Hacker News: frabonacci

You can play Snake while ChatGPT generates your image

frabonacci — Fri, 11 Sep 2026 03:25:22 +0000

Article URL: https://twitter.com/francedot/status/2098249540268888546

Comments URL: https://news.ycombinator.com/item?id=49653196

Points: 1

# Comments: 0

Keeping Open Source Open in the Age of Coding Agents

frabonacci — Wed, 02 Sep 2026 07:35:25 +0000

Article URL: https://twitter.com/francedot/status/2095052001667408295

Comments URL: https://news.ycombinator.com/item?id=49533004

Points: 2

# Comments: 0

Run OpenClaw on Cua Cloud Fleet

frabonacci — Mon, 31 Aug 2026 17:43:04 +0000

Article URL: https://cua.ai/docs/how-to-guides/sandbox/run-openclaw-on-cloud-fleet

Comments URL: https://news.ycombinator.com/item?id=49512549

Points: 1

# Comments: 0

Omarchy Desktops on Cua Cloud Fleets

frabonacci — Fri, 28 Aug 2026 18:35:17 +0000

Article URL: https://cua.ai/docs/how-to-guides/sandbox/run-omarchy-on-cloud-fleet

Comments URL: https://news.ycombinator.com/item?id=49482640

Points: 2

# Comments: 0

Cua's open-source alternative to Codex Computer History, encrypted locally

frabonacci — Tue, 18 Aug 2026 18:14:56 +0000

Article URL: https://github.com/trycua/cua/blob/main/libs/cua-driver/docs/computer-history-preview.md

Comments URL: https://news.ycombinator.com/item?id=49350030

Points: 1

# Comments: 0

New comment by frabonacci in "Apple Silicon and macOS VMs: Faster LLM Inference with llama.cpp"

frabonacci — Tue, 11 Aug 2026 22:32:15 +0000

We've also seen similar improvements on a M5 max. no M1 pro or M3 pro results yet though - would love to see someone try those

New comment by frabonacci in "Apple Silicon and macOS VMs: Faster LLM Inference with llama.cpp"

frabonacci — Tue, 11 Aug 2026 21:22:43 +0000

my guess is apple chose a conservative profile for compatibility across different chips and guest releases

New comment by frabonacci in "Apple Silicon and macOS VMs: 11–16× Faster LLM Inference with Llama.cpp"

frabonacci — Tue, 11 Aug 2026 19:18:57 +0000

agreed on the title. added more context below on the exact scope and why this is really a VM capability-reporting issue: https://news.ycombinator.com/item?id=49260087

New comment by frabonacci in "Apple Silicon and macOS VMs: Faster LLM Inference with llama.cpp"

frabonacci — Tue, 11 Aug 2026 19:17:04 +0000

apple silicon is what made us start Lume in the first place last year. the hardware is so good (M1 is now 6 years old!) that people keep pushing through the gaps in the platform. just yesterday we ran a fully offline computer-use agent with Cua Driver and Muse Glimmer, all locally on Apple Silicon, an now the same kind of agent can run isolated inside a macOS VM and use Apple’s GPU path too

New comment by frabonacci in "Apple Silicon and macOS VMs: Faster LLM Inference with llama.cpp"

frabonacci — Tue, 11 Aug 2026 19:05:06 +0000

the better question is why a throwaway account is doing capitalization forensics

New comment by frabonacci in "Apple Silicon and macOS VMs: Faster LLM Inference with llama.cpp"

frabonacci — Tue, 11 Aug 2026 17:41:37 +0000

yeah the naming is confusing. Apple family 9 isnt M9, it's a Metal GPU feature family. Apple maps family 7 to M1, family 8 to M2, family 9 to M3/M4, and family 10 to M5

New comment by frabonacci in "Apple Silicon and macOS VMs: 11–16× Faster LLM Inference with Llama.cpp"

frabonacci — Tue, 11 Aug 2026 15:53:56 +0000

RunAnywhere or Conifer?

