Nvidia have delayed the last piece of its GPU puzzle. After two delays in getting its Compute Unified Device Architecture (CUDA) out the door in November and then again in January, reports say Nvidia has pushed the architecture code-named Fermi later into next year.
Fermi was the last piece in Nvidia’s GPU computing puzzle, analysts say. The CUDA architecture brings the high-performance piece to the table. High-performance computing applications are expected to benefit from the first GPU to have an on-chip cache and off-chip shared memory.
Fermi’s graphics capabilities are expected to bring improvements in game play, multimedia encoding and enhancement, and other PC applications, according to the company.
Nvidia first demonstrated Fermi’s power at its GPU Technology Conference in September and again at the SCO9 event in November. Observers believed Fermi would be ready for mass production in January to coincide with the 2010 Consumer Electronics Show.
Nvidia executives, however, said the architecture is not delayed.
“I don’t believe there are any changes from anything we’ve already said, which was the first quarter,” said Henry Marinez, a Nvidia spokesperson. “We are still on schedule for shipping in Q1.”
Analysts believe Fermi is worth waiting for.
Fermi has increased performance startup by eight times over Nvidia’s last-generation GPU. The increase is critical for high-performance computing applications such as quantum chemistry, linear algebra, and numerical simulation, according to Nvidia. Fermi also provides supercomputing features and performance at one-tenth the cost and one-twentieth the power of other CPU-only servers.
Performance increases are seen in ray tracing, physics, high-precision scientific computing, sparse linear algebra, and sorting and search algorithms, according to Nvidia.
Fermi’s arrival was expected to coincide with two new industry standards, including one driven by Apple and Nvidia called Open Computing Language (OpenCL). Nvidia was also behind a second Direct Compute…