Feed aggregator

Show HN: Free, serverless Redis clone on Durable Objects (works w Upstash SDKs)

Hacker News - Mon, 03/09/2026 - 2:08pm

meowdis gives you an Upstash compatible Redis API that you can deploy yourself on Cloudflare + the Workers free plan gives you 5 GBs of free storage! Supports both single commands and pipelined requests.

Supported Redis commands:

strings: GET, SET, DEL, EXISTS, INCR, INCRBY, DECR, DECRBY, MGET, MSET

expiry: EXPIRE, EXPIREAT, TTL, PTTL, PERSIST

hashes: HGET, HSET, HDEL, HGETALL, HEXISTS, HKEYS, HVALS

lists: LPUSH, RPUSH, LPOP, RPOP, LRANGE, LLEN

sets: SADD, SREM, SMEMBERS, SISMEMBER, SCARD

utility: PING, DBSIZE, FLUSHDB, KEYS

SET options: NX, XX, GET, EX, PX, EXAT, PXAT, KEEPTTL

EXPIRE / EXPIREAT options: NX, XX, GT, LT

LPOP / RPOP options: count

Repository / deployment instructions: https://github.com/zion-off/meowdis

How it works:

1) there’s a compute layer (this can be anything) and a storage layer (a cloudflare durable object instance) 2) you send redis commands to the compute layer using either a) rest b) or upstash sdks (see upstash docs) 3) compute layer translates redis commands to sqlite queries 4) storage layer executes them in a transaction 5)results are returned back to you

post script disclaimer -- this started out as a side project where i was trying to work around fun constriants i made up (serverless, free, etc.). if you try it out and run into problems, bug reports are welcome !!

Comments URL: https://news.ycombinator.com/item?id=47312949

Points: 2

# Comments: 0

Categories: Hacker News

Show HN: Tmuzika – terminal music player written in C (ncurses and GStreamer)

Hacker News - Mon, 03/09/2026 - 2:08pm

Hi HN,

I've been working on a small terminal music player for Linux called tmuzika.

It's written in C, uses ncurses for the terminal interface and GStreamer for audio playback. I started it mainly as a learning project to explore terminal UI development and multimedia handling on Linux.

Main features: - ncurses terminal interface - playlist support - shuffle and repeat modes - playlist search - radio stream playback

Source code: https://github.com/ivanjeka/tmuzika

Feedback from developers, suggestions, or contributions are very welcome.

Comments URL: https://news.ycombinator.com/item?id=47312939

Points: 1

# Comments: 0

Categories: Hacker News

The US tech giant is bidding for a £323m contract to replace the Post Office’s flawed Horizon IT system, a decade after its previous attempt was abandoned due to complexities

Computer Weekly Feed - Mon, 03/09/2026 - 1:37pm
The US tech giant is bidding for a £323m contract to replace the Post Office’s flawed Horizon IT system, a decade after its previous attempt was abandoned due to complexities
Categories: Computer Weekly

Ask HN: Is Cloudflare Down Again?

Hacker News - Mon, 03/09/2026 - 1:28pm

Tons of stuff offline right now at https://downdetector.com/

Comments URL: https://news.ycombinator.com/item?id=47312218

Points: 1

# Comments: 0

Categories: Hacker News

Show HN: ROLV – 20x faster MoE FFN inference on Llama 4 Maverick vs. cuBLAS

Hacker News - Mon, 03/09/2026 - 1:27pm

I benchmarked ROLV against dense cuBLAS on the actual Llama 4 Maverick MoE expert FFN layer (up_proj, 16384×5120, bfloat16) pulled directly from HuggingFace (model-00001-of-00084.safetensors). Numbers (Batch=512, 1000 iters, NVIDIA B200): Tokens/s: 369K (cuBLAS) → 7.66M (ROLV) — 20.7x faster TFLOPS (effective): 62 → 1,285 — 20.7x Time to First Token: 64.8ms → 0.37ms — 177x faster Energy: 232J → 43J — 81.5% savings ROLV exploits structured sparsity in MoE expert weights to skip large blocks of computation entirely, while producing canonically equivalent output (hash-verified). The TFLOPS figure is "effective" — computed as if doing the full dense multiply — so the 1285 TFLOPS isn't violating hardware peak; it's reflecting how much work was avoided. The TTFT speedup (177x) is especially relevant for interactive inference: MoE models spend a huge fraction of first-token latency in these expert projections, and collapsing that from 65ms to 0.4ms per layer changes what's possible for real-time applications. Verified with norm hashes at both ends (baseline and ROLV output) and a canonical check. Weights are real, not synthetic. Setup: PyTorch 2.8.0+cu128, CUDA 12.8, Python 3.12, NVIDIA B200.

Comments URL: https://news.ycombinator.com/item?id=47312202

Points: 1

# Comments: 0

Categories: Hacker News

GNU, and the AI Reimplementations

Hacker News - Mon, 03/09/2026 - 1:23pm

Article URL: https://antirez.com/news/162

Comments URL: https://news.ycombinator.com/item?id=47312143

Points: 2

# Comments: 0

Categories: Hacker News

Pages