Hackernews posts about FP16

RSS feed for FP16

01
new100% GPU Utilization, 39% of Rated FP16medium.com
▲ 2birdculture 21h discuss
02
newFP16 GEMM Optimization on AMD RDNA 3 and RDNA 4 Architecturesrocm.blogs.amd.com
▲ 2gmays 17d discuss
03
newShow HN: Running EmbeddingGemma2 on the Browsersearchyourcollection.com
▲ 1glpcc 26d discuss
04
newAhead of IPOs, AI companies are sketching plans to share their wealthapnews.com
▲ 8devonnull 17d 💬 4
05
newONNX Runtime and CoreML May Silently Convert Your Model to FP16ym2132.github.io
▲ 98Two_hands 320d 💬 17
06
newTurboOCR: 270–1200 img/s OCR with Paddle and TensorRT (C++/CUDA, FP16)github.com
▲ 7pfdomizer 207d 💬 4
07
newRunning the Deepseek-R1 671B Model at FP16 Fidelity on AMD EPYC CPUswww.servethehome.com
▲ 5tanelpoder 581d 💬 2
08
new90T/s on my iPhone llama3.2-1B-fp16www.reddit.com
▲ 5darkolorin 584d 💬 1
09
newShow HN: TurboOCR up to 1200 pages/s with Paddle and TensorRT (C++/CUDA, FP16)github.com
▲ 3pfdomizer 205d discuss
10
newNative vLLM and ROCm 7.15 for RX 6000 (RDNA2) on Windows 11 – 26 Tflops FP16github.com
▲ 2esseba-dev 81d discuss
11
newNumber to GPU Float Converter: FP32, FP16, BF16, FP8, and FP4 Explainedwww.bestgpusforai.com
▲ 2javaeeeee 114d discuss
12
newTTS engines: WebSocket vs. sync is 5.5x, INT8 slower than fp16 on M4ai.gopubby.com
▲ 2KirMoisha 205d discuss
13
newDefeating the Training-Inference Mismatch via FP16arxiv.org
▲ 2matt_d 371d discuss
14
newPyTorch 2.6 Delivers FP16 Support for x86 CPUs, Better Intel GPU Experiencewww.phoronix.com
▲ 2rbanffy 646d discuss
15
newDefeating the Training-Inference Mismatch via FP16arxiv.org
▲ 1billyzs 369d discuss
16
newPyTorch 2.6 Delivers FP16 Support for x86 CPUs, Better Intel GPU Experiencewww.phoronix.com
▲ 1mikece 646d discuss
17
newShow HN: Kitten TTS – 25MB CPU-Only, Open-Source TTS Modelgithub.com
▲ 1003divamgupta 458d 💬 361
18
newShow HN: Three new Kitten TTS models – smallest less than 25MBgithub.com
▲ 561rohan_joshi 232d 💬 181
19
newShow HN: Appear as anyone in video calls like zoom or Google meetswww.phazr.ai
▲ 110michaelphi 587d 💬 51
20
newShow HN: ZSE – Open-source LLM inference engine with 3.9s cold startsgithub.com
▲ 58zyoralabs 254d 💬 9
21
newShow HN: GPULlama3.java Llama Compilied to PTX/OpenCL Now Integrated in Quarkus
▲ 24mikepapadim 330d 💬 6
22
newShow HN: Run 500B+ Parameter LLMs Locally on a Mac Minigithub.com
▲ 17fatihturker 243d 💬 10
23
newShow HN: OpenGraviton – Run 500B+ parameter models on a consumer Mac Miniopengraviton.github.io
▲ 13fatihturker 244d 💬 5
24
newShow HN: Lightweight Llama3 Inference Engine – CUDA Cgithub.com
▲ 12abhisheknair10 671d discuss
25
newShow HN: OS Megakernel that match M5 Max Tok/w at 2x the Throughput on RTX 3090github.com
▲ 6GreenGames 212d 💬 1
26
newShow HN: 1B Embeddings
▲ 4INVARIAN 216d 💬 2
27
newShow HN: Train 300M/32-Layer Model in 1.5GB RAM on Base M1 Mac
▲ 3vlad_kalinkin 69d discuss
28
newTq-KV – Rust implementation of TurboQuant that works on GGUF models
▲ 3onurgokyildiz 220d discuss
29
newShow HN: HyperAI GPU Leaderboard – A benchmark comparison site for AI workloadshyper.ai
▲ 2Ada_trying 286d 💬 2
30
newShow HN: Efficient LLM Architectures for 32GB RAM (Ternary and Sparse Inference)github.com
▲ 2fatihturker 242d 💬 1
31
newShow HN: Loft CLI – Fine-tune and run LLMs (1–3B) on 8 GB MacBook Air, no GPUs
▲ 2dips2umar 474d 💬 1
32
newShow HN: Optimized runtimes for three VLAs on Jetson Thorgithub.com
▲ 2hhuytho 45d discuss
33
newShow HN: I made Qwen3.5-4B 13% smarter by compressing it to 4-bithuggingface.co
▲ 2singularity_max 243d discuss
34
newShow HN: Ghost Engine – generate weights on the flygithub.com
▲ 1saznlamsal 291d 💬 1