| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
| World Community Grid Forums
|
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 3
|
|
| Author |
|
|
kylelyk
Cruncher Joined: Apr 19, 2011 Post Count: 1 Status: Offline Project Badges:
|
So I am trying to increase my cache size because my internet will randomly shut off while I am at work and never turn back on without manual intervention. I changed the BOINC cache size from 0.5 to 1.5 days. Unfortunately, I still have only about 8 hours per core stocked up (including the tasks currently running) after changing the setting almost 24 hours ago. Does it take more time than that for the changes to take effect? Or is this not the correct option to change? I have "updated" the project several times since then.
My second problem is that I downloaded the newest (and stable) version of the WCG version of the BOINC client and it does not "replace" the older version. This means that the old version runs at start-up. How do I disable the old version (6.10) and enable the new version to start up when the computer does? I am running Windows 7, 64-bit. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Did not know there was a newest (and stable) version of the WCG client, so what is that exact version, as knreed sure will be interested to learn of the non-replacement of the previous 6.10 release [which exactly... was it 6.10.58?].
Been testing the WCG builds and discussing offline with techs, now a work fetch issue popping up too, a fix just checked in by devs. Basics is, that the buffer is sized in-amongst to the hours a device is on and connected in say last 10 days. These values are stored in the below 5 [sample] fraction values, in the client_state.xml: <on_frac>0.938831</on_frac> <connected_frac>0.979211</connected_frac> <cpu_and_network_available_frac>0.978738</cpu_and_network_available_frac> <active_frac>0.999512</active_frac> <gpu_active_frac>0.999512</gpu_active_frac> This says essentially that this host is on 93% and when on, virtually computes 100%. Work is buffered on that basis. There's actually more issues, such as slow adjusting of projected runtimes. E.g. I want 2 days, but to get that I have to set the cache to 3 days [because my host is much much faster than FLOPS in the WU headers suggest it will take]. My message is; play with the cache values to get what you want [till we really have a Long Term Support] release, which is as I understand it, for many 7.0.xx, but could also be fast-tracked to become 7.2.xx and that does not even exist yet, except in a feature complete alpha alpha 7.1.15 build. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Errata: On the install front, run the 6.10.xx installer and select uninstall, then rerun the 7.0 installer again. That will ensure you get to run what you want.
As noted I've been testing and had an issue on W8 first time, but could not reproduce, probably because I did not go blindly with the installer default selections. Have installed over various versions, flip-flopped 'upgrading' from Berkeley 64 bits to WCG 32 bits, and it always did this right, replacing previous release, now on test build 7.0.67 on my W7-64 pro. |
||
|
|
|