MTT S80 vs. GT 1030
BullLabs have a review of the first commercially available Chinese GPU called MTT S80.
The card was manufactured by a company called Moore Threads, and it is one of the most important developments in the GPU space since Intel Arc launch. This is indeed a fourth player entering the game of discrete gaming GPUs.
The S80 is no doubt a significant milestone for Moore Threads, a second generation GPU based on MUSA architecture. This card features 4096 FP32 cores, 12nm GPU and very interesting specs matching competing products. The most important one being the 16GB VRAM and the other feature is PCIe Gen5 support, which is not available with any modern GPU yet.
There were other reviews posted earlier, but they were clearly commissioned by the manufacturer where each Chinese media outlet followed the guidelines to the letter. This means the same software stack in each review with barely anything more.
The good news is that the card has now been tested by BullsLab, a South Korean media outlet, who bought one of these cards from China. These guys are known for comprehensive reviews, so this is as good as it gets for now.
They explain the installation process which is not as easy as any GPU from competing companies. The S80 has a list of hardware restrictions, as it only works with certain motherboards and monitors. Furthermore, MTT S80 does not support the latest Windows and many of the modern games. This card is currently restricted to DX9 games, and it doesn’t even support tessellation as per other reviews.
The card was put against GeForce GTX 1060 and GT 1030, both based on the same Pascal architecture. Despite their age, these cards are much better options than S80 for gaming, although the Chinese card is advertised as such. Although MTT’s and NVIDIA’s architectures are years apart, the 30W GT 1030 which is a passive design outperforms this 250W card in nearly all game benchmarks.
BullsLab Jay explains all the drawbacks of the architecture, such as high-power usage (even at idle state), lack of compatibility with professional workloads (although it supports NVIDIA CUDA) or the fact that it doesn’t even support AVI decoding with YouTube, although AV1 support is advertised everywhere.
Check the full review below, it has English captions.