The Tasks and Stats graphs use the predefined 20 colors. If there is a larger number of projects, then all over the 20th share the last color.
Could please the colors be used again from beginning, starting with the 21st and 41st project etc.?
Quote from: Pepo on January 14, 2011, 04:34:14 PM
The Tasks and Stats graphs use the predefined 20 colors. If there is a larger number of projects, then all over the 20th share the last color.
Could please the colors be used again from beginning, starting with the 21st and 41st project etc.?
Not to introduce too much complexity ::) but does the graphing toolkit allow for things like broken lines or maybe alternating strips of color for higher "blocks" of projects beyond 20?
Quote from: jjwhalen on January 14, 2011, 07:03:53 PM
Quote from: Pepo on January 14, 2011, 04:34:14 PM
The Tasks and Stats graphs use the predefined 20 colors. If there is a larger number of projects, then all over the 20th share the last color.
Could please the colors be used again from beginning, starting with the 21st and 41st project etc.?
Not to introduce too much complexity ::) but does the graphing toolkit allow for things like broken lines or maybe alternating strips of color for higher "blocks" of projects beyond 20?
Toolkit, I'm not using any toolkits, but I still can draw a different line ;D.
Is on the list, with the repeating of colors.
The stdoutdae.txt contains messages 14-Jan-2011 14:04:16 [Enigma@Home] Enigma 0.76b is not available for Microsoft Windows running on an AMD x86_64 or Intel EM64T CPU.
15-Jan-2011 20:06:05 [SETI@home] No work available for the applications you have selected. Please check your project preferences on the web site.
BOINC Manager's Messages dialog displays similar (but "enhanced") red messages 14.01.2011 14:04:16 | Enigma@Home | Message from server: Enigma 0.76b is not available for Microsoft Windows running on an AMD x86_64 or Intel EM64T CPU.
15.01.2011 20:06:05 | SETI@home | Message from server: No work available for the applications you have selected. Please check your project preferences on the web site.
BOINC Manager's Notices tab displays corresponding notices with headlines "SETI@home/Enigma@Home: Notice from server". [edit]BT displays the same notices as translated (because of the existing matching translation strings in the .btlang's translation part), but BT is completely missing the projects' headlines - just a thick empty line appear there.[/edit]
The .btlang's translation part contains just similar "Message from server" (which was initially marked as translatable in BT's Messages and is correctly translated afterwards in the Messages). So, either the BOINC Manager is guessing and generating the missing headline (with "Notice"), or BT is loosing an existing headline.
Any clue how is it actually?
The "Include general messages" tag is missing in .btlang files.
I am getting the following error on Windows 7 64 bit.
Fatal error in language file: C:\Program files\eFMer\BOINCTasks\lang\BOINCTasks_ENU.btlang
Look in c:\Program Files\eFMerBOINCTasks\lang\BOINCTasks_ENU_error.txt for more details.
The referenced file does not exist.
Quote from: John McLeod VII on January 17, 2011, 05:24:51 PM
I am getting the following error on Windows 7 64 bit.
Fatal error in language file: C:\Program files\eFMer\BOINCTasks\lang\BOINCTasks_ENU.btlang
Look in c:\Program Files\eFMerBOINCTasks\lang\BOINCTasks_ENU_error.txt for more details.
The referenced file does not exist.
A simple fix is delete the C:\Program files\eFMer\BOINCTasks\lang\BOINCTasks_ENU.btlang and run the installer again.
The error file can't be stored in teh program folder. So you have to run BT as administrator once.
Or copy the BT folder to a temp folder and run BT from there.
Quote from: Pepo on January 17, 2011, 02:00:20 PM
The stdoutdae.txt contains messages 14-Jan-2011 14:04:16 [Enigma@Home] Enigma 0.76b is not available for Microsoft Windows running on an AMD x86_64 or Intel EM64T CPU.
15-Jan-2011 20:06:05 [SETI@home] No work available for the applications you have selected. Please check your project preferences on the web site.
BOINC Manager's Messages dialog displays similar (but "enhanced") red messages 14.01.2011 14:04:16 | Enigma@Home | Message from server: Enigma 0.76b is not available for Microsoft Windows running on an AMD x86_64 or Intel EM64T CPU.
15.01.2011 20:06:05 | SETI@home | Message from server: No work available for the applications you have selected. Please check your project preferences on the web site.
