Posts by mmonnin

1) Message boards : News : CUDA app testing (Message 5150)
Posted 17 days ago by mmonnin
Post:
Down to 1 user in past 24 hours with 162 tasks still out there.
2) Message boards : Number crunching : Gromacs GPU work units now available (Message 5142)
Posted 20 days ago by mmonnin
Post:
Wonder whats different with this version to make it a new app name.

v1 execute one task, v2 is 6 tasks (5 very short, prepare data for sixth) (look at job.xml).


Kind of hard to look at job.xml if I can't get any. They were only sent to just several hosts like a week ago and still aren't finished.
3) Message boards : News : CUDA app testing (Message 5139)
Posted 22 days ago by mmonnin
Post:
Just 5 people got the last batch of several hundred v2 tasks?
4) Message boards : Number crunching : Gromacs GPU work units now available (Message 5131)
Posted 26 days ago by mmonnin
Post:
Wonder whats different with this version to make it a new app name.
5) Questions and Answers : Windows : Building Gromacs on Windows. (Message 5128)
Posted 26 days ago by mmonnin
Post:
How does it compared to the stock app?
6) Message boards : Number crunching : Gromacs GPU work units now available (Message 5126)
Posted 26 days ago by mmonnin
Post:
Whats with the huge disparity in points?
12480736 11644869 18 Oct 2017, 15:35:17 UTC 21 Oct 2017, 2:30:45 UTC Completed and validated 20,386.33 20,250.11 830,742.42 gromacs v0.23 (cuda23)
windows_x86_64
12479110 11649199 18 Oct 2017, 15:30:02 UTC 21 Oct 2017, 13:34:58 UTC Completed and validated 20,364.12 20,222.16 22,291.11 gromacs v0.23 (cuda23)
windows_x86_64
7) Questions and Answers : Unix/Linux : Linux Gromacs compilation (Message 5117)
Posted 27 days ago by mmonnin
Post:
You have to use app_info (or app_config, can't remember) to use it as "anonymous" platform... I never do this, but probably some users can help.


app_info to run a custom exe file instead of the one provided by a project. I've never set one up myself. SETI Lunatics is an installer that sets one up if a user wants an example.
8) Message boards : News : CUDA app testing (Message 5103)
Posted 28 days ago by mmonnin
Post:
Progress is lost if the task becomes paused

No, isn't, just at the moment checkpoints are not reported to Manager.

I will update application for proper progress bar and checkpoints reporting later today.


Progress bar is the thing that seems to work. At least I've seen it progress.

Points were cut by a 1/10?
9) Message boards : Number crunching : Gromacs GPU work units now available (Message 5099)
Posted 28 days ago by mmonnin
Post:
<cpu_usage>0.2</cpu_usage>

You have set that single work unit can use only 20% of CPU thread. This value should by "1".


This has absolutely zero impact on how much CPU is used to feed the GPU. It only changes how many CPU threads are removed from doing CPU work. If its 1 then your 8 core CPU can only run 7 CPU tasks. If anything less than 1 BOINC will still run CPU tasks along with 1 GPU task. Then its up to the OS to give priority to feed the GPU. Process Lasso in windows and Linux has no issues properly feed GPUs.
10) Message boards : News : CUDA app testing (Message 5083)
Posted 29 days ago by mmonnin
Post:
No, you have CUDA 6.5 where 8.0 is minimum, you need to update your CUDA driver.


Why is the plan class cuda23 if cuda80 is needed? Just call it cuda65 or cuda80 to match the requirement.

So just AVX? You said AVX256 and AVX and AVX2 added some 256 bit commands so it wasn't really specific enough.


Cuda 2.3 is "Compute capabilty" where Cuda 8.0 is driver version used in Gromacs compilation. Just CUDA capabilities are not same as CUDA runtime version...

For this app CUDA 2.3 version is need with CUDA 8.0 version driver.


Other projects put in the 6.5, 8.0, 9.0 number in the plan class. SETI and GPUGrid do. Thought it was that number.
11) Message boards : News : CUDA app testing (Message 5081)
Posted 29 days ago by mmonnin
Post:
No, you have CUDA 6.5 where 8.0 is minimum, you need to update your CUDA driver.


