Monday, April 5th 2021
Tianshu Zhixin Big Island GPU is a 37 TeraFLOP FP32 Computing Monster
Tianshu Zhixin, a Chinese startup company dedicated to designing advanced processors for accelerating various kinds of tasks, has officially entered the production of its latest GPGPU design. Called "Big Island" GPU, it is the company's entry into the GPU market, currently dominated by AMD, NVIDIA, and soon Intel. So what is so special about Tianshu Zhixin's Big Island GPU, making it so important? Firstly, it represents China's attempt of independence from the outside processor suppliers, ensuring maximum security at all times. Secondly, it is an interesting feat to enter a market that is controlled by big players and attempt to grab a piece of that cake. To be successful, the GPU needs to represent a great design.
And great it is, at least on paper. The specifications list that Big Island is currently being manufactured on TSMC's 7 nm node using CoWoS packaging technology, enabling the die to feature over 24 billion transistors. When it comes to performance, the company claims that the GPU is capable of crunching 37 TeraFLOPs of single-precision FP32 data. At FP16/BF16 half-precision, the chip is capable of outputting 147 TeraFLOPs. When it comes to integer performance, it can achieve 317, 147, and 295 TOPS in INT32, INT16, and IN8 respectively. There is no data on double-precision floating-point numbers, so the chip is optimized for single-precision workloads. There is also 32 GB of HBM2 memory present, and it has 1.2 TB of bandwidth. If we compare the chip to the competing offers like NVIDIA A100 or AMD MI100, the new Big Island GPU outperforms both at single-precision FP32 compute tasks, for which it is designed.
Pictures of possible solutions follow.
Source:
via VideoCardz
And great it is, at least on paper. The specifications list that Big Island is currently being manufactured on TSMC's 7 nm node using CoWoS packaging technology, enabling the die to feature over 24 billion transistors. When it comes to performance, the company claims that the GPU is capable of crunching 37 TeraFLOPs of single-precision FP32 data. At FP16/BF16 half-precision, the chip is capable of outputting 147 TeraFLOPs. When it comes to integer performance, it can achieve 317, 147, and 295 TOPS in INT32, INT16, and IN8 respectively. There is no data on double-precision floating-point numbers, so the chip is optimized for single-precision workloads. There is also 32 GB of HBM2 memory present, and it has 1.2 TB of bandwidth. If we compare the chip to the competing offers like NVIDIA A100 or AMD MI100, the new Big Island GPU outperforms both at single-precision FP32 compute tasks, for which it is designed.
41 Comments on Tianshu Zhixin Big Island GPU is a 37 TeraFLOP FP32 Computing Monster
This is Big Navi on "stereoids", and whatever limitation AMD placed in the design (check the licensed Chinese EPYC v1 CPU vs AMD EPYC 7xx1 CPU), is gone. This is a very high performing part.
This was done before in the 90's with companies like Tseng Labs, Rendition, Number Nine, 3D Labs, etc.... They all promised the Moon and delivered...well...
The problem is software. I reckon some battles shall be fought there. Lots of ban hammers can be swung...
The hardware... They will make it. We already saw some low end Zen ripoffs.
Time will tell how good these are, but it's clear China is pretty good at building big islands. ;)
I will believe the performance claims when they are independently tested. I hope they are excellent at mining and really cheap. Then the miners will leave the video cards to gamers.
Pretty amazing how they could pull something like that out of their hat so fast!!?!?!?!
This isn't the licenced CPU, the Chinese version of which is actually the lower performant , limited device.
And no where have I heard about AMD licencing rDNA or cDNA ,which this server part would fight with, so as the Op says this is a from scratch developed GPU not a Chinese copy of an American design, time and the lawyers will tell how true that is ,if required.