I have uploaded a new acemdbeta application for Linux and some workunits to test.

Mainly bug fixes.

gdf

No. This is a cuda3.1 app.

yet I don't understand what that means. In the middle of what?

gdf

The first results are fine. We are going to produce today a beta for Windows.
This applications will substitute all production apps already this week, if all goes well
gdf

Windows application is out.

gdf

acemdbeta_6.38_windows_intelx86__cuda31 - Application Error

The exception unknown software exception (0xc0000005) occurred in the application at location 0x0040258c.

Click OK to terminate the program
Click on CANCEL to debug the program

System: 2003 Server x64 i7-2600K, 8GB DDR3, 2TB, GTX470 (native clocks, increased fan speeds)

Task ran for 1h45min but stayed at 0% complete, apparently going through a loop.

My only other two Betas also failed after 7 and 15sec.

Worth noting that if there is a pop-up error message and you don't select to end the task it will continue running indefinately.
So if anyone has such a message do something about it or you will just keep running the same erroneuos Beta task.

No, I am trying with a quick change in few minutes to see if it works. Otherwise, it will take more time to debug it.

gdf

acemdbeta_6.39 substitutes acemdbeta_6.38 hopefully with better results...

gdf

This 6.39 task reached 10% complete in 17min 43sec on a stock GTX470 (System), so estimated run time is 3h.
98% GPU Utilization, 315MB video memory usage.

Looks good so far... 20%

4069069 2520078 9 Jun 2011 17:12:15 UTC 9 Jun 2011 20:10:53 UTC Completed and validated 10,510.59 10,444.27 7,491.18 11,236.77 ACEMD beta version v6.39 (cuda31)

Well, this task ran and finished (2h 55min) without error on the 6.39 app. Hopefully other results are just as positive.

Any problem?

gdf

Uhmm, don't really know why. It seems a driver thing, as it crashes quickly.

So far, all the 6.38 and 6.39 Betas crashed with computational errors on my computers, and yes, the computers were sluggish when running these units.

nenym's No heartbeat from core client for 30 sec - exiting

This can be caused by the Boinc client, the app, another CPU project, or something else running on the system.
What else are you crunching/running? This is common when the CPU is over-taxed, often by other CPU projects.
Are you using recommended settings (freeing up at least 1 CPU core/thread per GPU and using SWAN_SYNC=0)?

It may be worth noting that GDF's latest tasks use 98% of the GPU, so previous overclocks (Bedrich 755MHz GTX480) that never encountered such high utilization may no longer be reliable.
It's best to Beta test with default/factory settings and recommended GPUGrid settings. Also worth keeping an eye on temperature - I saw a rise of a couple of degrees, but in some cases that might be enough to push the card over the edge.

Any problem with the higher priority?

gdf

Not personally, but I suppose if the GPUGrid priority was too high it might under some circumstances promote a Boinc client to crash. That said the system would still need to be heavily taxed by other CPU processes.

Just after doing a system restart and got this 6.38 error message:

Event Type: Information
Event Source: Application Error
Event Category: (100)
Event ID: 1004
Date: 10/06/2011
Time: 13:15:29
User: N/A
Computer: S
Description:
Reporting queued error: faulting application acemdbeta_6.38_windows_intelx86__cuda31, version 0.0.0.0, faulting module acemdbeta_6.38_windows_intelx86__cuda31, version 0.0.0.0, fault address 0x00002c58.

For more information, see Help and Support Center at http://go.microsoft.com/fwlink/events.asp.
Data:
0000: 41 70 70 6c 69 63 61 74 Applicat
0008: 69 6f 6e 20 46 61 69 6c ion Fail
0010: 75 72 65 20 20 61 63 65 ure ace
0018: 6d 64 62 65 74 61 5f 36 mdbeta_6
0020: 2e 33 38 5f 77 69 6e 64 .38_wind
0028: 6f 77 73 5f 69 6e 74 65 ows_inte
0030: 6c 78 38 36 5f 5f 63 75 lx86__cu
0038: 64 61 33 31 20 30 2e 30 da31 0.0
0040: 2e 30 2e 30 20 69 6e 20 .0.0 in
0048: 61 63 65 6d 64 62 65 74 acemdbet
0050: 61 5f 36 2e 33 38 5f 77 a_6.38_w
0058: 69 6e 64 6f 77 73 5f 69 indows_i
0060: 6e 74 65 6c 78 38 36 5f ntelx86_
0068: 5f 63 75 64 61 33 31 20 _cuda31
0070: 30 2e 30 2e 30 2e 30 20 0.0.0.0
0078: 61 74 20 6f 66 66 73 65 at offse
0080: 74 20 30 30 30 30 32 63 t 00002c
0088: 35 38 58


I was running a Tony long task, no Beta's. The Tony task is still running after I closed the Windows pop-up Error message.