BOINC Manager's Notices tab displays corresponding notices with headlines "SETI@home/Enigma@Home: Notice from server". [edit]BT displays the same notices as translated (because of the existing matching translation strings in the .btlang's translation part), but BT is completely missing the projects' headlines - just a thick empty line appear there.[/edit]
The .btlang's translation part contains just similar "Message from server" (which was initially marked as translatable in BT's Messages and is correctly translated afterwards in the Messages). So, either the BOINC Manager is guessing and generating the missing headline (with "Notice"), or BT is loosing an existing headline.
Any clue how is it actually?
The "Include general messages" tag is missing in .btlang files.
The latest version probably has an additional flag, for the received from. In the todo list.
You can check if there is something wrong with the translation by removing the check at window -> "translate BOINC text". That way the translation text is disregarded, as in English.
Quote from: fred on January 17, 2011, 06:20:01 PM
You can check if there is something wrong with the translation by removing the check at window -> "translate BOINC text". That way the translation text is disregarded, as in English.
Yes. After unchecking it, the two lines got immediately untranslated (
:D) back to English.
Nevertheless, they still missed their headline in
Notices, so it was really not influenced by some translation mismatch.
(Forgot to note, my client is now 6.12.10).
Switched to 0.97 when it came out and since then have been getting consistent stalling of BT (updating........ for minutes, taking minutes even to exit). Sometimes it worked OK, more often not so much. Decided to switch back to 0.96 a few hours ago and since then BT is operating much better. I think the problem might be related to a couple WiFi attached boxes that don't have the most solid connection. BT 0.96 seems to handle it, 0.97 doesn't. Has there been a change that might have caused this?
Also have noticed (as mentioned before) increasing CPU usage for BT. Is this due to adding features or? Any possibility of a BT Lite? Don't need or want graphs, widgets or gadgets. Just want to be able to control BOINC on a local network without a lot of resource usage. Limiting history was suggested before, I keep 1 day history, lowering doesn't seem to make much of a difference. Thanks!
Quote from: Beyond on January 19, 2011, 05:10:50 PM
Switched to 0.97 when it came out and since then have been getting consistent stalling of BT (updating........ for minutes, taking minutes even to exit). Sometimes it worked OK, more often not so much. Decided to switch back to 0.96 a few hours ago and since then BT is operating much better. I think the problem might be related to a couple WiFi attached boxes that don't have the most solid connection. BT 0.96 seems to handle it, 0.97 doesn't. Has there been a change that might have caused this?
Also have noticed (as mentioned before) increasing CPU usage for BT. Is this due to adding features or? Any possibility of a BT Lite? Don't need or want graphs, widgets or gadgets. Just want to be able to control BOINC on a local network without a lot of resource usage. Limiting history was suggested before, I keep 1 day history, lowering doesn't seem to make much of a difference. Thanks!
Strange, there are no changes from 96->97 that could relate to the slow updates. But the BOINC client is a low priority process, so when on of the computers is doing something the response time may go way down.
Al the extra things like graphic and gadgets do nothing when you don't use them.
In the BT setting set "Smart mode" and a "Maximum update time" of 120 (default 30).
Set no rules.
But a large amount of WU's may still cause some overhead, they must all be read. Especially when all computers have tasks close to completion.
Quote from: fred on January 19, 2011, 08:12:52 PM
Quote from: Beyond on January 19, 2011, 05:10:50 PM
Switched to 0.97 when it came out and since then have been getting consistent stalling of BT (updating........ for minutes, taking minutes even to exit). Sometimes it worked OK, more often not so much. Decided to switch back to 0.96 a few hours ago and since then BT is operating much better. I think the problem might be related to a couple WiFi attached boxes that don't have the most solid connection. BT 0.96 seems to handle it, 0.97 doesn't. Has there been a change that might have caused this?
Strange, there are no changes from 96->97 that could relate to the slow updates. But the BOINC client is a low priority process, so when on of the computers is doing something the response time may go way down.
Al the extra things like graphic and gadgets do nothing when you don't use them.
In the BT setting set "Smart mode" and a "Maximum update time" of 120 (default 30).
Set no rules.
But a large amount of WU's may still cause some overhead, they must all be read. Especially when all computers have tasks close to completion.
