Researchers from Hewlett-Packard Enterprise (HPE) have unveiled what they claim is a breakthrough in computing, with a new machine capable of handling vast amounts of data at supercomputing speeds.

The prototype, named "the Machine", uses a new approach to computer architecture that the company says can be adapted for a range of big-data applications, handling tasks at thousands of times the speed of existing devices.

The system is called "memory-driven computing" and uses light waves to transmit data instead of electrical impulses travelling over silicon, bypassing what engineers say is an obstacle to boosting speeds.

Sharad Singhal, who heads machine applications for HPE, said previous efforts to boost computing power "were running into a brick wall into computation" because computing needs were increasing beyond the capacity of existing chips.

Singhal said the project was an effort "to rethink computers from the ground up". This means that instead of a silicon chip at the heart of the computer, "we are putting data at the centre", the researcher said.

The prototype, unveiled on Tuesday, contains 160 terabytes of memory, capable of simultaneously working with the contents of about 160-million books, a task never before possible in a single unit.

Singhal said one area in which the system could be useful was in healthcare, where powerful computing can analyse health studies, genetics and the potential for personalising medical treatment.

"These kinds of things can be done a lot faster on the architecture we are talking about." He said this approach could help shorten the time in which medicines developed by better analysis of their effectiveness and side effects.

"The secrets to the next great scientific breakthrough, industry-changing innovation, or life-altering technology hide in plain sight behind the mountains of data we create every day," HPE CE Meg Whitman said.

"To realise this promise, we can’t rely on the technologies of the past, we need a computer built for the big data era."


Please sign in or register to comment.