Recently there have been several reports at Milkyway@Home that Milkyway tasks always run at high priority, leaving no time for other tasks. I experienced this myself, and after some investigation have come to the conclusion that this is a malfunction of the BOINC 7 client. If I read the code correctly, the scheduler always falls through into high priority mode if a project sets the dont_use_dcf flag, so other projects beside Milkyway should experience the same problem. I can't verify this though, as I don't know of more projects that use dont_use_dcf. For Milkyway, this patch solved the issue for me:

diff -Naur boinc_core_release_7_0_25/client/cpu_sched.cpp boinc_core_release_7_0_25_patched/client/cpu_sched.cpp
--- boinc_core_release_7_0_25/client/cpu_sched.cpp	2012-04-14 10:02:12.000000000 +0200
+++ boinc_core_release_7_0_25_patched/client/cpu_sched.cpp	2012-04-22 02:14:30.000000000 +0200
@@ -470,7 +470,7 @@
 
         // treat projects with DCF>90 as if they had deadline misses
         //
-        if (!p->dont_use_dcf && p->duration_correction_factor < 90.0) {
+        if (p->dont_use_dcf || p->duration_correction_factor < 90.0) {
             if (p->rsc_pwf[rsc_type].deadlines_missed_copy <= 0) {
                 continue;
             }

It is against the 7.0.25 release code, but should also apply to the current development code.

An unrelated side note: When i accidentally "tested" BOINC 7.0.26, I noticed the RAM size of my GPUs was bogus wherever I saw it displayed (log file, scheduler requests). It were very large numbers, possibly a 64/32 bit issue or a type conversion error. I haven't looked any more into this though, since I am now happily back to my (patched) 7.0.25. Some brief but possibly relevant infos about my system: Linux 64 bit, NVIDIA GPUs.

I hope the developers read this, else could someone please inform them. I would prefer not to join the development mailing list for a single message. Thanks.

@the HP problem: The developers already know and I thought that the fix was in 7.0.26s code.

---

Actually this was not fixed, but the suggest patch fixes it - thanks!
-- David

---