Smart mode was on, changed max update time from 30 to 120, have no rules set. Decided to change the 2 WiFi clients to longer range adapters and that seems to have helpped a lot. V0.97 is running much better now (0.96 was starting to do the same as 0.97 but never as bad, probably just chance). Seems when the connection to any client is even a little flaky, BT chokes. Think that should be addressed. With the 10 CPUs (40 cores) and 16 GPUs running there's an average of 5 WUs a minute completing, so there's always many close to completion. That's probably quite a bit more than most BT users are seeing and I'm sure that's contributing to the problem. Thanks very much for the help, Fred!
Quote from: Beyond on January 19, 2011, 09:33:10 PM
Strange, there are no changes from 96->97 that could relate to the slow updates. But the BOINC client is a low priority process, so when on of the computers is doing something the response time may go way down.
Al the extra things like graphic and gadgets do nothing when you don't use them.
In the BT setting set "Smart mode" and a "Maximum update time" of 120 (default 30).
Set no rules.
But a large amount of WU's may still cause some overhead, they must all be read. Especially when all computers have tasks close to completion.
Smart mode was on, changed max update time from 30 to 120, have no rules set. Decided to change the 2 WiFi clients to longer range adapters and that seems to have helpped a lot. V0.97 is running much better now (0.96 was starting to do the same as 0.97 but never as bad, probably just chance). Seems when the connection to any client is even a little flaky, BT chokes. Think that should be addressed. With the 10 CPUs (40 cores) and 16 GPUs running there's an average of 5 WUs a minute completing, so there's always many close to completion. That's probably quite a bit more than most BT users are seeing and I'm sure that's contributing to the problem. Thanks very much for the help, Fred!
[/quote]
I made a note to add an lean option in a version to come.
You probably already have a recent release version of BOINC The later version cause less history overhead.
And it's always the computer that is slow in responding that holds up the rest. Because every computer has to reply before the next command can be handled.
Normally at home I've a similar number of CPU's and GPU's running.
Of course closing BT to the taskbar gives the least overhead. Bringing it up with a hot-key is the best way.
Quote from: fred on January 19, 2011, 08:12:52 PM
Quote from: Beyond on January 19, 2011, 09:33:10 PM
Quote from: fred on January 19, 2011, 09:51:56 PM
Strange, there are no changes from 96->97 that could relate to the slow updates. But the BOINC client is a low priority process, so when on of the computers is doing something the response time may go way down.
[...]
But a large amount of WU's may still cause some overhead, they must all be read. Especially when all computers have tasks close to completion.
[...] Decided to change the 2 WiFi clients to longer range adapters and that seems to have helpped a lot. V0.97 is running much better now (...). Seems when the connection to any client is even a little flaky, BT chokes. Think that should be addressed. With the 10 CPUs (40 cores) and 16 GPUs running there's an average of 5 WUs a minute completing, so there's always many close to completion. That's probably quite a bit more than most BT users are seeing and I'm sure that's contributing to the problem.
I made a note to add an lean option in a version to come.
[...]
And it's always the computer that is slow in responding that holds up the rest. Because every computer has to reply before the next command can be handled.
Could maybe the single hosts be communicated asynchronously? And at each screen refresh, use the most recent data at hand from each computer?
From BT 0.96 thread:
Quote from: fred on January 17, 2011, 06:11:26 PM
Quote from: John McLeod VII on January 17, 2011, 05:27:51 PM
New messages prompt won't go away. If I visit the messages page, all of the messages should be marked as "not new". BOINC Tasks keeps flipping to the foreground (yuck, please do NOT do this as it takes the keyboard away from whatever I am doing) - this may be related to the new messages problem.
I store the last received message time. When a message with a later time stamp appears the icon message should popup.
But sometimes the client keeps refreshing the received time.
But when does BT comes to the foreground.
Today I've switched to my (already logged-in, a few days ago, if it matters) account on the family computer. BT was already running since days. (The last time I've visited it was possibly two days ago. If it matters.) I've seen BOINC Manager's greeting about new Notices, a minute later the BT's one, about 3 new notices. Clicked, read, closed BT. But the balloon kept coming and coming, even if immediately clicked, again in 4 seconds. If not clicked, then it disappeared itself with some delay and immediately appeared again. After around 30 :o I've ceased to count it, a bit more later even to click on it. But it still popped up a handful more times. Then it somehow left closed ::)
BOINC Manager displayed it just the one time.
