lepoulsdumonde.com is one of the many independent Mastodon servers you can use to participate in the fediverse.
Small french Mastodon instance for friends, family and useful bots

Administered by:

Server stats:

52
active users

#amd

4 posts4 participants0 posts today

Starting to think that API likes Vulkan are on purpose limited 😒

Testing a RDNA3 RX 7600 XT, it's supposed to have 128KB of LDS per WGP. They say in the ISA, "A single work-group may allocate up to 64kB of LDS space". Also later, if the dispatch is done in WGP Mode (but I can't tell how to achieve this in Vulkan), it can access the full 128KB.

Now, Vulkan reports only 32KB accessible! Why so little? I don't think ROCm is limited for instance

Anybody from AMD around?

Replied in thread

@jerry

Somewhere in a meeting room in AMD headquarters there's a manager asking his engineers, "Okay, we did 192 cores per socket. Did Jerry max that out? Where can we go from here?"

#AMD#CPU#core

The current #gpu situation is a joke, I do not understand why no tech reviewers are talking about the #amd 9070 pricing compared to the 7800 pricing.

The 9070 is basically the 7800 of this generation, that should be about £500 or less. Don't get me wrong its a good generational leap but its a mid range card commanding enthusiast pricing at the moment.

Because #nvidia hasnt provided any generational uplift we seem to have forgotten the 9070 is a mid range card, and should be priced as such.

I just found out that the NVIDIA driver on Linux simply cannot utilise the system RAM when your GPU is running out of VRAM. Allocations will simply fail.

That fucking sucks.

Windows is apparently remarkably good at managing video memory, so many games simply rely on that and don't take into account how much VRAM is used or available.

Seems like AMD's drivers (Mesa) can deal with this, but I'm not sure.

#NVIDIA#Linux#RAM