GPU AND FPGA SOLUTIONS
GPGPUs and FPGAs have been at forefront of accelerating HPC solutions for quite a few years now. They accelerate varied workloads by providing either a purpose-built architecture (FPGA) or a highly parallelised multi-core environment.
With the advent of new-age workloads – AI and Machine learning, these accelerators have evolved to suit these workloads.
MPCL provides solutions built around GPGPU from NVIDIA and FPGA from Intel. NVIDIA solutions comprise of cards populated on standard x64 servers or DGX systems with 8 or 16 GPGPUs giving a tremendous boost to the AI/ML applications. MPCL has the skills to deploy the DGX systems and provide L1 support and diagnosis.
Parallel and TensorFlow applications
NVIDIA V100 and NVIDIA T4 are offerings from NVIDIA. While NVIDIA V100 features Cuda cores as well as Tensor cores, T4 features Tensor cores. Based on application profile, we propose these GPUs to our customers.
Programmable ASIC like environment
Intel Stratix and Arria FPGAs provide ASIC like environment which can be programmed to execute specific workloads. Another brownie point served by FPGAs is very low power consumption.