Home > News content

The performance of tachyum self-developed CPU is compatible with X86, arm and rsic-v instructions

via:快科技     time:2020/8/6 19:13:24     readed:866

Speaking of tachyum, most people have never heard of this company, but in the field of processors, they can make a lot of changes. Previously, they developed a CPU architecture with 128 cores at most. Now, the 64 core version of tachyum processor is compatible with X86, arm and risc-v instruction sets.


China Telecom Tianyi cloud S2 virtual machine limit 1, the minimum is only 88 yuan / year

Alibaba cloud welfare service ECS as low as 102 yuan / year

In the field of CPU, in addition to high-performance x86 and low-power arm, risc-v has also gained momentum recently. These three instruction sets are the most promising in the future. Tachyum is a latecomer, which was founded in Slovakia in the last two years. However, the main founders are Americans with rich experience.

Tachyum has developed a new CPU architecture, prodigy, in a single silicon chip,It integrates general purpose processor, high performance computing, AI Artificial Intelligence, DML deep mechanical learning, explainable AI, bio AI, etc. based on parallel multiprocessor environment, the programming model and environment can be simplified.

The most advanced prodigy supports up to 128 CPU cores, 12 channel DDR4 / 5 memory, 64 core t864, th24, and 32 core t432 processors. The specifications are simplified to varying degrees, but the overall configuration is still amazing and luxurious.

A few days ago, tachyum announced the latest development of its CPUProdigy processor has successfully completed the binary simulation for x86, arm and risc-v architecturesThis is a milestone of prodigy processor, which means that it can be compatible with the above three ecology, and its performance is better than current and future arm and risc-v processors.

Tachyum means,In the 9-18 months of transition to prodigy platform, customers can run 100% local programs, and their performance exceeds the fastest Xeon processor.

According to tachyum, the power consumption of prodigy processor in the data center is 10 times lower than that of Xeon. The performance of prodigy processor on AI and HPC is better than that of NVIDIA's fastest GPU. 125 prodigy cabinets can provide 32 exalfops tensor performance, that is, 320 billion times.

At the same time, the cost per MIPS of prodigy processor is 3 times lower, the power conversion is 10 times lower, and the total cost of TCO is 4 times lower.

However, prodigy processor's progress has also been delayed, saying that it would quit within this year,It will not be available until 2021The company is also developing a native Linux distribution to match the prodigy processor.

China IT News APP

Download China IT News APP

Please rate this news

The average score will be displayed after you score.

Post comment

Do not see clearly? Click for a new code.

User comments