Showing posts with label gpgpu. Show all posts
Showing posts with label gpgpu. Show all posts

10.31.2022

TSMC Suspends Production of Chinese Biren Technology 7nm Accelerators Due to US Sanctions

TSMC Suspends Production of Chinese Biren Technology 7nm Accelerators Due to US Sanctions

TSMC Suspends Production of Chinese Biren Technology 7nm Accelerators Due to US Sanctions

A day ago, Bloomberg confidently reported that the computing capabilities of the Chinese company's Biren BR100 gas pedals are slightly worse than NVIDIA A100, and therefore the US sanctions will not prevent their further production.
Today, the source sharply changed the rhetoric, saying that TSMC decided to suspend the shipment of these 7nm gas pedals to Chinese customers.Image source: Biren TechnologyIn summer, Biren Technology was not shy about stating that its BR100 gas pedal could challenge the NVIDIA A100.
Features of Chinese product were really impressive: 77 billion transistors, 7nm process technology and spatial layout of 2.5D CoWoS, 64 GB of memory, support for PCI Express 5.0 and CXL interfaces.
However, when speaking about a new wave of anti-Chinese sanctions, Biren representatives preferred to remain anonymous and claimed that their chip does not meet NVIDIA A100 level and therefore would not fall under new export tunnels.
Today Bloomberg reports that TSMC refused to take risks and stopped production of Biren Technology 7nm chip gas pedals.
The Taiwanese contractor still hasn't come to a final conclusion about the belonging of Biren products to the banned category of products, but the shipments will be stopped \"just in case\".
Biren gas pedals are designed to work in machine vision systems, processing of natural speech and its synthesis using artificial intelligence.
Representatives of the Chinese company have not yet commented on the news, according to Bloomberg.

9.05.2022

U.S.Authorities Restrict Shipments of Advanced AMD and NVIDIA Computing Accelerators to China and Russia

U.S.Authorities Restrict Shipments of Advanced AMD and NVIDIA Computing Accelerators to China and Russia

U.S.Authorities Restrict Shipments of Advanced AMD and NVIDIA Computing Accelerators to China and Russia

The last day of the calendar summer brought unpleasant news to Chinese and Russian AMD and NVIDIA customers.
The U.S.
government just banned both companies from shipping GPU computing gas pedals to those countries that meet certain performance criteria.
In the case of NVIDIA refers to products series A100 and H100, while AMD attributed to the list of \"sanctioned\" Accelerators Instinct MI250.Image source: AMDAccording to AMD representatives, as explained by Reuters, the supply of Accelerators Instinct MI100 did not fall under the ban.
At the same time, AMD believes that the new restrictions will not have a significant impact on the company's business, as trade turnover in this area were not large enough.
NVIDIA shares on the background of this news lost more than 6% in price, in the case of AMD reduction did not exceed 3.8%.
It is believed that the level of performance that determines the need for a ban on the supply of gas pedal in China and Russia, took the speed solution NVIDIA A100 with Ampere architecture.On the idea of American officials, advanced components of American origin should not be used by Chinese and Russian customers to improve military technology.
AMD and NVIDIA computing gas pedals, which are under the ban, can theoretically be used in speech recognition systems for eavesdropping and recognition of military objects in the images of terrain received from reconnaissance satellites.

1.20.2021

China has created another GPU that threatens NVIDIA and AMD in the data centre

China has created another GPU that threatens NVIDIA and AMD in the data centre

China has created another GPU that threatens NVIDIA and AMD in the data centre

The GPU market is getting crowded. We are not talking about graphics cards for PC yet, although the Chinese have already made a bid to squeeze NVIDIA and AMD GPUs out of the entry-level segment. Attempts have been made on the data centre accelerator market, where NVIDIA and AMD GPUs have been successfully present for a long time. Chinese company unexpectedly presented GPGPU of its own design, built on the most advanced 7nm processor.

GPGPU Big Island (BI). Image source: Tianshu Zhixin

The Big Island (BI) general purpose graphics processing unit (GPGPU) is reportedly developed by Shanghai Tianshu Zhixin Semiconductor Co., Ltd. The project was launched in 2018 and relies solely on the original graphics architecture. The chip is apparently manufactured by TSMC, as the solution uses the 2.5D CoWoS packaging practiced by this Taiwanese contract manufacturer. The GPGPU Big Island digital design was ready in May last year. In silicon, the solution with 24 billion transistors appeared in December and has been comprehensively tested ever since. The company does not disclose all the specifications of the new product, but claims, for example, that for FP16 operations Big Island achieves a performance of 147 TFLOPS. By comparison, under similar conditions, the NVIDIA A100 GPU achieves 77.97 TFLOPS and the AMD Instinct MI100 & ; 184.6 TFLOPS. In addition to FP16 operations, the Big Island GPGPU supports a host of other compute formats, including FP32, BF16, INT32, INT16 and INT8. Thereby, the Chinese novelty for data centres can offer a very attractive performance/cost ratio. First of all, as we can see from the list of supported formats, we are talking about machine learning (training) and tasks related to artificial intelligence. It looks like «Skynet» will have Chinese as its native language.