level 33

“In the past, we designed supercomputers to do physics. That’s why rapid FLOPS [floating-decimal-point operations-per-second] are so important, because there’s a lot of math to do. But this new kind of test measures memory access and the ability to marshal huge data sets efficiently,” he said. “Graph500 is a test for a totally new area.”
Applications include following the huge number of barrels of oil in transit around the world in ships, or keeping track of the medical records of every patient in the U.S.
In fact, he said, the goal is a transformation of the field, from using supercomputers to simulate a hypothesis to building supercomputers capable of generating a hypothesis.
The transition to exascale computing, or one million trillion operations per second, will be challenging, Murphy said. “Unlike the tera-to-petascale transition, we know we can’t just scale commodity architectures: the barriers have to do with fundamental physics. Perhaps even more significantly, the tasks we want the computer to achieve are changing: it’s not just 3-D physics anymore.
http://www.graph500.org/specifications
There are five problem classes defined by their input size:
toy
17GB or around 10^10 bytes, which we also call level 10,
mini
140GB (10^11 bytes, level 11),
small
1TB (10^12 bytes, level 12),
medium
17TB (10^13 bytes, level 13),
large
140TB (10^14 bytes, level 14), and
huge
1.1PB (10^15 bytes, level 15).
В штатах уже есть компании, активно работающие
с данными пятнадцатого уровня в бизнес-целях.
Пытаются то есть, потому что данные есть, получается не очень.
Людей потому что знающих как надо правильно
это делать нет, вообще то есть нет.
|
</> |