There are also two types of data files: short and long. The short data files (l1_XXXX.X) are from the LIGO Livingston Observatory, and are about 4.5MB in size. The long data files (h1_XXXX.X) are from LIGO Hanford and are about 16MB in size. Note: once your computer downloads one of these data files, it should be able to do many workunits for that same file.´
There is a way to get only short WUs ?
einstein preferences perhaps ?
The 1st S5 WU I downloaded, used up to 6 hours of my dialup @ 28800 bps
This, was possible only after
I pysically replace the HD, and reinstall windows 2000 on a new, larger HD
*Before that I was getting "no work from project" -> Insufficient disk space
*Sure, all S4 WUs , must already had been done, so no one could be sent to me.
btw: On this computer I was crunching 1 S4 WU each hour and half,
(using akos optz app 3Dnow! ), and getting granted about 55 credits for each WU
This S5 WU is crunching *slooower* , and my CPU keeps cool -:(
When downloaded, BoincView was predicting 6 hours to complete
*After 43 minutes of CPU time, it is only 3.7% done, and Boincview is now
predicting 7 hours to finish ... and this "prediction" keeps increasing ...
So, I am aborting current WU and
requesting "no more work" for einstein on this computer,
until things get "FIXED" -> download time, disk space need, crunch speed
Carlos, S5R1 large h1 files are 16MB downloads. Even on 28.8k dial-up, six hours is unrealistic. You must have a local problem over there.
Also, if you're internet connection can't sustain a boinc project with large file transfers, you should switch to a project with smaller transfers. HashClash would be a good one.
Whew! Can't take any time off around here! :-)
So we've been around the credit garden again, I see! :-)
Anyhows, it's real good to know that we are into real serious data now, team, as the S5 is from the LIGO's operating at design spec!
So let's get enthused, excited and a-quiver by thinking of the ( historic ) waves just lurking in those downloads!
I just know one of my boxes will nail one....... :-)
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
Anyhows, it's real good to know that we are into real serious data now, team, as the S5 is from the LIGO's operating at design spec!
So let's get enthused, excited and a-quiver by thinking of the ( historic ) waves just lurking in those downloads!
I just know one of my boxes will nail one....... :-)
Cheers, Mike.
This will be such an awesome discovery when it happens!
From the POV (point of view) of most science bang for the CPU seconds buck, and the fact that deciding where to spend/use scarce (development, akos, ...) resources always involves trade-offs, I'm curious to know what those who (likely) have the experience and knowledge to be able to comment would say concerning how much faster the S5 data will be crunched, given {insert realistic specifiation/description/definition of incremental resource allocation here} ... (if that isn't too convoluted a sentence/request).
This will be such an awesome discovery when it happens!
Yeah, that's why I ( politely ) suggest that people don't worry much about the credit! :-)
I'll also change back to the standard BOINC client ( I had Truxoft + Akos Albert's for a lot of S4 ), and update it to 5.4.9 while I'm at it. Probably go for service installs also.
Doesn't the 5.4.9 have real nice graphing on the stats pane? A pleasure to view. Neat & snazzy .... congrats to the dev's :-)
I should be getting some new boxes on line shortly as well, with these new long WU's they'll come in handy. I think we should call these WU's thumpers or somesuch. As in 'my box is crunching so much it thumps' ..... visions of 19th century mills with rows of Babbage Difference Engines :-)
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
Carlos, S5R1 large h1 files are 16MB downloads. Even on 28.8k dial-up, six hours is unrealistic. You must have a local problem over there.
Also, if you're internet connection can't sustain a boinc project with large file transfers, you should switch to a project with smaller transfers. HashClash would be a good one.
Yeah ! I will do just it, I already cruch HashClash too,
but I will not place all my computing power on only 1 project.
I prefer then Rosetta, that may lead to some Ills start having cure.
*Better use for computing resources !
Why I have to download 16 MB WUs, on dialup,
when there are 4.5 MB (S5 WUs) that are sent instead,
to the ones with a very fast Internet Link ?
About unrealistic download times,
I suggest u travelling here, and sitting in front of my computer with a watch
in hands, to measure exactly download times.
Clearly, large / small WUs
is a user preference,
and so, users must be able to select, via einstein preferences
While a "BOINC" download occurs, the link is clogged, and u cannot surf Internet -:(
Boinc was designed to download two files at a time (simultaneus)
and is this that block users from using Internet during a WU download (on dialup)
*and cause the need to be connected doubled time
btw: my last telephone bill was $600,00
Ignoring the issue of credit (which is always a touchy subject), there's another issue which may drive some crunchers away...
