I should think a fairly easy remedy—at least for the shorter term—would be to extend the deadlines. It doesn’t seem very reasonable to expect the S5R2 tasks, which apparently take several times as long to process as the S5R1 & S5RI, to get returned on the same schedule.
The problem is that doing so also increases the delay for abandoned WUs getting a quorum. That's a much more serious problem towards 'customer statisfaction' than the loss of a small number of boxes so old that thier total contribution is essentially negligable.
Well so what? It doesn't matter how long a result stays pending to your performance metrics over the long haul.
So why should my old timers be eliminated from participation just to please the false perceptions of the RAC hounds? They may be slow, but they're on 24/7 because of their primary function and turn in results just as valid and useful as the fastest C2Q.
LOL... that's why I sleazed in the followup edit's.
BTW, what genus are your K6's?
Ooooh .... err.. K6 500Mhz with '3D' I think - I call it "Baby" 'cos it's the slowest on the patch, and is on a baby AT board. It's a legacy/homage/nostalgia job really, the first one I got after my really fast first Pentium 90!! I never had a 486, but stepped up from a 33 Mhz 386 WITH, mind you, a 387 FP co-processor by it's side ... ahhh, those were the days of linear address spaces and easy-peasy segment:offset manipulations ..... :-)
Quote:
I've never heard of anyone getting that day off. I think they have to settle on being off Labor Day like the "common" man.
In beaurocratic Canberra, they never schedule any meetings on Wednesdays ...... just so don't they spoil two weekends in a row ...... :-)
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
LOL... that's why I sleazed in the followup edit's.
BTW, what genus are your K6's?
Ooooh .... err.. K6 500Mhz with '3D' I think - I call it "Baby" 'cos it's the slowest on the patch, and is on a baby AT board. It's a legacy/homage/nostalgia job really, the first one I got after my really fast first Pentium 90!! I never had a 486, but stepped up from a 33 Mhz 386 WITH, mind you, a 387 FP co-processor by it's side ... ahhh, those were the days of linear address spaces and easy-peasy segment:offset manipulations ..... :-)
Quote:
I've never heard of anyone getting that day off. I think they have to settle on being off Labor Day like the "common" man.
In beaurocratic Canberra, they never schedule any meetings on Wednesdays ...... just so don't they spoil two weekends in a row ...... :-)
Cheers, Mike.
Ah OK, that would make it a K6-2/500. I have three of them, so feel free to use them as comparo units for yours. I also have a -III/450 I'm looking for the right Super 7 board for to round out my K6 family.
Interestingly I also have a Pentium 90, but its 'resting' at the moment. The reason is I haven't thought of something for it to do yet, but I will. ;-)
Hmmm, sounds like there isn't much difference between Canberra and DC, or any state capital here in the States for that matter. Especially once golf season starts! :-)
LOL... that's why I sleazed in the followup edit's.
BTW, what genus are your K6's?
Ooooh .... err.. K6 500Mhz with '3D' I think - I call it "Baby" 'cos it's the slowest on the patch, and is on a baby AT board. It's a legacy/homage/nostalgia job really, the first one I got after my really fast first Pentium 90!! I never had a 486, but stepped up from a 33 Mhz 386 WITH, mind you, a 387 FP co-processor by it's side ... ahhh, those were the days of linear address spaces and easy-peasy segment:offset manipulations ..... :-)
Quote:
I've never heard of anyone getting that day off. I think they have to settle on being off Labor Day like the "common" man.
In beaurocratic Canberra, they never schedule any meetings on Wednesdays ...... just so don't they spoil two weekends in a row ...... :-)
Cheers, Mike.
Ah OK, that would make it a K6-2/500. I have three of them, so feel free to use them as comparo units for yours. I also have a -III/450 I'm looking for the right Super 7 board for to round out my K6 family.
Interestingly I also have a Pentium 90, but its 'resting' at the moment. The reason is I haven't thought of something for it to do yet, but I will. ;-)
Hmmm, sounds like there isn't much difference between Canberra and DC, or any state capital here in the States for that matter. Especially once golf season starts! :-)
Alinator
Ah, the K-6s. Great chips at a fair price, I haven't bought an Intel in probably close to 10 years and this is the very first time that I feel I might be at a disadvantage, and you know what, I don't care. AMD has saved computer users millions if not billions of dollars and I'm not about to let this little glitch, that after I have gotten more results back, may be less than 5%.
Gary
In the beginning the Universe was created. This has made a lot of people very angry and been widely regarded as a bad move.....Douglas Adams
I think we just got a definition of the word 'pedantic', by example. :-)
Damn, exposed again! I must have overestimated the amount of cover my use of “wannabe� would provide.
You should be grateful that I only perform by invitation. ;)
Quote:
As Ambassador Carlos said, "...We all depend on language for so very much, but are any of us its master?". ;-)
“ ‘When I use a word,’ Humpty Dumpty said, in rather a scornful tone, ‘it means just what I choose it to mean—neither more nor less.’
“ ‘The question is,’ said Alice, ‘whether you can make words mean so many different things.’