Boinc log:
10/06/2011 13:15:48 | | Starting BOINC client version 6.12.28 for windows_x86_64
10/06/2011 13:15:48 | | Config: report completed tasks immediately
10/06/2011 13:15:48 | | Config: use all coprocessors
10/06/2011 13:15:48 | | Config: zero long-term debts on startup
10/06/2011 13:15:48 | | log flags: file_xfer, sched_ops, task, coproc_debug, cpu_sched_debug, dcf_debug
10/06/2011 13:15:48 | | log flags: debt_debug, rr_simulation, sched_op_debug
10/06/2011 13:15:48 | | Libraries: libcurl/7.19.7 OpenSSL/0.9.8l zlib/1.2.5
10/06/2011 13:15:48 | | Data directory: C:\Documents and Settings\All Users\Application Data\BOINC
10/06/2011 13:15:48 | | Running under account Administrator
10/06/2011 13:15:48 | | Processor: 8 GenuineIntel Intel(R) Core(TM) i7-2600K CPU @ 3.40GHz [Family 6 Model 42 Stepping 7]
10/06/2011 13:15:48 | | Processor: 256.00 KB cache
10/06/2011 13:15:48 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 cx16 sse4_1 sse4_2 syscall nx lm vmx tm2 popcnt aes pbe
10/06/2011 13:15:48 | | OS: Microsoft Windows Server 2003 "R2": Standard Server x64 Edition, Service Pack 2, (05.02.3790.00)
10/06/2011 13:15:48 | | Memory: 7.98 GB physical, 9.56 GB virtual
10/06/2011 13:15:48 | | Disk: 1.82 TB total, 1.43 TB free
10/06/2011 13:15:48 | | Local time is UTC +1 hours
10/06/2011 13:15:48 | | NVIDIA GPU 0: GeForce GTX 470 (driver version 27533, CUDA version 4000, compute capability 2.0, 1280MB, 1089 GFLOPS peak)
10/06/2011 13:15:48 | | NVIDIA library reports 1 GPU
10/06/2011 13:15:48 | | No ATI library found.
10/06/2011 13:15:48 | GPUGRID | URL http://www.gpugrid.net/; Computer ID 91249; resource share 1000
10/06/2011 13:15:48 | | General prefs: using separate prefs for home
10/06/2011 13:15:48 | | Reading preferences override file
10/06/2011 13:15:48 | | Preferences:
10/06/2011 13:15:48 | | max memory usage when active: 7357.62MB
10/06/2011 13:15:48 | | max memory usage when idle: 7357.62MB
10/06/2011 13:15:48 | | max disk usage: 100.00GB
10/06/2011 13:15:48 | | max CPUs used: 6
10/06/2011 13:15:48 | | (to change preferences, visit the web site of an attached project, or select Preferences in the Manager)
10/06/2011 13:15:48 | | [cpu_sched] Request CPU reschedule: Prefs update
10/06/2011 13:15:48 | | [cpu_sched] Request CPU reschedule: Startup
10/06/2011 13:15:48 | | Not using a proxy
10/06/2011 13:16:19 | | [cpu_sched] Request CPU reschedule: Idle state change
10/06/2011 13:16:19 | | [cpu_sched] Request CPU reschedule: periodic CPU scheduling
10/06/2011 13:16:19 | GPUGRID | [debt] CPU ineligible; LTD 0.00
10/06/2011 13:16:19 | | [debt] CPU LTD: adding offset 0.000000
10/06/2011 13:16:19 | GPUGRID | [debt] NVIDIA GPU LTD 0.00 delta 0.00 (0.50*0.00 - 0.00)/1
10/06/2011 13:16:19 | | [debt] NVIDIA GPU LTD: adding offset 0.000000
10/06/2011 13:16:19 | | [cpu_sched] schedule_cpus(): start
10/06/2011 13:16:19 | | [rr_sim] rr_sim start: work_buf_total 4320.86 on_frac 0.975 active_frac 0.897
10/06/2011 13:16:19 | GPUGRID | [rr_sim] 0.00: starting A429-TONI_AGGdense1-8-100-RND0011_0 (0.14 CPU + 1.00 NV)
10/06/2011 13:16:19 | GPUGRID | [rr_sim] 26954.51: A429-TONI_AGGdense1-8-100-RND0011_0 finishes after 3489.82 (151482.68G/43.41G)
10/06/2011 13:16:19 | GPUGRID | [cpu_sched] scheduling A429-TONI_AGGdense1-8-100-RND0011_0 (coprocessor job, FIFO)
10/06/2011 13:16:19 | | [cpu_sched_debug] reserving 1.000000 of coproc CUDA
10/06/2011 13:16:19 | | [cpu_sched] enforce_schedule(): start
10/06/2011 13:16:19 | | [cpu_sched] preliminary job list:
10/06/2011 13:16:19 | GPUGRID | [cpu_sched] 0: A429-TONI_AGGdense1-8-100-RND0011_0 (MD: no; UTS: no)
10/06/2011 13:16:19 | | [cpu_sched] final job list:
10/06/2011 13:16:19 | GPUGRID | [cpu_sched] 6: A429-TONI_AGGdense1-8-100-RND0011_0 (MD: no; UTS: no)
10/06/2011 13:16:19 | GPUGRID | [coproc] Assigning CUDA instance 0 to A429-TONI_AGGdense1-8-100-RND0011_0
10/06/2011 13:16:19 | GPUGRID | [cpu_sched] scheduling A429-TONI_AGGdense1-8-100-RND0011_0
10/06/2011 13:16:19 | GPUGRID | [cpu_sched] A429-TONI_AGGdense1-8-100-RND0011_0 sched state 1 next 2 task state 0
10/06/2011 13:16:19 | GPUGRID | Restarting task A429-TONI_AGGdense1-8-100-RND0011_0 using acemdlong version 613
10/06/2011 13:16:19 | | [cpu_sched] app startup took 21.218748 secs
10/06/2011 13:16:19 | | [cpu_sched] Request CPU reschedule: slow app startup
10/06/2011 13:16:19 | | [cpu_sched] enforce_schedule: end
10/06/2011 13:16:42 | GPUGRID | [debt] CPU ineligible; LTD 0.00
10/06/2011 13:16:42 | | [debt] CPU LTD: adding offset -21.776753
10/06/2011 13:16:42 | GPUGRID | [debt] NVIDIA GPU LTD -11.11 delta -11.11 (0.50*22.22 - 22.22)/1
10/06/2011 13:16:42 | | [debt] NVIDIA GPU LTD: adding offset -11.109374
10/06/2011 13:16:42 | | [cpu_sched] schedule_cpus(): start
10/06/2011 13:16:42 | | [rr_sim] rr_sim start: work_buf_total 4320.86 on_frac 0.975 active_frac 0.897
10/06/2011 13:16:42 | GPUGRID | [rr_sim] 0.00: starting A429-TONI_AGGdense1-8-100-RND0011_0 (0.14 CPU + 1.00 NV)
10/06/2011 13:16:42 | GPUGRID | [rr_sim] 26954.51: A429-TONI_AGGdense1-8-100-RND0011_0 finishes after 3489.82 (151482.68G/43.41G)
10/06/2011 13:16:42 | GPUGRID | [cpu_sched] scheduling A429-TONI_AGGdense1-8-100-RND0011_0 (coprocessor job, FIFO)

