Posts by entigy

1) Message boards : Number crunching : Computation Errors (Message 1070)
Posted 21 Sep 2018 by entigy
Post:
So, my GTX 1070 system is successfully completing WUs in under 8 minutes.
I have a 'tweaked' config file but NO app_config file.

My GTX 1060 system, on the other hand, flatly refuses to run any Collatz WU if there's any sort of optimized config file
This is what it contains at the moment (copied from your GTX 1050 settings):

verbose=1
kernels_per_reduction=48
threads=9
lut_size=17
sieve_size=27
cache_sieve=1
sleep=0

If the config file is blank then the WUs complete successfully - but they take about 20 minutes.
They always used to finish after 11 - 12 minutes, but I can't seem to find any optimized settings that it's happy with now....

Edit

Scratch that..
I've copied over the GTX 1070 config and I've managed to complete a unit in around 12 minutes.
Let's hope that's sorted now....
2) Message boards : Number crunching : Computation Errors (Message 1068)
Posted 21 Sep 2018 by entigy
Post:
Just removed/clean installed the Nvidia software on my GTX 1070 machine - it's made no difference whatsoever:

<core_client_version>7.12.1</core_client_version>
<![CDATA[
<message>
(unknown error) - exit code 9999 (0x270f)</message>
<stderr_txt>
Collatz Conjecture Sieve 1.30 Windows x86_64 for OpenCL
Written by Slicker (Jon Sonntag) of team SETI.USA
Based on the AMD Brook+ kernels by Gipsel of team Planet 3DNow!
Sieve code and OpenCL optimization provided by Sosiris of team BOINC@Taiwan
Collatz Config Settings:
verbose 1 (yes)
kernels/reduction 48
threads 2^9 (512)
lut_size 17 (1048576 bytes)
sieve_size 2^30 (51085096 bytes)
sleep 1
cache_sieve 1 (yes)
reducecpu 0 (no)
Processor Type NVIDIA
Max Dimensions 3
Max Work Items 1024 1024 64
Max Work Groups 1024
Max Kernel Threads 256
Device Vendor NVIDIA Corporation
Name GeForce GTX 1070
Driver Version 411.63
OpenCL Version OpenCL 1.2 CUDA
Device Vendor NVIDIA Corporation
Name GeForce GTX 1070
Driver Version 411.63
OpenCL Version OpenCL 1.2 CUDA
actual threads 512
Error: GPU steps do not match CPU steps. Workunit processing aborted.
18:56:49 (6924): called boinc_finish(9999)

</stderr_txt>
]]>

Seti and Amicable Numbers also use OpenCL and those workunits are completing with no problem....

Edit
Just tweaked my Collatz config file and now I have some successful results.
Curiouser and curiouser...
3) Message boards : Number crunching : Computation Errors (Message 1062)
Posted 20 Sep 2018 by entigy
Post:
Yeah, I've used DDU many times.....

BOINC seems happy enough that I have a GPU installed:

20/09/2018 19:12:35 | | CUDA: NVIDIA GPU 0: GeForce GTX 1060 6GB (driver version 411.63, CUDA version 10.0, compute capability 6.1, 4096MB, 3562MB available, 4730 GFLOPS peak)
20/09/2018 19:12:35 | | OpenCL: NVIDIA GPU 0: GeForce GTX 1060 6GB (driver version 411.63, device version OpenCL 1.2 CUDA, 6144MB, 3562MB available, 4730 GFLOPS peak)

I started one WU as a test and it failed with the following error:

<core_client_version>7.12.1</core_client_version>
<![CDATA[
<message>
Access is denied.
(0x5) - exit code 5 (0x5)</message>
<stderr_txt>
Collatz Conjecture Sieve 1.30 Windows x86_64 for OpenCL
Written by Slicker (Jon Sonntag) of team SETI.USA
Based on the AMD Brook+ kernels by Gipsel of team Planet 3DNow!
Sieve code and OpenCL optimization provided by Sosiris of team BOINC@Taiwan
Processor Type NVIDIA
Device Vendor NVIDIA Corporation
Name GeForce GTX 1060 6GB
Driver Version 411.63
OpenCL Version OpenCL 1.2 CUDA
Error: GPU steps do not match CPU steps. Workunit processing aborted.
19:13:03 (9228): called boinc_finish(5)

</stderr_txt>
]]>


Incidentally, my other PC (which has a GTX 1070, and which also updated its drivers last night) is merrily chugging through some Seti units at the moment....
4) Message boards : Number crunching : Computation Errors (Message 1060)
Posted 20 Sep 2018 by entigy
Post:
GeForce Experience alerted me to the new drivers, and I did a custom installation via that (not a 'clean install' though).

I'm running Win 10 Pro, and have 'driver updates' disabled via Group Policy.
5) Message boards : Number crunching : Computation Errors (Message 1054)
Posted 19 Sep 2018 by entigy
Post:
Started getting errors tonight - WUs are failing after just 6 or 7 seconds.

19/09/2018 22:50:01 | collatz | Starting task collatz_sieve_1cf6ab45-abae-4dea-b0a5-9ee61bb157bc_0
19/09/2018 22:50:07 | collatz | Computation for task collatz_sieve_1cf6ab45-abae-4dea-b0a5-9ee61bb157bc_0 finished
19/09/2018 22:50:07 | collatz | Output file collatz_sieve_1cf6ab45-abae-4dea-b0a5-9ee61bb157bc_0_r575015586_0 for task collatz_sieve_1cf6ab45-abae-4dea-b0a5-9ee61bb157bc_0 absent
19/09/2018 22:52:08 | collatz | Sending scheduler request: To report completed tasks.
19/09/2018 22:52:08 | collatz | Reporting 1 completed tasks

This is with a GTX 1060 6Gb card.
My machine did update the Nvidia drivers to 411.63 tonight - could that be the culprit ???
6) Message boards : News : Use at your own risk (Message 11)
Posted 18 Apr 2018 by entigy
Post:
Well I have had two units validated, and the points credited.
They are taking longer to crunch though - approx 15 mins on a 1070, and around 22 mins on a 1060...
Update
Forgot to edit the config files... Whoops! They've speeded up now...




©2018 Jon Sonntag; All rights reserved