Hackernews posts about H100

H100 is a high-performance graphics processing unit (GPU) accelerator developed by Nvidia, designed for large-scale artificial intelligence and deep learning workloads.

RSS feed for H100

Related:Nvidia AMD A100 MI300X Google Tesla Intel IBM Hopper LLM MI300 L40S

31
newIt Ain't AGI Until H100s Go for $30 an Hourcorbt.com
▲ 26gmays 15d 💬 8
32
newShow HN: We ported FLA's Triton kernels to Jax Pallas; they're faster on H100github.com
▲ 1KarLexAI 9d discuss
33
newTraining a 180B-parameter model on a single 8×H100 nodephysionlabs.ai
▲ 1ljhnick 16d discuss
34
newShow HN: Offline Tiny-LMs 25MB for personal goals, career and pitch deck advicegithub.com
▲ 2andyvane 37h 💬 1
35
newShow HN: The first open-weight model for general deterministic taskhuggingface.co
▲ 2aparup2003 18d discuss
36
newChinese AI chipmaker Iluvatar CoreX says new GPU matches Nvidia's H200www.scmp.com
▲ 8yogthos 31h 💬 3
37
newCorvus – A lightweight native Signal client in Qtgithub.com
▲ 2ada-h00k 2d discuss
38
newNintendo's worst nightmare begins as Switch 2 games start running on PC and Macvideocardz.com
▲ 28ada-h00k 3d 💬 10
39
newBottles creator claims Raven will run windows apps faster than WINEtwitter.com
▲ 3ada-h00k 7d 💬 2
40
newHonor-Based Value Transfer: A Walk Through Hawalaxn--gckvb8fzb.com
▲ 2ada-h00k 6d discuss
41
newRun Qwen 3.8 Flash Next (125B) on consumer hardware (RTX 4090) at 100T/sgithub.com
▲ 940snehesht 27d 💬 420
42
newStackOverflow now gets fewer than 1000 questions a monthdata.stackexchange.com
▲ 842babelfish 3d 💬 752
43
newAsk HN: If the models are so good, why doesn't OpenAI just start 100 companies?
▲ 681ex-aws-dude 9d 💬 755
44
newDeepSeek search and fetch – $0.75 per 1000 successful requestsapi-docs.deepseek.com
▲ 607fagnerbrack 13d 💬 282
45
newMedicare Will Pay $50k for Your Heart Attack. It Should Pay $100 to Prevent Itwww.medpagetoday.com
▲ 303brandonb 38h 💬 285
46
newRemoving Next.js Got My Blog to 100 on Every Lighthouse Categorykorindiehacker.bloffy.blog
▲ 198textlanes33 6d 💬 133
47
new100+ reactions to 100+ solutionsproofsandprompts.com
▲ 106jacobedawson 22d 💬 104
48
newAnthropic Console overcharged my card 100x due to internationalization input bug
▲ 91rsorto 15d 💬 36
49
newShow HN: Parseable, an open observability datalake, handles 100M time-series/minwww.parseable.com
▲ 90yashdotrv 25d 💬 25
50
newBank of America hired 1000 contractors to patch software flaws exposed by Mythoswww.cnbc.com
▲ 38mirzap 17d 💬 14
51
newSpaceX activated 100mph speed-limit restriction on Starlink terminals in Ukrainebsky.app
▲ 35ck2 17d 💬 16
52
new100 days later: Microsoft still steers Windows and Copilot users to Edgeblog.mozilla.org
▲ 34thimabi 24d 💬 6
53
newHuawei, once hobbled by US sanctions, passes 100M Kirin chip milestonewww.scmp.com
▲ 30yogthos 18d 💬 3
54
newShredding JSON columns gave Logfire up to 1000x faster queriespydantic.dev
▲ 29acossta 20d 💬 30
55
new100 years of student radio history in the DLARC college radio collectionsblog.archive.org
▲ 26HieronymusBosch 29d 💬 4
56
newUltramarine 1 Flash: Frontier Model 100X Cheaperwww.paperinstruments.com
▲ 24dzonga 17d 💬 1
57
newI've created an open source 100% Rust clean-room clone of 7 of Adobe's appswww.reddit.com
▲ 18robenkleene 29d 💬 2
58
newOpenAI: We are doubling all usage limits for all Pro 100/200/500 tierstwitter.com
▲ 17tosh 2d 💬 1
59
newBritish intelligence misplaced 1000s of storage devices still can't say how manywww.theregister.com
▲ 12beardyw 38h 💬 1
60
newThe Top 100 Gen AI Consumer Apps — 7th Editiona16z.com
▲ 12pseudolus 24d 💬 1

More stories →