Since SETI's internet connection is (more or less) permanently maxed out, I'm trying to run SETI and Rosetta together on my Athlon X2 with equal ressource share. So I'd expect, that BOINC runs one task from each of this projects all the time, eventually for some short periods of time also 2 tasks from the same project, if the other has got more CPU time for whatever reason. That works quite OK so far.

What I don't understand, is why BOINC fills the cache with Rosetta tasks and has only 1-2 SETI tasks in cache. Right now for example the 2.5 days cache consists of about 100 hours of Rosetta tasks and 4 hours of SETI work.

It's like that now since 3-4 weeks, first I thought it might be because of the large negative "CPU work fetch priority" of Rosetta, so I restarted BOINC with zero_debts in hope that will solve the problem. Nope, the numbers changed, but the weird caching behaviour is the same.

If that helps, here is the current output from work_fetch_debug:

12/08/2012 11:35:34				[work_fetch]: work fetch start
12/08/2012 11:35:34				[work_fetch] ------- start work fetch state -------
12/08/2012 11:35:34				[work_fetch] target work buffer: 12096.00 + 241920.00 sec
12/08/2012 11:35:34				[work_fetch] CPU: shortfall 0.00 nidle 0.00 saturated 531652.80 busy 0.00 RS fetchable 100.00 runnable 200.00
12/08/2012 11:35:34	rosetta@home		[work_fetch] CPU: fetch share 1.00 LTD -5624.64 backoff dt 0.00 int 0.00
12/08/2012 11:35:34	Collatz Conjecture	[work_fetch] CPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (no new tasks) (blocked by prefs)
12/08/2012 11:35:34	Milkyway@Home		[work_fetch] CPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (comm deferred) (blocked by prefs)
12/08/2012 11:35:34	SETI@home		[work_fetch] CPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00
12/08/2012 11:35:34				[work_fetch] ATI GPU: shortfall 229187.41 nidle 0.00 saturated 24793.21 busy 0.00 RS fetchable 0.00 runnable 100.00
12/08/2012 11:35:34	rosetta@home		[work_fetch] ATI GPU: fetch share 0.00 LTD 0.00 backoff dt 79229.77 int 86400.00
12/08/2012 11:35:34	Collatz Conjecture	[work_fetch] ATI GPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (no new tasks)
12/08/2012 11:35:34	Milkyway@Home		[work_fetch] ATI GPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (comm deferred)
12/08/2012 11:35:34	SETI@home		[work_fetch] ATI GPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (blocked by prefs)
12/08/2012 11:35:34	rosetta@home		[work_fetch] overall LTD -180949.63
12/08/2012 11:35:34	Collatz Conjecture	[work_fetch] overall LTD 0.00
12/08/2012 11:35:34	Milkyway@Home		[work_fetch] overall LTD -283580.73
12/08/2012 11:35:34	SETI@home		[work_fetch] overall LTD -7179.42
12/08/2012 11:35:34				[work_fetch] ------- end work fetch state -------
12/08/2012 11:35:34				[work_fetch] No project chosen for work fetch

Any ideas? Any additional info needed for ideas?

---

When I was running S@H Beta on a 4 core machine I wanted to give it 25% of the CPU's, 1 CPU. However it would sometimes fill the cache 100% with beta work and other times 100% regular work. Not what I was wanting.

The only solution I could come up with that worked was to run each project in a separate instance of BOINC. The instance for S@H was configured for 3 CPU's and the other for S@H Beta was configured for 1 CPU.

The primary disadvantage of this setup is that if one project runs out of work you have idle resources.

I think they would need to add a configuration to specify the minimum resources a project can use while it actively has work to work like you, and many others, would want.
<minimum_procs>
<url>project_URL</url>
[<mp_cpus>N</mp_cpus>]
[<mp_gpus>N</mp_gpus>]
</minimum_procs>

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

2 CPU cores, 2 projects 50:50 - how difficult can that be?

+1

When I was running S@H Beta on a 4 core machine I wanted to give it 25% of the CPU's, 1 CPU. However it would sometimes fill the cache 100% with beta work and other times 100% regular work. Not what I was wanting.

