![]() |
![]() |
#1 |
Aug 2002
Victoria, BC, Canada
32 Posts |
![]()
I've made a successful conversion over to v25 and have lost no work doing so - always make backups before you update anything!
Now, in cheking my v25 assignment details in my account I've noticed that where I have multi-core CPUs (2) or multi-CPU boxes (1), the Assignment Details report is only showing assignments for n-1 of my multi-core/cpu computers (where 'n' = number of core's/cpu's). i.e. I have a Core2-Quad and I'm only seeing assignments for core's 1, 2 & 3 (no core 0 [4?]) .. my Core2-Duo is the same, I only see work assignment for core 1 (nada for core 0 [2?]). Ditto for my dual-CPU single-core Xeon box - only Xeon CPU 1 is showing assigned work (nothing for CPU 0 [2?]). I have work assigned to ALL my core's and CPU's and v25 has been running nicely for at least 4-5 days (so work for all core's and CPUs has been reported to the Primenet Server) Anyone else experiencing this in their assignment reports? Last fiddled with by worknplay on 2008-11-04 at 09:14 |
![]() |
![]() |
![]() |
#2 |
Apr 2008
Oslo, Norway
7×31 Posts |
![]()
Are you sure it's not saying that it's starting helper threads on 3 CPUs and starting 4 threads? :)
In other words, it's starting 1 main worker thread and 3 helper threads for that worker. That is what is happening on my computer and I can see that it utilizing approx. 100% CPU. [Nov 4 08:02] Using all-complex FFT length 1536K, 4 threads ^-- Starting 4 threads. [Nov 4 08:02] Setting affinity to run helper thread 1 on any logical CPU. [Nov 4 08:02] Setting affinity to run helper thread 2 on any logical CPU. [Nov 4 08:02] Setting affinity to run helper thread 3 on any logical CPU. ^--- Starting 3 helper threads. |
![]() |
![]() |
![]() |
#3 |
Nov 2008
5 Posts |
![]()
I see the same behavior you described.
|
![]() |
![]() |
![]() |
#4 |
Aug 2002
Victoria, BC, Canada
32 Posts |
![]()
After doing some investigation, I've come to the conclusion that exponents originally assigned by the v4 server will show up as assignments to v4_computers despite the exponents having been moved to the v25 client and that work being registered with the v5 server.
For TF exponents, this will usually clear itself within a few days depending on how much work has been queued. For LL exponents assigned by the v4 server, I expect this anomaly will clear once the exponent has completed testing and a new LL exponent is assigned by the v5 server. Part of the problem also is that in converting my computers to the v25 client, I had done so without fully realizing what the v25 client was capable of (i.e. running separate threads from one client). Having two multi-core and one dual-CPU computers, I simply backed-up each v24 client folder and dropped the new v25 client files into each folder and started the clients up - on a Core2-Quad the result was each of the four v25 clients now running detected the presence of a quad-core CPU and thus requested work for four threads each .. you do the math. ![]() Quickly recognizing what I had done, I stopped all four clients, "decommissioned" three of four (on the Quad-core) of my v24 folders (with the v25 client) and transferred the work they had requested from the v5 server to the now primary v25 client's worktodo.txt file (had to do this twice over for my dual-CPU server and Core2-Duo workstation having upgraded them to the v25 client before recognizing my mistake) .. ![]() It will be a couple weeks before any of my assigned LL exponents are completed testing, so my conclusion won't be proven (or disproved) till then - unless of course someone else has a similar situation where their work is completed sooner. Cheers, ![]() |
![]() |
![]() |
![]() |
Thread Tools | |
![]() |
||||
Thread | Thread Starter | Forum | Replies | Last Post |
multi-core TF | dbaugh | Information & Answers | 3 | 2011-09-06 01:32 |
GMP-Fermat and multi-core processors | ixfd64 | Factoring | 1 | 2011-03-16 16:07 |
Automatic multi-core detect and use? | remaker | Software | 11 | 2007-04-02 20:18 |
CGPU multi core x86, Intel Larrabee | dsouza123 | Hardware | 2 | 2007-02-14 18:00 |
My prediction about multi-core chips | jasong | Hardware | 6 | 2006-02-14 16:07 |