| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: FightAIDS@Home Phase 2
Thread: Odd usage of "unlimited" parameter in Device Profiles |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 8
|
|
| Author |
|
|
captainjack
Advanced Cruncher Joined: Apr 14, 2008 Post Count: 144 Status: Offline Project Badges: 
|
On Nov. 28, 2018, knreed posted a message that said crunchers could select the maximum number of tasks to download for each research topic.
knreed post When I select the default profile with the maximum number of tasks for FAH2 = unlimited, BOINC will download enough tasks to keep my computer busy for the amount of time selected in the profile. When I select the home profile with the maximum number of tasks for FAH2 = unlimited, BOINC will only download the number of tasks that matches the number of threads available on the machine. IIRC, this same restriction applies to the school profile. Seems like the maximum number of tasks parameter is operating differently based on the profile selected. If you have more questions, please let me know. |
||
|
|
Steve W
Advanced Cruncher Joined: Dec 9, 2005 Post Count: 110 Status: Offline Project Badges: 
|
What are the work unit cache settings on all of those profiles? If BOINC is only requesting work for the available threads then it seems that the cache is set to 0.0 so it only requests tasks for available threads.
If the cache is > 0.0 then it requests extra work for the specified time. Worth double checking the settings are the same on all profiles. |
||
|
|
Crystal Pellet
Veteran Cruncher Joined: May 21, 2008 Post Count: 1317 Status: Offline Project Badges: 
|
Are you sure those FAH2's are not BETA's from batches FAH2_002222 up to 002229?
---------------------------------------- |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
FAH2 always downloads the number of work units that matches your number of threads at maximum. The reason is, that there chains of workunits for this project, where the next workunit depends on the results of the preceeding. WCG wants to make sure therefore, that WUs are crunched as fast as possible and do not remain in cache too long. Only by doing so can computation of a complete chain of WUs be done in a reasonable amount of time.
In the device profile you can only reduce the number of FAH2 WUs further, below the number of threads. So "unlimited" in this instance means, that the preset limit of the project applies, you only do not limit it further. |
||
|
|
captainjack
Advanced Cruncher Joined: Apr 14, 2008 Post Count: 144 Status: Offline Project Badges:
|
When I select the Home profile with 0.2 days of work with unlimited FAH2 tasks on an 8 thread machine, it gets 8 FAH2 tasks.
When I select the Default profile with 0.2 days of work with unlimited FAH@ tasks on an 8 thread machine, it gets 36 FAH2 tasks. None of them are FAH2 beta tasks. |
||
|
|
Crystal Pellet
Veteran Cruncher Joined: May 21, 2008 Post Count: 1317 Status: Offline Project Badges:
|
When I select the Home profile with 0.2 days of work with unlimited FAH2 tasks on an 8 thread machine, it gets 8 FAH2 tasks. What happens when you choose e.g. 16 tasks for your Home profile? |
||
|
|
captainjack
Advanced Cruncher Joined: Apr 14, 2008 Post Count: 144 Status: Offline Project Badges:
|
When I select the School profile with 0.2 days of work with FAH2 tasks set to 64 on a 16 thread machine, it gets 64 FAH@ tasks.
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Is there a limit on the total number of WUs that can be distributed under the "unlimited" category? Just brought up a server with 2 AMD EPYC 7551 32-core processors and I can only get 64 WUs. Machine has 128 threads available. This is regardless of NCPU setting (NCPU has been set to 128)
----------------------------------------Thu 11 Jul 2019 07:48:52 AM CDT | | Starting BOINC client version 7.14.2 for x86_64-pc-linux-gnu Thu 11 Jul 2019 07:48:52 AM CDT | | log flags: file_xfer, sched_ops, task Thu 11 Jul 2019 07:48:52 AM CDT | | Libraries: libcurl/7.64.0 OpenSSL/1.1.1b zlib/1.2.11 libidn2/2.0.5 libpsl/0.20.2 (+libidn2/2.0.5) libssh/0.8.6/openssl/zlib nghttp2/1.36.0 librtmp/2.3 Thu 11 Jul 2019 07:48:52 AM CDT | | Data directory: /var/lib/boinc-client Thu 11 Jul 2019 07:48:52 AM CDT | | No usable GPUs found Thu 11 Jul 2019 07:48:52 AM CDT | | [libc detection] gathered: 2.29, Ubuntu GLIBC 2.29-0ubuntu2 Thu 11 Jul 2019 07:48:52 AM CDT | | Host name: golf Thu 11 Jul 2019 07:48:52 AM CDT | | Processor: 128 AuthenticAMD AMD EPYC 7551 32-Core Processor [Family 23 Model 1 Stepping 2] Thu 11 Jul 2019 07:48:52 AM CDT | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nopl nonstop_tsc cpuid extd_apicid amd_dcm aperfmperf pni pclmulqdq monitor ssse3 fma cx16 sse4_1 sse4_2 movbe popcnt aes xsave avx f16c rdrand lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw skinit wdt tce topoext perfctr_core perfctr_nb bpext perfctr_llc mwaitx cpb hw_pstate sme ssbd sev ibpb vmmcall fsgsbase bmi1 avx2 smep bmi2 rdseed adx smap clflushopt sha_ni xsaveopt xsavec xgetbv1 xsaves clzero irperf xsaveerptr arat npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold avic v_vmsave_vmload vgif overflow_recov succor smca Thu 11 Jul 2019 07:48:52 AM CDT | | OS: Linux Ubuntu: Ubuntu 19.04 [5.0.0-20-generic|libc 2.29 (Ubuntu GLIBC 2.29-0ubuntu2)] Thu 11 Jul 2019 07:48:52 AM CDT | | Memory: 125.51 GB physical, 8.00 GB virtual Thu 11 Jul 2019 07:48:52 AM CDT | | Disk: 823.71 GB total, 770.47 GB free Additionally, have observed a strange anomaly concerning memory usage. Each FAH2 WU shows using 12G (12431MB) Virtual Memory and 1.2GB (1238MB) resident memory. Resident memory is twice what is showing for FAH2 on any of my Intel processors. Also an AMD APU also shows the same as Intel. Have used LINUX commands to verify what I'm seeing in BOINCTasks and they verify. All systems (Intel and AMD) are on the same OS release and BOINC Client version. This is only happening on the AMD EPYC processors. Haven't had a chance to check other sub projects yet to see if it happens to them also. Just an observation. Update: Added MCM1 to the mix and immediately downloaded another 64 WUs. All 128 threads are active. That means there must be a limit of 64 FAH2 and HST1 WUs as I was not able to get any for over 24 hours. MCM1 memory seems to be behaving as expected. Approx 76MB Virtual and 35MB Resident. There must be something about the BEDAM program that causes it to use more memory on the EPYC processors. [Edit 1 times, last edit by Doneske at Jul 11, 2019 4:42:08 PM] |
||
|
|
|