Every week, I get another email asking if I'd review some random AI box from a random company—I don't know if this Pocket Lab was offered for review at some point, but it sounds very similar to others:<p><pre><code> - Commodity Arm SoC (or sometimes N100 or N150 x86)
- 8/16/32 GB of LPDDR5x RAM
- 'NPU' (usually unspecified) with ambiguous 'TOPS' number (like 20, 40, 80)
</code></pre>
Usually specifics aren't provided, and TOPS is never defined in a technically useful way. The few times it is, are from more established companies (e.g. Asus or Raspberry Pi integrating a well-known NPU chip into one of their products).<p>It's worse at this point than the peak of the crypto boom, when I was getting emails touting the next chain-of-proof software, or ledger-this/ledger-that. Now that there are a few actual use cases for this hardware, it requires more nuance to separate the wheat from the chaff.<p>And for me, I spend weeks, typically, with any hardware I _do_ review, running as many models and test runs as I can (and documenting everything on GitHub, in depth, with scripts so other people can verify). Most reviewers (like those with publications named in this post) either don't have the time, or sadly, the understanding, to test these devices in a meaningful way.<p>Therefore, random blog posts (which are getting harder and harder to find, amidst the AI-laden first 2-4 pages of DuckDuckGo and Google results) are the best source of information. Or sometimes a post on Mastodon, which is never easy to find since search isn't a thing there.<p>Edit: Ah, they did reach out around CES time. Funny seeing their pitch deck including a note on Dr. Miles Mi, with a row of logos on that page including Apple, MIT, Berkeley, DJI, VIVO, Tuya, and a few others, as if they were using this project or something?
I genuinely feel disrespected if AI is used to write an article and it's not disclosed in the first paragraph. It's not <i>really</i> that big of a deal, tbh, it's like saying "I took a picture, I didn't paint it."<p>Which is fine, but please disclose it. Otherwise, like in this case, I'm going to assume the author is a moron that can't write for shit who thinks their readers are morons that can't read for shit.
Someone posted this analysis to their Kickstarter comments (they dodged)<p><a href="https://www.kickstarter.com/projects/tiinyai/tiiny-ai-pocket-lab/comments?comment=Q29tbWVudC00NjkyOTIxOQ%3D%3D" rel="nofollow">https://www.kickstarter.com/projects/tiinyai/tiiny-ai-pocket...</a>
Tl;Dr it's actually a CIX p1 + 32gb (similar to orange pi 6) and a "160TOPS" NPU accelerator with 48gb - attached via NVME. models will either have to fit in one pool or deal with shuttling data over m.2, the company has some optimizations regarding this but it's still a serious limitation.<p>There you go, two sentences without burying the lede.<p>Is it maybe competitive value anyways though? Even if you only think of the accelerators, 48gb+160TOPS seems comparable to some Strix Halo mini PCS with 64gb - lower memory bandwidth but a few hundred dollars cheaper. If they sold just the accelerator card for $800 or something that would be potentially very interesting.
Previously (23 points, 6 days ago, 6 comments) <a href="https://news.ycombinator.com/item?id=47395786">https://news.ycombinator.com/item?id=47395786</a><p>Including questions of LLM origin. Seems like the OP might have submitted that one (47431685) although there's another copy now (beyond this SCP entry from 3 days ago)
Great research and write-up, maybe a bit <i>too</i> elaborate.<p>Will be interesting to see if a public outcry will happen once these boxes start arriving at those who funded the kickstarter.
I bet they picked the name to be confused with tinygrad
>No cloud. No GPU. No subscriptions. Private, offline, always on.<p>Flagged.