+1

The only solution I could come up with that worked was to run each project in a separate instance of BOINC. The instance for S@H was configured for 3 CPU's and the other for S@H Beta was configured for 1 CPU.

I see... so basically BOINC is not doing what it was intended for: running more projects than just SETI.

I also found one other oddity which, no matter what I do, lets BOINC end up with a full cache from Rosetta: when BOINC has work from 2 projects it only wants to keep enough work for one core, when it has work only from one project, it wants to have enough for both cores. So once it finishes all SETI WUs, it uploads the result, requests new tasks... now if it gets "no tasks available", it asks from Rosetta for a full cache for both cores, while when it was running SETI i was going down to cache for one core, i.e. half of what it actually should have. Seen that 2-3 times already, here for example:

13/08/2012 09:26:29	SETI@home	Computation for task 27jn10aa.26515.15200.15.10.51_2 finished
13/08/2012 09:26:29	rosetta@home	Resuming task Tc735_ab_relax_cst_domain2_t000___dekim_cstwt_2.0_IGNORE_THE_REST_03_16_52590_530_0 using minirosetta version 331
13/08/2012 09:26:31	SETI@home	Started upload of 27jn10aa.26515.15200.15.10.51_2_0
13/08/2012 09:26:43	SETI@home	Finished upload of 27jn10aa.26515.15200.15.10.51_2_0
13/08/2012 09:26:44	SETI@home	Sending scheduler request: To fetch work.
13/08/2012 09:26:44	SETI@home	Reporting 1 completed tasks, requesting new tasks for CPU
13/08/2012 09:26:48	SETI@home	Scheduler request completed: got 0 new tasks
13/08/2012 09:26:48	SETI@home	Project has no tasks available
13/08/2012 09:26:53	rosetta@home	Sending scheduler request: To fetch work.
13/08/2012 09:26:53	rosetta@home	Requesting new tasks for CPU
13/08/2012 09:27:01	rosetta@home	Scheduler request completed: got 1 new tasks
13/08/2012 09:27:03	rosetta@home	Started download of Tc717_ab_relax_cst_t000___dekim_07_12.zip
13/08/2012 09:31:06	rosetta@home	Sending scheduler request: To fetch work.
13/08/2012 09:31:06	rosetta@home	Requesting new tasks for CPU
13/08/2012 09:31:25	rosetta@home	Scheduler request completed: got 1 new tasks

Note: I use 18 hours runtime preference on Rosetta. So while working on the last SETI WU, it was not asking for more work, when it could not get more from SETI, 18 hours was not enough, it asked for more. Don't understand that... I mean, it should know all the time it has two cores.

Anyway thanks for the answer, I thought I have configured something wrong.

---

I also saw similar behavior when I was running both SETI and Rosetta.

It would download days and days worth of Rosetta work, and sometimes one or two SETI work units (a couple of hours worth). Then, it would attempt to contact SETI (once), get the "no work available" message, so go and download several hundred hours of Rosetta work. Rinse and repeat.

My experience was in versions from several years ago (before they removed the messages tab).

---

It would download days and days worth of Rosetta work, and sometimes one or two SETI work units (a couple of hours worth). Then, it would attempt to contact SETI (once), get the "no work available" message, so go and download several hundred hours of Rosetta work. Rinse and repeat.

Exactly what's happening here. What I understand even less ist why BOINC keeps enough work for just one core (according to cache settings), when it has work from both projects but once it has no work from SETI at all, it fills up the cache for both cores from Rosetta. Also if I set Rosetta to NNT, it will suddenly discover it has 2 cores and will cache quite a lot work from SETI.

My experience was in versions from several years ago (before they removed the messages tab).

