TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

Andromeda Cluster: 10 Exaflops* for Startups from Nat Friedman and Daniel Gross

165 pointsby mobileexpertalmost 2 years ago

12 comments

robwwilliamsalmost 2 years ago
Forget LinPack and friends. Jack Dongarra is going to need to switch to the new metric for supercomputers—-kilograms of H100 GPUs—- about 3,300 give or take a few grams for this system.
评论 #36314807 未加载
paxysalmost 2 years ago
&gt; For use by startup investments of Nat Friedman and Daniel Gross<p>&gt; Reach out if you want access<p>I&#x27;m confused by the last two bullet points. Is this website only meant to be used by these &quot;startup investments&quot; or can anyone fill out the linked form?
评论 #36315173 未加载
评论 #36314881 未加载
dekhnalmost 2 years ago
Can the creators explain in more detail: how is this different from (for example) the OpenAI cluster that MSFT built in Azure? Is it hosted in an existing cloud provider, or in a data center? Which data center? Who admins the system, is there an SRE team in case it goes down during training? And can you attempt ot run the same benchmarks that Top500 uses to determine what your <i>double precision flops are</i> and give that number in addition to your &quot;10 exaflops&quot; (which I believe is single precision).
评论 #36314366 未加载
mobileexpertalmost 2 years ago
Emad from Stability estimates this at 4M&#x2F;month. <a href="https:&#x2F;&#x2F;twitter.com&#x2F;emostaque&#x2F;status&#x2F;1668666509298745344" rel="nofollow noreferrer">https:&#x2F;&#x2F;twitter.com&#x2F;emostaque&#x2F;status&#x2F;1668666509298745344</a>
评论 #36313910 未加载
bigdictalmost 2 years ago
lmao are they trolling with the naming<p><a href="https:&#x2F;&#x2F;www.cerebras.net&#x2F;andromeda&#x2F;" rel="nofollow noreferrer">https:&#x2F;&#x2F;www.cerebras.net&#x2F;andromeda&#x2F;</a>
评论 #36315513 未加载
MR4Dalmost 2 years ago
The mainframe is dead. Long love the new mainframe. We just call it DGX because it’s cool.<p>Even the leasing model has made a comeback!
ryanwaggoneralmost 2 years ago
Same guys behind <a href="https:&#x2F;&#x2F;aigrant.org" rel="nofollow noreferrer">https:&#x2F;&#x2F;aigrant.org</a>, maybe it&#x27;s mainly as a way to get dealflow?
评论 #36314579 未加载
评论 #36314746 未加载
Ameoalmost 2 years ago
Would be very cool to see some pictures of the cluster; sounds like an impressive build!
评论 #36313874 未加载
braindead_inalmost 2 years ago
Someone please start the GPT 5 training before the regulation kicks in.
005almost 2 years ago
Looks like they&#x27;ve reserved a bunch of compute from Lambda Labs?<p>Edit: Based off this tweet, looks very similar <a href="https:&#x2F;&#x2F;twitter.com&#x2F;LambdaAPI&#x2F;status&#x2F;1668676838044868620" rel="nofollow noreferrer">https:&#x2F;&#x2F;twitter.com&#x2F;LambdaAPI&#x2F;status&#x2F;1668676838044868620</a>
评论 #36315901 未加载
brucethemoose2almost 2 years ago
&gt; Big enough to train llama 65B in ~10 days<p>Y&#x27;all could totally eat Meta&#x27;s lunch and train an open LLM with all the innovations that have come since LLaMA&#x27;s release. Other startups are trying, but they all seem bottlenecked by training time&#x2F;resources.<p>This could be where the next Stable Diffusion 1.5 comes from.
redshalmost 2 years ago
The new lean startup. 3,291 kg