Why is the plan class cuda23 if cuda80 is needed? Just call it cuda65 or cuda80 to match the requirement.

So just AVX? You said AVX256 and AVX and AVX2 added some 256 bit commands so it wasn't really specific enough.
12) Message boards : Number crunching : Gromacs GPU work units now available (Message 5060)
Posted 18 Oct 2017 by mmonnin
Post:
How long should these Wu's run ??? the 2 I have running are up to 23+ Hr's !

EDIT: Wu on ID 181 still running but the Wu on 5044 erred out when I clicked okay on the point of entry error message, I detached & reattached computer ID 5038, I'm back to output file errors again.

PBOYZTOYNP9873

3058 DrugDiscovery@Home 10/18/2017 4:30:39 AM Scheduler request completed: got 5 new tasks
3059 DrugDiscovery@Home 10/18/2017 4:30:41 AM Started download of gmx_GMX-T025_XXXX_TTTTT_tasks.gmx_58_1000.in
3060 DrugDiscovery@Home 10/18/2017 4:30:41 AM Started download of gmx_GMX-T025_XXXX_TTTTT_tasks.gmx_462_1000.in
3061 DrugDiscovery@Home 10/18/2017 4:30:41 AM Started download of gmx_GMX-T025_XXXX_TTTTT_tasks.gmx_490_1000.in
3062 DrugDiscovery@Home 10/18/2017 4:30:41 AM Started download of gmx_GMX-T025_XXXX_TTTTT_tasks.gmx_503_1000.in
3063 DrugDiscovery@Home 10/18/2017 4:30:41 AM Started download of gmx_GMX-T025_XXXX_TTTTT_tasks.gmx_504_1000.in
3064 DrugDiscovery@Home 10/18/2017 4:30:46 AM Finished download of gmx_GMX-T025_XXXX_TTTTT_tasks.gmx_462_1000.in
3065 DrugDiscovery@Home 10/18/2017 4:30:48 AM Starting task gmx_GMX-T025_XXXX_TTTTT_tasks.gmx_462_1000.in_2
3066 DrugDiscovery@Home 10/18/2017 4:30:50 AM Sending scheduler request: To fetch work.
3067 DrugDiscovery@Home 10/18/2017 4:30:50 AM Requesting new tasks for CPU and NVIDIA GPU
3068 DrugDiscovery@Home 10/18/2017 4:30:51 AM Scheduler request completed: got 0 new tasks
3069 DrugDiscovery@Home 10/18/2017 4:30:51 AM No tasks sent
3070 DrugDiscovery@Home 10/18/2017 4:30:51 AM No tasks are available for smina_mol2
3071 DrugDiscovery@Home 10/18/2017 4:30:51 AM No tasks are available for smina_sdf
3072 DrugDiscovery@Home 10/18/2017 4:30:51 AM No tasks are available for smina_pdbqt
3073 DrugDiscovery@Home 10/18/2017 4:30:53 AM Computation for task gmx_GMX-T025_XXXX_TTTTT_tasks.gmx_462_1000.in_2 finished
3074 DrugDiscovery@Home 10/18/2017 4:30:53 AM Output file gmx_GMX-T025_XXXX_TTTTT_tasks.gmx_462_1000.in_2_r2129845200_2 for task gmx_GMX-T025_XXXX_TTTTT_tasks.gmx_462_1000.in_2 absent
3075 DrugDiscovery@Home 10/18/2017 4:30:53 AM Output file gmx_GMX-T025_XXXX_TTTTT_tasks.gmx_462_1000.in_2_r2129845200_5 for task gmx_GMX-T025_XXXX_TTTTT_tasks.gmx_462_1000.in_2 absent


Probably depends on the speed of the CPU. I have 3x 'running' on a 970 but there is zero GPU usage so its my 1950x doing all the work. Longest any task has taken was about 11.5 hours.
13) Message boards : News : CUDA app testing (Message 5035)
Posted 17 Oct 2017 by mmonnin
Post:
No, there is two "AVX256" extensions...
AVX_256 and
AVX2_256

OK, does that mean that Ivy Bridge (i7-3770 for example) is good enough to support the Windows version?


