Nvidia has debuted an advanced line of Fermi-based Tesla GPUs for the high performance computing (HPC) market.
An Nvidia spokesperson told TG Daily that the Tesla 20 GPU series significantly accelerates a number of processor intensive applications, including ray tracing, 3D cloud computing, video encoding and computer-aided engineering.
?”At their core, Tesla GPUs are based on the massively parallel CUDA computing architecture that offers developers a parallel computing model that is easier to understand and program than any of the alternatives developed over the last 50 years,” the spokesperson explained.
He added that Tesla 20 GPUs offer support for double precision floating points (IEEE 754-2008), standard error correcting codes (ECC) and a multi-level cache hierarchy (L1 and L2).
The GPUs also support:
- Up to 1 terabyte of memory.
- Concurrent kernel execution.
- Fast context switching.
- 10x faster atomic instructions.
- 64-bit virtual address space.
- System calls and recursive functions.
The Tesla C2050 and C2070 GPUs (520GFlops – 630 GFlops) are expected to retail for $2,500 and $4,000, while the Tesla S2050 and S2070 (2.1 TFlops – 2.5 TFlops) will have $13,000 and $19,000 price tags.
All four GPUs will be officially released in Q2 2010.
Nvidia CEO says he is “all” Apple
Nvidia: Fermi production schedule remains on track
Nvidia may be moving into X86 market
Nvidia: Our nForce chipsets are “better” than Intel’s
Nvidia: Windows 7 “dramatically improves” GPU performance