I'm using 6.12.34 on this particular machine, I hope that should be new enough for to feed a dual core CPU.

---

It would download days and days worth of Rosetta work, and sometimes one or two SETI work units (a couple of hours worth). Then, it would attempt to contact SETI (once), get the "no work available" message, so go and download several hundred hours of Rosetta work. Rinse and repeat.

Actually even without "no work available" from SETI it doesn't work as it should. I have ATM a cache for 1.94 days, so BOINC should keep enough work for about 4 days for one core.

Now it had about 11 hours of SETI work and 34 hours of Rosetta work. What it does? It requests work from SETI, gets 1 shortie worth about 1h15m and requests instantly after that work from Rosetta (and gets an 18-hours WU from there).

Well, I'll let it run like that for now, it respects at least the actuall resource share, so it makes up for the times without SETI WUs by running two of them at once every now and than. The thing I really don't like is that when it starts two Rosetta WUs (because it didn't care to cache enough from SETI), once it has SETI WUs and switches to them, it usually suspends the first Rosetta WU, i.e. the one which would finish sooner, once it suspended the one that needed about 10 minutes to finish and crunched on the one that needed 17 hours (not high priority or something).

---

Actually even without "no work available" from SETI it doesn't work as it should.

Perhaps that should read "doesn't work as I expect it to"?

Well, I'll let it run like that for now, it respects at least the actuall resource share, so it makes up for the times without SETI WUs by running two of them at once every now and than.

Mine didn't even do that. It would run about 99% Rosetta. When it got low on work, it would make one request to SETI, get the "no tasks" message, and immediately download enough Rosetta work that it went into "panic mode". Long term debt values were something like 30-40 million when I stopped waiting for it to "figure it out", and did some manual intervention.

---

Actually even without "no work available" from SETI it doesn't work as it should.

Perhaps that should read "doesn't work as I expect it to"?

I don't think there's any difference. 2 days cache on a dual core machine means 4 days of work in case of single thread applications.

Current status here: no SETI WUs, only 38 hours of Rosetta WUs left (about 136800 seconds, for a dual core machine that means actually only half of that) but according to work_fetch_debug BOINC thinks it has enough for 183000 seconds (50.8 hours for one core) and does not ask for work.

So I decided to do a little experiment, I suspended work fetch for Rosetta...