Quote from: Pepo on January 19, 2011, 10:01:18 PM
Today I've switched to my (already logged-in, a few days ago, if it matters) account on the family computer. BT was already running since days. (The last time I've visited it was possibly two days ago. If it matters.) I've seen BOINC Manager's greeting about new Notices, a minute later the BT's one, about 3 new notices. Clicked, read, closed BT. But the balloon kept coming and coming, even if immediately clicked, again in 4 seconds. If not clicked, then it disappeared itself with some delay and immediately appeared again. After around 30 :o I've ceased to count it, a bit more later even to click on it. But it still popped up a handful more times. Then it somehow left closed ::)
BOINC Manager displayed it just the one time.
BT uses this time: HKEY_CURRENT_USER\Software\eFMer\BoincTasks\notices\localhost_time.
Only messages past this time should be counted.
But I've seen Windows loop the message for some reason, but normally only when the icon isn't shown. So hidden in Windows 7.
So testing is closing BT removing the key localhost_time and starting BT.
Quote from: fred on January 19, 2011, 10:28:57 PM
Quote from: Pepo on January 19, 2011, 10:01:18 PM
Today I've switched to my (already logged-in, a few days ago, if it matters) account on the family computer. BT was already running since days. (The last time I've visited it was possibly two days ago. If it matters.) I've seen BOINC Manager's greeting about new Notices, a minute later the BT's one, about 3 new notices. Clicked, read, closed BT. But the balloon kept coming and coming, even if immediately clicked, again in 4 seconds...
BT uses this time: HKEY_CURRENT_USER\Software\eFMer\BoincTasks\notices\localhost_time.
Only messages past this time should be counted.
But I've seen Windows loop the message for some reason, but normally only when the icon isn't shown. So hidden in Windows 7.
It happened in WinXPand BT's icon was visible, but this probably does not matter :)
Quote
So testing is closing BT removing the key localhost_time and starting BT.
OK, the next time I'll see this, I'll restart BT and if it will still be popping up balloons, I'll use the stop-deletekey-start woodoo method ;D
While BT kept running, I've updated the client 6.12.10->6.12.11. BT lost connection to it (it was the only connected client, local) and did not get it again even many minutes after the client (in service mode) started. I remember already having seen this in the past.
After clicking through all tabs, Tasks and Messages and Gadget were emptied, but Projects keep displayed (for a not-connected client). And, the tray tooltip still displays "machinename, Running: 2".
After a couple of minutes, I've "edited" the local computer's IP definition (it is localhost, I've just clicked on it and pressed Enter). The connection was promptly created.
From the log: 19 január 2011 - 23:02:56 GetLocalDirectory ---- C:\Program Files\eFMer\BoincTasks
19 január 2011 - 23:03:56 GetLocalDirectory ---- C:\Program Files\eFMer\BoincTasks
19 január 2011 - 23:29:22 Connect ---- Lost connection to: Host: localhost, vetroplach, Port: 31416
19 január 2011 - 23:40:43 BoincTasks is closing down all threads
19 január 2011 - 23:40:43 BoincTasks has closed down all threads
19 január 2011 - 23:40:45 BoincTasks version: 0.97
19 január 2011 - 23:40:45 Maximum number of computers: 60
19 január 2011 - 23:40:45 Computers ---- localhost get local password: (C:\Documents and Settings\All .... aplikací\BOINC\\gui_rpc_auth.cfg) blahblahblabla
19 január 2011 - 23:40:46 Computers ---- vetroplach , localhost
19 január 2011 - 23:40:46 Computers ---- pavilon6 , pavilon6
19 január 2011 - 23:40:46 GetLocalDirectory ---- C:\Program Files\eFMer\BoincTasks
19 január 2011 - 23:40:46 GetLocalDirectory ---- C:\Program Files\eFMer\BoincTasks
19 január 2011 - 23:40:46 Heap checking ---- WARNING: is enabled
19 január 2011 - 23:40:46 Heap checking ---- WARNING: is enabled
19 január 2011 - 23:40:46 Connect ---- Connected to: Host: localhost, vetroplach, Port: 31416, BOINC Version: 6.12.11
19 január 2011 - 23:40:47 GetLocalDirectory ---- C:\Program Files\eFMer\BoincTasks
19 január 2011 - 23:40:49 Connect, init ---- Host: pavilon6, pavilon6,Port: 31416, Couldn't resolve hostname
Filter based on plan class (too):
Primegrid just today deployed a new GPU application user-friendly name Cullen/Woodall Prime Search (Sieve) (cuda23) as a companion to the CPU version which has been around for awhile. The .exe's (on a 64-bit machine) are primegrid_gcwsieve_1.12_windows_intelx86__cuda23 and primegrid_gcwsieve_1.12_windows_x86_64 respectively.
