Live data from Hacker News

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

phoronix.com

451–460 of 722 posts

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

#451

Earlier quoted context omitted.

Nvidia's $30,000 is a 90% margin product at scale. They could charge 1/3 that and still be very profitable. There has rarely been such a profitable large corporation in terms of the combo of profit & margin. Their last quarter was $35b in sales and $26b in gross profit ($21.8b op income; 62% op income margin vs sales). Visa is notorious for their extreme margin (66% op income margin vs sales) due to being basically a…

> And indeed that's exactly what will be required for any serious attempt to cut into their monopoly position. You misunderstand why and how Nvidia is a monopoly. Many companies make GPUs, and all those GPUs can be used for computation if you develop compute shaders for them. This part is not the problem, you can already go buy cheaper hardware that outperforms Nvidia if price is your only concern. Software is the is…

> The only way the entire industry can fuck over Nvidia is if they choose to invest in a complete CUDA replacement like OpenCL. That is the only way that Nvidia's value can be actually deposed without any path of recourse for their business, and it will never happen because every single one of Nvidia's competitors hate each other's guts and would rather watch each other die

Intel seems to have thrown their weight behind SYCL, which is an open standard intended to compete with CUDA. Its not clear there has been much interest from other hardware vendors though.

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

#452
post #345

Earlier quoted context omitted.

It is open and they regularly merge PRs. https://github.com/ROCm/ROCm/pulls?q=is%3Apr+is%3Aclosed

AMD GPUs aren't very attractive to ML folks because they don't outshine Nvidia in any single aspect. Blasting lots of RAM onto a GPU would make it attractive immediately with lots of attention from devs occupied with more interesting things.

does the 7900xt outperform the 3090ti? if so, there's already a market because those are the same price. I don't mean in theory are there any workloads that the 7900xt can do better? Even if they're practically equal performance you get a warranty and support with your new 7900xt.

also i didn't know there was a 192GB amd GPU.

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

#453

I wonder how many transistors it has and what the chip size it is. For power, it's 190W compared to 4060's 115 W. EDIT: from [1]: B580 has 21.7 billion transistors at 406 mm² die area, compared to 4060's 18.9 billion and 146 mm². That's a big die. [1] https://www.techpowerup.com/gpu-specs/arc-b580.c4244

Those numbers are identical to the A770, and don't match the numbers from the preview[0], so I think that's a copy paste error.

If we use the numbers from the preview:

    |        |Arc A770|Arc B580|RTX 4060|
    |--------|--------|--------|--------|
    |Process |N6      |N5      |N5      |
    |Die Size|406mm^2 |272mm^2 |159mm^2 |
    |Trans.  |21.7B   |19.6B   |18.9B   |
    |Mem Bus |256 bit |192 bit |128 bit |
    |TDP     |225W    |190W    |115W    |
    |~Perf   |90%     |110%    |100%    |
In terms of performance per die area it's a big improvement over A770 but still far behind Nvidia. It's interesting that the transistor density is so much lower than the 4060 despite having the same (or at least similar) process node. Speculating about why that may be:

- Nvidia has better layout. - Intel is using higher performance (larger) transistor libraries or layout in order to hit the higher boost frequencies (2800 vs 2460). - Wider bus interface takes up more space. - The B580 may have 1 render slice and 64-bits of memory bus disabled, and they're not including those transistors in the count, but they still take up area.

[0]: https://www.techpowerup.com/review/intel-arc-b580-battlemage...

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

#454

I'm really curious to see if these still rely heavily on resizable BAR. Putting these in old computers in linux without reBAR support makes the driver crash with literally any load rendering the cards completely unusable. It's a real shame, the single slot a380 is a great performance for price light gaming and general use card for small machines.

Yes, the slide deck mentions reBAR is still required.

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

#455
From the gaming side of things, I'm disappointed that Intel and AMD are focusing on the midrange market going forwards. I'm on Linux with a 6900XT and wasn't going to upgrade until there's a compatible option with acceptable raytracing performance (and when HDR is finally sorted out). The 4090 and other high tier cards are absurdly expensive, would be good to have competition in that segment.

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

#456

Earlier quoted context omitted.

The reason is AMD and Nvidia don't is that they don't want to cannibalize their high end AI market. Intel doesn't have a high end AI market to protect.

There are products like this one: https://www.intel.com/content/www/us/en/products/sku/232592/... As far as I understand it, it gives you 64 GiB of HBM per socket.

that's a CPU. We are talking about GPUs here for highly parallel matrix multiplication problems. 2 different beasts.

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

#457

12GB max is a non-starter for ML work now. Why not come out with a reasonably priced 24gb card even if it isn't the fastest and target it at the ML dev world? Am I missing something here?

> Am I missing something here? This is a graphics card.

Sir, this is a Wendy's

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

#458

Earlier quoted context omitted.

I have been trying to hold my slurs in reading this thread. These ML AI Macbook people are legit insane. Desktops and gaming is ugly and complex to them (because lego is hard and macbook look nice unga bunga), yet it is a mass market Intel wants to move in on. People here complain because Intel is not making a cheap GPU to "make AI" on when that's a market of maybe 1000 people. This Intel card is perfect for an espor…

Intel GPUs don't sell well to gamers. They've been on the market for years now. >market of maybe 1000 people The market of people interested in local ai inference is in the millions. If it's cheap enough the data center market is at least 10 million.

> Intel GPUs don't sell well to gamers. They've been on the market for years now.

Intel has only had discrete GPUs on the market for 2 years. I guess that is a plural number of years, but only barely.

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

#459

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 don't know the numbers but the manufacture of the chip and the cards can't be that expensive...the design was probably much more expensive. Hopefully they are at least breaking even but hopefully making money. Nobody goes into business to lose money. Shareholders would be pissed!

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

#460

Earlier quoted context omitted.

You use at least half of this stack for desktop setups. You need copying daemons, the ecosystem support (docker-nvidia, etc.), some of the libraries, etc. even when you're on a single system. If you're doing inference on a server; MIG comes into play. If you're doing inference on a larger cloud, GPU-direct storage comes into play. It's all modular.

It's possible you're underestimating the open source community. If there's a competing platform that hobbyists can tinker with, the ecosystem can improve quite rapidly, especially when the competing platform is completely closed and hobbyists basically are locked out and have no alternative.

That would require competently produced documentation. Intel can't do that for any of their side projects because their MBAs don't get a bonus if the tech writers are treated as a valuable asset.
Post reply on HN