Nvidia’s and its CUDA C-like programming for GPUs marks a turning point for multicore.  While there are several similar multicore devices with very high computation rates available from a variety of large and small companies, Nvidia offers the first which is guaranteed to be in high volume production.  Nvidia has also offered a credible software solution with CUDA.  Nvidia and similar devices from AMD / ATI will be a significant challenge to all other emerging multicore devices.

Nvidia CUDA Helps Bring Parallel Programming to the Classroom

 

From CNN Money:  Microsoft Unveils Windows HPC Server 2008

From Ars Technica:  Multicore, 64-bit x86 shakes up Top 500 Supercomputer List

AMD has announced its new GPU device with 320 cores and an estimated 500 GFLOPs performance:  AMD Delivers First Stream Processor with Double Precision Floating Point Technology

SiCoretex has announced its low-power multicore supercomputer cluster.  It makes use of a custom MIPS64-based multicore device containing six processors and drawing only 10 Watts of power per device.  The SC5832 has 5832 processors, giving a combined performance of 5.8 TFLOPS.  Total power consumption is less than 20 kWatts.

MIT professor Charles Leiserson is commercializing the cilk language with an eye on multicore:  Cilk Arts collects first funding from Stata Ventures

The EE Times reports that RapidMind is expanding its tool set to x86 multicore processors:  Tool speeds apps on multicore CPUs  RapidMind began with tools for GPU programming but has also provided support for the IBM / Sony / Toshiba Cell processor.