SoylentNews

1000-core MIMD Microprocessor Built

posted by takyon on Wednesday June 22 2016, @05:38AM   
from the more-core dept.

butthurt writes:

Motherboard reports on a press release by the University of California Davis, where researchers designed a multiple instruction, multiple data (MIMD) microprocessor. Unlike a GPU, each core can run distinct instructions on distinct data.

According to the researchers the chip has a greater number of cores than any other "fabricated programmable many-core [chip]," exceeding the 336 cores of the Ambric Am2045, which was produced commercially.

IBM was commissioned to fabricate the processor in 32 nm partially depleted silicon-on-insulator (PD-SOI). It is claimed that the device can "process 115 billion instructions per second while dissipating only 1.3 watts." or, when operating at greater supply voltage and clock rate, "execute 1 trillion instructions/sec while dissipating 13.1 W."

---

Original Submission #1  Original Submission #2 

• Re:Kilocore! What is it good for? (Score: 0) by Anonymous Coward on Wednesday June 22 2016, @09:18PM

  by Anonymous Coward on Wednesday June 22 2016, @09:18PM (#364024)

  That's not a C++ problem

  It usually is.

  that's a program organization problem

  Yes, C++ makes organising a program harder than many other languages.

  Just break the code up into a bunch of independent libraries and it will compile quickly.

  Interfaces are brittle in C++, which means library recompilation is often required, more so than in better-designed languages. Longer development cycle, more bugs...

  This also facilitates code reuse...though not as much as it logically ought to.

  Because C++ interfaces are complex and brittle. Templates? Exceptions? Compiler versions?

  Most of the libraries are likely to end up only being used in one project.

  Use a language with proper support for modules, no pre-processor and a proper ABI.

  Parent