I have a strong preference for projects with work units that don't take more than 2-3 hours to compute on a reasonably fast CPU. There are other like me (scary thought?)
SETI massively increased their WU length. It was one of the reasons I stopped my involvement in that project and came back to Einstein. With AKOS apps, my slowest computer would take 2 hours at worst for a "long" work unit (and the fastest would do them in half an hour). Great!
If S5 work units are going to take over 9 hours on one of my fastest computers (Opteron 248), it is a disincentive for me to stay with Einstein when I can get similar credit elsewhere on work units that don't take as long.
Just offering my point of view on work unit length and how it may affect some people's choice of project, driving them away from Einstein.
You make sense here. I wont address the credit issue, but one thought on WU times, is that the project has to take into consideration the potential of optimized apps. Consider that these S5 apps are basically the S4 beta apps, and consider that the windows beta was considerably slower than Akos' apps, eventually as either the project or others release further optimized apps, the time to complete will reduce significantly. If these initial official apps were 2-3 hours, and if Akos' optimizations are any indication of the possibility of S5 optimizations, long S5 WUs would be even faster than long S4 WUs. Having work last 45 minutes for long and 10 minutes for short WUs seems too frequent and a waste.
I guess it's about trying to get the lenght just right, and that is different for different people. I believe we'll see further optimizations in the future. Bernd has said he will be working on them in this thread. I understand gettin put off by long completion times, but I still remember the days of taking 14 hours to complete a WU, so it doesn't bother me, especially cuz the science interests me.
So maybe you'll leave and come back when we have faster apps. Happy crunching wherever you be.
S4 WUs used to take about 70000 s on my Pentium II running SoSE Linux. My first S5 result should finish in about 40000 s. SETI WUs are getting impossibly long times on a 400 MHz CPU (about 150000 s). I suggest to people on dialup to go to QMC@home, a quantum chemistry program. Downloads are smaller.
Tullio
@ Carlos. I'm and always was into CPDN. I've suggested people with slow connection to put BOINC on flashdisk, download WUs they want to crunch somewhere with fast connection (school, job, i-net cafe), bring it home and run there until results are ready to send back. CPDN just fits fine, 'cause it takes a while to complete a climate model.
If you are keen on Einstein, you may do so once a week (deadlines are 14 weeks) if it is possible for you.
(Rosetta is not good for dial-up users - too much data, too short deadline).
You may take a look at my table, where it summarize HW demands, traffic, crunch times etc. so all BOINC projects; it is designed to help people choose was may suits them if they have any constrains (slow connection, outdated HW etc.). http://www.boinc.cz/img/BOINC_Projects_table.png
RE: There are also two
)
There is a way to get only short WUs ?
einstein preferences perhaps ?
The 1st S5 WU I downloaded, used up to 6 hours of my dialup @ 28800 bps
This, was possible only after
I pysically replace the HD, and reinstall windows 2000 on a new, larger HD
*Before that I was getting "no work from project" -> Insufficient disk space
*Sure, all S4 WUs , must already had been done, so no one could be sent to me.
btw: On this computer I was crunching 1 S4 WU each hour and half,
(using akos optz app 3Dnow! ), and getting granted about 55 credits for each WU
This S5 WU is crunching *slooower* , and my CPU keeps cool -:(
When downloaded, BoincView was predicting 6 hours to complete
*After 43 minutes of CPU time, it is only 3.7% done, and Boincview is now
predicting 7 hours to finish ... and this "prediction" keeps increasing ...
So, I am aborting current WU and
requesting "no more work" for einstein on this computer,
until things get "FIXED" -> download time, disk space need, crunch speed
Thanks
Click signature for global team stats
Carlos, S5R1 large h1 files
)
Carlos, S5R1 large h1 files are 16MB downloads. Even on 28.8k dial-up, six hours is unrealistic. You must have a local problem over there.
Also, if you're internet connection can't sustain a boinc project with large file transfers, you should switch to a project with smaller transfers. HashClash would be a good one.
Whew! Can't take any time off
)
Whew! Can't take any time off around here! :-)
So we've been around the credit garden again, I see! :-)
Anyhows, it's real good to know that we are into real serious data now, team, as the S5 is from the LIGO's operating at design spec!