With work onboard for both apps and the task filter off, individual tasks are correctly displayed as CPU or GPU. With the filter on, both types are combined into a single line entry :(
Request you refine the filter parameters to keep the two plan classes separate. You'll recall we went through this with SETI & SETI Beta tasks also, when both projects were using the same app version number ;) Note that in the present case, version numbers are also the same (1.12).
Quote from: Pepo on January 19, 2011, 10:49:18 PM
While BT kept running, I've updated the client 6.12.10->6.12.11. BT lost connection to it (it was the only connected client, local) and did not get it again even many minutes after the client (in service mode) started. I remember already having seen this in the past.
Interesting. I also updated localhost today 6.12.10->6.12.11 (all hosts actually) with BT still running. but BT had no trouble reconnecting to local
or remotes ??? But I'm not running in protected (service) mode.
To add to my connection problem: the reconnect time is IIRC set to 710 seconds. I guess this does matter.
Quote from: Pepo on January 19, 2011, 10:49:18 PM
While BT kept running, I've updated the client 6.12.10->6.12.11. BT lost connection to it (it was the only connected client, local) and did not get it again even many minutes after the client (in service mode) started. I remember already having seen this in the past.
After clicking through all tabs, Tasks and Messages and Gadget were emptied, but Projects keep displayed (for a not-connected client). And, the tray tooltip still displays "machinename, Running: 2".
After a couple of minutes, I've "edited" the local computer's IP definition (it is localhost, I've just clicked on it and pressed Enter). The connection was promptly created.
BT probably isn't able to reconnect, most likely not allowed. Changing the name is almost the same as closing down BT and restarting it.
You may want to disable the HEAP checking, it will slow down things considerably.
Quote from: jjwhalen on January 19, 2011, 10:51:14 PM
Filter based on plan class (too):
Primegrid just today deployed a new GPU application user-friendly name Cullen/Woodall Prime Search (Sieve) (cuda23) as a companion to the CPU version which has been around for awhile. The .exe's (on a 64-bit machine) are primegrid_gcwsieve_1.12_windows_intelx86__cuda23 and primegrid_gcwsieve_1.12_windows_x86_64 respectively.
With work onboard for both apps and the task filter off, individual tasks are correctly displayed as CPU or GPU. With the filter on, both types are combined into a single line entry :(
Request you refine the filter parameters to keep the two plan classes separate. You'll recall we went through this with SETI & SETI Beta tasks also, when both projects were using the same app version number ;) Note that in the present case, version numbers are also the same (1.12).
What's in the column Use and Application in both cases.
Quote from: fred on January 19, 2011, 11:20:17 PM
What's in the column Use and Application in both cases.
CPU App:
Application = 1.12 Cullen/Woodall Prime Search (Sieve)
Use = (null field)
GPU App:
Application = 1.12 Cullen/Woodall Prime Search (Sieve) (cuda23)
Use = 0.18C + 1NV (condensed Use column :))
Quote from: fred on January 19, 2011, 11:17:16 PM
Quote from: Pepo on January 19, 2011, 10:49:18 PM
While BT kept running, I've updated the client 6.12.10->6.12.11. BT lost connection to it (it was the only connected client, local) and did not get it again even many minutes after the client (in service mode) started.
After a couple of minutes, I've "edited" the local computer's IP definition (it is localhost, I've just clicked on it and pressed Enter). The connection was promptly created.
BT probably isn't able to reconnect, most likely not allowed. Changing the name is almost the same as closing down BT and restarting it.
What does mean "not allowed"? By what (rule)?
Quote from: Pepo on January 19, 2011, 10:49:18 PM
After clicking through all tabs, Tasks and Messages and Gadget were emptied, but Projects kept displayed (for a not-connected client). And, the tray tooltip still displays "machinename, Running: 2".
What about these?
Quote from: Pepo on January 19, 2011, 11:07:47 PM
To add to my connection problem: the reconnect time is IIRC set to 710 seconds. I guess this does matter.
In such situation in the past, I've also tried to lower the reconnect interval to few seconds (while BT kept running, without IP editing), with no influence. Could it be that the (710 sec, in my case) countdown still kept running without reset?
