Thursday, October 19th 2023
Moore Threads Prepares S90 and S4000 GPUs for Gaming and Data Center
Moore Threads Technology (MTT), a Chinese GPU manufacturer, is reportedly testing its next-generation graphics processors for client PCs and data centers. The products under scrutiny are the MTT S90 for client/gaming computers and the MTT S4000 for data centers. Characterized by their Device IDs, 0301 and 0323, this could imply that these GPUs belong to MTT's 3rd generation GPU lineup. While few details about these GPUs are available, the new Device IDs suggest a possible introduction of a novel microarchitecture following the MTT Chunxiao GPU series. The current generation Chunxiao series, featuring the MTT S70, MTT S80, and MTT S3000, failed to compete effectively with AMD, Intel, and NVIDIA GPUs.
Thanks to @Löschzwerg who found the Device Hunt submission, we see hardware identifiers in PCI ID and USB ID repositories earlier than launch, as this often signals the testing of new chips or drivers by various companies. In the case of MTT, the latest developments are complicated by its recent inclusion on the U.S. Entity List, limiting its access to US-made technologies. This introduces a problem for the company, as they can't access TSMC's facilities for chip production, and will have to turn to domestic production in the likely case, with SMIC being the only leading option to consider.
Sources:
Device Hunt, @Loeschzwerg_3DC (X/Twitter), via Tom's Hardware
Thanks to @Löschzwerg who found the Device Hunt submission, we see hardware identifiers in PCI ID and USB ID repositories earlier than launch, as this often signals the testing of new chips or drivers by various companies. In the case of MTT, the latest developments are complicated by its recent inclusion on the U.S. Entity List, limiting its access to US-made technologies. This introduces a problem for the company, as they can't access TSMC's facilities for chip production, and will have to turn to domestic production in the likely case, with SMIC being the only leading option to consider.
9 Comments on Moore Threads Prepares S90 and S4000 GPUs for Gaming and Data Center
The only possible use of something so inefficient is if you don't have access to other options.
Sorry but your comment is just very shortsighted, these steps are needed to reach something that will compete.
There's a reason there's a duopoly in the GPU market, AMD and Nvidia control enough IP that no one can compete with them. Even Intel pays AMD to not incur patent infringement.
Tesla thrived in a market with no competition, no strings attached. It's a meaningless comparison
I also have no clue why you think that having to pay to use a patent means you cannot compete.....that is such an, I have trouble putting into words how silly that chain of thought is, its like being aware and unaware of how things work at the same time.
To embrace the challenge of competing with AMD and Nvidia and their thousands of engineers, as well as their billions in capital, several key factors must be in place:
First, you need the capital to cover the hundreds of millions required to develop a chip using the latest manufacturing process, as competing without being in the same process is not feasible, assuming there is available capacity.
Second, you have to procure volumes equivalent to those of AMD and Nvidia, as larger scale translates to lower costs per unit, and these companies receive special discounts from TSMC for obvious reasons.
Third, you must provide software support (drivers) and collaborate with gaming companies throughout the planned lifespan of your product. This will entail additional expenditure.
Fourth, you will need to perform exceptionally well to succeed on all the above points, pay for third-party IP, and still maintain sustainable competitiveness. Based on a superficial overview, this company is already just using imagination IP, which isn't a big deal. It's easier to wait for water to turn into gasoline than to wait for a company like this to bring a GPU capable of facing the titans. Consider Intel, which, despite having the financial resources to withstand potential setbacks and losses, faces challenges in this regard.
To come into the CPU and especially the GPU market is a monumental task. Even if your 1st gen GPU somehow ends up being better than whatever Nvidia or AMD has, which would be an amazing feat of engineering to do so without infringing on patents or just in general, consumers aren't likely to purchase your product because proprietary features like CUDA, Reflex, ect mean that your card can't be used in any professional capacity and will lack a lot of software features compared to competitors simply because many games utilize features that only work on a single GPU brand. Your GPU might as well be a brick in adobe software, which alone is going to loose you 20-30% of your marketshare. In addition, as a new market player you are at a disadvantage as existing players can control game features and performance to an extent with game sponsorships.
The level of entrenchment in the GPU market essentially makes it near impossible to enter. The fact that Intel is having trouble when it already has experience making GPUs, with sponsorships, and with implementing software features just goes to show you how difficult it is to enter.
PS. Almost same but better exposed in wcc*kek.