New comment by frabonacci in "Apple Silicon and macOS VMs: Faster LLM Inference with llama.cpp"

frabonacci — Tue, 11 Aug 2026 15:43:10 +0000

> this won't speed up llama.cpp for everyone, just for users running it in this particular kind of Virtualization.framework VM.

correct. these figures apply to llama.cpp inside the macOS guest configuration we tested. Lume is the VM frontend we used, while Apple's Virtualization.framework provides the virtual GPU. bare-metal llama.cpp is unaffected.

> The fix here works around a problem where the VM was causing llama.cpp to select the wrong kernels.

mostly, with one nuance: llama.cpp is selecting the correct kernels for the capability answers it receives. the stock guest reports an older Apple GPU family and a 32 KB threadgroup memory limit, so llama.cpp chooses slower kernels. Our process-scoped layer reports the tested Apple 9 and 64 KB values while allowing llama.cpp to select newer paths that the paravirtual GPU successfully execute

the layer itself though works at the Metal API boundary, independently of llama.cpp. other Metal compute and graphics apps now may select newer paths from the same capability answers, although this is still preliminary and each app needs separate testing. for example, MLX-LM stayed flat in our tests

historically related limitations have been coming up across Apple Silicon VM frontends for a while e.g. Tart tracked MPS/GPU support back in 2023: - https://github.com/openai/tart/issues/501 - https://github.com/openai/tart/issues/1032

UTM also has related cases where apps detect the Apple paravirtual Metal device but falls back to software rendering: https://github.com/utmapp/UTM/issues/7671

New comment by frabonacci in "Apple Silicon and macOS VMs: 11–16× Faster LLM Inference with Llama.cpp"

frabonacci — Tue, 11 Aug 2026 15:27:18 +0000

yeah fair point. it's always tricky to get the whole idea across within HN's title limit. tldr: we ran the same workload in the same Lume macOS VM on the same Apple Silicon host, first with stock Metal capability reporting and then with our process-scoped dynamic library. The 11.08x figure is prompt processing, while 16.36x is token generation. the mechanism technically extends to graphics workloads too but these figures are specifically from llama.cpp

Apple Silicon and macOS VMs: Faster LLM Inference with llama.cpp

frabonacci — Tue, 11 Aug 2026 14:50:33 +0000

Article URL: https://github.com/trycua/cua/blob/main/blog/gpu-passthrough-macos-vms.md

Comments URL: https://news.ycombinator.com/item?id=49259339

Points: 307

# Comments: 41

Windows computer-use: synthetic cursors for background agents

frabonacci — Wed, 27 May 2026 18:48:20 +0000

Article URL: https://github.com/trycua/cua/blob/main/blog/inside-windows-computer-use.md

Comments URL: https://news.ycombinator.com/item?id=48298678

Points: 3

# Comments: 0

New comment by frabonacci in "MacBook Neo Deep Dive: Benchmarks, Wafer Economics, and the 8GB Gamble"

frabonacci — Thu, 14 May 2026 12:49:16 +0000

i bought this for my girlfriend as an entrypoint laptop considering she is coming from Windows - and overall satisfied. the battery though could be improved especially considering for a couple of hundred bucks more we could have gotten a used macBook air

Computer-Use in Hermes Agent v2.0 [video]

frabonacci — Thu, 14 May 2026 12:43:39 +0000

Article URL: https://www.youtube.com/watch?v=Gx2joHxUhgg

Comments URL: https://news.ycombinator.com/item?id=48134594

Points: 2

# Comments: 0

Clicky Background Computer-Use

frabonacci — Tue, 05 May 2026 01:31:52 +0000

Article URL: https://twitter.com/farzatv/status/2051454940326097220

Comments URL: https://news.ycombinator.com/item?id=48017026

Points: 3

# Comments: 0

New comment by frabonacci in "Show HN: Drive any macOS app in the background without stealing the cursor"

frabonacci — Wed, 29 Apr 2026 05:59:53 +0000

A few examples i'm excited about:

- Closing the coding feedback loop by having agents verify their own changes in a real app

- Automating repetitive workflows across apps that don't have good APIs

- Agents recording product demos of them using software. One compelling use case here: https://x.com/trycua/status/2047383207612645426

- Creating CLI and APIs for apps by reverse implementing their GUI, e.g. see: https://github.com/HKUDS/CLI-Anything