The world's first 7nm GPU! AMD Announces Radeon Instinct MI60\x2fMI50 Computing Card

Home > Sci-Tech

The world's first 7nm GPU! AMD Announces Radeon Instinct MI60\x2fMI50 Computing Card

2018-11-07 10:25:09 142 ℃

In addition to the CPU processor turning to the 7nm process, AMD GPU graphics computing also ushered in the 7nm era, but the first product is not the Radeon RX game card, but the Radeon Instinct computing card!

AMD today officially announced the new generation of Radeon Instinct MI60 and Radeon Instinct MI50, both based on the upgraded Vega architecture core of the 7nm process, the world's first 7nm GPU.

AMD said that the 7nm Vega core architecture has flexible high performance and the world's fastest FP64/FP32 PCI-E floating point performance, machine learning training and reasoning, leading video memory and scalability, unique hardware virtualization, end-to-end ECC error correction protection.

The 7nm Vega core integrates 13.2 billion transistors, an increase of 6.4% over the current 14nm Vega 1.25 billion, while the core area is 331 square millimeters, which is a full 31.6% smaller than the current 484 square millimeters. The transistor density has doubled.

With the same power consumption, the new core performance is improved by more than 25%, while at the same frequency, the new core power consumption is reduced by 50%.

The memory match is up to 32GB HBM2, the bandwidth is 1TB/s, and the built-in ECC error correction.

7nm Vega is also the world's first to support PCI-E 4.0 technology The GPU has a bidirectional bandwidth of 64GB/s and supports four GPUs in parallel with an Infinity Fabric system bus with a bandwidth of up to 100GB/s. The scalability is excellent, the dual-channel performance is improved by 99%, and the four-way is up by 298%. Eight roads are 664% better than single roads and are almost perfect.

Performance, MI60 FP64 double precision floating point up to 7.4TFlops, FP32 single precision up to 14.7TFlops, INT integer up to 118Tops, compared to the same 10nm Vega core MI25 professional performance up to 8.8 times At the same time, it surpasses or basically equals NVIDIA's most powerful Tesla V100.

AMD also announced the open source computing platform ROCm 2.0 and continues to commit to full open source.

In the future, AMD will continue to introduce the next generation of MI-NEXT, which is more powerful , better connectivity, better software compatibility - perhaps the Navi architecture?

HD Image:< /p>