFree/X.Org were effectively fixated on how some other vendors did things ( cough intel cough ), or involved things that effectively nobody wanted to spend engineering to fix properly (like rebu…

Well, Wayland support was a sore point for Nvidia for a very long time. And it's the prime example of why they don't see Linux as their normal support target. Otherwise they would have worked with upstream a long time ago, instead of just trying to start now.

Wayland wasn't in scope of "use cases actually paying for Linux support" (whether directly or transitively) and I wouldn't be surprised if majority of targeted supported applications don't support running under Wayland. The professional market is only now starting to do its toes into dealing with Wayland and that's mostly because RHEL9 defaults to it.

Also, it's not like Wayland offered a concrete target to support, different approaches of how to actually provide device context to applications were from beginning a "now draw the rest of the owl" thing.

Re: I was at AMD in the mid-late 2000s helping design CPU/APU/GPUs

#139

Post by shmerl »

p_l wrote: post #138
shmerl wrote: post #137

Earlier quoted context omitted.

Well, Wayland support was a sore point for Nvidia for a very long time. And it's the prime example of why they don't see Linux as their normal support target. Otherwise they would have worked with upstream a long time ago, instead of just trying to start now.

Wayland wasn't in scope of "use cases actually paying for Linux support" (whether directly or transitively) and I wouldn't be surprised if majority of targeted supported applications don't support running under Wayland. The professional market is only now starting to do its toes into dealing with Wayland and that's mostly because RHEL9 defaults to it. Also, it's not like Wayland offered a concrete target to support,…

That's the point. They don't care about Linux users and proper working with upstream despite having pools of money. So I'd avoid them on principle as a Linux user.
Post reply on HN