This is a most excellent place for technology news and articles.
It's a pretty decent value when stacked up against RTX 4000 and RX 7000 GPUs.
But we're only a month or two from the next generation of Nvidia & AMD cards.
Those companies could even shit the bed for a second generation in a row on price-to-performance improvements, and the B580 will probably just end up being in-line with those offerings.
Yeah,but by the time the 5060 is available, the tarrifs will have it at $450+.
Yeah, I'll be curious to see how that all plays out.
Current GPU pricing still seems to have the 2019-2020 25% GPU tariff price baked-in. Note how prices didn't drop 25% when those were rescinded.
Do Nvidia & AMD factor those in their pricing and give consumer a break? Or do they just jack up prices again and aim for mega-profits?
Hell, will the tariffs even happen? At one point, those tariffs were supposedly contingent on U.S. Federal income taxes being abolished, and being used to replace that government tax income. The income tax part seems to have been dropped from the narrative ever since the election.
How would this compare to an AMD RX580? I can't find this in the usual charts
I compared my Vega 56 with the RX 7900 GRE, which would be a 2.5x to 3x performance upgrade. I'd imagine the RX 580 to B580 swap would be in the same ballpark.
Looking at Vega's release reviews though, it was 40% faster than the RX 580. I assume your gains would be higher than 200%.
How is compatibility with older games now?
Because I'm not buying a GPU unless it works with everything.
We'll see come launch, but even the original Arc cards work totally fine with basically all DX9 games now. Arc fell victim to half baked drivers because Intel frankly didn't know what they were doing. That's a few years behind them now.
Intel designed their uarch to be DX 11/12/Vulkan based and not support hardware level DX9 and older drawcalls, which is a reasonable choice for a ground-up implementation- however it does also mean that it only runs older graphics interpreters using a translation/emulation layer, turning DX9 into DX12. And driver emulation is an always imperfect science.
A lot of it will have been because half the game optimisation code was often inside the drivers.
So Intel devs may not know what they were doing, but game devs are often worse.
All these weird numeric names. I'm gonna build a GPU and name it Jonathan.
Jonathan, what are you doing?
The Arc cards actually have a really fun generational naming mechanic.
It's RPG classes. First gen was Alchemist. Second (what the article is about) is Battlemage. I'm guessing we're getting Cleric, Druid, etc.
It's Celestial, so sayeth Steve
Pls no.
Hello, My name is Roger!
sorry, apple already took that one. call it Jeff or something.
Just don't name it Steve. You're in for a world of troubles with GPU Steve.
If they double up the VRAM with a 24GB card, this would be great for a "self hosted LLM" home server.
3060, 3090 prices have been rising like crazy because Nvidia is vram gouging and AMD inexplicably refuses to compete. Even ancient P40s (double vram 1080 TIs with no display) are getting expensive. 16GB on the A770 is kinda meager, but 24GB is the point where you can fit the Qwen 2.5 32B models that are starting to perform like the big corporate API ones.
And if they could fit 48GB with new ICs... Well, it would sell like mad.
I always wondered who they were making those mid- and low-end cards with a ridiculous amount of VRAM for... It was you.
All this time I thought they were scam cards to fool people who believe that bigger number always = better.
Yeah, AMD and Intel should be running high VRAM SKUs for hobbyists. I doubt it'll cost them that much to double the RAM, and they could mark them up a bit.
I'd buy the B580 if it had 24GB RAM, at 12GB, I'll probably give it a pass because my 6650 XT is still fine.
Don’t you need nvidia cards to run ai stuff?
Not at all
Nah, ollama works w/ AMD just fine, just need a model w/ enough VRAM.
I'm guessing someone would get Intel to work as well if they had enough VRAM.
Also "ridiculously" is relative lol.
The Llm/workstation crowd would buy a 48GB 4060 without even blinking, if that were possible. These workloads are basically completely vram constrained.
An LLM card with quicksync would be the kick I need to turn my n100 mini into a router. Right now, my only drive to move is that my storage is connected via usb. SATA is just not enough value for a whole new box. £300 for Ollama, much faster ml in immich etc and all the the transcodes I could want would be a "buy now figure the rest out later" moment.
Oh also you might look at Strix Halo from AMD in 2025?
Its IGP is beefy enough for LLMs, and it will be WAY lower power than any dGPU setup, with enough vram to be "sloppy" and run stuff in parallel with a good LLM.
*adds to wishlist
You could get that with 2x B580s in a single server I guess, though yoi could have already done that with the A770s.
... That's nuts. I only just graduated to a mini from a pi, I didnt consider a dual GPU setup. Arbitrary budget aside, I should have added an "idle power" constraint too. Reasonable to assume that as soon as LLMs get involved all concept of "power efficient" goes out the window. Don't mind me, just wishing for a unicorn.
Strix Halo is your unicorn, idle power should be very low (assuming AMD VCE is OK over quicksync)
Intel GPU claims are NEVER true.
Meh, I ended up with an A770 for a repurposed PC and it's been pretty solid, especially for the discounted price I got. I get that there were some driver growing pains, but I'm not in a hurry to replace that thing, it was a solid gamble.
The A770 was definitely a "fine wine" card from the start. Its raw silicon specs were way stronger than the competition, it just needed to grow into it.
This ones a bit smaller though...
Funny the Radeon RX 480 came out in 2016 at a similar price. Is that a coincidence?
Incidentally the last great generation offering a midrange GPU at a midrange price. The Nvidia 1060 was great too, and the 1080 is claimed to maybe be one of the best offers of all time. Since then everything has been overpriced.
The RX 480 was later replaced by the 580 which was a slight upgrade at great value too. But then the crypto wave hit, and soon a measly 580 cost nearly $1000!!! Things have never returned quite back to normal since. Too little competition with only Nvidia and AMD.
30 series started to look like a return to good priced cards. Then crypto hit and ruined that. Now we have AI to keep that gravy train going. Once the AI hype dies down maybe we'll see cards return to sane pricing.
I’m reserving judgement of course to see how things actually play out but I do want to throw a cheapest pc together for my nephew and that card would make a fine centerpiece.
As someone with a 6650 XT, which is a little slower than the 6700 or 4060, I doubt the increased vram, which is of course still nice, is enough to push it for 1440p. I struggle even in 1080p in some games, but I guess if you're okay with ~40 FPS then you could go that high.
Unfortunately, if the 4060 is roughly the target here, that's still far below what I'm interested in, which is more the upper midrange stuff (and I'd love one with 16 GB vram at least).
At least the price is much more attractive now.
Dunno, realisticly speaking it is a slightly cheaper 7600, hardly a market shake-up.
