Unified Memory on Apple Silicon: Designing for One Pool, Not Two Heaps
Apple Silicon integrates CPU, GPU, and Neural Engine into one unified memory pool. Learn how to implement zero-copy data sharing and manage bandwidth contention.
Loading…
ReadMeFeed / Community knowledge
#apple-silicon
Real questions. Useful conversations. Find the people who know your stack.
appleRunning x86_64 VMs on Apple Silicon with VMware Fusion 13: What You Gain and What You Lose
VMware Fusion 13 runs x86_64 VMs on Apple Silicon via Rosetta 2 translation, Metal graphics, and virtio‑fs shared folders. This post measures the real CPU, GPU, and I/O trade‑offs, shows a concrete fio benchmark, and gives a checklist for deciding whether Fusion fits your workflow.
Metal Compute on Apple Silicon: Argument Buffers, Shared Memory, and the MPS Decision
A practical look at structuring Metal compute pipelines on Apple Silicon—when to use Argument Buffers, how UMA changes synchronization, and why MPS may beat hand-written MSL for standard operations.
Unified Memory on Apple Silicon: What Changes When CPU and GPU Share the Same Pool
Apple Silicon's Unified Memory Architecture lets CPU, GPU, and Neural Engine share one DRAM pool—no PCIe copies, no blits. Here's how to exploit it in Metal, where the traps are, and how to verify zero-copy behavior with Xcode Instruments.
Unified Memory on Apple Silicon: Kernel Insights and Profiling Tips
Apple's M-series unified memory architecture lets CPU and GPU share a single RAM pool, but performance hinges on fault-in behavior. Learn how to diagnose with sysctl and Instruments, and tune a Metal kernel for better throughput.
How to run amd64 docker image on arm64 host platform?
I have an m1 mac and I am trying to run a amd64 based docker image on my arm64 based host platform. However, when I try to do so (with docker run) I get the following error: WARNING: The requested image's platform (linux/amd64) does not match the detected host platform (linux/arm64/v8) and no specific platform was requested. When I try adding the tag --platf
dockerScaling parity for Apple Silicon-based cloud instances
Migrating low-traffic workloads to Apple Silicon-based instances (M1/M2/M3) offers significant performance-per-watt advantages compared to traditional x86 architectures. While ARM64-based serverless containers can reduce idle costs, the architectural behavior regarding dynamic scaling remains inconsistent across cloud providers. Specifically, there is a docu
anthoscli ERROR on installing Google Cloud SDK on Apple Silicon / Big Sur
I am searching for the solution for some weeks now. A couple weeks ago I received the new MacBook Pro with the Apple Silicon processor and with Big Sur installed on it. When I tried to install Google Cloud SDK, because I need that for my work, I get my first problem. I opened the terminal and used "curl https://sdk.cloud.google.com | bash" to start the insta
Homebrew prefix transition from /usr/local to /opt/homebrew breaks CI build scripts
Homebrew changed its default installation prefix when Apple Silicon was introduced, moving from /usr/local to /opt/homebrew . Build scripts and application configurations that hardcode the legacy Intel path continue to work on older machines but fail when the same repository is built on an arm64 CI runner. The documented brew shellenv command is intended to
homebrew