Nvidia Corp. has quietly revealed specifications of its next-generation Tesla cards designed for high-performance computing. Rather surprisingly, the flagship computing processor boards will not feature the GF100/Tesla T20 graphics processing unit (GPU) with all stream processors enabled: both Tesla C2050 and C2070 will only feature 448 stream processors.

As reported previously, Nvidia Tesla C2050 and C2070 computing processor boards are single-chip cards with 3GB and 6GB (respectively) of on-board GDDR5 memory (with ECC enabled, user available memory will be 2.625GB for a C2050 and to 5.25GB for a C2070) with 384-bit interface operating in the range between 1.80GHz and 2.0GHz. The cards are mostly designed for desktop-based high-performance computing and other compute-intensive applications and are not meant to do perform graphics tasks. Nvidia stated that its new Tesla solutions will deliver double precision performance in the range of 520GFlops – 630GFlops.

Previously it was widely believed that the new Nvidia Tesla C2000 boards based on the Tesla T20 GPUs will feature 512 stream processors at 1.25GHz – 1.40GHz clock-speeds, the fully-fledged configuration of the company’s GF100 chip (also known as NV60, G300, GT300, etc) based on Fermi architecture at slightly reduced clock-speeds. However, Nvidia decided to cut down the Tesla C20 GPU for some undisclosed reasons.

Perhaps, Nvidia had to lower performance of its flagship single-chip computing solution for desktops in order to keep thermal design power in certain envelope. Nvidia declares 225W TDP for Tesla C2000 boards, meanwhile, the company’s current cards – GeForce GTX 285, dual-chip GeForce GTX 295 and Tecla C1050 – based on 55nm G200b (GT200b, NV55b) have maximum heat dissipation at 204W, 289W and 187.8W, respectively. Both Tesla C2000 boards will have one 8-pin PCI Express power connector and one 6-pin PCI Express power connector.

The Tesla C2050 and C2070 computing processor cards will retail for $2499 and $3999. Products will be available in Q2 2010.

The flagship Nvidia Fermi “GF100” 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 (for GPUs) 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.

Tags: Nvidia, Fermi, Tesla, GF100, NV60, GT300, 40nm, Geforce, Quadro