IBM has brought together Linux, blade technology, supercomputer cluster management and storage software to produce what it says is a low-cost supercomputing environment.
Big Blue has re-designed its eServer Cluster 1350, introduced last September, to accommodate 84 two-way blade servers (168 processors) per rack in the six-rack cabinet.
It has also introduced a Linux version of its cluster management software (CSM) - developed originally for its Deep Blue chess-playing supercomputer - with added autonomic computing techniques.
"What this means for customers is the density of 1U servers has gone from 72 to 168 processors in the same floor space, with cabling reduced by 90 per cent," said Nick Davis, IBM EMEA xSeries Linux sales manager.
"With Linux this means ridiculously cheap supercomputing. Jobs you previously thought were impossible have now hit a price-point where the impossible has become possible."
Davis added that CMS auto-maintained the cluster, allocating processors to jobs and handling errors including failover with minimal user intervention.
But IBM is yet to detail actual pricing for the mix, which will be available on 6 June.
Martin Hingley, vice president of European systems group at IDC, said it was a huge advantage that the system could manage everything together.
"But no company is going to throw away what they've got, so the low cost really applies to a forward-looking strategy," he said.
Hingley added that cable-reduction alone could produce a huge cost saving, but said that a fully-populated cabinet could give rise to heat problems.
The Cluster 1350 can comprise any combination of the Intel Xeon-based eServer x335 1U-high and x345 2U servers, or use the x345 or 2-4 processor x360, which contains disk bays. One x345 is used as a management node.
Trump proposes a $1.3bn fine and a round of firings to un-bork ZTE
Findings could mean new optical frequencies to transmit more data along optical cables
Findings made by reconstructing its orbit by numerical simulation
3D printer was specially adapted to build therapeutic biomaterials from multiple materials