It looks like more than a few people are still having a problem where clients are aborting workunits early because of max time limit elapsed.

If you could post here with any information about your client (and if you're using an anonymous platform or anything like that), I'd appreciate it -- so it's all in one place for us to look over.

Is anyone having this happen for the CPU applications, or is it just limited to the GPU applications?

---

The two most recent configurations (from about an hour ago) for me:

ATI HD4850
BOINC 6.10.58

One -- Windows 7 64 bit
The other - Windows XP SP3 32 bit

What I saw on both was computation errors.

---

HD4770 Driver version 11.2 Windows 7 x64 BOINC 6.10.58
&
HD4770 Driver version 11.4 Ubuntu 11.04 x64 BOINC 6.10.58

2011-05-08 09:04:54 Milkyway@home Aborting task de_separation_13_3s_fix20_1_457402_1304833085_1: exceeded elapsed time limit 143.041835

WinXP 32Bit SP3
ATI HD3850
BOINC 6.10.58

Seems only an GPU App Problem.

---

DSKAG Austria Research Team: http://www.research.dskag.at

I've tried increasing the RSC_FPOPS_BOUND of newly generated workunits. I'm hoping this might increase the bounds the clients are calculating high enough so they won't automatically be canceling workunits. Let me know if it helps.

---

ok i will try now. Wait some minutes for my answer ^^

Edit:
After 9:11 Minutes ~ 75%
I think it runs longer then last times (~55%?), but still not long enough.



<core_client_version>6.10.58</core_client_version>
<![CDATA[
<message>
Maximum elapsed time exceeded
</message>
<stderr_txt>
<search_application> milkywayathome_client separation 0.62 Windows x86 double CAL++ </search_application>
Found 1 CAL devices
Chose device 0

Device target: CAL_TARGET_670
Revision: 41
CAL Version: 1.4.1016
Engine clock: 720 Mhz
Memory clock: 950 Mhz
GPU RAM: 512
Wavefront size: 64
Double precision: CAL_TRUE
Compute shader: CAL_FALSE
Number SIMD: 4
Number shader engines: 1
Pitch alignment: 256
Surface alignment: 256
Max size 2D: { 8192, 8192 }

Estimated iteration time 740.316840 ms
Target frequency 30.000000 Hz, polling mode 1, using responsiveness factor of 1.000000
Dividing into 25 chunks
Integration range: { nu_steps = 640, mu_steps = 1600, r_steps = 1400 }
Using { 1, 25 } chunk(s) of size { 1400, 64 }


Unhandled Exception Detected...

- Unhandled Exception Record -
Reason: Breakpoint Encountered (0x80000003) at address 0x7C91120E

Engaging BOINC Windows Runtime Debugger...

---

DSKAG Austria Research Team: http://www.research.dskag.at

I am wondering what kind of points are allotted for the N-body projects. I've noticed that they start grabbing processor threads before they even start running. This shuts down processing on my other projects and reduces productivity on my machines. This type of project grabs all the processing threads of my 12 core system. If the points are high enough or the project important enough then OK. Please advise.
Mark W. Patton

I am running only GPU

windows7, ultimate x64
12 GB ram, i7 960
2 ATI 5870, driver 11.3 BOINC 6.10.58

The first time I had errors was May 7 in the afternoon when I started the rig after a few days off.

---

Greetings from,
TJ

I am wondering what kind of points are allotted for the N-body projects. I've noticed that they start grabbing processor threads before they even start running. This shuts down processing on my other projects and reduces productivity on my machines. This type of project grabs all the processing threads of my 12 core system. If the points are high enough or the project important enough then OK. Please advise.
Mark W. Patton

The n-body workunits are multi-threaded -- so they'll use all cores that are available AFAIK.

---

The first 4 have made it through my 5870 CF rig. Your fix seems to worked. One question. Why are they running High Priority, they are only 2 minute tasks.

Dan

Running CPU only, have had several workunits fail.

My client is shown in my stats, and the errored workunits listed, as of this post.

---

It's NOT just a name, it's an attitude.

Originally, I was experiencing the "timed out" error and the WU's only had an estimated time of 0:00:06 to 0:00:10 each. There was a fix applied and the times then went up to around normal - around 3:30 to 4:00 each.

I had approximately 10 old WU's that were downloaded around 2:15 a.m. EDT (5/8). I opened one up for work and it timed out as before, and I received new 2 WU's around 1:30 p.m. EDT (5/8). The first one downloaded and was completed, as well as the second one.

When the second one was completed, I was given ONE new WU instead of having the usual limit of 12. I'm not complaining about not getting any work, just advising of what's happened.

Here's my rig's information:

BOINC Version 6.10.58 - all programs are NOT OPTIMIZED
Platform: Microsoft Windows7 64-bit Home Premium
8Gb RAM
Intel Core2 Quad Q8400
ATI HD4850

I am also attached to PrimeGrid and Collatz, both using GPU.

Paul

---

BOINC 6.10.58
Radeon 4850, Catalyst 11.2
Windows 7 x64

All WUs fail with computational error about 65% through

http://pastebin.com/raw.php?i=3idYGNvE

Hello Travis,

This is about GPU

I see a few strange things:
1. The WU's run complete again without error but are 2 to 4 times longer. 2 to 4 minutes instead of 1.12 it was last week.

2. I got 24 new but they were not going to MW. I did manually. After all complete I got one new and one was runnig, so one ATI card has nothing to do. After half an hour I got 24 new ones, but they are not moved and replaced by new ones as last week.

3. The runtime is very variable but the cpu time is almost the same variing around 6 secs. Credits seem te be granted by cpu use and not the total running time.

4. High priority is gone now.

5. BOINC has an estimated run time of 2 hours, 21 minutes, 40 seconds for the WU's who take between 2 and 4 minutes now.

Windwos7, ultimate x64 12Gb RAM with 2ATI 5870 driver 11.3 BOINC 6.10.58 stock app, no modifications or tweaks and no anonymous client.

---

Greetings from,
TJ

Have no problems at all with my HD6970 (cath 11.4 preview)
I only cruch Milkyway on GPU ( 2 iterations using xml file)
There is some lag, but I don't care.

Win 7 64
AMD Phenom(tm) II X6 1090T (all 6 cores crunch other projects) 5% OC

My GPU WUs are completing and validating now.

The n-body WUs are failing due to maximum time limit elapsed. Everything else is fine. I have Win7 64-bit, an Intel Core i7 980X, and a Sapphire Radeon HD 6970. I am using BOINC 6.10.60. My Catalyst drivers are the 11.4 production drivers. I am using stock everything, no app_info.xml, and no overclocking whatsoever.

BOINC 6.10.58
Radeon HD3850, Catalyst 10.8
Windows XP SP3
Athlon XP 2600+

All WUs fail after 2 seconds.

I think the problem for my computer is the problem related here:
http://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=2345

My CPU don't support SSE2.
Can you do something pleas?

nbody WUs continue to error out in a few seconds on my CPU.

---

BOINC 6.10.58
Radeon HD3850, Catalyst 10.8
Windows XP SP3
Athlon XP 2600+

All WUs fail after 2 seconds.

I think the problem for my computer is the problem related here:
http://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=2345

My CPU don't support SSE2.
Can you do something pleas?

update to 11.3