Quote from: fred on January 19, 2011, 11:17:16 PM
Quote from: Pepo on January 19, 2011, 10:49:18 PM
From the log: 19 január 2011 - 23:40:46 Heap checking ---- WARNING: is enabled
19 január 2011 - 23:40:46 Heap checking ---- WARNING: is enabled
You may want to disable the HEAP checking, it will slow down things considerably.
And, where? If I did it, I've already forgot when and where :-[
(looking for .....) I've found
<heap_check> mentioned just in
C:\Program Files\eFMer\BoincTasks\examples\log\log.xml, my
C:\Program Files\eFMer\BoincTasks\log\log.xml does not contain it. Or is the heap checking active by default and I have to explicitly set it =0 to disable?
Quote from: fred on January 19, 2011, 09:51:56 PM
Quote from: Beyond on January 19, 2011, 09:33:10 PM
Smart mode was on, changed max update time from 30 to 120, have no rules set. Decided to change the 2 WiFi clients to longer range adapters and that seems to have helpped a lot. V0.97 is running much better now (0.96 was starting to do the same as 0.97 but never as bad, probably just chance). Seems when the connection to any client is even a little flaky, BT chokes. Think that should be addressed. With the 10 CPUs (40 cores) and 16 GPUs running there's an average of 5 WUs a minute completing, so there's always many close to completion. That's probably quite a bit more than most BT users are seeing and I'm sure that's contributing to the problem. Thanks very much for the help, Fred!
I made a note to add an lean option in a version to come.
You probably already have a recent release version of BOINC The later version cause less history overhead.
Of course closing BT to the taskbar gives the least overhead. Bringing it up with a hot-key is the best way.
I had a mixture of BOINC 6.12.6, 6.12.10 and 6.12.11. Updated them all to 6.12.11 now and always have kept BT closed to the taskbar. BT resource use also seems better since updating to longer range adapters on the 2 WiFi boxes. Thanks again!
Quote from: Pepo on January 20, 2011, 09:11:28 AM
And, where? If I did it, I've already forgot when and where :-[
(looking for .....) I've found <heap_check> mentioned just in C:\Program Files\eFMer\BoincTasks\examples\log\log.xml, my C:\Program Files\eFMer\BoincTasks\log\log.xml does not contain it. Or is the heap checking active by default and I have to explicitly set it =0 to disable?
It's default off.
When it's switched on you get a message, 1 per computer.
C:\Program Files\eFMer\BoincTasks\log
BT needs to be restarted for this to become effective.
<log>
<options>
<heap_check> 0 </heap_check>
</options>
</log>
Quote from: fred on January 20, 2011, 07:32:25 PM
Quote from: Pepo on January 20, 2011, 09:11:28 AM
And, where? If I did it, I've already forgot when and where :-[
(looking for .....) I've found <heap_check> mentioned just in C:\Program Files\eFMer\BoincTasks\examples\log\log.xml, my C:\Program Files\eFMer\BoincTasks\log\log.xml does not contain it. Or is the heap checking active by default and I have to explicitly set it =0 to disable?
It's default off.
When it's switched on you get a message, 1 per computer.
C:\Program Files\eFMer\BoincTasks\log
I beg to disagree. It must be on per default. I had to create the non-existing log.xml with explicit
<heap_check> 0 </heap_check> to be able to disable it. (To clarify, on my two machines, one had and other one did not have
C:\Program Files\eFMer\BoincTasks\log\log.xml,
without mentioning
<heap_check>, and I had to define it and set to zero to switch it off.)
Never mind, it's already off.
Quote from: Pepo on January 20, 2011, 08:00:58 PM
I beg to disagree. It must be on per default. I had to create the non-existing log.xml with explicit <heap_check> 0 </heap_check> to be able to disable it. (To clarify, on my two machines, one had and other one did not have C:\Program Files\eFMer\BoincTasks\log\log.xml, without mentioning <heap_check>, and I had to define it and set to zero to switch it off.)
Never mind, it's already off.
You are of course right, this may cause BT to seriously slow down.
Fixed in 0.98
Quote from: fred on January 17, 2011, 06:20:01 PM
Quote from: Pepo on January 17, 2011, 02:00:20 PM
BOINC Manager's Notices tab displays corresponding notices with headlines "SETI@home/Enigma@Home: Notice from server". BT displays the same notices as translated (because of the existing matching translation strings in the .btlang's translation part), but BT is completely missing the projects' headlines - just a thick empty line appear there.
So, either the BOINC Manager is guessing and generating the missing headline (with "Notice"), or BT is loosing an existing headline. Any clue how is it actually?