20/08/2012 23:02:57				[work_fetch]: work fetch start
20/08/2012 23:02:57				[work_fetch] ------- start work fetch state -------
20/08/2012 23:02:57				[work_fetch] target work buffer: 3456.00 + 164160.00 sec
20/08/2012 23:02:57				[work_fetch] CPU: shortfall 0.00 nidle 0.00 saturated 182179.21 busy 0.00 RS fetchable 200.00 runnable 200.00
20/08/2012 23:02:57	rosetta@home		[work_fetch] CPU: fetch share 0.50 LTD -4388.37 backoff dt 0.00 int 0.00
20/08/2012 23:02:57	Collatz Conjecture	[work_fetch] CPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (no new tasks) (blocked by prefs)
20/08/2012 23:02:57	Milkyway@Home		[work_fetch] CPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (blocked by prefs)
20/08/2012 23:02:57	SETI@home		[work_fetch] CPU: fetch share 0.50 LTD 0.00 backoff dt 0.00 int 0.00
20/08/2012 23:02:57				[work_fetch] ATI GPU: shortfall 158328.72 nidle 0.00 saturated 8710.90 busy 0.00 RS fetchable 0.00 runnable 100.00
20/08/2012 23:02:57	rosetta@home		[work_fetch] ATI GPU: fetch share 0.00 LTD 0.00 backoff dt 88576.83 int 86400.00
20/08/2012 23:02:57	Collatz Conjecture	[work_fetch] ATI GPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (no new tasks)
20/08/2012 23:02:57	Milkyway@Home		[work_fetch] ATI GPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00
20/08/2012 23:02:57	SETI@home		[work_fetch] ATI GPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (blocked by prefs)
20/08/2012 23:02:57	rosetta@home		[work_fetch] overall LTD -71398.56
20/08/2012 23:02:57	Collatz Conjecture	[work_fetch] overall LTD 0.00
20/08/2012 23:02:57	Milkyway@Home		[work_fetch] overall LTD -106932.51
20/08/2012 23:02:57	SETI@home		[work_fetch] overall LTD 0.00
20/08/2012 23:02:57				[work_fetch] ------- end work fetch state -------
20/08/2012 23:02:57				[work_fetch] No project chosen for work fetch
20/08/2012 23:03:04	rosetta@home		work fetch suspended by user
20/08/2012 23:03:57				[work_fetch]: work fetch start
20/08/2012 23:03:57	SETI@home		[work_fetch] chosen: starved CPU: 1.00 inst, 1.00 sec
20/08/2012 23:03:57				[work_fetch] ------- start work fetch state -------
20/08/2012 23:03:57				[work_fetch] target work buffer: 3456.00 + 164160.00 sec
20/08/2012 23:03:57				[work_fetch] CPU: shortfall 0.00 nidle 0.00 saturated 182028.77 busy 0.00 RS fetchable 100.00 runnable 200.00
20/08/2012 23:03:57	rosetta@home		[work_fetch] CPU: fetch share 0.00 LTD -4448.47 backoff dt 0.00 int 0.00 (no new tasks)
20/08/2012 23:03:57	Collatz Conjecture	[work_fetch] CPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (no new tasks) (blocked by prefs)
20/08/2012 23:03:57	Milkyway@Home		[work_fetch] CPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (blocked by prefs)
20/08/2012 23:03:57	SETI@home		[work_fetch] CPU: fetch share 1.00 LTD 0.00 backoff dt 0.00 int 0.00
20/08/2012 23:03:57				[work_fetch] ATI GPU: shortfall 158419.84 nidle 0.00 saturated 8619.95 busy 0.00 RS fetchable 0.00 runnable 100.00
20/08/2012 23:03:57	rosetta@home		[work_fetch] ATI GPU: fetch share 0.00 LTD 0.00 backoff dt 88516.73 int 86400.00 (no new tasks)
20/08/2012 23:03:57	Collatz Conjecture	[work_fetch] ATI GPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (no new tasks)
20/08/2012 23:03:57	Milkyway@Home		[work_fetch] ATI GPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00
20/08/2012 23:03:57	SETI@home		[work_fetch] ATI GPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (blocked by prefs)
20/08/2012 23:03:57	rosetta@home		[work_fetch] overall LTD -71404.13
20/08/2012 23:03:57	Collatz Conjecture	[work_fetch] overall LTD 0.00
20/08/2012 23:03:57	Milkyway@Home		[work_fetch] overall LTD -105887.23
20/08/2012 23:03:57	SETI@home		[work_fetch] overall LTD 0.00
20/08/2012 23:03:57				[work_fetch] ------- end work fetch state -------