“ ‘The question is,’ said Humpty Dumpty, ‘which is to be the master—that’s all.’ �
—Lewis Carroll, Through the Looking Glass
Quote:
LOL, my second K6/300 (which is my personal BDC) just started its first S5R2. Talk about pessimistic, it's telling me it thinks it's going 1640 hours to run it! The PDC K6/300 is saying I can do one in 575.
My Mac G4/733 is looking at about sixty hours, based on a task that’s about one-quarter complete. That’s still within my comfort zone, although having a 20% resource share for E@h (IIRC) it won’t have much time to spare before the deadline.
My li'l Baby ( speed ~ 440 MFlops ) is estimating 5 days, 17 hours and a bit in total ( ~ 137 ) for it's first R2 finish. Because of Bruce's earlier 'flush out' of the remaining R1's, it got to start this R2 about 4 days ago. Credit estimate is ~ 245, so that's ~ 1.7 credits/hour.
Now the R1 it did just immediately prior was 72.65 hours chugging a credit of 53.77 credits. So that was ~ 0.75 credits/hour!
Interesting ... :-)
Cheers, Mike.
( edit ) Mind you the first is an estimate on an uncompleted unit, while the second is an actual awarded one ....
( edit ) and the report deadline is 05/05/2007, so plenty of time.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
My li'l Baby ( speed ~ 440 MFlops ) is estimating 5 days, 17 hours and a bit in total ( ~ 137 ) for it's first R2 finish. Because of Bruce's earlier 'flush out' of the remaining R1's, it got to start this R2 about 4 days ago. Credit estimate is ~ 245, so that's ~ 1.7 credits/hour.
Where did you get that credit estimate from, Mike?
Your quorum partner claimed 175.72, using BOINC v5.4.9
Edit - but it's still a valid calculation, and even on a revised estimate of 175 credits in 137 hours, that's 1.28 Cr/hr - a big improvement. Need to double-check that estimated time to completion, though.
It prompted me to do the same comparison on my Celeron 400:
Last S5RI unit: 54.25 credits in 36.58 hours, @ 1.49 Cr/hr
First S5R2 unit: 44.82 hours so far, 47.968 %done, quorum partner claimed 179.06
That works out at 1.92 Cr/hr, the only assumption being that the second half will crunch at the same rate as the first half.
So the credit awards for old, slow crunchers (1 AMD, 1 Intel) have been substantially increased by the S5R2 app.
I assume that, basically, old/slow/simple processors couldn't make any use of the sophisticated optimisations in the old app. Temporarily, new feature-rich processors are running as if all their extended instructions (MMX/SSEx) have been turned off. So as the debugging concludes, and the optimisation begins, expect substantial speed increases on newer processors.
Sounds good, but then we always expected that there was some potential for optimization in the new app. Should be interesting to see the next version of the app in action.
On another note, I'm still having a few problems with my VM. With the help of some fellow students I got it to run, connect to the network via NAT and install Debian, but no BOINC so far. Both the manager and core client seem to run allright but I can't get the manager to connect to the core client. Does anyone know this problem or have a useful idea?
Where did you get that credit estimate from, Mike?
Sorry, forgot to mention - BoincView 1.2.5, gives a read on that.
Now that you query it, I guess the question is how does it know? :-)
Quote:
Your quorum partner claimed 175.72, using BOINC v5.4.9
Edit - but it's still a valid calculation, and even on a revised estimate of 175 credits in 137 hours, that's 1.28 Cr/hr - a big improvement. Need to double-check that estimated time to completion, though.
Time so far : 4d 8h 3m
To completion : 1d 8h 3m
Total : 5d 16h 6m ( that estimate has come down by about an hour during the last 3 hours )
so ~ 79% done as we speak.
Quote:
It prompted me to do the same comparison on my Celeron 400:
Last S5RI unit: 54.25 credits in 36.58 hours, @ 1.49 Cr/hr
First S5R2 unit: 44.82 hours so far, 47.968 %done, quorum partner claimed 179.06
That works out at 1.92 Cr/hr, the only assumption being that the second half will crunch at the same rate as the first half.
So the credit awards for old, slow crunchers (1 AMD, 1 Intel) have been substantially increased by the S5R2 app.
I assume that, basically, old/slow/simple processors couldn't make any use of the sophisticated optimisations in the old app. Temporarily, new feature-rich processors are running as if all their extended instructions (MMX/SSEx) have been turned off. So as the debugging concludes, and the optimisation begins, expect substantial speed increases on newer processors.
Bingo, eh!!
They were my suspicions too ... time will tell though. :-)
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
RE: RE: I should think a
)
Well so what? It doesn't matter how long a result stays pending to your performance metrics over the long haul.
So why should my old timers be eliminated from participation just to please the false perceptions of the RAC hounds? They may be slow, but they're on 24/7 because of their primary function and turn in results just as valid and useful as the fastest C2Q.
Alinator
RE: Now [ just to totally
)
I've never heard of anyone getting that day off. I think they have to settle on being off Labor Day like the "common" man.
