For Windows: By default they go into C:\\Documents and Settings\\All Users\\BOINC Data\\
I must say, if you want to test this version, why aren't you looking at what the installer does? BOINC 6 separates the data from the executables to make sure it's compatible with 'higher security' operating systems such as Vista.
This one finished a little while ago in 35,361.01 seconds. That's 16% faster than v4.36 - comparing to resultid=96594697 which took 41,177.77 seconds on the same host. (Now, let's hope it validates.)
This one finished a little while ago in 35,361.01 seconds. That's 16% faster than v4.36 - comparing to resultid=96594697 which took 41,177.77 seconds on the same host. (Now, let's hope it validates.)
Hi!
The speed difference is only caused by the "natural" variations of runtime for different WUs. When executed on a modern SSE capable PCs, the 4.46 app should behave exactly like the 4.36 app, because it's actually executing the exact same 4.36 executable. So speed is not the focus of this beta test, but the switching mechanism that will run either the "optimized" or "legacy-PC" app version. So likewise if you see a speed decrease of the 4.36, please don't stop testing this app, 4.36 would not have been faster.
Of course it will validate ;-). There is absolutely nothing unusual about the two crunch times. The new quick result had a seq# of __96 which is virtually right on one of the fastest possible parts of the cycle. The old slow result had a seq# of __207 which is quite close to one of the slowest parts of the cycle.
I've seen a very disturbing BOINC/Einstein behavior on one host running 4.46 in the last 24 hours. It quite possibly has nothing to do with 4.46, or even 4.36, or perhaps even BOINC, but I thought I'd post it here in case any others see something similar.
I noticed yesterday afternoon that my copy of boinc (I was running 5.10.30) was starting up four results of Einstein 4.46 about once a minute. The pattern was that they would run until accumulating between 20 and 60 CPU seconds (which on my Q6600 host means they never got out of the high I/O phase at startup) then disappeared from my task list simultaneously (I use Process Explorer, I assume I'd have seen the same thing in regular Windows Task Manager). After something like 20 seconds of down time (boincmgr and boinc still active) a fresh set of four Einsteins would start, only to die together again.
As I generally could not gain access to the user screens of BOINC Manager, I could not make much intervention or diagnosis.
As on reboot Windows XP decided to chkdsk my boot drive, and found a couple of screens of errors to fix, I assumed this was the outcome of a system problem, which had put one or more files in the BOINC directories into a bad state which the code happened not to recover from gracefully.
I moved all my BOINC stuff to a backup directory, and did a fresh install, starting with BOINC 5.10.45, and "signing up" afresh for my two projects, SETI and Einstein. As the stock science aps ran OK for an hour or so (no more 20 seconds startup to death cycle) I stopped BOINC and installed the same enhanced aps I had been running (Windows 4.46 for Einstein, AK_v8_win_SSSE3x for SETI).
The newly installed enhanced aps picked up in-process work units started with the stock aps, and appeared to run normally for the couple of hours until I went to bed, but I woke this morning to the same minute-long startup and crash cycle, which persisted through a full cold reboot of the host.
I did chance on a moment when I was allowed into the BOINC mgr user interface, and used it to suspend Einstein. So far four SETI's have run without events for over twenty minutes.
I don't have any really good ideas either of how to test the questionable hypothesis that 4.46 is involved in my troubles or to restore my host to good working order. If anyone has suggestions on the latter that are off thread topic, I'd be happy to get advice by Private Message. I have some not so good ideas for probing 4.46 involvement which I'll act on after a couple of successful SETI task cycles (a bit under three hours from now).
How can I overcome the message reached daily quota of 4 results so i can use the new app before tomorrow?
thanks in advance
The only way to increase your quota is to return good/valid WUs. The fact that you are restricted to 4 WUs per day indicates you have a problem with your system and it's returning bad results. Find/fix the problem before you attempt using the new app.
I got my first workunits and started crunching them with the new 4.46 app. It seems faster than the 4.36 app by 20%. It crunched almost 3% in 10 minutes, which means some 4 hours while with the previus app it took around five hours.
Gravity increases significantly in Autumn, because apples fall in large numbers during that time!
I got my first workunits and started crunching them with the new 4.46 app. It seems faster than the 4.36 app by 20%. ...
That would be really surprising given the fact that on a SSE capable PC, 4.46 is actually calling the 4.36 app executable :-). The switching mechanism is what this beta test is about, not any speed improvements.
.............
As on reboot Windows XP decided to chkdsk my boot drive, and found a couple of screens of errors to fix, I assumed this was the outcome of a system problem, which had put one or more files in the BOINC directories into a bad state which the code happened not to recover from gracefully.
..............
quote]
@archae86,
if it isn't a problem with the new App, it could be a problem with the Harddisk or Power Supply. My last failures with Harddisk looks like similar.
24h/7 Cruncher, HD's ~3Years old Samsung. HD-Led burning allways, Computer frozen. Reset, System starts with chkdsk and seems to run until next freeze. Some weeks later, after reseting System, Windows makes chkdsk and drives directly in a Blue-Screen. Reset, lost of HD in Bios. Waiting some time (1h or so) System boots up normaly until.... Copying the HD on a new one (with Acronis, ignoring all read errors while copying) System works fine again.
