AMD’s latest promises to boost gaming performance for laptops. The trick: VGM and a little AI

Just a few months ago, AMD announced its proposal to enter the emerging world of AI PCs: the AMD Ryzen AI 300. These processors, whose NPU reaches 50 TOPS, are largely designed for tasks related to artificial intelligence, but it is no secret that there will be users who want to use them for gaming. In order to improve performance, the company just announced two very interesting novelties: VGM, acronym for Variable Graphics Memory (variable graphics memory) and AFMF 2.

Let’s go step by step.

The Ryzen AI 300. Although we break them down better in this article, let’s do a quick review of what it offers. On the one hand, the CPU has Zen 5 architecture and will have up to 12 cores (4 Zen 5, 8 efficient Zen 5C) and 24 threads, up to 5.1 GHz clock speed and up to 36 MB cache.

The GPU, meanwhile, is a Radeon 880M (12 CU) or a Radeon 890M (16 CU) with RDNA 3.5 architecture. Thirdly, the NPU has XDNA 2 architecture and goes from the 10 TOPs of the previous XDNA to 50 TOPS, a huge jump and superior, on paper, to that of Qualcomm’s Snapdragon X Elite. They are not gaming components by any means, but the gaming experience should be good.

Why are they important? Because the technologies we are dealing with today are only available, for now, on these processors and, moreover, in the testing phase. To access them, you will have to download a driver, although they will soon be able to be activated from the Adrenalin software.

Shall we share? The first novelty that AMD brings is VGM, an acronym for Variable Graphics Memory (variable graphics memory). This feature allows up to 75% of the system’s RAM to be converted into dedicated graphics memory aka vRAM. According to AMD:

“By default, the integrated graphics have a “dedicated” graphics allocation of 512MB, and AMD recommends leaving at least 16GB of RAM for the CPU. This means that if you set Variable Graphics Memory (VGM) to “medium”, on a 32GB system, it will allocate 8GB of that reservation and convert it to 8GB of vRAM for the AMD Radeon 890M integrated graphics after a reboot, which is an increase of 7.5GB over the default “dedicated” vRAM allocation of 512MB.”

That is, we went from having a computer with 32 GB of RAM and 512 MB of vRAM to one with 24 GB of RAM and 8 GB of vRAM. This is particularly useful for being able to use those games that are limited to a minimum of vRAM.

In The Verge have been able to test this technology on an ASUS Zenbook S 16 Ryzen AI 9 HX 370 with this feature and the result is mixed. On the one hand, there are games like ‘Control‘ (54 vs 65 FPS) and ‘Far Cry 6’ (56 vs 62 FPS) which benefit greatly from this technology, but there are others, such as ‘Hitman 3’ (62 v 56 FPS) which suffer. In other titles, the majority really, the increase is just one or two FPS.

The key: combine it with AI. VGM is clearly boosted by AFMF 2, the new frame generation technology. It is an evolution of the previous version that, according to AMD, uses AI optimizations to improve quality and performance, as well as reduce latency. By combining AMF2, VGM and FSR, AMD promises a fairly substantial performance improvement both in FullHD and higher resolutions.

These are features that are currently in the testing phase and only for the Ryzen AI 300 series processors. However, it is always good news that these types of options are available to give an FPS boost to games on devices such as laptops.

Image | Richyart with Midjourney

At Xataka | AMD will put all its money where its mouth is: in AI

(function() { window._JS_MODULES = window._JS_MODULES || {}; var headElement = document.getElementsByTagName(‘head’)[0]; if (_JS_MODULES.instagram) { var instagramScript = document.createElement(‘script’); instagramScript.src=”https://platform.instagram.com/en_US/embeds.js”; instagramScript.async = true; instagramScript.defer = true; headElement.appendChild(instagramScript); } })();

–
The news

AMD’s latest promises to boost gaming performance for laptops. The trick: VGM and a little AI

was originally published in

Xataka

by Jose Garcia .

Source link