20/08/2012 23:03:57	SETI@home		[work_fetch] request: CPU (1.00 sec, 1.00 inst) ATI GPU (0.00 sec, 0.00 inst)
20/08/2012 23:03:57	SETI@home		Sending scheduler request: To fetch work.
20/08/2012 23:03:57	SETI@home		Reporting 4 completed tasks, requesting new tasks for CPU
20/08/2012 23:04:05	SETI@home		Scheduler request completed: got 1 new tasks
20/08/2012 23:04:05				[work_fetch] Request work fetch: RPC complete
20/08/2012 23:04:10				[work_fetch]: work fetch start
20/08/2012 23:04:10				[work_fetch] ------- start work fetch state -------
20/08/2012 23:04:10				[work_fetch] target work buffer: 3456.00 + 164160.00 sec
20/08/2012 23:04:10				[work_fetch] CPU: shortfall 135349.63 nidle 0.00 saturated 90970.56 busy 0.00 RS fetchable 0.00 runnable 100.00
20/08/2012 23:04:10	rosetta@home		[work_fetch] CPU: fetch share 0.00 LTD -4448.47 backoff dt 0.00 int 0.00 (no new tasks)
20/08/2012 23:04:10	Collatz Conjecture	[work_fetch] CPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (no new tasks) (blocked by prefs)
20/08/2012 23:04:10	Milkyway@Home		[work_fetch] CPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (blocked by prefs)
20/08/2012 23:04:10	SETI@home		[work_fetch] CPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (comm deferred)
20/08/2012 23:04:10				[work_fetch] ATI GPU: shortfall 158439.68 nidle 0.00 saturated 8601.20 busy 0.00 RS fetchable 0.00 runnable 100.00
20/08/2012 23:04:10	rosetta@home		[work_fetch] ATI GPU: fetch share 0.00 LTD 0.00 backoff dt 88503.47 int 86400.00 (no new tasks)
20/08/2012 23:04:10	Collatz Conjecture	[work_fetch] ATI GPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (no new tasks)
20/08/2012 23:04:10	Milkyway@Home		[work_fetch] ATI GPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00
20/08/2012 23:04:10	SETI@home		[work_fetch] ATI GPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (comm deferred) (blocked by prefs)
20/08/2012 23:04:10	rosetta@home		[work_fetch] overall LTD -71384.77
20/08/2012 23:04:10	Collatz Conjecture	[work_fetch] overall LTD 0.00
20/08/2012 23:04:10	Milkyway@Home		[work_fetch] overall LTD -105659.35
20/08/2012 23:04:10	SETI@home		[work_fetch] overall LTD -7661.41
20/08/2012 23:04:10				[work_fetch] ------- end work fetch state -------
20/08/2012 23:04:10				[work_fetch] No project chosen for work fetch
20/08/2012 23:04:22	rosetta@home		work fetch resumed by user
20/08/2012 23:04:22				[work_fetch] Request work fetch: project allowed to fetch work by user
20/08/2012 23:04:25				[work_fetch]: work fetch start
20/08/2012 23:04:25	rosetta@home		[work_fetch] chosen: minor shortfall CPU: 0.00 inst, 135381.78 sec
20/08/2012 23:04:25				[work_fetch] ------- start work fetch state -------
20/08/2012 23:04:25				[work_fetch] target work buffer: 3456.00 + 164160.00 sec
20/08/2012 23:04:25				[work_fetch] CPU: shortfall 135381.78 nidle 0.00 saturated 90970.87 busy 0.00 RS fetchable 100.00 runnable 100.00
20/08/2012 23:04:25	rosetta@home		[work_fetch] CPU: fetch share 1.00 LTD -4448.47 backoff dt 0.00 int 0.00
20/08/2012 23:04:25	Collatz Conjecture	[work_fetch] CPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (no new tasks) (blocked by prefs)
20/08/2012 23:04:25	Milkyway@Home		[work_fetch] CPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (blocked by prefs)
20/08/2012 23:04:25	SETI@home		[work_fetch] CPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (comm deferred)
20/08/2012 23:04:25				[work_fetch] ATI GPU: shortfall 158462.44 nidle 0.00 saturated 8578.46 busy 0.00 RS fetchable 0.00 runnable 100.00
20/08/2012 23:04:25	rosetta@home		[work_fetch] ATI GPU: fetch share 0.00 LTD 0.00 backoff dt 88488.37 int 86400.00
20/08/2012 23:04:25	Collatz Conjecture	[work_fetch] ATI GPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (no new tasks)
