aychocho

55 posts

aychocho banner
aychocho

aychocho

@ChosenOne229

Building @quok_it | CS @pennengineers | @franklin_dao dev + investments

Katılım Mayıs 2022
194 Takip Edilen118 Takipçiler
aychocho
aychocho@ChosenOne229·
@theZachChen psmux, it’s basically tmux but on powershell
English
2
0
1
31
Zach Chen
Zach Chen@theZachChen·
best way to run 4+ terminals at once? I'm on windows using powershell. But down to try other things if it can boost my agentic coding
English
2
0
1
215
Zach Chen
Zach Chen@theZachChen·
if you're using codex / claude code via terminal, what is your way of editing files quickly in a terminal setting? without using the AI (editing .env, etc)
English
3
0
0
164
aychocho
aychocho@ChosenOne229·
All I want for Christmas is a Claude code plugin for vim
English
0
0
1
68
aychocho
aychocho@ChosenOne229·
GPUs, for some reason.
GIF
English
2
1
2
91
aychocho
aychocho@ChosenOne229·
GPU falling off the bus (XID79) can’t be hotfixed… can it??? 👀👀👀
English
1
1
2
75
aychocho
aychocho@ChosenOne229·
We just bought a bunch of GPUs, anyone have a rig we can test them on this weekend?
aychocho tweet media
English
0
0
0
57
aychocho
aychocho@ChosenOne229·
Hey look it’s me
aychocho tweet media
English
0
0
0
39
aychocho
aychocho@ChosenOne229·
@teromee @HotAisle Not sure about the AMD/Intel NPUs, but China seems to be betting on Huawei’s NPU in a push to nationalize their AI supply chain
English
0
0
0
18
teromee
teromee@teromee·
You know I do have a question. Has anyone actually tried to write a usable program for any of the npus that AMD and Intel have put in some of their latest generation chips? Also, I do agree with you. There's just something really important. We all got to remember. There's always useful. Compute you just got to be willing to spend your time learning how to access it.
English
1
0
0
23
Hot Aisle
Hot Aisle@HotAisle·
Ya'll are missing the elephant in the room. Doesn't matter how TCO efficient a TPU is. Nobody training anything significant is going to upload their proprietary training corpus to Google Cloud. No data, no training. End of story. 🥱
English
10
3
61
5.8K
aychocho
aychocho@ChosenOne229·
@HotAisle @BenBajarin How come? AFAIK, between OEMs the chips r mostly the same (save for mobo), warranties & support r similar, and the only differences I can rly see r the software and airflow. Are iDRAC and Dell’s chassis design worth that much of an upcharge? I’ve got to be missing something here
English
0
0
0
16
Hot Aisle
Hot Aisle@HotAisle·
@BenBajarin $DELL is the best. $SMCI is the worst. People are finally waking up to what we've understood for a while now. This is why I laugh at all the NeoClouds that went the cheap route.
English
1
0
2
701
Ben Bajarin
Ben Bajarin@BenBajarin·
Ok, just want to throw this out there. Double whammy of an insight. Updated my GW tracker by name to reflect newest announcements in capacity. Note the neoclouds... A lot of those AI buidouts will be filled with $DELL $NVDA racks.
Ben Bajarin tweet media
English
18
18
120
22.5K
aychocho
aychocho@ChosenOne229·
And typically these RMA requests result in either a repair of a non-gpu component (negligible cost) or a full replacement in gpu failure situations. i'd imagine that the extension of useful life is made to account for operation past a node's warranty's expiration
English
0
0
0
67
aychocho
aychocho@ChosenOne229·
So this is what 2 million pounds of freight looks like #SC25
aychocho tweet media
English
0
0
0
105
aychocho
aychocho@ChosenOne229·
Generated a pretty neat visualization of the infiniband network topology with an ibtopotool-based visualizer I built. While it's cool that they can afford to be rail-optimized, this is a LOT more than I should be seeing lol.
aychocho tweet media
English
1
0
1
180
aychocho retweetledi
AGAT 💎 GOLIK
AGAT 💎 GOLIK@GolikAgata·
If you feel like your GPU’s been scamming you - now you can check it 😎 @quok_it tests cloud GPUs from Vast, Octa, Hyperbolic & more to see who really delivers. @gensynai members get early access 🧠⚡️ I made a setup guide here 👉 github.com/AgataGolik/quo… Drop me a ⭐️ there! :)
AGAT 💎 GOLIK tweet media
English
1
5
14
1.1K
aychocho
aychocho@ChosenOne229·
🔥
FELLOW 🪽 ⚔️@fellow_cm

𝐇𝐎𝐖 𝐓𝐎 𝐔𝐒𝐄 𝐐𝐔𝐎𝐊.𝐈𝐓 𝐂𝐋𝐈 𝐓𝐎 𝐓𝐄𝐒𝐓 𝐘𝐎𝐔𝐑 𝐆𝐏𝐔 ➲ step 1: register at quok.it/gensyn fill in your discord username and email and make sure the selected options and click "get early access" ➲ step 2: after registering, you'll receive the login details for your account and your 𝐀𝐏𝐈 𝐊𝐄𝐘 (very important) in your email after login, reset your password when prompted to fully access your dashboard. ➲ step 3: for @octa_space users, go to your sessions and click on it to reveal the popup modal then click on "access web ssh" to open the command line screen on a new tab ➲ step 4: run the following commands IN ORDER in the command line ⤏ sudo install -d -m 0755 /etc/apt/keyrings && curl -fsSL repo.quok.it/quok.asc | gpg --dearmor | sudo tee /etc/apt/keyrings/quok.gpg >/dev/null && echo 'deb [signed-by=/etc/apt/keyrings/quok.gpg] repo.quok.it cross main' | sudo tee /etc/apt/sources.list.d/quok-stable.list >/dev/null && sudo chmod 0644 /etc/apt/keyrings/quok.gpg /etc/apt/sources.list.d/quok-stable.list && sudo apt-get update ⤏ sudo apt install quok ⤏ quok init then input your API KEY (the one sent to your email) when prompted ps: you can also find these commands under the "get started" tab on your dashboard

ART
0
0
1
55
aychocho retweetledi
shuod 黄豹
shuod 黄豹@shuod_a·
The Cloud GPU Battle has begun The @gensynai just teamed up with @quok_it a tool that benchmarks real GPU performance across cloud platforms like Octa, Hyperbolic, Vast, and more I’ve joined the challenge to see which provider truly delivers top performance Gensyn members get early free access to the CLI tool just run one command to test your rented GPUs for speed, stability, and network quality Let’s find out who’s really on top 👉 quok.it/gensyn
shuod 黄豹 tweet media
English
14
1
40
341
Hot Aisle
Hot Aisle@HotAisle·
Ooof. Not looking very reliable there...
Hot Aisle tweet media
English
5
0
17
2K