Live data from Hacker News

Intel announces Arc B-series "Battlemage" discrete graphics with Linux support

phoronix.com

51–60 of 722 posts

Re: Intel announces Arc B-series "Battlemage" discrete graphics with Linux support

#51

Why don't they just release a basic GPU with 128GB RAM and eat NVidia's local generative AI lunch? The networking effect of all devs porting their LLMs etc. to that card would instantly put them as a major CUDA threat. But beancounters running the company would never get such an idea...

48 GB is at the tail end of what's reasonale for normal GPUs. The IO requires a lot of die space. And intel's architecture is not very space efficient right now compared to nvidia's

> The IO requires a lot of die space.

And even if you spend a lot of die space on memory controllers, you can only fit so many GDDR chips around the GPU core while maintaining signal integrity. HBM sidesteps that issue but it's still too expensive for anything but the highest end accelerators, and the ordinary LPDDR that Apple uses is lacking in bandwidth compared to GDDR, so they have to compensate with ginormous amounts of IO silicon. The M4 Ultra is expected to have similar bandwidth to a 4090 but the former will need a 1024bit bus to get there while the latter is only 384bit.

Re: Intel announces Arc B-series "Battlemage" discrete graphics with Linux support

#52
post #9

I put an a360 Card into an old machine I turned into a plex server. It turned it into a transcoding powerhouse. I can do multiple indepdent streams now without it skipping a beat. Price-performance ratio was off the chart

Any idea how that compares to Apple Silicon for that job? I bought the $599 MacBook Air with M1 as my plex server for this reason. Transcodes 4k HEVC and doesn’t even need a fan. Sips watts.

All Intel arc even the $99 A310 has HW accel h265 and AV1 encoding.

Re: Intel announces Arc B-series "Battlemage" discrete graphics with Linux support

#53

Why don't they just release a basic GPU with 128GB RAM and eat NVidia's local generative AI lunch? The networking effect of all devs porting their LLMs etc. to that card would instantly put them as a major CUDA threat. But beancounters running the company would never get such an idea...

Just how "basic" do you think a GPU can be while having the capability to interface with that much DRAM? Getting there with GDDR6 would require a really wide memory bus even if you could get it to operate with multiple ranks. Getting to 128GB with LPDDR5x would be possible with the 256-bit bus width they used on the top parts of the last generation, but would result in having half the bandwidth of an already mediocre card. "Just add more RAM" doesn't work the way you wish it could.

Re: Intel announces Arc B-series "Battlemage" discrete graphics with Linux support

#54

Why don't they just release a basic GPU with 128GB RAM and eat NVidia's local generative AI lunch? The networking effect of all devs porting their LLMs etc. to that card would instantly put them as a major CUDA threat. But beancounters running the company would never get such an idea...

GDDR isnt like the ram that connects to cpu, it's much more difficult and expensive to add more. You can get up to 48GB with some expensive stacked gddr, but if you wanted to add more stacks you'd need to solve some serious signal timing related headaches that most users wouldn't benefit from.

I think the high memory local inference stuff is going to come from "AI enabled" cpus that share the memory in your computer. Apple is doing this now, but cheaper options are on the way. As a shape its just suboptimal for graphics, so it doesn't make sense for any of the gpu vendors to do it.

Re: Intel announces Arc B-series "Battlemage" discrete graphics with Linux support

#55
post #7

We'll have to wait for first-party benchmarks, but they seem decent so far. A 4060 equivalent $200-$250 isn't bad at all. for I'm curious if we'll get a B750 or B770 and how they'll perform. At the very least, it's nice to have some decent BUDGET cards now. The ~$200 segment has been totally dead for years. I have a feeling Intel is losing a fair chunk of $ on each card though, just to enter the market.

I'd love to see their GPGPU software support under Linux.

The keywords you're looking for are Intel basekit, oneapi, and ipex.

https://christianjmills.com/posts/intel-pytorch-extension-tu...

https://chsasank.com/intel-arc-gpu-driver-oneapi-installatio...

Re: Intel announces Arc B-series "Battlemage" discrete graphics with Linux support

#56

Why don't they just release a basic GPU with 128GB RAM and eat NVidia's local generative AI lunch? The networking effect of all devs porting their LLMs etc. to that card would instantly put them as a major CUDA threat. But beancounters running the company would never get such an idea...

They are probably held back by same reason thats preventing AMD and nVidia from doing it either.

Re: Intel announces Arc B-series "Battlemage" discrete graphics with Linux support

#57
I like Intel's aggressive pricing against entry/mid level GPUs, which hopefully puts downward pressure on all GPUs. Overall, their biggest concern is software support. We've had reports of certain DX11/12 games failing to run properly on Proton, and the actual performance of the A series varied greatly between games even on Windows. I suspect we'll see the same issues when the B580 gets proper third party benchmarking.

Their dedication to Linux Support, combined with their good pricing makes this a potential buy for me in future versions. To be frank, I won't be replacing my 7900 XTX with this. Intel needs to provide more raw power in their cards and third parties need to improve their software support before this captures my business.

Re: Intel announces Arc B-series "Battlemage" discrete graphics with Linux support

#58

Why don't they just release a basic GPU with 128GB RAM and eat NVidia's local generative AI lunch? The networking effect of all devs porting their LLMs etc. to that card would instantly put them as a major CUDA threat. But beancounters running the company would never get such an idea...

Because if they could just do that and it would rival what NVidia has, they would just do it.

But obvoiusly they don't.

And for reasons: NVidia has worked on CUDA for ages, do you believe they just replace this whole thing in no time?

Re: Intel announces Arc B-series "Battlemage" discrete graphics with Linux support

#59

Why don't they just release a basic GPU with 128GB RAM and eat NVidia's local generative AI lunch? The networking effect of all devs porting their LLMs etc. to that card would instantly put them as a major CUDA threat. But beancounters running the company would never get such an idea...

Judging by the number of 16 GB laptops I see around, 128 GB of RAM would probably cost a bajillion dollars

[flagged]

Re: Intel announces Arc B-series "Battlemage" discrete graphics with Linux support

#60

Why don't they just release a basic GPU with 128GB RAM and eat NVidia's local generative AI lunch? The networking effect of all devs porting their LLMs etc. to that card would instantly put them as a major CUDA threat. But beancounters running the company would never get such an idea...

Because if they could just do that and it would rival what NVidia has, they would just do it. But obvoiusly they don't. And for reasons: NVidia has worked on CUDA for ages, do you believe they just replace this whole thing in no time?

llama.cpp and its derivatives say yes.
Post reply on HN