20/08/2012 23:04:25	Milkyway@Home		[work_fetch] ATI GPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00
20/08/2012 23:04:25	SETI@home		[work_fetch] ATI GPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (comm deferred) (blocked by prefs)
20/08/2012 23:04:25	rosetta@home		[work_fetch] overall LTD -71374.77
20/08/2012 23:04:25	Collatz Conjecture	[work_fetch] overall LTD 0.00
20/08/2012 23:04:25	Milkyway@Home		[work_fetch] overall LTD -105398.48
20/08/2012 23:04:25	SETI@home		[work_fetch] overall LTD -7661.41
20/08/2012 23:04:25				[work_fetch] ------- end work fetch state -------
20/08/2012 23:04:25	rosetta@home		[work_fetch] request: CPU (135381.78 sec, 0.00 inst) ATI GPU (0.00 sec, 0.00 inst)
20/08/2012 23:04:25	rosetta@home		Sending scheduler request: To fetch work.
20/08/2012 23:04:25	rosetta@home		Requesting new tasks for CPU
20/08/2012 23:04:38	rosetta@home		Scheduler request completed: got 2 new tasks
20/08/2012 23:04:38				[work_fetch] Request work fetch: RPC complete
20/08/2012 23:04:43				[work_fetch]: work fetch start
20/08/2012 23:04:43				[work_fetch] ------- start work fetch state -------
20/08/2012 23:04:43				[work_fetch] target work buffer: 3456.00 + 164160.00 sec
20/08/2012 23:04:43				[work_fetch] CPU: shortfall 0.00 nidle 0.00 saturated 187173.00 busy 0.00 RS fetchable 0.00 runnable 100.00
20/08/2012 23:04:43	rosetta@home		[work_fetch] CPU: fetch share 0.00 LTD -4191.58 backoff dt 0.00 int 0.00 (comm deferred)
20/08/2012 23:04:43	Collatz Conjecture	[work_fetch] CPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (no new tasks) (blocked by prefs)
20/08/2012 23:04:43	Milkyway@Home		[work_fetch] CPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (blocked by prefs)
20/08/2012 23:04:43	SETI@home		[work_fetch] CPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (comm deferred)
20/08/2012 23:04:43				[work_fetch] ATI GPU: shortfall 158490.93 nidle 0.00 saturated 8549.98 busy 0.00 RS fetchable 0.00 runnable 100.00
20/08/2012 23:04:43	rosetta@home		[work_fetch] ATI GPU: fetch share 0.00 LTD 0.00 backoff dt 88470.17 int 86400.00 (comm deferred)
20/08/2012 23:04:43	Collatz Conjecture	[work_fetch] ATI GPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (no new tasks)
20/08/2012 23:04:43	Milkyway@Home		[work_fetch] ATI GPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00
20/08/2012 23:04:43	SETI@home		[work_fetch] ATI GPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (comm deferred) (blocked by prefs)
20/08/2012 23:04:43	rosetta@home		[work_fetch] overall LTD -135551.02
20/08/2012 23:04:43	Collatz Conjecture	[work_fetch] overall LTD 0.00
20/08/2012 23:04:43	Milkyway@Home		[work_fetch] overall LTD -105071.68
20/08/2012 23:04:43	SETI@home		[work_fetch] overall LTD -7661.41
20/08/2012 23:04:43				[work_fetch] ------- end work fetch state -------
20/08/2012 23:04:43				[work_fetch] No project chosen for work fetch

There are two things I completely don't understand in this part of the log:

1. Why does BOINC request work from SETI directly after I supended work fetch from Rosetta, even if it still thinks it has more than enough?

2. Why does BOINC forget, that the machine has 2 CPU cores and apparently remembers that again after the first request? Note the decrease in predicted busy time for CPU (which is still wrong according to the expected runtimes of the WUs, but anyway, suddenly it's only half of the previous prediction). Also resuming work fetch for Rosetta does not make BOINC forget it again, that will probably happen again slowly like it did before, when the cache get smaller and smaller.


The whole thing seems broken to me... unless we consider the possibility, that BOINC learned in the past few weeks (or so) that it's quite impossible to download anything from SETI, so why ask for it.

---