Introduction As part of CUDA unified memory, the addresses used to access memory on both of the CPU and GPU side are identical. However, only… Read More »Beyond GPU unified memory: fully unified address spaces

In the prior post, we highlighted why PCIe BARs should be mapped as device memory on 64-bit Arm platforms. What is unified memory in a… Read More »How does NVIDIA implement unified memory on PCIe GPUs and Tegra platforms?

Introduction On 64-bit Arm machines today running Linux, PCIe remote memory device address ranges are mapped as normal instead of device memory. What does the… Read More »The case for mapping PCIe BARs as device memory on arm64

The GPGPU ecosystem is built on an assumption that the virtual memory address space is unified between the host and the device(s). That has been… Read More »The biggest gap in the graphics APIs for GPGPU workloads

GPU virtualisation is becoming more and more essential with time, on both server and desktop scenarios. And automotive too, but that’s a write-up for another… Read More »Virtualising GPUs: the different approaches

This blog post is a collection of tweets on the information known about Secure Launch today. This will be hopefully helpful to some of the… Read More »Tweet collection on Secure Launch

Apple operating systems explicitly do not support CPU pinning in a shipping configuration. This is unlike almost every other shipping OS, so do not even… Read More »CPU Pinning on macOS

Note: most of this post applies to iOS 16 too. Since macOS Ventura, Rapid Security Response is supported. This allows for lightweight updates that do… Read More »Quick look at macOS Rapid Security Response

Let’s start by plugging in a USB stick (using the FAT, ExFAT or NTFS file systems) on macOS Ventura… and see what happens. The NTFS… Read More »Quick look at user-mode file systems on macOS Ventura

Quick look at Rosetta on Linux.