Asus Releases Desktop-Sized Supercomputer

angry tapir writes "Asustek has unveiled its first supercomputer, the desktop computer-sized ESC 1000, which uses Nvidia graphics processors to attain speeds up to 1.1 teraflops. Asus's ESC 1000 comes with a 3.33GHz Intel LGA1366 Xeon W3580 microprocessor designed for servers, along with 960 graphics processing cores from Nvidia inside three Tesla c1060 Computing Processors and one Quadro FX5800."

Comments

But how can you trust the results?

[HalfFlat]

The Tesla c1060 processor boards sound like a very efficient way of packing in compute power, but unless they're neglecting to mention it, the 4GB of GDDR3 RAM each has on board has no error correction. Given the rates of correctable errors observed e.g. here, I could never recommend using it for computing simulations that matter. A flipped bit in a floating point number can have a disproportionate affect on the outcome of calculations that rely upon it, and short of running the whole simulation a second or third time, one couldn't be confident that such an error did not occur.

Large compute-intensive simulations can take weeks, and are used to justify engineering and business decisions that involve the disposition of large amounts of money and other resources — it is important that the computational part of the process can be relied upon.

Re:But how can you trust the results?

[Anonymous Coward]

I'm a student at the University of Washington and once talked to a representative for Cray about using GPU's a a cheaper supercomputer and he told me that they generally have a nontrivial error rate. The issue with using ECC memory is that the GPU's are also libel for errors within their computations, making the ECC RAM pointless. A weird pixel in one frame of a game is no problem, but an error when performing a large simulation creates problems if the algorithm isn't designed to compensate for that noise.