A Case Against That Other Paper

By: Mark Roulo (nothanks.delete@this.xxx.com), May 18, 2021 3:32 pm
Room: Moderated Discussions
Brendan (btrotter.delete@this.gmail.com) on May 18, 2021 12:37 pm wrote:
> A Case Against That Other Paper
> By Brendan and a Rubber Duck
> Introduction
> CPUs with lots of hardware threads suffer severe problems for all forms of caching (including branch
> prediction and TLBs) while also destroying any hope of effective hardware prefetching (for instruction,
> data and TLB) as there's no hope of any kind of locality between disparate threads; causing cache
> thrashing, and exacerbating the "1000+ channel memory controller doesn't exist" problem. With significantly
> fewer hardware threads (no more than 4) we believe a CPU can actually work properly, leading to
> an order of magnitude better performance for embarrassingly parallel workloads and 2 or more orders
> of magnitude better performance for anything subject to Amdahl's law.
> Currently, all of the CPUs that provide many hardware threads get fundamental parts of task switching wrong.

The paper proposes that the SOFTWARE control the task switching, which is quite different from today's SMT/Hyperthreading.

The idea is that task switching will be faster if each task stores its register state in H/W rather than in memory (cache or DRAM). The OS, presumably, would still be responsible for selecting the thread(s) to run at any one time, but the cost of a task switch would be low.

I'm envisioning something like SPARCs register windows, just used for a different purpose and MUCH larger. Or like the Z80 banked registers, just much more so.
< Previous Post in ThreadNext Post in Thread >
TopicPosted ByDate
A Case Against (Most) Context SwitchesLittle Horn2021/05/17 05:03 PM
  A Case Against (Most) Context Switchesrwessel2021/05/17 06:55 PM
  A Case Against (Most) Context SwitchesFoo_2021/05/18 01:58 AM
    A Case Against (Most) Context SwitchesDoug S2021/05/18 08:45 AM
      A Case Against (Most) Context SwitchesKonrad Schwarz2021/05/19 07:35 AM
  A Case Against (Most) Context SwitchesEtienne Lorrain2021/05/18 03:11 AM
  A Case Against (Most) Context SwitchesAndrey2021/05/18 06:58 AM
  A Case Against (Most) Context Switchesgallier22021/05/18 08:41 AM
  A Case Against (Most) Context Switches---2021/05/18 09:00 AM
  A Case Against That Other PaperBrendan2021/05/18 12:37 PM
    A Case Against That Other PaperMark Roulo2021/05/18 03:32 PM
      A Case Against That Other PaperBrendan2021/05/18 11:05 PM
        A Case Against That Other PaperMark Roulo2021/05/19 01:09 PM
  A Case Against (Most) Context SwitchesRomain Dolbeau2021/05/19 04:05 AM
    A Case Against (Most) Context SwitchesBjörn Ragnar Björnsson2021/05/19 01:13 PM
      A Case Against ... authors show zero awareness of Cray-MTABjörn Ragnar Björnsson2021/05/19 06:18 PM
    Cray MTA avoided cachesPaul A. Clayton2021/05/20 06:36 AM
      Cray MTA avoided cachesdmcq2021/05/20 10:09 AM
        Cray MTA avoided cachesRayla2021/05/20 10:28 AM
      A LONG response to the paperPaul A. Clayton2021/05/22 06:15 AM
        A LONG response to the paperAdrian2021/05/22 09:18 AM
          Thank you for the note of appreciationPaul A. Clayton2021/05/24 05:06 AM
  A Case Against (Most) Context Switchesdmcq2021/05/19 01:47 PM
Reply to this Topic
Body: No Text
How do you spell tangerine? 🍊