By: pgerassi (pgerassi2012.delete@this.wi.rr.com), August 13, 2018 3:16 pm
Room: Moderated Discussions
> Linus Torvalds (torvalds.delete@this.linux-foundation.org) on August 13, 2018 2:12 pm wrote:
> wumpus (lost.delete@this.in.a.cave) on August 13, 2018 9:32 am wrote:
> > Per Hesselgren (perhesselgren.delete@this.yahoo.se) on August 13, 2018 8:50 am wrote:
> > > Threadripper on Anandtech
> >
> > Looks like coherency isn't an issue for 32 cores, but obviously not everything scales perfectly.
>
> Even loads that should scale really well (like the chromium build) seem to have hit the
> point of negative scaling here, with a 32c/64t 2990WX doing worse than a 16c/32t 2950X.
>
> I do suspect it's mostly the load that doesn't scale any more at that point, with the negative scaling coming
> from more memory contention and perhaps half of the cores being further away from the memory controller.
>
> Somebody who knows the Chromium build better might be able to tell how far they'd expect
> the build to scale, but even for simple things like a parallel build, you often hit
> serialization points (eg linking phases after you've built all your library files).
> And writing Makefiles that expose all the parallelism isn't trivial either.
>
> But some of that may be "nobody spent any effort to scale past X cores because they are so
> rare, and build farms have their own scaling issues where IO may be the bottleneck".
>
> So some of the scalability issues may just be "people haven't even
> tried to get that much parallelism". Others may be more fundamental.
>
> Linus
Here are some reviews from Phoronix:
<a href="http://">2990WX on Linux
<a href="http://">2990WX on Windows 10 Pro vs 4 Linux distros stock
Essentially it seems that Linux distros don't have a scaling problem to 32/64 cores/threads. Windows 10 Pro is the problem.
Pete
> wumpus (lost.delete@this.in.a.cave) on August 13, 2018 9:32 am wrote:
> > Per Hesselgren (perhesselgren.delete@this.yahoo.se) on August 13, 2018 8:50 am wrote:
> > > Threadripper on Anandtech
> >
> > Looks like coherency isn't an issue for 32 cores, but obviously not everything scales perfectly.
>
> Even loads that should scale really well (like the chromium build) seem to have hit the
> point of negative scaling here, with a 32c/64t 2990WX doing worse than a 16c/32t 2950X.
>
> I do suspect it's mostly the load that doesn't scale any more at that point, with the negative scaling coming
> from more memory contention and perhaps half of the cores being further away from the memory controller.
>
> Somebody who knows the Chromium build better might be able to tell how far they'd expect
> the build to scale, but even for simple things like a parallel build, you often hit
> serialization points (eg linking phases after you've built all your library files).
> And writing Makefiles that expose all the parallelism isn't trivial either.
>
> But some of that may be "nobody spent any effort to scale past X cores because they are so
> rare, and build farms have their own scaling issues where IO may be the bottleneck".
>
> So some of the scalability issues may just be "people haven't even
> tried to get that much parallelism". Others may be more fundamental.
>
> Linus
Here are some reviews from Phoronix:
<a href="http://">2990WX on Linux
<a href="http://">2990WX on Windows 10 Pro vs 4 Linux distros stock
Essentially it seems that Linux distros don't have a scaling problem to 32/64 cores/threads. Windows 10 Pro is the problem.
Pete
Topic | Posted By | Date |
---|---|---|
Threadripper 32 core review | Per Hesselgren | 2018/08/13 08:50 AM |
Threadripper 32 core review | wumpus | 2018/08/13 09:32 AM |
Threadripper 32 core review | Linus Torvalds | 2018/08/13 02:12 PM |
Threadripper 32 core review | pgerassi | 2018/08/13 03:16 PM |
Threadripper 32 core review | Linus Torvalds | 2018/08/13 03:48 PM |
Less Power Hungry Part? | Passing Through | 2018/08/13 04:19 PM |
Threadripper 32 core review | Groo | 2018/08/13 04:50 PM |
Threadripper 32 core review | Passing Through | 2018/08/14 12:45 PM |
Threadripper 32 core review | pgerassi | 2018/08/13 06:41 PM |
Threadripper 32 core review | juanrga | 2018/08/14 04:44 AM |
Threadripper 32 core review | Linus Torvalds | 2018/08/14 09:37 AM |
Threadripper 32 core review | juanrga | 2018/08/14 11:18 AM |
Threadripper 32 core review | juanrga | 2018/08/16 02:16 AM |
Threadripper 32 core review | Jukka Larja | 2018/08/14 08:41 AM |
[side topic] personal supercomputer (Threadripper 32 core review) | hobold | 2018/08/16 04:47 AM |
Threadripper 32 core review | Alberto | 2018/08/13 11:11 AM |
Threadripper 32 core review | Maynard Handley | 2018/08/13 01:27 PM |
Threadripper 32 core review | Doug S | 2018/08/13 02:01 PM |
Threadripper 32 core review | Alberto | 2018/08/14 02:50 AM |
Threadripper 32 core review | none | 2018/08/14 05:08 AM |
On revenues and right choices | AM | 2018/08/14 08:33 AM |
Threadripper 32 core review | Alberto | 2018/08/14 12:57 AM |
Threadripper 32 core review | Jeff S. | 2018/08/14 07:00 AM |
Shrinks and power density | AM | 2018/08/14 08:49 AM |
Shrinks and power density | David Hess | 2018/08/18 01:30 PM |
Shrinks and power density | AM | 2018/08/19 08:12 AM |
Shrinks and power density | anonymou5 | 2018/08/19 09:13 PM |
Shrinks and power density | John H | 2018/08/20 05:16 PM |
Threadripper 32 core review | David Hess | 2018/08/18 01:43 PM |
Threadripper 32 core review | anon | 2018/08/13 02:48 PM |
Threadripper 32 core review | anon | 2018/08/13 03:25 PM |
*while (NT) | anon | 2018/08/13 03:26 PM |
Threadripper 32 core review | pgerassi | 2018/08/13 07:34 PM |
Threadripper 32 core review | anon | 2018/08/14 05:40 AM |
Threadripper 32 core review | Gabriele Svelto | 2018/08/14 04:22 AM |
Threadripper 32 core review | anon | 2018/08/14 05:44 AM |
Flock of Chickens. | Jouni Osmala | 2018/08/13 08:04 PM |
Flock of Chickens. | none | 2018/08/14 01:43 AM |
AnandTech review clarity | Jeff S. | 2018/08/14 05:06 AM |
AnandTech review clarity | none | 2018/08/14 05:15 AM |
AnandTech review clarity | Jeff S. | 2018/08/14 05:36 AM |
AnandTech review clarity | none | 2018/08/14 05:44 AM |
AnandTech review clarity | none | 2018/08/14 05:53 AM |