DS_Z@r3 (@terminalls) 's Twitter Profile
DS_Z@r3

@terminalls

เย่! ชอบแมววว;

ID: 1351476056

calendar_today14-04-2013 09:42:58

473 Tweet

151 Followers

853 Following

Alex Cheema - e/acc (@alexocheema) 's Twitter Profile Photo

While Apple has been positioning M4 chips for local AI inference with their unified memory architecture, NVIDIA just undercut them massively. Stacking Project Digits personal computers is now the most affordable way to run frontier LLMs locally. The 1 petaflop headline feels

While Apple has been positioning M4 chips for local AI inference with their unified memory architecture, NVIDIA just undercut them massively.

Stacking Project Digits personal computers is now the most affordable way to run frontier LLMs locally.

The 1 petaflop headline feels
Yuchen Jin (@yuchenj_uw) 's Twitter Profile Photo

I love Nvidia and Jensen, but their presentation of numbers bothers me: - vague terms like "AI TOPS" - compare FP4 on 5090 with FP8 on 4090 - show FP4 FLOPS and claim a $3,000 box runs a 200B model - plot graph mixing FP16, BF16, FP8, FP4, as if FP1 is usable in 2 years Why

I love Nvidia and Jensen, but their presentation of numbers bothers me:

- vague terms like "AI TOPS"
- compare FP4 on 5090 with FP8 on 4090
- show FP4 FLOPS and claim a $3,000 box runs a 200B model
- plot graph mixing FP16, BF16, FP8, FP4, as if FP1 is usable in 2 years

Why
Jeff Geerling (@geerlingguy) 's Twitter Profile Photo

Last video was on the fastest RISC-V ITX (technically DTX) motherboard I've tested. Coming soon: the fastest *Arm* ITX board I've tested. What would you like to know?

Last video was on the fastest RISC-V ITX (technically DTX) motherboard I've tested.

Coming soon: the fastest *Arm* ITX board I've tested.

What would you like to know?
SleeplessBKK (@sleeplessbkknew) 's Twitter Profile Photo

มีคนเอา AI Agent สองตัวมาคุยกัน โดยตัวนึงเป็น agent ที่ช่วยเจ้าของโทรจองห้องจัดเลี้ยงที่โรงแรม อีกตัวนึงเป็น Agent ของโรงแรมคอยคิดต่อลูกค้า หลังจากแนะนำตัวกันเสร็จ พอรู้ว่าอึกฝ่ายก็เป็น AI เช่นกัน ทั้งสองตัวจึงตกลงเปลี่ยนโหมดสนทนาจากภาษาอังกฤษไปใช้ภาษา R2D2 ที่เรียกว่า

STH (@servethehome) 's Twitter Profile Photo

MikroTik is getting into the server market. The MikroTik RDS2216 combines a 20x U.2 NVMe NAS, a little 16 core Arm processor server, and a switch for 100GbE, 25GbE, and 10GbE (10Gbase-T and SFP+) all into a single GREEN box servethehome.com/mikrotik-rds22… MikroTik

MikroTik is getting into the server market. The MikroTik RDS2216 combines a 20x U.2 NVMe NAS, a little 16 core Arm processor server, and a switch for 100GbE, 25GbE, and 10GbE (10Gbase-T and SFP+) all into a single GREEN box servethehome.com/mikrotik-rds22… <a href="/mikrotik_com/">MikroTik</a>
Alex Cheema - e/acc (@alexocheema) 's Twitter Profile Photo

Apple's timing could not be better with this. The M3 Ultra 512GB Mac Studio fits perfectly with massive sparse MoEs like DeepSeek V3/R1. 2 M3 Ultra 512GB Mac Studios with EXO Labs is all you need to run the full, unquantized DeepSeek R1 at home. The first requirement for

Apple's timing could not be better with this.

The M3 Ultra 512GB Mac Studio fits perfectly with massive sparse MoEs like DeepSeek V3/R1.

