Taalas HC1: 17,000 tokens/sec on Llama 3.1 8B vs Nvidia H200’s 233 tokens/sec. 73x faster at one-tenth the power. Each chip runs ONE model, hardwired into the transistors.
This sounds great to me. Anything that would increase supply of AI processing could lower demand on the GPU supply. I want to be able to upgrade my gaming computer again someday!
AI really needs dedicated hardware, I feel like if there was more chip manufacturing in the west we might have more diverse chips.
Frankly I’m really confused as to why this llm demand on ram isn’t encouraging new companies to manufacture ram. If this is a bubble then we all just wait it out, if it’s not a bubble then someone else would swoop in to take up the market.
It’s not easy to scale up chip production, because it relies on extremely precise machines, which take a long time to build, and many different steps on the way from raw materials to a finished chip.
If you’d want to set up a new RAM chip factory with competitive performance, it’d be an investment of over a billion USD at the bare minimum, and it’d take a few years to set up all the processes because the first chips can roll off the assembly line.
If the bubble popped by then, then your new factory would probably run at a loss because it’s nearly impossible to complete with companies who have had decades to optimise their production processes.
Even if the bubble didn’t pop by then, then the next problem will likely be the wafer supply. Because just like how there are only a few companies with the infrastructure to build modern, high-performance computer chips, there are only a few companies with the infrastructure to build silicon wafers of a high enough quality to build those chips with. And they have only just enough capacity to supply their current customers.
So to then solve the wafer problem, someone needs to be willing to invest at least a few hundreds of millions of USD to build a new factory for those, which again would struggle to complete in a post-scarcity market. And wafers are far from being the only resource with that issue.
TL;DR: It’s be a huge investment and a huge gamble, and would likely end up just moving the problem anyway.
Each chip runs ONE model, hardwired into the transistors.
That’s… that’s an ASIC. That’s literally just an ASIC… with all the tradeoffs and compromises that come with it.
ASIC just means “specialized chip”. They don’t claim anything else.
Shh you’ll pop the bubble if you start talking sensibly. It’s not an ASIC—it’s a specialized piece of hardware optimized to execute a model with unparalleled performance. Now buy my entire stock of them and all the supply for the next two years please.
(Figuring out the compose combination for an emdash took longer than I’d like to admit lol)
Dedicated, single purpose, chip designs are always going to be faster and more efficient to run than general purpose ones. The question will be what the environmental, and financial costs will be of updating to a new model. With a general purpose design it’s just a case of liading sone new code. With a model that’s baked into the silicon you have to design and manufacture new chips, then install them.
I can see this being useful in certain niche usecases where requirements are not going to change, but it sounds rather limiting in the general case.
A lot of the models we have are about as good as they are going to get. I mean, ChatGPT 5 isn’t appreciably better than ChatGPT 4. Hook one of those models or even one not as strong to a purpose-built RAG pipeline and a controller to run as mesh of interconnected prompts and agents, and you’ll blow away general purpose chatbots in niche areas in terms of cost, efficiency, and performance.
The question then becomes, to what purpose can you put this super fast, dedicated machine that performs certain small-scopes, simple tasks really well, but also fucks up often enough that you can’t depend on it. To what tasks could you set a bot that does stuff with minimal competence let’s say 90% of the time, and the other 10%, doesn’t create even bigger problems?
That domain exists, but it’s thin and narrow.
To what tasks could you set a bot that does stuff with minimal competence let’s say 90% of the time, and the other 10%, doesn’t create even bigger problems?
Sounds like a typical human to me.
A chip like this would be perfect for an autonomous robot. Drone, humanoid, whatever - something that still needs to be able to handle itself when it’s cut off from outside control. Always nice to have an internet connection to draw on a bigger, more capable “brain” somewhere else, but if that connection is lost you want it to be able to carry on with whatever it’s doing and not just flop over limply.
Sure. It excels in cases where 60-90% success rate is better than nothing. If you have a smart mine that doesn’t detonate on civilians, 50% success is better than 0. It reduces civilian casualties by 50%, which is still awful, but if you’re going to plant mines it’s better than entirely indiscriminate. Use cases definitely exist. A false positive means it doesn’t detonate on one soldier but might on the next — still an effective deterrent. A false negative means it blows up a kid, which a dumb mine would also do anyway.
It’s just generally not in situations most people are generally thinking about. You have to imagine cases where there is some upside and no downside. It doesn’t work in a context of say, auto-breaking a car if a pedestrian is detected because a false positive is going to cause accidents and probably kill people even if in other circumstances it does save lives.
Why doesn’t it work in those contexts? It’s better than nothing in those contexts too. I’d rather have a car with onboard intelligence to take over than an uncontrolled one.
I think you’re letting the perfect be the enemy of the good, here. There are plenty of situations where you don’t need a robot to behave perfectly. People don’t behave perfectly.
No, it doesn’t work in this context because false positive is worse than nothing. False negative is better than nothing. Zero sum. Obviously it depends where you set the threshold of false positive and false negative. I imagined a very simple scenario the first time.
If even only .001% of the time, you’re going to cause a shit load accidents. You’re going to average a car slamming on the breaks for no reason like every… 2 minutes would be .12, 20 would be 1.2, 200 would be 12% 800 would be 48%, so you’re going to have every car slam on their breaks every 12-15 hours of drive time. That would be an absolute mess.
I have no idea what you’re thinking the scenario is here. The alternative is an uncontrolled car, I think I’d rather it had at least some brains behind the decisions it’s making.
How does it decide the car is uncontrolled? That’s a failure scenario, too.
I’m not even sure what you’re arguing. I said from the get go that there are niche cases where AI is nothing but positive. You seem to be arguing that there are a bunch more cases. Fine. Maybe the niche is slightly less thin and narrow than I think. Cool.
