nvkhkhr
Prime VIP
IBM has struck an important deal with the graphics giant Nvidia that has resulted in the creation of a GPGPU-enabled IBM datacenter. This major milestone helps IBM customers that crunch lots of numbers take advantage of a tremendous horsepower provided by the Nvidia Tesla supercomputing processors.
Dubbed the iDataPlex DX360 M3, this new IBM datacenter server features dual PCIe X16 slots that house two Nvidia Tesla GPGPU boards, Bright Side of News reported. The publication noted that IBM will ship the DX360 M3 with two Tesla C2050 cards installed. For customers that need even more power, IBM will begin shipping in the third quarter the Tesla C2070 board with 6GB GDDR5 memory. The publication noted how IBM cleverly circumvented datacenters’ lack of easy upgrade for individual nodes:
We were not surprised to see that the Tesla boards are actually positioned in the front of the system, rather than being in the back. This way, datacenter administrators can easily access and replace not just the four hard drives, but the GPGPU boards as well. Given that upgrading a datacenter is nothing less than a daunting task, IBM is offering very neat future-proofing option, when either 28nm or 22nm GPUs appear on the scene.
But what’s such a powerful graphics technology doing inside a datacenter, you ask. I asked Theo Valich, the author of the report, the same question. He responded via email that IBM was smart enough to provide its customers with a GPGPU route rather than push the alternatives like the Power or Cell chips. It’s all about GPGPU technology coming of age, he said:
Performance difference between GPGPU and CPU code can often be measured in orders of magnitude – and you’re saving millions of dollars in cooling and electricity cost at the same time. From scientific research to Hollywood movies to your home, GPGPU performs tasks much faster than a conventional CPU.
Hardly a buzzword, GPGPU is here to stay for a long, long time, Valich said. Scalability and performance of GPUs far outweighs the costs associated with engineering high-performance systems around general-purpose CPUs. A quantum leap in GPU performance over CPU takes us places “we couldn’t have gone before.” Most importantly, Valich said, GPGPU computing “creates new types of jobs in the same way personal computer revolution in the 80s and 90s helped create the world we know today.”
