purchasable Ø

Longing tops since '17. Tweets are not financial advice.

[purchasable Ø @isdeploying]

is FTX crashing? $BTC #btc #bitcoin

[purchasable Ø @isdeploying]

@jun_song How many base mac mini m4 cluster for this setup? 16?

[purchasable Ø @isdeploying]

@hxiao I remember you did wired limit 512gb ram. So it went out of memory..

[purchasable Ø @isdeploying]

@ivanfioravanti @pcuenq So distributed LLM inference can actually increase the decode?

[purchasable Ø @isdeploying]

@TheAhmadOsman I’m running cluster Mac minis with tensor split on llama.cpp.. Will vLLM increase the tok/s?

[purchasable Ø @isdeploying]

@hxiao How did it not crash with wired limit at max ram? I allocated 2gb ram for my machines so they can at least run the OS and basic stuff.

[purchasable Ø @isdeploying]

@iamlukethedev @NousResearch Which LLM you running and hardware specs?

[purchasable Ø @isdeploying]

@ashen_one @AlexFinn Sir how many base mac minis do we cluster to achieve this? Haha

[purchasable Ø @isdeploying]

@leyten @Zai_org wow decentralized compute

[purchasable Ø @isdeploying]

@antirez Is ssd stream the same as mmap?

[purchasable Ø @isdeploying]

@jun_song Yo how many Mac mini m4 in cluster for this build? Rpc via llama cpp

[purchasable Ø @isdeploying]

@TheAhmadOsman Will my Mac mini cluster running on llama.cpp make it? 🥹

[purchasable Ø @isdeploying]

@Teknium What if I’ve already built the profile before this? Do I ask the agent to move it into the new directory?

[purchasable Ø @isdeploying]

@alexocheema @exolabs Why is it with exolabs I cannot customize the flags but I can with llama.cpp when clustering? Will you all start allowing customisations?

[purchasable Ø @isdeploying]

@predictdotfun @dingalingts Spot me $300 so I can tell ppl all you need is ask!

[purchasable Ø @isdeploying]

@radmadvlad A Mac mini allows any user to use it directly out of the box, and also as an infra for inference. This thing is cheaper, but not every user knows how to use it.

[purchasable Ø @isdeploying]

@arkham @machibigbrother can you spot me a $mil bug brother?

[purchasable Ø @isdeploying]

@LottoLabs @0xSero I’ve 3 base Mac mini m4 cluster running the qwen3.6 35b q4 kp. About 20 tok/s decode

[purchasable Ø @isdeploying]

@antirez Is this the same as - -mmap?

[purchasable Ø @isdeploying]

@Teknium I just make my Hermes Agent self clone itself to the new machine. No github repo, no fuss. Ez

[purchasable Ø @isdeploying]

@ItsmeAjayKV @franktheglock Gm bros, is running 2x 3060 more efficient in power consumption or better to run cluster Mac minis?