No HD problems in the smart event list, or so one. All HD's shown a similar failure picture (dont know the right english expression)
Another Q6600 B3 starts crashing WU's, but with no freeze but close the actual WU's. Opening the Power Supply, blown up Elko in the 12V Lane (CPU-Power). Because all of my Q6600 65nm are undervolted, only one defective elko results in a computation error.
RE: So... Where do the
)
For Windows: By default they go into C:\\Documents and Settings\\All Users\\BOINC Data\\
I must say, if you want to test this version, why aren't you looking at what the installer does? BOINC 6 separates the data from the executables to make sure it's compatible with 'higher security' operating systems such as Vista.
Do a search for "BOINC Data".
resultid=97045856 This one
)
resultid=97045856
This one finished a little while ago in 35,361.01 seconds. That's 16% faster than v4.36 - comparing to resultid=96594697 which took 41,177.77 seconds on the same host. (Now, let's hope it validates.)
RE: resultid=97045856 This
)
Hi!
The speed difference is only caused by the "natural" variations of runtime for different WUs. When executed on a modern SSE capable PCs, the 4.46 app should behave exactly like the 4.36 app, because it's actually executing the exact same 4.36 executable. So speed is not the focus of this beta test, but the switching mechanism that will run either the "optimized" or "legacy-PC" app version. So likewise if you see a speed decrease of the 4.36, please don't stop testing this app, 4.36 would not have been faster.
CU
Bikeman
RE: (Now, let's hope it
)
Of course it will validate ;-). There is absolutely nothing unusual about the two crunch times. The new quick result had a seq# of __96 which is virtually right on one of the fastest possible parts of the cycle. The old slow result had a seq# of __207 which is quite close to one of the slowest parts of the cycle.
Cheers,
Gary.
I've seen a very disturbing
)
I've seen a very disturbing BOINC/Einstein behavior on one host running 4.46 in the last 24 hours. It quite possibly has nothing to do with 4.46, or even 4.36, or perhaps even BOINC, but I thought I'd post it here in case any others see something similar.
I noticed yesterday afternoon that my copy of boinc (I was running 5.10.30) was starting up four results of Einstein 4.46 about once a minute. The pattern was that they would run until accumulating between 20 and 60 CPU seconds (which on my Q6600 host means they never got out of the high I/O phase at startup) then disappeared from my task list simultaneously (I use Process Explorer, I assume I'd have seen the same thing in regular Windows Task Manager). After something like 20 seconds of down time (boincmgr and boinc still active) a fresh set of four Einsteins would start, only to die together again.
As I generally could not gain access to the user screens of BOINC Manager, I could not make much intervention or diagnosis.
As on reboot Windows XP decided to chkdsk my boot drive, and found a couple of screens of errors to fix, I assumed this was the outcome of a system problem, which had put one or more files in the BOINC directories into a bad state which the code happened not to recover from gracefully.
I moved all my BOINC stuff to a backup directory, and did a fresh install, starting with BOINC 5.10.45, and "signing up" afresh for my two projects, SETI and Einstein. As the stock science aps ran OK for an hour or so (no more 20 seconds startup to death cycle) I stopped BOINC and installed the same enhanced aps I had been running (Windows 4.46 for Einstein, AK_v8_win_SSSE3x for SETI).
The newly installed enhanced aps picked up in-process work units started with the stock aps, and appeared to run normally for the couple of hours until I went to bed, but I woke this morning to the same minute-long startup and crash cycle, which persisted through a full cold reboot of the host.
I did chance on a moment when I was allowed into the BOINC mgr user interface, and used it to suspend Einstein. So far four SETI's have run without events for over twenty minutes.
I don't have any really good ideas either of how to test the questionable hypothesis that 4.46 is involved in my troubles or to restore my host to good working order. If anyone has suggestions on the latter that are off thread topic, I'd be happy to get advice by Private Message. I have some not so good ideas for probing 4.46 involvement which I'll act on after a couple of successful SETI task cycles (a bit under three hours from now).
How can I overcome the
)
How can I overcome the message reached daily quota of 4 results so i can use the new app before tomorrow?
thanks in advance
Gravity increases significantly in Autumn, because apples fall in large numbers during that time!
RE: How can I overcome the
)
The only way to increase your quota is to return good/valid WUs. The fact that you are restricted to 4 WUs per day indicates you have a problem with your system and it's returning bad results. Find/fix the problem before you attempt using the new app.
Seti Classic Final Total: 11446 WU.
I got my first workunits and
)
I got my first workunits and started crunching them with the new 4.46 app. It seems faster than the 4.36 app by 20%. It crunched almost 3% in 10 minutes, which means some 4 hours while with the previus app it took around five hours.
Gravity increases significantly in Autumn, because apples fall in large numbers during that time!
RE: I got my first
)
That would be really surprising given the fact that on a SSE capable PC, 4.46 is actually calling the 4.36 app executable :-). The switching mechanism is what this beta test is about, not any speed improvements.
CU
Bikeman
RE: ............. As on
)