frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

Forcing Flash Attention onto a TPU and Learning the Hard Way

https://archerzhang.me/forcing-flash-attention-onto-a-tpu
18•azhng•4d ago

Comments

gdiamos•59m ago
One of my lessons in using different accelerators, whether they be different NVIDIA versions, or GPU->TPU, etc is that someone needs to do this work of indexing, partitioning, mapping, scheduling, and benchmarking. That work is labor intensive.

In this case, google has already done it, and that will be true for high resourced accelerator companies like Google working with the most popular operations like attention.

As long as you use those operations, you are okay. But if you do something different, you need to be prepared to do all of this yourself.

refulgentis•54m ago
It broke my heart to have a visceral "I'm being slop'd" reaction reading this: it's such good work, and AI's barely used AFAICT, but there's enough odd transitions and copy-pasta'd markdown that you get the subconcious "this is AI" reaction regardless.

Many sentences are 3x as long as it normally would be in subtle ways (to wit: "My flash attention is 35x slower than the fused standard at n=4096. Not a little worse. Catastrophically worse."), it really wears on attention. (pun intended) It brings literary voice to a technical blog post, and a very difficult process-oriented technical blog post. I have to reallocate my unfortunately-limited brain cells from "maintaining state of where we are in the process" to "is this cutesy fluff or important" and I've never had to do that in 37 years with technical blog posts.

The Markdown gets bad. Bolding is used for important phrases (like a human would), then, all of a sudden, after the "Inside a TPU chip" header its being used every other sentence, on anything that is a proper noun/would have a Wikipedia article. It got so weird that at some point I was like "a human definitely didn't let this through...they must be links?" and tried clicking them.

It's doubly bad at that point, because markdown tables start coming in hot and heavy too. So you're left with "It's pretty apparent the LLM did it from here, and I can't keep trying to keep the state of the process in my head while trying to figure out if the bolding is important, reflexive close tab

Shall I implement it? No

https://gist.github.com/bretonium/291f4388e2de89a43b25c135b44e41f0
302•breton•1h ago•109 comments

Malus – Clean Room as a Service

https://malus.sh
912•microflash•8h ago•357 comments

Bubble Sorted Amen Break

https://parametricavocado.itch.io/amen-sorting
214•eieio•5h ago•74 comments

Reversing memory loss via gut-brain communication

https://med.stanford.edu/news/all-news/2026/03/gut-brain-cognitive-decline.html
168•mustaphah•5h ago•48 comments

ATMs didn't kill bank teller jobs, but the iPhone did

https://davidoks.blog/p/why-the-atm-didnt-kill-bank-teller
265•colinprince•7h ago•318 comments

Innocent woman jailed after being misidentified using AI facial recognition

https://www.grandforksherald.com/news/north-dakota/ai-error-jails-innocent-grandmother-for-months...
188•rectang•1h ago•102 comments

The Met releases high-def 3D scans of 140 famous art objects

https://www.openculture.com/2026/03/the-met-releases-high-definition-3d-scans-of-140-famous-art-o...
176•coloneltcb•6h ago•34 comments

Forcing Flash Attention onto a TPU and Learning the Hard Way

https://archerzhang.me/forcing-flash-attention-onto-a-tpu
20•azhng•4d ago•2 comments

Document poisoning in RAG systems: How attackers corrupt AI's sources

https://aminrj.com/posts/rag-document-poisoning/
18•aminerj•8h ago•7 comments

Show HN: OneCLI – Vault for AI Agents in Rust

https://github.com/onecli/onecli
104•guyb3•5h ago•36 comments

Bringing Chrome to ARM64 Linux Devices

https://blog.chromium.org/2026/03/bringing-chrome-to-arm64-linux-devices.html
27•ingve•2h ago•30 comments

Runners who churn butter on their runs

https://www.runnersworld.com/news/a70683169/how-to-make-butter-while-running/
51•randycupertino•3h ago•24 comments

Launch HN: IonRouter (YC W26) – High-throughput, low-cost inference

https://ionrouter.io
29•vshah1016•3h ago•13 comments

WolfIP: Lightweight TCP/IP stack with no dynamic memory allocations

https://github.com/wolfssl/wolfip
71•789c789c789c•6h ago•7 comments

An old photo of a large BBS (2022)

https://rachelbythebay.com/w/2022/01/26/swcbbs/
127•xbryanx•2h ago•92 comments

Dolphin Progress Release 2603

https://dolphin-emu.org/blog/2026/03/12/dolphin-progress-report-release-2603/
281•BitPirate•12h ago•47 comments

Converge (YC S23) Is Hiring a Founding Platform Engineer (NYC, Onsite)

https://www.runconverge.com/careers/founding-platform-engineer
1•thomashlvt•5h ago

Big data on the cheapest MacBook

https://duckdb.org/2026/03/11/big-data-on-the-cheapest-macbook
273•bcye•10h ago•241 comments

Show HN: Understudy – Teach a desktop agent by demonstrating a task once

https://github.com/understudy-ai/understudy
66•bayes-song•5h ago•18 comments

US private credit defaults hit record 9.2% in 2025, Fitch says

https://www.marketscreener.com/news/us-private-credit-defaults-hit-record-9-2-in-2025-fitch-says-...
171•JumpCrisscross•9h ago•304 comments

Show HN: Detect any object in satellite imagery using a text prompt

https://www.useful-ai-tools.com/tools/satellite-analysis-demo/
6•eyasu6464•4d ago•0 comments

Show HN: Axe – A 12MB binary that replaces your AI framework

https://github.com/jrswab/axe
120•jrswab•8h ago•85 comments

Show HN: OpenClaw-class agents on ESP32 (and the IDE that makes it possible)

https://pycoclaw.com/
4•pycoclaw•44m ago•1 comments

Are LLM merge rates not getting better?

https://entropicthoughts.com/no-swe-bench-improvement
88•4diii•10h ago•95 comments

The Cost of Indirection in Rust

https://blog.sebastiansastre.co/posts/cost-of-indirection-in-rust/
73•sebastianconcpt•3d ago•31 comments

Should hack-back be legal?

https://speculumx.at/blogpost/should-hack-back-be-legal
11•Vektorceraptor•1d ago•12 comments

The Road Not Taken: A World Where IPv4 Evolved

https://owl.billpg.com/ipv4x/
40•billpg•6h ago•67 comments

NASA's DART spacecraft changed an asteroid's orbit around the sun

https://www.sciencenews.org/article/spacecraft-changed-asteroid-orbit-nasa
91•pseudolus•3d ago•57 comments

Full Spectrum and Infrared Photography

https://timstr.website/blog/fullspectrumphotography.html
40•alter_igel•4d ago•22 comments

DDR4 Sdram – Initialization, Training and Calibration

https://www.systemverilog.io/design/ddr4-initialization-and-calibration/
47•todsacerdoti•2d ago•13 comments