So let's get enthused, excited and a-quiver by thinking of the ( historic ) waves just lurking in those downloads!
I just know one of my boxes will nail one....... :-)
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: Anyhows, it's real good
)
This will be such an awesome discovery when it happens!
From the POV (point of view)
)
From the POV (point of view) of most science bang for the CPU seconds buck, and the fact that deciding where to spend/use scarce (development, akos, ...) resources always involves trade-offs, I'm curious to know what those who (likely) have the experience and knowledge to be able to comment would say concerning how much faster the S5 data will be crunched, given {insert realistic specifiation/description/definition of incremental resource allocation here} ... (if that isn't too convoluted a sentence/request).
RE: This will be such an
)
Yeah, that's why I ( politely ) suggest that people don't worry much about the credit! :-)
I'll also change back to the standard BOINC client ( I had Truxoft + Akos Albert's for a lot of S4 ), and update it to 5.4.9 while I'm at it. Probably go for service installs also.
Doesn't the 5.4.9 have real nice graphing on the stats pane? A pleasure to view. Neat & snazzy .... congrats to the dev's :-)
I should be getting some new boxes on line shortly as well, with these new long WU's they'll come in handy. I think we should call these WU's thumpers or somesuch. As in 'my box is crunching so much it thumps' ..... visions of 19th century mills with rows of Babbage Difference Engines :-)
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: Carlos, S5R1 large h1
)
Yeah ! I will do just it, I already cruch HashClash too,
but I will not place all my computing power on only 1 project.
I prefer then Rosetta, that may lead to some Ills start having cure.
*Better use for computing resources !
Why I have to download 16 MB WUs, on dialup,
when there are 4.5 MB (S5 WUs) that are sent instead,
to the ones with a very fast Internet Link ?
About unrealistic download times,
I suggest u travelling here, and sitting in front of my computer with a watch
in hands, to measure exactly download times.
Clearly, large / small WUs
is a user preference,
and so, users must be able to select, via einstein preferences
While a "BOINC" download occurs, the link is clogged, and u cannot surf Internet -:(
Boinc was designed to download two files at a time (simultaneus)
and is this that block users from using Internet during a WU download (on dialup)
*and cause the need to be connected doubled time
btw: my last telephone bill was $600,00
Click signature for global team stats
RE: Ignoring the issue of
)
You make sense here. I wont address the credit issue, but one thought on WU times, is that the project has to take into consideration the potential of optimized apps. Consider that these S5 apps are basically the S4 beta apps, and consider that the windows beta was considerably slower than Akos' apps, eventually as either the project or others release further optimized apps, the time to complete will reduce significantly. If these initial official apps were 2-3 hours, and if Akos' optimizations are any indication of the possibility of S5 optimizations, long S5 WUs would be even faster than long S4 WUs. Having work last 45 minutes for long and 10 minutes for short WUs seems too frequent and a waste.
I guess it's about trying to get the lenght just right, and that is different for different people. I believe we'll see further optimizations in the future. Bernd has said he will be working on them in this thread. I understand gettin put off by long completion times, but I still remember the days of taking 14 hours to complete a WU, so it doesn't bother me, especially cuz the science interests me.
So maybe you'll leave and come back when we have faster apps. Happy crunching wherever you be.
Cheers.
M
S4 WUs used to take about
)
S4 WUs used to take about 70000 s on my Pentium II running SoSE Linux. My first S5 result should finish in about 40000 s. SETI WUs are getting impossibly long times on a 400 MHz CPU (about 150000 s). I suggest to people on dialup to go to QMC@home, a quantum chemistry program. Downloads are smaller.
Tullio
@ Carlos. I'm and always was
)
@ Carlos. I'm and always was into CPDN. I've suggested people with slow connection to put BOINC on flashdisk, download WUs they want to crunch somewhere with fast connection (school, job, i-net cafe), bring it home and run there until results are ready to send back. CPDN just fits fine, 'cause it takes a while to complete a climate model.
If you are keen on Einstein, you may do so once a week (deadlines are 14 weeks) if it is possible for you.
(Rosetta is not good for dial-up users - too much data, too short deadline).
You may take a look at my table, where it summarize HW demands, traffic, crunch times etc. so all BOINC projects; it is designed to help people choose was may suits them if they have any constrains (slow connection, outdated HW etc.).
http://www.boinc.cz/img/BOINC_Projects_table.png