AMD Radeon 800M iGPUs With 16 RDNA 3.5 Cores Scores Over 3600 Points In 3DMark Time Spy, Close To RTX 2050 Perf

AMD's upcoming Radeon 800M iGPUs for Ryzen AI 300 APUs, equipped with RDNA 3.5 iGPUs, should offer a nice uplift in graphics performance.

The latest information regarding AMD's RDNA 3.5 "Radeon 800M" iGPUs comes from Golden Pig Upgrade at Bilibili. The insider has shared details, stating that based on measurements conducted on MSI's Ryzen AI 300-powered laptops at Computex 2024, the next-generation integrated GPU is going to be quite the improvement over its RDNA 3 "Radeon 700M" predecessors.

We know for a fact that the MSI laptops showcased during Computex 2024 were not final designs and were using engineering silicon of the AMD Ryzen AI 300 chips. Most of the laptops shown are based around the Ryzen AI 9 HX 370 CPU which is the flagship configuration with 12 cores based on the Zen 5 core architecture, 24 threads, and the top Radeon 890M iGPU with 16 CUs (compute units).

Based on the measurements conducted, the AMD Radeon 800M (890M) with 16 compute units clocked at around 2900 MHz scored over 3600 points in the 3DMark Time Spy Graphics test. The Cinebench R23 performance was also conducted and here, the CPU scored over 2000 points in single-core and over 20,000+ points in the multi-threaded test. It is reported that the overall CPU performance has improved by over 20% versus the previous generation but no details regarding the TDP were mentioned.

A score of over 3600 points should put the AMD Radeon 890M iGPU in close distance to the NVIDIA GeForce RTX 2050 (Laptop) GPU. Since this is an early ES chip & we don't know anything about the TDP, it's easy to say that the final performance should be much higher.

We previously reported the expected performance of the high-end iGPUs can potentially exceed even 4000 points in the same 3DMark Time Spy tests. Overall, for just a small refreshed architecture, these are great gains & the extra compute units should help deliver some nice gains in gaming. These will be great for a range of applications such as thin and light platforms, Mini PCs, and handhelds.