New application coming up with ABOVE_NORMAL_PRIORITY_CLASS.
acemdbeta_6.40 is out.

gdf

I think the intention is to improve stability on some systems. A well optimized system is unlikely to see any benefit, but there are so many poorly configured/optimized systems around that such a potential for overall project improvement must be investigated. In the last year or so I have come across several projects (5 or 6) that have at some stage impacted on other Boinc projects, so some Kevlar skin might help. I do think that such bullet-proofing changes would be best organized through Berkeley but as soon as just one project bypasses a default setting (and many have) it messes with other projects.

The point of increasing slightly the priority is that sometime we see people with a GTX580 which are performing at less than half the speed. We assume that it is because they are overloading the CPU. Nowadays, the CPU is naturally overloaded due to hyper-threading, hopefully this will not impact the performance of their system but guarantee a decent responsiveness to the GPU.

If you are using SWAN_SYNC, this is not going to change anything, but using SWAN_SYNC as default is not practicable.

gdf

Running this 29-GIANNI_TESTDHFR10-0-5-RND5720_2 task on the 6.40 (cuda31) app.
No problems so far. Looks like it will take about 3h on a GTX470.
It's a fairly safe system; only using 6 of 8 threads for CPU crunching (various projects) and SWAN_SYNC=0 in use. I did suspend it and start another GPUGrid task before continuing the Beta, so it seems robust enough.
Again 98% GPU usage and 322MB GDDR.

- Completed and validated 10,414.70 10,378.75 7,491.18 11,236.77 ACEMD beta version v6.40 (cuda31)

So far so good.
If there is nothing else, I will pass it to production today.

gdf