Message boards :
Number crunching :
boinc --allow_multiple_clients
Message board moderation
Author | Message |
---|---|
Send message Joined: 22 Apr 10 Posts: 14 Credit: 149,472,464 RAC: 0 |
Hi, what's going on with the MW scheduling? My dual 5970(1000/1200, 1.25V) never seems to have enough WU. I tried setting up some new client dirs and running c: cd "c:\program files (x86)\boinc" boinc --allow_multiple_clients --detach --dir c:\programdata\boinc0 --gui_rpc_port 5000 boinc --allow_multiple_clients --detach --dir c:\programdata\boinc1 --gui_rpc_port 5001 boinc --allow_multiple_clients --detach --dir c:\programdata\boinc2 --gui_rpc_port 5002 boinc --allow_multiple_clients --detach --dir c:\programdata\boinc3 --gui_rpc_port 5003 Thinking that this would be like I had each GPU in a separate rig.. so would get a cache per GPU. Everything seems to work, but all 4 new boincs are seen as the same client(despite different IDs) when it comes to checking MW for work, as I get "last request too soon"(or similar) How can I tell MW to only check for work every 5 min? Thanks to the programmers Thx Blox |
Send message Joined: 24 Feb 09 Posts: 620 Credit: 100,587,625 RAC: 0 |
You are getting the default limited cache for the Project 36 WUs, which is scaled as 6 per cpu, so for a hex, 36. Its expressed as per cpu as BOINC server software cannot express limits as per gpu as yet. Doubt it will run as you hope anyway, extract from BOINC cc_config options: "..... If set, you must run each BOINC client in a separate BOINC data directory (if you run multiple clients in the same directory, this will not be detected, and mayhem will ensue)." EDIT: A thought ..... have you got HT turned on with the 980? If you can, it will see it as 12 cores and you'll get a 72 WU cache. Regards Zy |
Send message Joined: 22 Apr 10 Posts: 14 Credit: 149,472,464 RAC: 0 |
My 920 + 4870x2 often limits at 48 (6x8) tasks.. 980x + 2 x 5970 never gets to the limit I am running each new boinc in a separate dir, and have a cc_config and rpc_gui_auth in each. The cc_config ignores the 3 other GPUs, e.g. 0 ignores 1,2,3 HT is on, 12 cores available |
Send message Joined: 24 Feb 09 Posts: 620 Credit: 100,587,625 RAC: 0 |
For some reason its not seeing the HT on the 980 and only gets the 6 each for the main CPUs - could be that BOINC (or the server revision MW runs)has not yet caught up re recognising HT on the 980s There is probably still going to be an issue anyway as there is no command line line option or config option to assign specific GPUs to each BOINC instance, so as setup currently, each BOINC instance will fight over a GPU to use as they will all see the same cards, defaulting to 0 as first option - thats probably why you are getting the result you are at present. The multiple client option was put in for dev work in cpu days, so I reckon you'll be out of luck at the end of the day trying to run 4 instances for GPU use. Regards Zy |
Send message Joined: 27 Nov 09 Posts: 108 Credit: 430,760,953 RAC: 0 |
For some reason its not seeing the HT on the 980 and only gets the 6 each for the main CPUs I had this problem with a XEON 5540 when I first enabled HT. The message log may claim it sees N cores but dispatches for only N/2. Or if you turn off HT, it claims it sees N/2 cores but still dispatches for N. For a Windows client, use the Advanced menu and do Run CPU Benchmarks. This was sufficient to reset whatever BOINC was using internally to correct the core count. You may have to do this every time you change the enable/disable state of HT. There is apparently a command line "--run_benchmarks" switch but whether this quits after running it I don't know. |
Send message Joined: 22 Apr 10 Posts: 14 Credit: 149,472,464 RAC: 0 |
I had my BOINC settings at 50% CPUs so I was only getting 36 WU, upped it to 75%(9) and now get 54.. a big improvement... but why is it limited on no of CPU?.. we're in GPU land now. even with 54 WU, my 4 GPU@1Ghz can do that in 90 min. This doesn't help us contribute to the awesome MW project when the servers are down. We all need to have at least 24 hrs of WU we can crunch while the servers are not responding. It's wasting our time and resources. And my room was freezing this morning as WU = processing = waste heat = toasty room :) Please admins/progs, lets work to giving each client enough WU to survive a 24hr outage of the servers... this = more WU total, which is a good thing.. Unless we are going to run out of WU soon :( Hope you have a new project planned if that is the case Rock on |
Send message Joined: 25 Jun 10 Posts: 284 Credit: 260,490,091 RAC: 0 |
The subject of the cache size is often discussed in this forum. Here is one such thread: http://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=2025 -Mike |
Send message Joined: 22 Apr 10 Posts: 14 Credit: 149,472,464 RAC: 0 |
Thanks Mike.. I have been trying to think of a workaround.. this is my current idea http://www.overclock.net/blogs/blox/2046-bloxcache-boinc-caching-batch-file.html The main problem atm, is that the --exit_after_app_start switch doesn't seem to be hnored all the time.. and max value seems to be 90. Also any half processed wu are restarted at 0, so a bit of a waste. Any ideas? Cheers, Rick |
Send message Joined: 22 Apr 10 Posts: 14 Credit: 149,472,464 RAC: 0 |
I've written a guide.. and allowed non-members to view(I hope) http://www.overclock.net/blogs/blox/2050-bloxcache-boinc-caching-batch-file-initialisation.html http://www.overclock.net/blogs/blox/2051-bloxcache-boinc-caching-batch-file-running.html |
Send message Joined: 14 Feb 09 Posts: 999 Credit: 74,932,619 RAC: 0 |
I've written a guide.. and allowed non-members to view(I hope) Clickable |
©2024 Astroinformatics Group