Intel entered the GPU scene late with the Intel Arc , but has since accelerated innovation. In August, the company released its first Flex GPU series for data centers, and last week it released its Max GPU series for data centers.

The GPU Max Series for data centers is the highest density GPU in the industry, containing more than 100 billion transistors on 47 active modules and up to 128 Xe-HPC cores. The product maximizes throughput (up to 128 GB HBM2e), capacity (up to 408 MB of Rambo L2 cache) and memory (up to 64 MB of L1 cache).

Intel said its 408MB L2 cache will be able to deliver twice the performance of previous versions. The GPU solves a number of obstacles, such as code migration and refactoring, economic and technical difficulties associated with proprietary GPU environments which prevent portability between different GPU vendors, and finally, mismatches between CPU and GPU implementations, such as the CPU having too little memory.

Intel GPU Max is a product that maximizes bandwidth, processing power, developer productivity and efficiency. In addition, the entire Max Series, that is, both CPUs and GPUs, use oneAPI , an open-source programming model that allows developers to use a variety of accelerated architectures.

Data Center Max Series products are scheduled to launch in 2024.

Flex Series for Data Centers

The previous version of the Data Center GPU Flex Series was created to manage media streaming and cloud gaming, as well as to support AI visual output and virtual desktop infrastructure workloads. The device can be configured at different power levels to meet requirements ranging from basic AI needs to complex AI workloads.

Two weeks ago, Intel announced that the Flex series of data center GPUs has been added to the PluggableDevices family, also known as Intel’s extension to TensorFlow. The PluggableDevice architecture offers a pluggable module mechanism for registering devices with TensorFlow without requiring source code changes.

This new implementation will allow the use of Intel Data Center GPU Flex Series hardware and Intel Arch company graphics. However, it is believed to be compatible with Linux and the Windows subsystem for Linux when connected to oneAPI .