IBM Corp. said it will be using the Tesla graphic processing unit (GPU) chips made by Nvidia Corp in its Power System servers starting next year.
The use of Tesla graphic chips with Power central processing unit (CPU) in servers is meant to help users process and analyze data faster, according to Sean Tetpon, an IBM spokesman said. He said IBM intends to roll out racked mounted Power servers with Tesla GPU by early next year.
The first servers will combine the Nvidia chips with IBM’s 12-core Power8 chip. The Power8 chip, which ships next year, is up to three times faster than the current Power7 chip which was introduced in 2010 and used on IBM’s Watson supercomputer.
Power servers traditionally used vector co-processors, field-programmable gate arrays (FPGA) and other circuitry. However, IBM already uses Nvidia GPUs in System x servers which use Intel’s x86 processors.
The Power8 architecture is built around the PCI-Express 3.0 data transfer standard which is used for GPUs in PCs and x86 servers.