RE: LOL... that's why I
)
Ooooh .... err.. K6 500Mhz with '3D' I think - I call it "Baby" 'cos it's the slowest on the patch, and is on a baby AT board. It's a legacy/homage/nostalgia job really, the first one I got after my really fast first Pentium 90!! I never had a 486, but stepped up from a 33 Mhz 386 WITH, mind you, a 387 FP co-processor by it's side ... ahhh, those were the days of linear address spaces and easy-peasy segment:offset manipulations ..... :-)
In beaurocratic Canberra, they never schedule any meetings on Wednesdays ...... just so don't they spoil two weekends in a row ...... :-)
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
RE: RE: LOL... that's why
)
Ah OK, that would make it a K6-2/500. I have three of them, so feel free to use them as comparo units for yours. I also have a -III/450 I'm looking for the right Super 7 board for to round out my K6 family.
Interestingly I also have a Pentium 90, but its 'resting' at the moment. The reason is I haven't thought of something for it to do yet, but I will. ;-)
Hmmm, sounds like there isn't much difference between Canberra and DC, or any state capital here in the States for that matter. Especially once golf season starts! :-)
Alinator
RE: RE: RE: LOL...
)
Ah, the K-6s. Great chips at a fair price, I haven't bought an Intel in probably close to 10 years and this is the very first time that I feel I might be at a disadvantage, and you know what, I don't care. AMD has saved computer users millions if not billions of dollars and I'm not about to let this little glitch, that after I have gotten more results back, may be less than 5%.
Gary
In the beginning the Universe was created. This has made a lot of people very angry and been widely regarded as a bad move.....Douglas Adams
RE: I think we just got a
)
Damn, exposed again! I must have overestimated the amount of cover my use of “wannabe� would provide.
You should be grateful that I only perform by invitation. ;)
“ ‘When I use a word,’ Humpty Dumpty said, in rather a scornful tone, ‘it means just what I choose it to mean—neither more nor less.’
“ ‘The question is,’ said Alice, ‘whether you can make words mean so many different things.’
“ ‘The question is,’ said Humpty Dumpty, ‘which is to be the master—that’s all.’ �
—Lewis Carroll, Through the Looking Glass
My Mac G4/733 is looking at about sixty hours, based on a task that’s about one-quarter complete. That’s still within my comfort zone, although having a 20% resource share for E@h (IIRC) it won’t have much time to spare before the deadline.
My li'l Baby ( speed ~ 440
)
My li'l Baby ( speed ~ 440 MFlops ) is estimating 5 days, 17 hours and a bit in total ( ~ 137 ) for it's first R2 finish. Because of Bruce's earlier 'flush out' of the remaining R1's, it got to start this R2 about 4 days ago. Credit estimate is ~ 245, so that's ~ 1.7 credits/hour.
Now the R1 it did just immediately prior was 72.65 hours chugging a credit of 53.77 credits. So that was ~ 0.75 credits/hour!
Interesting ... :-)
Cheers, Mike.
( edit ) Mind you the first is an estimate on an uncompleted unit, while the second is an actual awarded one ....
( edit ) and the report deadline is 05/05/2007, so plenty of time.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
RE: My li'l Baby ( speed ~
)
Where did you get that credit estimate from, Mike?
Your quorum partner claimed 175.72, using BOINC v5.4.9
Edit - but it's still a valid calculation, and even on a revised estimate of 175 credits in 137 hours, that's 1.28 Cr/hr - a big improvement. Need to double-check that estimated time to completion, though.
It prompted me to do the same comparison on my Celeron 400:
Last S5RI unit: 54.25 credits in 36.58 hours, @ 1.49 Cr/hr
First S5R2 unit: 44.82 hours so far, 47.968 %done, quorum partner claimed 179.06
That works out at 1.92 Cr/hr, the only assumption being that the second half will crunch at the same rate as the first half.
So the credit awards for old, slow crunchers (1 AMD, 1 Intel) have been substantially increased by the S5R2 app.
I assume that, basically, old/slow/simple processors couldn't make any use of the sophisticated optimisations in the old app. Temporarily, new feature-rich processors are running as if all their extended instructions (MMX/SSEx) have been turned off. So as the debugging concludes, and the optimisation begins, expect substantial speed increases on newer processors.
Sounds good, but then we
)
Sounds good, but then we always expected that there was some potential for optimization in the new app. Should be interesting to see the next version of the app in action.
On another note, I'm still having a few problems with my VM. With the help of some fellow students I got it to run, connect to the network via NAT and install Debian, but no BOINC so far. Both the manager and core client seem to run allright but I can't get the manager to connect to the core client. Does anyone know this problem or have a useful idea?
RE: Where did you get that
)
Sorry, forgot to mention - BoincView 1.2.5, gives a read on that.
Now that you query it, I guess the question is how does it know? :-)
Time so far : 4d 8h 3m
To completion : 1d 8h 3m
Total : 5d 16h 6m ( that estimate has come down by about an hour during the last 3 hours )
so ~ 79% done as we speak.
Bingo, eh!!
They were my suspicions too ... time will tell though. :-)
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal