Just finished up my first couple of new WUs (MacPro, Intel).
Crunch time is down 4%, credit is down 19%. (Averages)
Slight adjustment needed.
At least for Core2 boxes like yours, the average crunch time is definitely down by more than just 4%, should actually be close to 40% or 50% reduction. Have a little patience and see runtime going down for other WUs (runtime variation is much greater now).
Just finished up my first couple of new WUs (MacPro, Intel).
Crunch time is down 4%, credit is down 19%. (Averages)
Slight adjustment needed.
At least for Core2 boxes like yours, the average crunch time is definitely down by more than just 4%, should actually be close to 40% or 50% reduction. Have a little patience and see runtime going down for other WUs (runtime variation is much greater now).
CU
Bikeman
Not Core2, Intel X5150 Xeon's in a Mac Pro.
Average of the first 8: they are averaging almost exactly the same times for the S5 WU's that they did for the S4's. That makes the 20% credit reduction hurt a little.
I'm thinking of pausing the Mac WU's and booting the machine under Linux to see if they run better.
Not Core2, Intel X5150 Xeon's in a Mac Pro.
Average of the first 8: they are averaging almost exactly the same times for the S5 WU's that they did for the S4's. That makes the 20% credit reduction hurt a little.
I'm thinking of pausing the Mac WU's and booting the machine under Linux to see if they run better.
Xeon is the marketing name, it's a Core2 generation microarchitecture CPU nevertheless.
The Mac OS apps are usually the fastest, so I don't have high hopes you'll get better performance under Linux. The credit output is under constant observation and once there are enought datapoints in for various CPUs AND various WU frequency ranges, Bernd will make a decision if/how to re-calibrate. At the moment it looks to me like there will be some adjustments. Please bear with us :-)
CU
Bikeman
EDIT: After inspecting your results, it looks like you were lucky to have a long run of results near the runtime minimun near the end of S5R4, which makes the comparison to S5R5 credit a bit "unfair". Note that some of your S5R4 results took longer than 28000 seconds for a fixed 222 credits, and they were not even at the expected maximum of runtime for that frequency range. As I said, more datapoints are needed.
Not Core2, Intel X5150 Xeon's in a Mac Pro.
Average of the first 8: they are averaging almost exactly the same times for the S5 WU's that they did for the S4's. That makes the 20% credit reduction hurt a little.
I'm thinking of pausing the Mac WU's and booting the machine under Linux to see if they run better.
Xeon is the marketing name, it's a Core2 generation microarchitecture CPU nevertheless.
The Mac OS apps are usually the fastest, so I don't have high hopes you'll get better performance under Linux. The credit output is under constant observation and once there are enought datapoints in for various CPUs AND various WU frequency ranges, Bernd will make a decision if/how to re-calibrate. At the moment it looks to me like there will be some adjustments. Please bear with us :-)
CU
Bikeman
EDIT: After inspecting your results, it looks like you were lucky to have a long run of results near the runtime minimun near the end of S5R4, which makes the comparison to S5R5 credit a bit "unfair". Note that some of your S5R4 results took longer than 28000 seconds for a fixed 222 credits, and they were not even at the expected maximum of runtime for that frequency range. As I said, more datapoints are needed.
I agree, trying to extrapolate data from that few points is dangerous.
Xeon is the marketing name, it's a Core2 generation microarchitecture CPU nevertheless.
The Mac OS apps are usually the fastest, so I don't have high hopes you'll get better performance under Linux. The credit output is under constant observation and once there are enought datapoints in for various CPUs AND various WU frequency ranges, Bernd will make a decision if/how to re-calibrate. At the moment it looks to me like there will be some adjustments. Please bear with us :-)
A lot of recalibration I would say. I did my own comparison, and I have my doubt about Mac OS apps being the fastest if you were to compare Windows and Linux crunch boxes on top 20 hosts, particularly with R5s... Peanut's Xeon for example, his R5 units are worst off than R4s. I'll soon see my RAC spiraling downwards...
It is early in the SR5 game. I haven't been watching things too close lately. But, I think my core duo (not core 2 duo) mac minis are running the new WUs better than my mac pro is. The mac minis seem to be running SR5 WUs faster than they ran the SR4 ones. My Pro does seem to be worse off with the new WUs. Kind of odd that the minis like them and the Pro seems not to. Oh well. With the variations in run times its hard to be sure anything I say is right.
Just a rough guess here, but it seems like a 10% boost in credit awards would keep cr/hr rates in line with S5R4 rates... I was getting about 25/hr with S5R4 and about 22/hr now...
It makes me depressive when I see the credits going down. On my quad it went from about 33 cr/h (S5R4)over 30 cr/h and 27 cr/h to 25 cr/h. look here and ongoing
The runtimes went down also.
It makes me depressive when I see the credits going down. On my quad it went from about 33 cr/h (S5R4)over 30 cr/h and 27 cr/h to 25 cr/h. look here and ongoing
The runtimes went down also.
How is the way the creditcalculation is made?
