I think AMDs offer was fair (full remote access to several test machines), then again just giving tinycorp the boxes on their terms with no strings attached as a kind of research grant would have earned them some goodwill with that corner of the community.
Either way both parties will continue making controversial decisions.
> Now, why don't they send me the two boxes? I understand when I was asking for firmware to be open sourced that that actually might be difficult for them, but the boxes are on eBay with a simple $$ cost. It was never about the boxes themselves, it was a test to see if software had any budget or power. And they failed super hard
I know this is someone else's reasoning, so you can't answer this question, but, doesn't this just test if they want to spend the budget on this specific thing?
If I ask a company for a $100,000 grant, and they're not willing, it doesn't seem like correct logic to assume that means they don't have the budget for it. Maybe they just don't want to spend $100,000 on me.
Why does this mean they don't have a budget or power?
He assumes the software department wants to do this, which - yes - seems to be flawed logic on his side.
Let's imagine he's indeed correct. He receives the hardware, get's hacking and solves all of AMDs problem, the stock surges and tinygrad becomes a major deep learning framework.
That would be a collosal embarrassment for AMDs software department.
Chip vendors regularly send out free hardware to software developers. In this case I don't think the cost is the issue; AMD simply doesn't want what Geohot is offering.
Considering that AMD is only really supporting their datacenter GPUs with ROCm, this is the worst possible response. It means compute on AMD GPUs is only meant for the elite of the elite and forever out of reach for the average consumer and that Nvidia is not only outcompeting AMD on quality but also on cost.
> He refused. It had to come from AMD. That's absurd and extortionist.
I'm on the wrong side of the Twitter wall to read the source, but that doesn't sound absurd. Extortionist, maybe. Hotz's major complaint (last time I checked, anyway) is pretty close to one I have - AMD appears to have between little and no strategic interest in consumer grade graphics cards having strong GPGPU support leading to random crashes from the kernel drivers and a certain attitude of "meh, whatever" from AMD corporate when dealing with that.
I doubt any specific boxes or testing regime are his complaint, he'd be much more worried about whether AMD management have any interest in companies like his succeeding. Third parties providing some support doesn't sound like it'd cut it. The process of being burned by AMD leaves one a little leery of any alleged support without some serious guarantees that more major changes are afoot in their management view.
> ...he'd be much more worried about whether AMD management have any interest in companies like his succeeding.
This reads as incredibly entitled. AMD owes him nothing, especially if he's opposed to the leadership's vision[1] and being belligerent about it.
There is maybe 1 or 2 companies with enough cachet to demand management changes at a supplier like AMD - and they have market caps in the trillions.
1. Lisa Su hasn't been shy about AMD being all about partnering with large partners who can move volume. My interpretation of this is AMD prefers dealing with Sony, Microsoft, hyperscalers, and HPC builders, then possibly tier II OEMs. Small startups are probably much further down the line, close to consumers at the tail end of AMD's attention queue. I don't like it as a consumer, but it seems like a sound strategy since the partners will shoulder most of the software effort, which is a weakness AMD has against Nvidia. They can focus on cranking out ok-to-great hardware at more-than-ok prices and build up a warchest for future investments, and who knows when this hype bubble will burst and take VC dollars with it, or someone invents an architecture that's less demanding on compute (if you're more optimistic)
Sure. But we hear a lot about Hotz because all the unentitled people rolled their eyes and went over to buy Nvidia cards. He's one of the major voices who are unreasonable enough to pipe up on Twitter and air dirty laundry.
I doubt AMD are going to listen to him. They're in a great spot and are probably going to tap into the market in a big way. But Hotz isn't crazy to test them in an odd way - although he'd probably be better off dropping AMD cards like most other people in his price range would.
> But Hotz isn't crazy to test them in an odd way..
He should have just read the Lisa Su interview from Q1 2024 where ahe laid out AMDs strategy without equivocating
> ... although he'd probably be better off dropping AMD cards
I think this is what's best for everyone. Looking at his recent track record[1], he seems like a person who's gets really excited by kicking things off and experiencing the exponentially growth phase, and then when it flattens out into a sigmoid curve, he dusts his hands and declares his work done, and moves to the next thing.
. 1. Hired by Elon to "fix" Twitter, CommaAI, and soon, Tiny
One might argue he's had a pattern for even longer. While he did do some early hypervisor glitching, even his PS3 root key release was basically just applying fail0verflow's ECDSA exploit (fail0verflow didn't release the keys specifically because they didn't want to get sued ... so that was a pretty dick move [1]).
For his projects, I think it's important to look at what he's done that's cool (eg, reversing 7900XTX [2], creating a user-space driver that completely bypasses AMD drivers for compute [3]) and separating it from his (super cringe) social media postings/self-hype.
Still, at the end of the day, here's hoping that someone at AMD realizes that having terrible consumer and workstation support will basically continue to be a huge albatross/handicap - it cuts them off basically all academic/research development (almost every single ML library and technique you can name/used in production is CUDA first because of this) and the non-hyperscaler enterprise market as well. Any dev can get a PO for a $500 Nvidia GPU (or has one on their workstation laptop already). What's the pathway for ROCm? (honestly, if I were in charge, my #1 priority would be to make sure ROCm is installed and works w/ every single APU installed, even the 2CU ones).
I don't really see why those companies would prefer AMD over Nvidia, they are not hurting for money and therefore able to spend that money on Nvidia or build their own hardware, like Google did.
Meta and Microsoft are big enough they could just build their own TPUs with a stable software stack and cut off Nvidia and AMD at the same time.
From this perspective, AMD only ever makes sense as an "also ran company" for a few niche use cases.
> This reads as incredibly entitled. AMD owes him nothing, especially if he's opposed to the leadership's vision[1] and being belligerent about it.
A generation ago, everyone in sales and developer relations understood that "the customer is always right". Remember a sweaty dude on stage jumping about screaming "developers! developers! developers"? It was exhausting dealing with all the free software and hardware sent to developers, not to mention the endless free conferences for even the most backwater developer community. But that's an ethos for boomers, I guess.
On the one hand "incredibly entitled" and on the other you talk about AMD's leadership vision. Your long closing paragraph shows that entitlement of a developer has nothing to do with anything and isn't relevant in the conversation (I can show you guys at OEMs who are incredibly arrogant and entitled or outright a$$holes but so what?). It's just an opinion based on your personal bias.
In reality, AMD simply doesn't care about small AI startups or developers as you've noted. They don't care about me wanting to run all my AI locally so that I can manage my dairy farm with a modest fleet of robots. If they cared, and they sent him MI300s immediately (or sent them to the other 8 startups that asked for them), you wouldn't be chastising him about being "incredibly entitled".
> AMD appears to have between little and no strategic interest in consumer grade graphics cards having strong GPGPU support leading to random crashes from the kernel drivers and a certain attitude of "meh, whatever" from AMD corporate when dealing with that.
AMD has little interest in software support in general.
Their Adrenalin software is riddled with bugs that have been here for years.
Having watched some of his streams on the topic, I think you've captured it well. He's basically saying he's done wasting time on AMD unless/until they get serious. It's not so much that he wants free hardware from them, rather he wants to see them put some skin in the game as they basically blew him off the last time he tried to engage with them.
We do care about software and acknowledge the gaps and will work hard to make it better. Please let me know any specific issues that are an issue for you and Im happy to push for it to get resolved or come back with why it isn't.
I think AMDs offer was fair (full remote access to several test machines), then again just giving tinycorp the boxes on their terms with no strings attached as a kind of research grant would have earned them some goodwill with that corner of the community.
Either way both parties will continue making controversial decisions.