As I head for bed, I'm 33 minutes into a result and am at 3.05% complete. Extrapolating out at the current rate to completion would put it at about 1082 minutes, or roughly 18.03 hours. If that ratio holds, then ftol() is rather slow compared to (I'm assuming) modf() that was being used by the "patched" 4.17 app. It would be nearly midway between the penalty of the "non-patched" 4.17 (approx 21.5 hours) and the "patched" app (15.1 hours).
Furthermore, if that holds true, AMD systems would still be being penalized by 20-25% for using Windows as an OS, as the "patched" way came significantly closer to parity (5-10%)
I know...I know... Don't look a gift horse in the mouth, etc, etc, etc...
Just stating facts...
IMO, YMMV, etc, etc, etc...
Edit: 44 minutes, 4.079% -> extrapolate -> 1078.7 minutes, so it is getting better, although at the end of the run of the last unit I did with the patched app, it took longer than the estimate.
Just re-attached the Windows incarnation of my Core Duo (you know, the one with the odd 40% Windows penalty) to Einstein and downloaded the new app. Let's see how this box does.
Just re-attached the Windows incarnation of my Core Duo (you know, the one with the odd 40% Windows penalty) to Einstein and downloaded the new app. Let's see how this box does.
Just noticed (on the BOINCStats page, too)... It must somehow have to do with attaching QMC on my Dad's box first, thus signing up under his account name... but while that always turned up in the stats, it never showed in the sig until today. Don't ask me why it does now.
Times have continued to improve, but not to the point of the "patched" app yet.
Currently I'm at 64.145% completed in 627 minutes (10h 27m).
100 / 64.145 * 627 = 977.47 minutes or ~977.5
977.5 * 60 = 58,650 seconds
The "patched" app did a result of the same frequency from the same datapack in 54,426 seconds. (58,650 - 54,426) / 60 = 70.4 minutes slower, or roughly 8% slower.
I haven't gotten to the point where the patched app slowed down either. I anticipate that this app will also slow down in a little while. Current completion estimate in the manager window is in 6h 15m for a total runtime of about 16h 51 minutes (1011 minutes).
First WU finished. The other results are done with the patched app. So the beta app is about 8% slower than the patched app and about 22% faster than the offical app.
First WU finished. The other results are done with the patched app. So the beta app is about 8% slower than the patched app and about 22% faster than the offical app.
cu,
Michael
As an added bonus, you cross-validated with Darwin...
As I head for bed, I'm 33
)
As I head for bed, I'm 33 minutes into a result and am at 3.05% complete. Extrapolating out at the current rate to completion would put it at about 1082 minutes, or roughly 18.03 hours. If that ratio holds, then ftol() is rather slow compared to (I'm assuming) modf() that was being used by the "patched" 4.17 app. It would be nearly midway between the penalty of the "non-patched" 4.17 (approx 21.5 hours) and the "patched" app (15.1 hours).
Furthermore, if that holds true, AMD systems would still be being penalized by 20-25% for using Windows as an OS, as the "patched" way came significantly closer to parity (5-10%)
I know...I know... Don't look a gift horse in the mouth, etc, etc, etc...
Just stating facts...
IMO, YMMV, etc, etc, etc...
Edit: 44 minutes, 4.079% -> extrapolate -> 1078.7 minutes, so it is getting better, although at the end of the run of the last unit I did with the patched app, it took longer than the estimate.
Installed it on my Pentium M
)
Installed it on my Pentium M under XP this morning.
So I guess first one to generate an error with a meaningful debug output wins :-).
Any idea what to try? Simulating any kind of common network problem might be useful. Using the screen-saver might also lead to some results.
CU
BRM
Just re-attached the Windows
)
Just re-attached the Windows incarnation of my Core Duo (you know, the one with the odd 40% Windows penalty) to Einstein and downloaded the new app. Let's see how this box does.
RE: Just re-attached the
)
BTW, your Sig has changed in mysterious ways
CU
BRM
Just noticed (on the
)
Just noticed (on the BOINCStats page, too)... It must somehow have to do with attaching QMC on my Dad's box first, thus signing up under his account name... but while that always turned up in the stats, it never showed in the sig until today. Don't ask me why it does now.
RE: BTW, your Sig has
)
LOL! That was a very polite way of putting it... ;-)
New update: Times have
)
New update:
Times have continued to improve, but not to the point of the "patched" app yet.
Currently I'm at 64.145% completed in 627 minutes (10h 27m).
100 / 64.145 * 627 = 977.47 minutes or ~977.5
977.5 * 60 = 58,650 seconds
The "patched" app did a result of the same frequency from the same datapack in 54,426 seconds. (58,650 - 54,426) / 60 = 70.4 minutes slower, or roughly 8% slower.
I haven't gotten to the point where the patched app slowed down either. I anticipate that this app will also slow down in a little while. Current completion estimate in the manager window is in 6h 15m for a total runtime of about 16h 51 minutes (1011 minutes).
First WU finished. The other
)
First WU finished. The other results are done with the patched app. So the beta app is about 8% slower than the patched app and about 22% faster than the offical app.
cu,
Michael
Both apps accelerate till
)
Both apps accelerate till they reach ~90%, then they slow down. So I guess you will also end up with -8%.
cu,
Michael
RE: First WU finished. The
)
As an added bonus, you cross-validated with Darwin...