Associate
- Joined
- 1 Sep 2013
- Posts
- 1,595
Whats the ai crap for?
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
The AMD Driver Thread
- Thread starter LtMatt
- Start date
Didn't you watch CES let me do a quick recap for you, AI AI AI AI AI AI new CPU that's 5% faster then the last one new APU's designed for AI and more AI AI AI AI
Installed, thanks. No issues.
It's not really crap if you are interested in that sort of thing, but yes not relevant for us gamers. Have a read of this if interested:
Soldato
- Joined
- 29 Aug 2010
- Posts
- 8,727
- Location
- Cornwall
Surely if you were interested in that sort of thing you'd have already found a way to make it work?
Unless it was impossible to do otherwise?
I don't think Nvidia have such a bundle is it not possible to run LLMs locally on an Nvidia GPU? (I've not looked into it)
Not something I've looked into either yeah, doesn't interest me. I'm not sure.
It was fiddly and annoying. The optional AI install now does all the heavy lifting for you so you can do local models with little effort. I don't see a problem with AMD offering an extra option for AI content people to use their products.
Last edited:
Soldato
- Joined
- 29 Aug 2010
- Posts
- 8,727
- Location
- Cornwall
Personally I'd rather they fix bugs than add this, but each to their own I guess.
It's probably the product of a different team, so is likely not impacting the bug-fixing team. It's doing things like ROCm to break the CUDA stranglehold of Nvidia. If more people buy AMD because the drivers support their professional needs, then more cards are sold, more money for more driver/software development, etc.
It's really no different than AMD spending time on CPUs and GPUs at the same time. It doesn’t mean that one is being sacrificed for the other, it means there's a bigger picture that needs supporting so that AMD can have a long term investment in one of their unique selling points, ie they can do both x86/x64 CPUs and GPUs.
One of the big reasons Nvidia has soared away in the professional market was their investment in a proprietary CUDA, and AMD has had a lot of flak for not doing the same, but it's difficult against an incumbent of that size.
Last edited:
Soldato
- Joined
- 29 Aug 2010
- Posts
- 8,727
- Location
- Cornwall
I get that but as discussed, this was already possible. They've just spent time and resources making it usable for the lazy people, and those are probably not the people using it professionally.
Nvidia didn't soar away with the market by including a 34GB AI bundle download in their drivers.
We don't know if the team that did this could've been working on driver fixes, maybe they could, maybe they couldn't. But AMD could've spent the money they spent hiring those people on people that could fix driver bugs.
Nvidia seem to be shifting all their people to AI too and I believe they have a lot of complaints about recent drivers possibly as a result.
I think we all know why AMD have done it, it's because "AI" is a buzzword and AMD want to use the buzzword as much as possible (see the recent CES stuff). Then we're the ones that get to suffer. Maybe if they just made the bundle available without including it in their drivers it'd feel a little like they're trying to force feed us AI. Could've spent time adding in FSR4 support for RDNA3 cards, but AI is a bigger buzzword. Also, wouldn't the time for doing all the ROCm work and including it in drivers to get it out to people have been before Nvidia dominated the AI scene? This feels like it might be a little late.
Maybe I'm just hoping it's too late and it fails to gain any real traction so AMD don't effectively pull out of the consumer graphics card market to focus on AI too.