I dont care if the credit is down a bit, but i do care about the runtime variation and the credit variation between WUs. Heres the seconds/credit for some of your results:
It makes me depressive when I see the credits going down. On my quad it went from about 33 cr/h (S5R4)over 30 cr/h and 27 cr/h to 25 cr/h. look here and ongoing
The runtimes went down also.
How is the way the creditcalculation is made?
There are sooooo many things that affect how many resources your computer gives to Boinc that there are many things you need to do first. Like rebooting the pc, checking for spyware, etc, etc, etc. Each of those things can slow down your pc. In Windows if you don't reboot about once a month it tends to have too many stuck resources that Windows isn't releasing, which in turn gives Boinc less and can cause a slowdown in crunching time.
SR5, 5.01 Credits down from SR4
)
At least for Core2 boxes like yours, the average crunch time is definitely down by more than just 4%, should actually be close to 40% or 50% reduction. Have a little patience and see runtime going down for other WUs (runtime variation is much greater now).
CU
Bikeman
RE: RE: Just finished up
)
Not Core2, Intel X5150 Xeon's in a Mac Pro.
Average of the first 8: they are averaging almost exactly the same times for the S5 WU's that they did for the S4's. That makes the 20% credit reduction hurt a little.
I'm thinking of pausing the Mac WU's and booting the machine under Linux to see if they run better.
RE: Not Core2, Intel X5150
)
Xeon is the marketing name, it's a Core2 generation microarchitecture CPU nevertheless.
The Mac OS apps are usually the fastest, so I don't have high hopes you'll get better performance under Linux. The credit output is under constant observation and once there are enought datapoints in for various CPUs AND various WU frequency ranges, Bernd will make a decision if/how to re-calibrate. At the moment it looks to me like there will be some adjustments. Please bear with us :-)
CU
Bikeman
EDIT: After inspecting your results, it looks like you were lucky to have a long run of results near the runtime minimun near the end of S5R4, which makes the comparison to S5R5 credit a bit "unfair". Note that some of your S5R4 results took longer than 28000 seconds for a fixed 222 credits, and they were not even at the expected maximum of runtime for that frequency range. As I said, more datapoints are needed.
RE: RE: Not Core2, Intel
)
I agree, trying to extrapolate data from that few points is dangerous.
RE: Xeon is the marketing
)
A lot of recalibration I would say. I did my own comparison, and I have my doubt about Mac OS apps being the fastest if you were to compare Windows and Linux crunch boxes on top 20 hosts, particularly with R5s... Peanut's Xeon for example, his R5 units are worst off than R4s. I'll soon see my RAC spiraling downwards...
It is early in the SR5 game.
)
It is early in the SR5 game. I haven't been watching things too close lately. But, I think my core duo (not core 2 duo) mac minis are running the new WUs better than my mac pro is. The mac minis seem to be running SR5 WUs faster than they ran the SR4 ones. My Pro does seem to be worse off with the new WUs. Kind of odd that the minis like them and the Pro seems not to. Oh well. With the variations in run times its hard to be sure anything I say is right.
Just a rough guess here, but
)
Just a rough guess here, but it seems like a 10% boost in credit awards would keep cr/hr rates in line with S5R4 rates... I was getting about 25/hr with S5R4 and about 22/hr now...
It makes me depressive when I
)
It makes me depressive when I see the credits going down. On my quad it went from about 33 cr/h (S5R4)over 30 cr/h and 27 cr/h to 25 cr/h.
look here and ongoing
The runtimes went down also.
How is the way the creditcalculation is made?
RE: It makes me depressive
)
I dont care if the credit is down a bit, but i do care about the runtime variation and the credit variation between WUs. Heres the seconds/credit for some of your results:
TaskID WU-ID CPUtime CreditGranted CPUSecPrCredit
115124389 47237042 14925.04 92.03 162.18
115120633 47179809 12520.16 102.97 121.59
115116352 47233275 14404.18 92.52 155.69
115113612 47216963 15116.29 93.51 161.65
115106226 47228509 14584.95 92.89 157.01
115074043 47213773 13856.07 93.97 147.45
115063739 47209043 14847.05 94.21 157.60
115043841 47199929 12279.29 95.83 128.14
115038561 47197496 11525.28 96.76 119.11
115031899 47190761 16104.83 98.11 164.15
115030568 47193797 15750.27 97.08 162.24
115030552 47193789 14310.52 97.42 146.90
115020998 47189336 12151.87 98.47 123.41
115020811 47189245 14968.00 98.83 151.45
115019898 47188807 15761.61 99.21 158.87
115016792 47187359 11606.37 99.59 116.54
115015949 47186962 16561.00 99.99 165.63
115012582 47185399 13464.71 100.39 134.12
Team Philippines
RE: It makes me depressive
)
There are sooooo many things that affect how many resources your computer gives to Boinc that there are many things you need to do first. Like rebooting the pc, checking for spyware, etc, etc, etc. Each of those things can slow down your pc. In Windows if you don't reboot about once a month it tends to have too many stuck resources that Windows isn't releasing, which in turn gives Boinc less and can cause a slowdown in crunching time.