Concurrency costs

By: Travis Downs (, July 6, 2020 7:26 pm
Room: Moderated Discussions
I wrote a post about concurrency costs.

It is not intended to be a "platform shootout" between hardware. Everything I tested seemed to do well enough. The contended tests are mostly is "100% contention", which tends to favor unfair platforms (which allow several operations to execute on a core once it gets the line), which is not necessarily something you want in other scenarios, so you shouldn't draw "HW a is better than b" conclusions here.

Feedback welcome.

 Next Post in Thread >
TopicPosted ByDate
Concurrency costsTravis Downs2020/07/06 07:26 PM
  Concurrency costsAnon2020/07/07 01:15 AM
    Concurrency costs - presentationMichael S2020/07/08 01:24 AM
    Concurrency costsTravis Downs2020/07/08 09:26 AM
      Concurrency costsAnne O. Nymous2020/07/08 10:12 AM
        Concurrency costsrwessel2020/07/08 10:28 AM
          race condition detected (NT)Travis Downs2020/07/08 10:29 AM
          Concurrency costsAnne O. Nymous2020/07/08 10:53 AM
        Concurrency costsTravis Downs2020/07/08 10:29 AM
  Concurrency costsj2020/07/07 04:19 AM
    Concurrency costsTravis Downs2020/07/08 09:42 AM
  Huge difference between chipsDavid Kanter2020/07/07 06:58 AM
    Huge difference between chipsMichael S2020/07/07 08:35 AM
      Huge difference between chipsTravis Downs2020/07/08 09:11 AM
        Huge difference between chipsnone2020/07/08 09:20 AM
          Huge difference between chipsTravis Downs2020/07/08 09:46 AM
            Huge difference between chipsnone2020/07/08 10:00 AM
          Huge difference between chipsMichael S2020/07/08 12:25 PM
            Huge difference between chipsLinus Torvalds2020/07/08 01:11 PM
              Huge difference between chipsMichael S2020/07/08 02:33 PM
  Fabulously detailed post!David Kanter2020/07/07 06:59 AM
  Concurrency costsMichael S2020/07/07 08:53 AM
    Concurrency costsTravis Downs2020/07/10 01:19 PM
  Concurrency costs: AMDAdrian2020/07/07 11:28 AM
    Concurrency costs: AMDTravis Downs2020/07/08 09:51 AM
      Concurrency costs: AMDAdrian2020/07/08 11:30 PM
        Concurrency costs: AMDMichael S2020/07/09 12:20 AM
          Concurrency costs: AMDanon22020/07/09 12:24 AM
            Concurrency costs: AMDAdrian2020/07/09 05:19 AM
          Concurrency costs: AMDAdrian2020/07/09 05:55 AM
            Concurrency costs: AMDMichael S2020/07/09 07:24 AM
              Concurrency costs: AMDTravis Downs2020/07/09 09:35 AM
                Concurrency costs: AMDMichael S2020/07/09 10:14 AM
        Concurrency costs: AMDTravis Downs2020/07/09 09:37 AM
          Concurrency costs: AMDAdrian2020/07/09 11:55 AM
            Concurrency costs: AMDTravis Downs2020/07/09 12:53 PM
  Regarding scalable counting and coherence trafficanon2020/07/07 12:59 PM
  Concurrency costsgpd2020/07/07 01:42 PM
    Concurrency costsTravis Downs2020/07/08 09:55 AM
  Concurrency costsLinus Torvalds2020/07/07 01:56 PM
    Concurrency costsTravis Downs2020/07/09 12:50 PM
      Should say: I did also try to measure this effect, although I did not WRITE ABOUT it. (NT)Travis Downs2020/07/09 12:51 PM
      Concurrency costsTravis Downs2020/07/09 01:06 PM
        Concurrency costsLinus Torvalds2020/07/09 06:11 PM
          Concurrency costsTravis Downs2020/07/09 06:44 PM
            Concurrency costsLinus Torvalds2020/07/09 07:24 PM
              Concurrency costsLinus Torvalds2020/07/09 08:23 PM
        Does type of LDADD matter?Michael S2020/07/10 02:20 AM
          Does type of LDADD matter?Travis Downs2020/07/10 08:56 AM
            Does type of LDADD matter?Michael S2020/07/10 09:18 AM
              Does type of LDADD matter?Travis Downs2020/07/10 01:04 PM
            Does type of LDADD matter?Linus Torvalds2020/07/10 09:28 AM
              Does type of LDADD matter?Linus Torvalds2020/07/10 09:35 AM
                Does type of LDADD matter?Travis Downs2020/07/10 01:10 PM
  sanity checkMichael S2020/07/07 02:27 PM
    sanity checkLinus Torvalds2020/07/07 03:05 PM
      sanity checkMichael S2020/07/07 03:23 PM
        sanity checkMichael S2020/07/08 02:50 PM
    sanity checkTravis Downs2020/07/08 09:19 AM
      sanity checkMichael S2020/07/08 12:03 PM
        sanity checkTravis Downs2020/07/09 09:30 AM
          sanity checkMichael S2020/07/09 12:27 PM
            sanity checkTravis Downs2020/07/09 01:10 PM
            lines bouncingMichael S2020/07/09 01:40 PM
              standard deviation == bad ideaMichael S2020/07/09 03:02 PM
                standard deviation == bad ideaj2020/07/10 01:57 AM
                  standard deviation == bad ideaMichael S2020/07/10 02:45 AM
  Concurrency costsanon22020/07/07 04:57 PM
    Concurrency costsTravis Downs2020/07/08 09:59 AM
  Concurrency costsLaurent2020/07/10 05:02 PM
    Thank you, fixed and credited (NT)Travis Downs2020/07/10 07:32 PM
Reply to this Topic
Body: No Text
How do you spell purple?