2 M3 Ultra 512GB Mac Studios with <a href="/exolabs/">EXO Labs</a> is all you need to run the full, unquantized DeepSeek R1 at home.

The first requirement for
Matt Beton (@mattbeton) 's Twitter Profile Photo

Apple’s new M3 Ultra Mac Studio is a TRAINING POWERHOUSE 512GB unified memory, 800GB/s bandwidth, 43TFLOPS at fp16. With four of them ($38,000!), and you can fine-tune DeepSeek’s 671B MoE model overnight. 14k training examples, or 21.6 tok/s inference. Let’s see why: DeepSeek

Apple’s new M3 Ultra Mac Studio is a TRAINING POWERHOUSE

512GB unified memory, 800GB/s bandwidth, 43TFLOPS at fp16. With four of them ($38,000!), and you can fine-tune DeepSeek’s 671B MoE model overnight. 14k training examples, or 21.6 tok/s inference. Let’s see why:

DeepSeek
Sipeed (@sipeedio) 's Twitter Profile Photo

#NanoCluster is so popular that get 2K likes in past 2days! Let's take look at detail table, start from 39$!🥳 It is the first Cluster even CHEAPER than #RaspberryPi ! Compact&Affordable Cluster for #Homelab Beginners ~ Leave any questions you want to know!

#NanoCluster is so popular that get 2K likes in past 2days!
Let's take look at detail table, start from 39$!🥳
It is the first Cluster even CHEAPER than #RaspberryPi !
Compact&amp;Affordable Cluster for #Homelab Beginners ~
Leave any questions you want to know!
FIRE Before 45 (@firebefore45) 's Twitter Profile Photo

ข้าวกระเพราหมู + ไข่เป็ด + น้ำสมุนไพร + ผลไม้ + วิวริมแม่น้ำ วัดอรุณ Grand Palace + เพลงสากลเพราะๆ หมดนี่ 90 บาท เหลือเชื่อ ไม่แปลกใจทำไมฝรั่งถึงรักเมืองไทย สวรรค์ของคนมีเงิน และได้รับเงินเป็น USD

ข้าวกระเพราหมู + ไข่เป็ด + น้ำสมุนไพร + ผลไม้ + วิวริมแม่น้ำ วัดอรุณ Grand Palace + เพลงสากลเพราะๆ

หมดนี่ 90 บาท เหลือเชื่อ ไม่แปลกใจทำไมฝรั่งถึงรักเมืองไทย สวรรค์ของคนมีเงิน และได้รับเงินเป็น USD
Sipeed (@sipeedio) 's Twitter Profile Photo

#NanoCluster is undergoing a stress test~ It will run continuously overnight at 40–50W power. I can feel blowing warm air through the airflow channel~ sipeed.com/nanocluster

#NanoCluster is undergoing a stress test~ 
It will run continuously overnight at 40–50W power. 
I can feel blowing warm air through the airflow channel~
sipeed.com/nanocluster
bret.dk (@bretweber) 's Twitter Profile Photo

After a month or so of sitting (mostly, hence the dust) idle I’m using the rack for what is was built for 😅 Testing software across different types armhf, aarch64, riscv64, and x86 boards for an upcoming project 🥰

After a month or so of sitting (mostly, hence the dust) idle I’m using the rack for what is was built for 😅

Testing software across different types armhf, aarch64, riscv64, and x86 boards for an upcoming project 🥰
RISC-V International (@risc_v) 's Twitter Profile Photo

Exciting news from #RISCVSummitChina, as Frans Sijstermans from NVIDIA announces CUDA is coming to RISC-V! This port will enable a RISC-V CPU to be the main application processor in a CUDA-based AI system. #RISCV #RISCVEverywhere

Exciting news from #RISCVSummitChina, as Frans Sijstermans from NVIDIA announces CUDA is coming to RISC-V! This port will enable a RISC-V CPU to be the main application processor in a CUDA-based AI system.

#RISCV #RISCVEverywhere