Cray bites the bullet on mother of all computers

The roll out of the Oak Ridge National Laboratory’s “Titan” super computer is being delayed because the Cray motherboards have too much gold in them.

According to Slashdot, the boards have to be sent back to Cray because the gold connectors have prevented it from working.

Titan was turned on in October and climbed to the top of the Top 500 list of the fastest supercomputers. But the testing was not finished because a fault was discovered.

The system was not stable enough to be used in anger and the schedules were delayed until the end of April.

Engineers found that the problems were with the connectors.

Over 200 cabinets had to be taken offline and the motherboards were shipped back to Cray for repairs.

There are about 20,000 connectors within the Titan, connecting the CPUs and GPUs. Each connector has about 100 pins.

The gold problem means that Titan works but cannot use its Nvidia K20X Kepler chips. It has managed to come up to speed just by using its 2.2 GHz 16-core AMD Opterons.

Apparently Cray has promised to fix the motherboards and seems to have put its hands up for the buck stopping at its door.