Hacker Newsnew | past | comments | ask | show | jobs | submit | roncesvalles's commentslogin

That slot is called USB-C. I can fully imagine inference ASICs coming in powerbank form factor that you'd just plug and play.

Like the chip-software in Gibson’s sprawl, from the micro-soft to the ROM cowboy to the Aleph, the endgame of computertool distribution is via single-use chunks of quasi-biological computronium

Michael Bay just read "computronium" and spawned an 8 movie franchise in his head.

This would be a hell of a hot power bank. It uses about as much power as my oven. So probably more like inside a huge cooling device outside the house. Or integrated into the heating system of the house.

(Still compelling!)


*the whole server uses 2.2kw or whatever, not a single board. I think that was for 8 boards or something.

Oh does it? Thanks for the clarification then. Their home page said 2.5kW so I assumed that's what it is.

To be fair, 2.5kW does sound too much for a single 3x3cm chip, it would probably melt.


More powwwwaaa!

Yeah, though I suppose once we get properly 3d silicon I would not be surprised at power rating for that, 3cm^3 would be something to behold.


> USB-C

With these speeds you can run it over USB2, though maybe power is limiting.


You would likely need external power anyway.

USB-C is just a form factor and has nothing to do with which protocol you run at which speeds.

I wasn't talking about the form factor.

Not if you need 200w power to run inference.

USB-C can do up to 240W. These days I power all my devices with a USB hub, even my Lipo charger.

Have you seen a device that can supply 240w and act as a data host? Or is the 240w only from dedicated chargers?

I haven't seen one, but I also don't tend to use it for anything other than a power supply, so I wouldn't know. Since the standard supports it, though, it's just a matter of the market needing a device like that.

Pretty sure it'd just be a thumbdrive. Are the Taalas chips particularly large in surface area?

The only product they've announced at the moment [0] is a PCI-e card. It's more like a small power bank than a big thumb drive.

But sure, the next generation could be much smaller. It doesn't require battery cells, (much) heat management, or ruggedization, all of which put hard limits on how much you can miniaturise power banks.

[0] https://taalas.com/the-path-to-ubiquitous-ai/


I wouldn't call that size a small power bank. That chip is in the same ballpark as gaming GPUs, and based on the VRMs in the picture it probably draws about as much power.

But as you said, the next generations are very likely to shrink (especially with them saying they want to do top of the line models in 2 generations), and with architecture improvements it could probably get much smaller.


I’m old enough to remember your typical computer filling warehouse-sized buildings.

Nowadays, your average cellphone has more computing power than those behemoths.

I have a micro SD card with 256GB capacity, and I think they are up to 2TB. On a device the size of a fingernail.


That is all definitely amazing, but data storage is a fundamentally different process with far fewer constraints than continuous computation.

It all uses the same miniaturization techniques, though.

800 mm2, about 90mm per side, if imagined as a square. Also, 250 W of power consumption.

The form factor should be anything but thumbdrive.


mmmhhhhh 800mm2 ~= (30mm)2, which is more like a (biggish) thumb drive.

Thanks!

I haven't had my coffee yet. ;)


Shit happens :D

always after the coffee :)

the radiator wouldn't be though

Well even programmable ASICs like Cerebras and Groq give many-multiples speedup over GPUs and the market has hardly reacted at all.

Seems both Nvidia (Groq) and OpenAI (Codex Spark) are now invested in the ASIC route one way or another.

The problem with groq was they only allowed LORA on llama 8b and 70b, and you had to have an enterprise contract it wasn't self service.

> market has hardly reacted at all

Guess who acqui-hired Groq to push this into GPUs?

The name GPU has been an anachronism for a couple of years now.


Cerebras gives a many multiple speedup but it's also many multiples more expensive.

Well it's less mental load. It's like Tesla's FSD. Am I a better driver than the FSD? For sure. But is it nice to just sit back and let it drive for a bit even if it's suboptimal and gets me there 10% slower, and maybe slightly pisses off the guy behind me? Yes, nice enough to shell out $99/mo. Code implementation takes a toll on you in the same way that driving does.

I think the method in TFA is overall less stressful for the dev. And you can always fix it up manually in the end; AI coding vs manual coding is not either-or.


Anthropic has been killing it with the marketing recently. Wouldn't put it past them for this to be one more brand campaign.

By and large "AI assistant" is not a real thing. Everyone talks about it but no one can point you to one, because it doesn't exist (at least not in a form that any fair non-disingenuous reading of that term would imply). It's one big collective hallucination.

Enterprise doesn't buy chat/meeting products without PSTN interop (dial-in dial-out to traditional phone line). Discord would probably need to double their dev team to add PSTN.

Building something like Slack or Teams to the level that a F500 company would make it their primary videoconferencing solution is a multi-thousand-employee project. It's not a little skunkworks project for 15-20 people in some corner of the office.

That's why TFA is hilariously flawed. When Altman says "tell us what we should build, we'll probably build it!", he's talking about driveways and backyard pools, not the Golden Gate Bridge. It's like asking mall Santa for a summer home in the Hamptons.


> multi-thousand-employee project

I know absolutely nothing about PSTN interop and I'm sure it's very complex to implement. However, at the end of the day, this is just software we're talking about right? Software is cheap and easy to produce these days and I doubt you need thousands of people to implement something that syncs your meeting's audio stream to a phone line especially given that it's a problem that has been solved before.


Hardly. You're going from analog to digital and vice-versa. You probably need specialized appliances. For every country in the world. And it's "solved" but only in proprietary contexts; I don't think there's a standard. Then you need to operate it - you need SREs, bug fixes, keeping up with downstream changes etc.

Adding PSTN to Discord is absolutely a Discord-sized problem.

>Software is cheap and easy to produce these days

Yeah todo list apps


There’s a size of enterprise where you can get away without PSTN integration but do need an answer for SSO and account provisioning/deprovisioning.

Exactly, that's why counting job postings is a terrible proxy for gauging market conditions. Companies may hire anywhere from 0 to 100s of people through the same JD.

>taking their oil

That's what Trump told you to sound badass and edgy. His advisors might have a more complicated rationale that's harder to explain to the public than a single 3-letter word.


Trump? The US has been doing it for over 100 years. Trump is just the latest figurehead, it was oil and other resources all along.

Foreign policy of the US has always been about orchestrating coups to create passive client states for US capitalists more efficiently extract natural resources, going back to 1953 in Iran. Only difference with Trump is he has done away with pretenses. He says the quiet part out loud. He says things like "we want the minerals in Ukraine", and then negotiates a mineral deal. He talks about conquering Panama, Greenland, Canada. He is an unabashed imperialist. It's been at least 70 years of this happening, catch up already. And it goes back even further, to the US controlling the Philippines in 1898, and the Monroe Doctrine in 1823.

Exactly. Posts that say "I got great results" are just advertisements. Tell me what you're doing that's working good for you. What is your workflow, tooling, what kind of projects have you made.

>Over the past year, I’ve been actively using Claude Code for development. Many people believed AI could already assist with programming—seemingly replacing programmers—but I never felt it brought any revolutionary change to the way I work.

Funny, because just last month, HN was drowning in blog posts saying Claude Code is what enables them to step away from the desk, is definitely going to replace programmers, and lets people code "all through chatting on [their] phone" (being able to code from your phone while sitting on the bus seems to be the magic threshold that makes all the datacenters worth it).


>only to have it completely obsoleted a few years later

Almost nothing goes obsolete in software; it just becomes unpopular. You can still write every website you see on the Internet with just jQuery. There are perfectly functional HTTP frameworks for Cobol.


obsolete in the software *industry


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: