Wait, it didn’t already or am I confusing it with the VESA support on Linux?

SDL has never supported DOS.

You might be thinking of Allegro?

https://github.com/superjamie/allegro-4.2.3.1-xc

So by the time Mythos is widely and generally available, it’ll be generations behind.

No thanks, but I’d rather not have mysterymeat patches from effectively impossible to replicate models.

Interesting to see them do this when the original study in Singapore did not (initially) enact the ban.

Did they just follow on from New Zealand?

Looks like Google is taking a page from Microsoft regarding embrace, extend, extinguish.

At this point, they’re at the Extinguish stage.

I see that Meta (if that’s still the case) is going for the gold here.

> Intel

For some workloads, the Arc Pro B70 actually does reasonably well when cached.

With some reasonable bring-up, it also seems to be more usable versus the 32gb R9700.

I have both of those cards. Llama.cpp with SYCL has thus far refused to work for me, and Vulkan is pretty slow. Hoping that some fixes come down the pipe for SYCL, because I have plenty of power for local models (on paper).

Hmm.

I had to rebuild llama.cpp from source with the SYCL and CPU specific backends.

Started with a barebones Ubuntu Server 24 LTS install, used the HWE kernel, pulled in the Intel dependencies for hardware support/oneapi/libze, then built llama.cpp with the Intel compiler (icx?) for the SYCL and NATIVE backends (CPU specific support).

In short, built it based mostly on the Intel instructions.

Which helps none, since that gates behind established actors.

Or melange, if we’re talking about Herbert’s Dune series.

s/bots/undesirables/

…at the end stage.