Was it this result?

http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=39615072

The output seems all garbled... probably why the validator didn't like it, because it couldn't parse out the likelihood value.

---

All my units are rated as invalid ....

---

Team Alliance francophone, boinc: 7.0.18

GA-P55-UD5, i7 860, Win 7 64 bits, 8g DDR3, GTX 470

Was it this result?

http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=39615072

The output seems all garbled... probably why the validator didn't like it, because it couldn't parse out the likelihood value.

Just an update, asking the BOINC mailing lists about this. It looks it's happening with clients version 6.12.27 only... is that an older client or a testing version by any chance?

---

All my units are rated as invalid ....

Looks like it was a problem with that version of the BOINC client. I added a quick fix server side which should make those results parse-able, so I think they'll validate now.

---

fortunately, despite the fact that all of the ps_test_1_xxx tasks i received this morning errored out, none of them showed up on the server as errors (which makes sense, since these tasks are for testing purposes only). i have no idea if my host has since crunched any ps_test_2_xxx, ps_test_3_xxx, or ps_test_4_xxx tasks yet (or even received any) b/c i had to leave for work shortly after noticing the ps_test_1_xxx errors this morning.

i did notice that the de_separation tasks are still running normally (as has already been mentioned in this thread), but i did have to manually contact the server to report the ps_test_1_xxx errors and fetch new work...otherwise i would have had to wait an hour or so for BOINC to count down and contact the MW@H server itself. this is the one thing i find a bit irritating about the random and intermittent distribution of these test WU's - if i get any errors with them, then the BOINC client doesn't seem to want to contact the server for new work every 60 seconds like it does when i have a steady flow of de_separation tasks coming through. instead, it tends to wait long periods of time between server contacts. and if i'm not there to manually update the project, then my host has to sit there and wait an hour or so (or perhaps even longer) for the BOINC client to fetch more MW@H tasks...during that dead time, my host could be crunching a significant amount of de_separation tasks.

i was going to say that "right now" would be a perfect example, b/c for the last hour or so, i've been monitoring my account on the server from work, and up until a few minutes ago, my last task received was right around 15:00 UTC - a bit over an hour ago. fortunately, it appears the count down between server contacts has run out, b/c according to the server, i received new work just minutes ago. i just wish i weren't looking at hour long down times for my host every time i return an errored test task...an hour lost is a significant amount of work not done...

btw, i'm now getting some ps_test_4_xxx tasks. i'll report back if i have any problems with them...

---

All my units are rated as invalid ....

Looks like it was a problem with that version of the BOINC client. I added a quick fix server side which should make those results parse-able, so I think they'll validate now.

Thank you.

yes I'm using version 6.12.27, for cons I've only had two units validated on ten I made ​​....

But as the stats do not stay visible lontemps, I can check the validation ....

---

Team Alliance francophone, boinc: 7.0.18

GA-P55-UD5, i7 860, Win 7 64 bits, 8g DDR3, GTX 470

I was handed out another test WU. Just crunched ps_test_4_2448_2 to completion with my GPU

ok, got two ps_test_4_xxx tasks, both of which completed successfully.

1) 171s run time; 6s CPU time; 213.76 points
2) 172s run time; 7s CPU time; 213.76 points

keep in mind these were run on an HD 5870 2GB GPU running 2 tasks at a time...so these would have taken ~86s to complete running one at a time.

platform: WinXP Pro 32-bit, BOINC v6.12.26.

---

2 more successful ps_test_4_xxx tasks:

1) 158s run time; 7s CPU time; 213.76 points
2) 155s run time; 7s CPU time; 213.76 points

luckily these ps_test_4_xxx tasks aren't causing errors on my host, or else i might be looking at more hour-long down times between server contacts.

---

I've had ps_test_4 passing correctly on two rigs now. Looks like all WUs made it.

I had some test-wu's, all failing. Sorry, it happened too fast; I could not check what version they were.
But work fetch behaviour is strange since then, need to update manually from time to time ..

I had some test-wu's, all failing. Sorry, it happened too fast; I could not check what version they were.
But work fetch behaviour is strange since then, need to update manually from time to time ..

same here...i tried to explain as best i could my work fetch problems several posts up, but the problem seems to have gone away ever since the ps_test_1_xxx error-prone tasks stopped coming. i've since verified the handful of ps_test_4_xxx tasks that have come my way as successful and valid, and have not since encountered the work fetch problems i was having earlier in the day (i.e. i've had smooth sailing since then - server contact/work fetch every 60 seconds since then).

---

ok, i got one that completed successfully, but won't validate:

Task 39801693

Name ps_test_4_4181_4
Workunit 27651481
Created 25 May 2011 | 20:01:52 UTC
Sent 25 May 2011 | 20:06:22 UTC
Received 25 May 2011 | 20:24:04 UTC
Server state Over
Outcome Success
Client state Done
Exit status 0 (0x0)
Computer ID 267861
Report deadline 2 Jun 2011 | 20:06:22 UTC
Run time 152.11
CPU time 6.83
Validate state Workunit error - check skipped
Credit 0.00
Application version MilkyWay@Home
Anonymous platform (ATI GPU)

Stderr output

<core_client_version>6.12.26</core_client_version>
<![CDATA[
<stderr_txt>
<search_application> milkywayathome_client separation 0.62 Windows x86 double CAL++ </search_application>
Found 2 CAL devices
Chose device 1

Device target: CAL_TARGET_CYPRESS
Revision: 2
CAL Version: 1.4.1332
Engine clock: 700 Mhz
Memory clock: 667 Mhz
GPU RAM: 2048
Wavefront size: 64
Double precision: CAL_TRUE
Compute shader: CAL_TRUE
Number SIMD: 20
Number shader engines: 2
Pitch alignment: 256
Surface alignment: 256
Max size 2D: { 16384, 16384 }

Estimated iteration time 152.293750 ms
Target frequency 30.000000 Hz, polling mode 1, using responsiveness factor of 1.000000
Dividing into 4 chunks
Integration range: { nu_steps = 640, mu_steps = 1600, r_steps = 1400 }
Using { 1, 4 } chunk(s) of size { 1400, 400 }
Integration time = 143.155698 s, average per iteration = 223.680778 ms
Integral 0 time = 145.571866 s
Likelihood time = 3.955354 s
<background_integral> 0.000333827035719 </background_integral>
<stream_integral> 273.754854896097240 102.609981040972190 2365.259365747790100 </stream_integral>
<background_likelihood> -3.638444956601465 </background_likelihood>
<stream_only_likelihood> -18.242006900048466 -4.996183039039412 -4.257230205921160 </stream_only_likelihood>
<search_likelihood> -3.149085805578304 </search_likelihood>
16:23:32 (2948): called boinc_finish

</stderr_txt>
]]>

note that this happened with BOINC v6.12.26, and not with either of the recent development versions v6.12.27 or v6.12.28.

---

^ ok, nevermind my ignorance lol. i just clicked on the "work unit details" link and discovered that my task can't be validated because other hosts have already returned "too many errors" & "too many results."

---

Hello.
Today i get my first test wu's.
All work fine on my 2 5850 cards.
Greetings from Germany.

I have some finished and validated test-wu's
http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=39673303

Work fetch seems to run normal now.

I just had a couple of test 4 WUs finish and validate on HD5870s.

-Dave

[Edit] Just saw a couple more validate...

---

All my WUs (ps_test4) finished correctly.

---

ps_test_4's running smoothly on my 4850. validating well too.

All my ps_test_4's are working fine on my ATI 5870, no problems to report here...