The latest version probably has an additional flag, for the received from. In the todo list.
I've noticed something very similar, although not exactly the same - no translation involved, different feed entries' date and I've not found corresponding lines in
Messages.
BT displays:
Quote from: BoincTasks
Upload Server Issues
Our upload server is having disk problems. The whole project is down until this is fixed. It may be a day or two.
19.01.11 00:03 viac... (http://setiathome.berkeley.edu/forum_thread.php?id=62843)
Note that there
is some headline (in contrast to previously reported incomplete notices), but there is additionally an empty leading line as if the headline would indeed be missing.
BOINC Manager's corresponding version is again enhanced:
Quote from: BOINC Manager
SETI@home: Upload Server Issues
Our upload server is having disk problems. The whole project is down until this is fixed. It may be a day or two.
20.01.2011 20:49:39 more... (http://setiathome.berkeley.edu/forum_thread.php?id=62843)
It again contains also the project name. Well, let's see why does it stay disclosed...
QuoteThe tag for Messages / Show project / "Include general messages" menu entry is missing in .btlang files.
This got possibly overlooked? Neither in 0.98, nor in the ToDo list... (Something like "
<GroupPopUpMe ssages> <ShowGeneralMessages> Include general messages </> </GroupPopUpMessages>"?)
Quote from: Pepo on January 21, 2011, 03:06:11 PM
QuoteThe tag for Messages / Show project / "Include general messages" menu entry is missing in .btlang files.
This got possibly overlooked? Neither in 0.98, nor in the ToDo list... (Something like "<GroupPopUpMe ssages> <ShowGeneralMessages> Include general messages </> </GroupPopUpMessages>"?)
Everything should be in the to do list.
Quote from: jjwhalen on January 19, 2011, 10:51:14 PM
Filter based on plan class (too):
Primegrid just today deployed a new GPU application user-friendly name Cullen/Woodall Prime Search (Sieve) (cuda23) as a companion to the CPU version which has been around for awhile. The .exe's (on a 64-bit machine) are primegrid_gcwsieve_1.12_windows_intelx86__cuda23 and primegrid_gcwsieve_1.12_windows_x86_64 respectively.
With work onboard for both apps and the task filter off, individual tasks are correctly displayed as CPU or GPU. With the filter on, both types are combined into a single line entry :(
Request you refine the filter parameters to keep the two plan classes separate. You'll recall we went through this with SETI & SETI Beta tasks also, when both projects were using the same app version number ;) Note that in the present case, version numbers are also the same (1.12).
I was able to confirm this behavior today with Primegrid's
other CPU/GPU application: 1.38 Proth Prime Search (Sieve) [+ (cuda23)] (exe's primegrid_tpsieve_1.38_windows_x86_64 and primegrid_tpsieve_1.38_windows_intelx86__cuda23). The last time I ran both of these on the same host back in December, the filter was not combining the 2 types together. But then the version numbers were different (1.36
vs. 1.37). With both now being 1.38, they are getting combined just like the gcwsieve (Cullen/Woodall).
Normally I don't run the CPU app on the same platform with a GPU except during a Challenge, since it's a dreadful waste of CPU time. But recently for technical reasons I've been forced to download CPU workunits also and then abort them.
Strange that the task filter isn't picking up the difference in plan class, because I thought we had that problem worked out ???
[EDIT] Now running BT 0.98, but the behavior is the same. [/EDIT]
Quote from: jjwhalen on January 21, 2011, 11:35:24 PM
I was able to confirm this behavior today with Primegrid's other CPU/GPU application: 1.38 Proth Prime Search (Sieve) [+ (cuda23)] (exe's primegrid_tpsieve_1.38_windows_x86_64 and primegrid_tpsieve_1.38_windows_intelx86__cuda23). The last time I ran both of these on the same host back in December, the filter was not combining the 2 types together. But then the version numbers were different (1.36 vs. 1.37). With both now being 1.38, they are getting combined just like the gcwsieve (Cullen/Woodall).
Normally I don't run the CPU app on the same platform with a GPU except during a Challenge, since it's a dreadful waste of CPU time. But recently for technical reasons I've been forced to download CPU workunits also and then abort them.
Strange that the task filter isn't picking up the difference in plan class, because I thought we had that problem worked out ???
[EDIT] Now running BT 0.98, but the behavior is the same. [/EDIT]
Fixed in V 0.99