Posts by mmonnin
|
1)
(Message 6583)
Posted 21 Apr 2020 by mmonnin Post: This question should be asked on the BOINC Forums. |
|
2)
(Message 6552)
Posted 10 Apr 2020 by mmonnin Post: The whole project had some issue yesterday. 10m credit yesterday, 120m today. My pending dropped a lot since yesterday, I first thought my own low credit on 4/9 was because I had a bunch of threads taken up by tasks that were not using any CPU on 2 different systems. Aborted those and have been fine since. It might have happened to others too. |
|
3)
(Message 6538)
Posted 4 Apr 2020 by mmonnin Post: What's going on with the new plan_class versions? There are specific versions for XP/Win7/Win10/Linux/OSX now. The ETAs are ludacris at 5 days on one PC instead of the older versions ETA of 3 hours. There seem to be no more AVX tasks as all PCs are getting SSE2/3. 100min run time vs 180min. |
|
4)
Posted 27 Mar 2020 by mmonnin Post: 7.5% of my tasks error out these days. Many instantly with the lc points. Especially when the server runs dry they all end up bad as they get cycled to a bunch of users. |
|
5)
(Message 6404)
Posted 19 Nov 2019 by mmonnin Post: Scroll down, there are several threads about being out of work, no new work, etc. |
|
6)
Message boards :
News :
Badges
Posted 24 Sep 2019 by mmonnin Post: Some projects have something like this if you find a Prime, special # or have crunch data used in research. PrimeGrid, GPUGrid, RakeSearch, DHEP, Amicable, Xansons, ODLK, Latin Squares (ODLK1), Pogs, SRBase are some that I see. |
|
7)
(Message 6341)
Posted 2 Sep 2019 by mmonnin Post: Turing is not supported. Several threads in Number Crunching forum and the project as a whole is slow to adjust support for new NV generations. |
|
8)
(Message 5851)
Posted 23 Jun 2018 by mmonnin Post: Basically all projects are chosen as possibilities unless there is sporadic work or its still beta testing. FB has been running for over a decade. I'd guess if a project admin requests their project to not be included it could be made so. Not sure what mining really has to do with Formula BOINC Sprint participants causing a shortage of work for a project. Two weeks ago there was a shortage for awhile as well and during the event the credit earned tripled for Latinsquares. Gridcoin does a huge amount of BOINC work but I don't recall that team having as large of spikes as other teams during FB Sprints. Gridcoin users carry on for the most part and the rest of us caused the outage. Disclosure: I don't agree with the historical reasoning for a crypto currency. Its mostly a waste of electricity. On the other hand w/o it, some people wouldn't be doing good with Curecoin, Foldingcoin and some of the projects Gridcoin whitelists. At least on the FAH side, foldingcoin does not have a team requirement unlike Curecoin and Gridcoin. As mentioned by xii5ku in the other thread, I don't know if Sébastien notifies admins judging by the reactions of some of the admins. But if any are concerned over the high load caused by the sprints, my guess the time for opting out is at the beginning of the year. It's like no complaints means its ok to be included I guess. |
|
9)
(Message 5840)
Posted 22 Jun 2018 by mmonnin Post: Basically all projects are chosen as possibilities unless there is sporadic work or its still beta testing. FB has been running for over a decade. I'd guess if a project admin requests their project to not be included it could be made so. |
|
10)
(Message 5804)
Posted 9 May 2018 by mmonnin Post: It took awhile for Pascal support so I guess there needs to be some manual intervention for it to work http://asteroidsathome.net/boinc/forum_thread.php?id=542#5016 |
|
11)
(Message 5110)
Posted 14 Jan 2017 by mmonnin Post: Not really. GPU applications that fully utilize the GPU will have screen lag. I mean like 99-100% full GPU utilization. I've seen it on several projects like Asteroids, Collatz and FAH. Others like E@H with a bit lower GPU util do not have screen lag. Basically, CPUs have interrupts but GPUs do not. Keyboard input is an interrupt that gets to move to the head of the work queue for CPUs. Since GPUs are meant to display live images and originally weren't designed as compute engines for other things besides drawing images on a screen, there is no GPU interrupt. Any input or change to the screen has to go through the work compute buffer before getting to your mouse movement. Also unlike CPUs you can't tell it to run at 90% either. If you're not gaming at the moment, the screen could be powered off the iGPU instead of the 980Ti. |
|
12)
(Message 5085)
Posted 6 Dec 2016 by mmonnin Post: Ahh yes I did post app_config.xml file which is 100% correct and works. I am running it now. I don't recall editing any app_info file but I use cc_config.xml file to limit projects to one card or another. I used it to run asteroids on a 970 and E@H on a 1070 in the same machine. Before Asteroids released its Pascal capable app. The same app_config.xml file works then and now. |
|
13)
(Message 5070)
Posted 30 Nov 2016 by mmonnin Post: <app_config>
<app>
<name>period_search</name>
<max_concurrent>4</max_concurrent>
<gpu_versions>
<gpu_usage>0.5</gpu_usage>
<cpu_usage>0.2</cpu_usage>
</gpu_versions>
</app>
</app_config>
|
|
14)
(Message 5035)
Posted 16 Oct 2016 by mmonnin Post: Your ETA should adjust as WUs complete. If they can't be completed then they could be aborted before hitting the deadline so they can be sent to someone else. |
|
15)
Posted 16 Oct 2016 by mmonnin Post: I haven't crunched before with Prime or Seti but I know that E@H does not fully utilize a GPU. Even with 3-4 WUs at once. The stuttering arises because the Asteroids app completely utilizes your GPU, as it does mine. I sit at 99-100% most of the time. Unlike CPUs, there is no interrupts for the GPUs. It expects to process data as it comes in, in real time. When the buffer is full of compute and then there is a request to change the screen there is lag. CPUs have interrupts to insert high priority tasks and moves it to the front of queue so to speak. There is no way to set GPU utilization % either. Best option would try to use iGPU or another card when using the computer for daily tasks and turn off Asteroids and use your 980 for gaming as an example. |
|
16)
(Message 5020)
Posted 10 Oct 2016 by mmonnin Post: Awesome! I'll be finishing E@H work and trying this on my 1070. I'm guessing 1060s have support as well? |
|
17)
(Message 5015)
Posted 9 Oct 2016 by mmonnin Post: I don't believe Asteroids supports Pascal GPUs yet. |
|
18)
(Message 4966)
Posted 24 Aug 2016 by mmonnin Post: Pascal is not yet supported. Dev is working on it. |
|
19)
(Message 4961)
Posted 13 Aug 2016 by mmonnin Post: Pascal is not supported yet. http://asteroidsathome.net/boinc/forum_thread.php?id=516 I have a 1070 myself and I am using it to run E@H and Asteroids on only my 970 due to no support yet. :( |