Apple’s homegrown chip could be the end of AMD GPUs on Macs

Tram Ho

At a developer seminar during WWDC 2020, as well as in the developer support document, Apple made it clear that the CPUs it cultivated will also support its own GPUs. The company now uses these GPUs on other ARM-based devices, like iPads and iPhones. If Apple wants to part ways with Intel, who is sure they won’t do the same for AMD, which is the manufacturer of GPUs for Macs? There is no evidence to suggest that, or vice versa, and Apple itself is quite tight-lipped about the issue, so everything to this point is speculation.

Apple-powered Macs (called Apple Silicon Macs) have Apple-designed GPUs, while Intel-powered Macs have GPUs from Intel, AMD, and Nvidia ” – Gokhan Avkarogullari, Apple’s GPU software manager, speaking at WWDC.

That is not surprising. Apple confirmed in a keynote post at WWDC that the company will abandon Intel processors to switch to its own system-on-chip (SoC) with Apple Silicon, and at the same time, switch to using it. The integrated GPU is their own! What is not clear is what these changes mean to the future of discrete GPUs on Macs? Apple officially stopped supporting Nvidia GPUs when launching macOS Mojave in 2018, but continued to launch a series of Macs with AMD GPUs.

In the near future, AMD discrete GPUs will still be trusted. Apple recently introduced a Mac Pro desktop version using the AMD Radeon Pro 5500X GPU, and the company said during the WWDC event that it will release more Intel versions. But one of the things the company pointed out during the developer conference is the difference between Apple GPUs and third-party GPUs. Apple GPU architecture is TBDR (tile-based deferred renderer), while Intel, Nvidia, and AMD are IMR (immediate mode renderer).

TBDR captures the entire frame before starting to render, dividing it into several small areas, ie tiles, which are processed separately, so it processes information fairly quickly and does not require a lot of tape. throughput memory. This architecture will not actually render the scene until it removes any blocked pixels.

IMR, in contrast, erects the entire frame before deciding which pixels need to be removed. You may have guessed: this method is less efficient, but that’s the way modern discrete GPUs work, and to do that, they need a lot of bandwidth.

Chip “nhà trồng” của Apple có thể là dấu chấm hết cho GPU AMD trên máy Mac - Ảnh 1.

For Apple Silicon’s ARM architecture, TBDR is much better because it focuses on speed and low power consumption – not to mention the GPU is on the same chip as the CPU, explaining why we have the SoC. This is probably the reason why Apple wrote that ” Don’t think a discrete GPU will provide better performance ” in developer support documentation.

That may also be the reason why the Shadow of the Tomb Raider demo (running on Rosetta 2) that Apple showed in the keynote article looks so great. If Apple supports game port developers not only for ARM architecture but also for GPU architecture, then they will have a great weapon to enter the gaming market. And if that happens, the Mac could become a true gaming machine by benchmarks.

The price of Apple-powered Macs is still unknown, especially when you can assemble or buy a better-configured PC but at a much lower price than a Mac. There is also the issue of “DIY culture” that is closely tied to the Windows PC market. Apple often forces customers to rely solely on the company for hardware-related issues or to upgrade their configurations, and if it wants to attract more developers to write games for their hardware and platforms. I, understand the gaming PC culture will be of great help. Anyway, if Apple’s GPU is better, that won’t be a problem either.

Like Intel, AMD will stay with Apple for as long as possible, until Apple is confident it can survive without third-party hardware components. At that time, the closed wall surrounding the Apple ecosystem was probably completely completed.

Reference: Gizmodo

Share the news now

Source : Genk