This is a new App for MacOS PPC that uses the same "App switching" mechanism we have in the 4.38 Linux App, here to distinguish between AltiVec- and G3 machines.
My first result was validated. G5, OS X10.4, ECC RAM, Boinc 6.1.16.
One done with 4.43 on a G3 iMac. It has not been validated yet: amazingly, I'm actually waiting for my wingman. That doesn't happen too often with this machine.
Running fine on a G4 400mhz, seems to be faster than the 4.29 but it'll take a while until the RAC verifies that.
RAC is a very inaccurate way to judge the speed of the new app. In your case you are going to get around a 20% to 30% speedup on average but you are not going to see that with the next couple of tasks and you may end up quite disappointed for a while. If you look at the next task after the one that is currently crunching, it's going to be actually slower than the last task you crunched with the old app. This is because of the cyclic nature of crunch times and good old lady luck.
Your last 4.29 task had a sequence number (seq#) of 581 which is not very far away from seq#s where you would expect the fastest crunching tasks to be for your search frequency. Your first 4.43 task had a seq# of 482 which is heading towards the area where the slowest crunching tasks will be. Your next two tasks have seq#s of 467 and 440. That last one is going to be about as slow as you will get. As you continue to get lower seq#s, you will start to see a good speed improvement and if you get one around 330, it will be a very fast one and then you will see a better indication of the capabilities of the new app.
Ok two another WU finished and validated (about 52 000 seconds for 236,81 Credits)
PS: so you see Bernd that altivec optimization boost G5 performance in the same way compared to G4. (the G5 FPU change nothing in altivec optimization)
One task finished and validated with 4.43 and 5.10.20 of Boinc client. Time seemed pretty good considering I have the G5 on automatic processor setting to keep the noise down since I'm using it for internet radio in my bedroom.
Ok a second WU reported ; 52
)
Ok a second WU reported ; 52 000 seconds so it's a long WU for 236 credits so it seems to be a bit slower (27-28%) than previous estimation of 30 %
waiting validation of this two WU .
ok the seconds WU of 52 000
)
ok the seconds WU of 52 000 seconds is validated ! waiting for the first WU
ok the first WU is validated
)
ok the first WU is validated too.
RE: This is a new App for
)
My first result was validated. G5, OS X10.4, ECC RAM, Boinc 6.1.16.
http://einsteinathome.org/task/96162723
Running fine on a G4 400mhz,
)
Running fine on a G4 400mhz, seems to be faster than the 4.29 but it'll take a while until the RAC verifies that.
One done with 4.43 on a G3
)
One done with 4.43 on a G3 iMac. It has not been validated yet: amazingly, I'm actually waiting for my wingman. That doesn't happen too often with this machine.
RE: Running fine on a G4
)
RAC is a very inaccurate way to judge the speed of the new app. In your case you are going to get around a 20% to 30% speedup on average but you are not going to see that with the next couple of tasks and you may end up quite disappointed for a while. If you look at the next task after the one that is currently crunching, it's going to be actually slower than the last task you crunched with the old app. This is because of the cyclic nature of crunch times and good old lady luck.
Your last 4.29 task had a sequence number (seq#) of 581 which is not very far away from seq#s where you would expect the fastest crunching tasks to be for your search frequency. Your first 4.43 task had a seq# of 482 which is heading towards the area where the slowest crunching tasks will be. Your next two tasks have seq#s of 467 and 440. That last one is going to be about as slow as you will get. As you continue to get lower seq#s, you will start to see a good speed improvement and if you get one around 330, it will be a very fast one and then you will see a better indication of the capabilities of the new app.
Cheers,
Gary.
A second work unit done with
)
A second work unit done with 4.43. This one has been validated.
Ok two another WU finished
)
Ok two another WU finished and validated (about 52 000 seconds for 236,81 Credits)
PS: so you see Bernd that altivec optimization boost G5 performance in the same way compared to G4. (the G5 FPU change nothing in altivec optimization)
One task finished and
)
One task finished and validated with 4.43 and 5.10.20 of Boinc client. Time seemed pretty good considering I have the G5 on automatic processor setting to keep the noise down since I'm using it for internet radio in my bedroom.