From 4228be047621b0d8ad5e287ae0903648bc6c2256 Mon Sep 17 00:00:00 2001 From: TUNA_BIN <65464657+TUNA-NOPE@users.noreply.github.com> Date: Tue, 10 Nov 2026 09:12:05 +0000 Subject: [PATCH] docs: a 3x simulator now runs at 120 fps, the screen's full rate The 2026-11-10 lab runs of the current build measured 118.7-120.9 fps (one Sim, GlassBench), up from the 117 written before. README, setup, performance, llms.txt and the chart say 120. --- README.md | 4 ++-- docs/images/fps-dark.svg | 8 ++++---- docs/images/fps-light.svg | 8 ++++---- docs/images/make-chart.py | 4 ++-- docs/performance.md | 6 +++--- docs/setup.md | 2 +- docs/site/llms.txt | 2 +- 7 files changed, 17 insertions(+), 17 deletions(-) diff --git a/README.md b/README.md index 7aa0aad..5df8a21 100644 --- a/README.md +++ b/README.md @@ -6,7 +6,7 @@ have, with no Mac and no cloud service. Claude Code, Codex and T3 Code each get simulator to build an app, run it, tap through it, take screenshots and record video, on their own. It's fast: an iPhone 17 Pro simulator at full 3x resolution renders a heavy Liquid Glass screen at -**117 fps on an Intel integrated GPU**. Before macbox's patches, the same GPU managed 17. +**120 fps on an Intel integrated GPU**, the screen's full refresh rate. Before macbox's patches, the same GPU managed 17. **Website:** [tuna-nope.github.io/macbox](https://tuna-nope.github.io/macbox/) @@ -80,7 +80,7 @@ VM when you say so. By hand: [docs/setup.md, Updating](docs/setup.md#updating). - iOS Simulator at 3x, Liquid Glass benchmark: 17.2 fps as upstream shipped it, 39.0 with the October 28 macbox build, 50.6 on November 5, 70.0 with Coherent, 117 with the build macbox makes now + iOS Simulator at 3x, Liquid Glass benchmark: 17.2 fps as upstream shipped it, 39.0 with the October 28 macbox build, 50.6 on November 5, 70 with Coherent, 120 with the build macbox makes now Measured on an Intel Core Ultra 7 265 (Arrow Lake iGPU) with an iPhone 17 Pro simulator at 3x diff --git a/docs/images/fps-dark.svg b/docs/images/fps-dark.svg index 69d4660..c1e6866 100644 --- a/docs/images/fps-dark.svg +++ b/docs/images/fps-dark.svg @@ -28,7 +28,7 @@ macbox, Oct 28 in place rendering -39.0 +39 macbox, Nov 5 cohwl: flat32 @@ -36,9 +36,9 @@ + Coherent zero-copy surfaces, Nov 5 -70.0 +70 macbox now Coherent + grouped completion - -117.0 + +120 diff --git a/docs/images/fps-light.svg b/docs/images/fps-light.svg index debc91c..c7f3ba5 100644 --- a/docs/images/fps-light.svg +++ b/docs/images/fps-light.svg @@ -28,7 +28,7 @@ macbox, Oct 28 in place rendering -39.0 +39 macbox, Nov 5 cohwl: flat32 @@ -36,9 +36,9 @@ + Coherent zero-copy surfaces, Nov 5 -70.0 +70 macbox now Coherent + grouped completion - -117.0 + +120 diff --git a/docs/images/make-chart.py b/docs/images/make-chart.py index 12ab1e2..61959ed 100644 --- a/docs/images/make-chart.py +++ b/docs/images/make-chart.py @@ -9,7 +9,7 @@ ("macbox, Oct 28", "in place rendering", (38.5 + 39.5) / 2, "after"), ("macbox, Nov 5", "cohwl: flat32", 50.6, "after"), ("+ Coherent", "zero-copy surfaces, Nov 5", 70.0, "after"), - ("macbox now", "Coherent + grouped completion", 117.0, "now"), + ("macbox now", "Coherent + grouped completion", 120.0, "now"), ] MODES = { "light": dict(surface="#fcfcfb", text="#0b0b0b", text2="#52514e", grid="#e4e3df", before="#eb6834", after="#2a78d6"), @@ -49,7 +49,7 @@ def svg(m): o.append(f'{label}') o.append(f'{sub}') o.append(bar(y, v, col, 1.0 if kind in ("before", "now") else 0.55)) - o.append(f'{v:.1f}') + o.append(f'{v:g}') y += BAR + GAP o.append("") return "\n".join(o) + "\n" diff --git a/docs/performance.md b/docs/performance.md index 96e8c41..19087ad 100644 --- a/docs/performance.md +++ b/docs/performance.md @@ -43,11 +43,11 @@ Pro at 3x (1206x2622), frames per second counted at the simulator's own screen: | macbox, 2026-10-28 (in