Ivy only has AVX. No AVX2.

CPUs with AVX2 per Wiki

Intel
Haswell processor, Q2 2013
Haswell E processor, Q3 2014
Broadwell processor, Q4 2014
Broadwell E processor, Q3 2016
Skylake processor, Q3 2015
Kaby Lake processor, Q3 2016(ULV mobile)/Q1 2017(desktop/mobile)
Coffee Lake processor, Q3 2017
Cannonlake processor, expected in 2018
Cascade Lake processor, expected in 2018
AMD
Excavator processor, Q2 2015
Zen processor, Q1 2017
14) Message boards : Number crunching : Gromacs GPU work units now available (Message 5015)
Posted 17 Oct 2017 by mmonnin
Post:
What would be the correct app_name and plan_class to use in the app_config.xml file for the Gromacs application?

Thx


FYI, if you don't know enter something and the event log will tell you the known app names for a given project. For some projects its the easiest way as the name isn't listed on the app or server status page.
15) Message boards : Number crunching : Gromacs GPU work units now available (Message 5014)
Posted 17 Oct 2017 by mmonnin
Post:
Looks like this is limited to the 1st GPU slot in a computer?

Have you tried a cc_config.xml file placed in the /etc/boinc-client folder?

<cc_config> 
    <options> 
    <use_all_gpus>1</use_all_gpus>
    </options> 
</cc_config> 


Yes, my GPUs work in other projects. And it's not just me. Its the project. Enigma only worked on GPU 0 when it first released its GPU app.

This has def slowed down GPU 0 running SETI even though BM shows it running on GPU 1 with no GPU utilization. Bout a 25% increase in run times and its taking up a GPU and using a CPU thread. I don't see the point in this app.
16) Message boards : Number crunching : Gromacs GPU work units now available (Message 5008)
Posted 16 Oct 2017 by mmonnin
Post:
Looks like this is limited to the 1st GPU slot in a computer?

And no check pointing? Lost like 3 hours pushing my two tasks to 1 GPU. The GPU with no load.
17) Message boards : Number crunching : Gromacs GPU work units now available (Message 5007)
Posted 16 Oct 2017 by mmonnin
Post:
Ran sudo apt install gromacs and it is "running" on a 1070 and 970 in Ubuntu. v0.14

The 1070 is showing 25-37 GPU load with 1-2% PCI-E bandwidth. The 970 is showing zero GPU load. None, nothing. Not parallel enough and doesn't work at all. :(

A full CPU core is being used as well. For both GPUs even though the 970 isn't even doing anything.
18) Message boards : Number crunching : Gromacs GPU work units now available (Message 4976)
Posted 15 Oct 2017 by mmonnin
Post:
my gpu utilization is only about 30 to 40 percent
how can i change this ?


Run several tasks at once with app_config file.

I moved a GPU over to a Linux computer. Just need some tasks now.
19) Questions and Answers : Web site : Task Purge (Message 4925)
Posted 9 Sep 2017 by mmonnin
Post:
Are tasks being purged? When I click the tasks link on the user home page I get an error:

Fatal error: Allowed memory size of 134217728 bytes exhausted (tried to allocate 78 bytes) in /home/boincadm/projects/drugdiscovery/html/inc/db_conn.inc on line 125


Going to view computers then just their tasks I can see them. Some have over 30k. Overall I've probably done over several hundred thousand at nearly 1mil points and just several points per task with nearly all of them beind mol2.
20) Message boards : News : Maintenance break (Message 4917)
Posted 13 Aug 2017 by mmonnin
Post:
I am only getting resends atm from others invalids but they are invalid for me too. Must be just bad tasks that are left. Aborting what I have left as it's a waste of time.


Next 20


©2017 All rights reserved | Design by Digital BioPharm Ltd