PS I thought you guys were experts?
Nvidia Corp. this week formally announced its next-generation Tesla solutions for high-performance computing markets based on code-named Fermi architecture. The company will only be able to start shipments of Tesla 20-series in the second quarter of next year, which further emphasizes that the new Fermi-G300 chip is dramatically late-to-market.
The family of Tesla 20-series solution includes:
- Tesla C2050 and C2070 GPU computing processors:
- Single GPU PCI Express 2.0 cards for workstation configurations;
- Up to 3GB and 6GB (respectively) on-board GDDR5 memory. With ECC enabled, user available memory will be 2.625GB for a C2050 and to 5.25GB for a C2070;
- Double precision performance in the range of 520GFlops – 630GFlops.
- Tesla S2050 & S2070 GPU computing systems:
- Four Tesla GPUs in a 1U system product for cluster and datacenter deployments;
- Up to 12 GB and 24 GB (respectively) total system memory on board GDDR5 memory. With ECC enabled, user available memory will be 10.5GB for an S2050 and 21GB for an S2070;
- Double precision performance in the range of 2.1TFlops – 2.5TFlops.
The Tesla C2050 and C2070 products will retail for $2499 and $3999 and the Tesla S2050 and S2070 will retail for $12 995 and $18 995. Products will be available in Q2 2010. For more information about the new Tesla 20-series products, visit the Tesla product pages.
The flagship Fermi graphics processor will feature 512 stream processing engines (which are organized as 16 streaming multi-processors with 32 cores in each) that support a type of multi-threading technology to maximize utilization of cores and deliver extreme computing power. Each stream processor has a fully pipelined integer arithmetic logic unit (ALU) and floating point unit (FPU). The top-of-the-range chip contains 3 billion of transistors, features 384-bit memory GDDR5 memory controller with ECC and features rather unprecedented 768KB unified level-two cache as well as rather complex cache hierarchy in general. Naturally, the Fermi family is compatible with DirectX 11, OpenGL 3.x and OpenCL 1.x application programming interfaces (APIs). The new chips will be made using 40nm process technology at TSMC.
It is interesting to note that Nvidia has always claimed that its new graphics processor – which is known under G300, GT300 and NV60 code-names – was due to be released commercially already in Q4 2010. However, first Nvidia chief executive officer said that the company would only ramp up production of Fermi graphics processing units (GPUs) in the first quarter of the company’s fiscal year 2011, and then this week Nvidia said that “as previously announced, the first Fermi-based consumer (GeForce) products are expected to be available first quarter 2010”.
There is a clear reason why Nvidia is announcing the new Tesla solutions up to six month before actual shipments: the company needs to attract maximum attention to its high-performance computing options. Both Advanced Micro Devices and Intel Corp. – the companies which processors power the vast majority of super computers now – are either readying or offering GPUs for HPC market segment and they may potentially have a number of advantages over Nvidia.
Despite of the fact that Nvidia is generally a newbie on the market of super computers, it does have followers.
“Nvidia has deployed a highly attractive architecture in Fermi, with a feature set that opens the technology up to the entire computing industry,” said Jack Dongarra, director of the Innovative Computing Laboratory at the University of Tennessee and co-author of LINPACK and LAPACK.