place rendering, 0004) | 38.5 / 39.5 | 25.0 ms/frame | | macbox cohwl, Coherent off (flat32), 2026-11-05 | 50.6 (p95 21.4 ms, p99 25.8 ms) | 97% busy | | macbox cohwl, Coherent on, 2026-11-05 | 70.0 (p95 17.7 ms, p99 22.0 ms) | 70% busy | -| **macbox now** (Coherent and the switches below, 2026-11-09) | **117** (114–121 over runs) | 7.7 ms/frame | +| **macbox now** (Coherent and the switches below, 2026-11-10) | **120** (118.7–120.9 over runs) | 7.7 ms/frame | -The iPhone 17 Pro's screen runs at up to 120 Hz, so 117 is close to the most a single simulator can +The iPhone 17 Pro's screen runs at up to 120 Hz, so 120 is the most a single simulator can show; GlassBench's own content animates at 60 Hz, and the simulator redraws on top of it. The -cohwl rows are one A/B in one window; the last row comes from the soak and A/B runs of 2026-11-09 +cohwl rows are one A/B in one window; the last row comes from the soak, A/B and lab runs of 2026-11-09 and -10 (fresh headless boots, Simulator.app closed). flat32 also has an A/B of its own against the build before it: 41.6 → 48–50 fps. diff --git a/docs/setup.md b/docs/setup.md index 2c5963a..1cbd206 100644 --- a/docs/setup.md +++ b/docs/setup.md @@ -38,7 +38,7 @@ node), makes an SSH key `~/.ssh/macbox_ed25519`, and adds a `Host macbox` block | `MACBOX_CPUSET` | empty | pin QEMU to some cores, e.g. a hybrid CPU's performance cores (`0-7`) | | `MACBOX_BUILD_CPUSET` | empty | cores for `macbox build`, e.g. the efficiency cores | | `MACBOX_DISK_CACHE` | none | `writeback` panicked macOS's NVMe driver under heavy swap; `unsafe` is faster, but a crash of the host itself can corrupt the VM's disk. Applies at the next `macbox up` | -| `MACBOX_COHERENT` | auto | `auto`: on with an Arrow Lake iGPU. Coherent, zero-copy surfaces, and the switches measured with it: a 3x simulator 50 → 117 fps there ([performance.md](performance.md#coherent-and-the-switches-measured-with-it-on-by-default-on-arrow-lake)); `on` with any GPU, `off` | +| `MACBOX_COHERENT` | auto | `auto`: on with an Arrow Lake iGPU. Coherent, zero-copy surfaces, and the switches measured with it: a 3x simulator 50 → 120 fps there ([performance.md](performance.md#coherent-and-the-switches-measured-with-it-on-by-default-on-arrow-lake)); `on` with any GPU, `off` | | `MACBOX_HOME` | ~/.local/share/macbox | everything macbox keeps | Then: diff --git a/docs/site/llms.txt b/docs/site/llms.txt index 4e9e142..d6e284d 100644 --- a/docs/site/llms.txt +++ b/docs/site/llms.txt @@ -1,6 +1,6 @@ # macbox -> Run macOS, Xcode and the real iOS Simulator on a Linux host, in Docker, with GPU acceleration (Metal through Reims vGPU and metal2vulkan on the host's Vulkan GPU), for AI coding agents such as Claude Code, Codex and T3 Code. An iPhone 17 Pro simulator at 3x runs a Liquid Glass benchmark at 117 fps on an Intel Arrow Lake iGPU (17 before macbox's patches). No Mac and no cloud service: one CLI, `macbox`, builds and runs it, and `mac-sim` gives each agent checkout its own iOS Simulator to build, run, tap, screenshot, record and measure at 60 fps. +> Run macOS, Xcode and the real iOS Simulator on a Linux host, in Docker, with GPU acceleration (Metal through Reims vGPU and metal2vulkan on the host's Vulkan GPU), for AI coding agents such as Claude Code, Codex and T3 Code. An iPhone 17 Pro simulator at 3x runs a Liquid Glass benchmark at 120 fps on an Intel Arrow Lake iGPU (17 before macbox's patches). No Mac and no cloud service: one CLI, `macbox`, builds and runs it, and `mac-sim` gives each agent checkout its own iOS Simulator to build, run, tap, screenshot, record and measure at 60 fps. License: PolyForm Noncommercial 1.0.0 for macbox's own code (noncommercial use; commercial use needs permission); the patches stay GPL-2.0-or-later OR LGPL-3.0-or-later (GPL-2.0-or-later for the QEMU ones).