Advanced Micro Devices (AMD) is joining forces with Baidu (BIDU) to work on GPU computing in data centers.
They are teaming up to enhance optimization for applications that use Radeon
Instinct accelerators. Advanced Micro Devices is eyeing the high-growing neural
network and machine learning market with its current Radeon Instinct efforts.
The company is targeting the machine learning and
neural networks market with its new GPU architecture, Vega. It's finally
targeting the right market. The company's new architecture supports FP16
and INT8 modes, which are highly beneficial for machine learning
performance.
The most significant take away from the recent
announcement is the software optimization efforts. One of the widely cited cons
for Advanced Micro Devices' data center offerings is the lack of optimization
and software support.
No comments:
Post a Comment