BTW: A RAMDISK could be a fine solution. I'm running diskless for quite a while now, running Boinc in RAM. But it surely has it limitations (less available space, so less cache; less available RAM for computations, but that's not that important for EAH; power outage results in loss of results, which was for me another reason for a small cache).
What program do you use, for your RAM DISK? Are you aware of any free programs for that use? I will search, but if you know, it would be great. Thanks Lt.
At the moment I'm running Knoppix, so I'm not running another programm but another OS. Sorry ;) It loads from CD into RAM. After loading, you're able to write to the RAMdisk it created. But it's generally easy to use.
The hard disks i have lost have been well over 5 years old. If every hard disk lasts me 3-5 years i will be a very very happy man lol. from reading other posts here the reason my disks don't seem to die may just be the fact they never power down and run at the same temperature constantly.
Looks similar to the arguments for shutting down your PC when not in use vs leaving it on, or even leaving it at high load (BOINC).
Seriously who here still has a hard disk in use thats less than 1gb? and the 1.5Gb was marginal even for a Linux box so i wasn't really upset when they failed. Actually even the 3-4gb drives i have are fast approaching there used by date for the simple fact they ain't SATA and wont be compatible with new motherboards before too much longer.
Ive thrown more hardware away because i don't own a working computer that can still use it anymore than i have due to failures.
You know you can get ide - sata adaptor/convertors, if you think it is worth while.
Andy
Sadly i have one of these IDE - SATA converters. It set me back $40 and didn't do anything for me :(. It wouldn't talk to my BOIS or my OS it just timed out. So wasted money. It could be i got a bad one but i don't plan on buying more of them to find out lol. Im better off buying a new hard disk for $80 and having a faster one with a lot more space.
NB: I originally brought the adapter due to not being able to find a SATA DVD burner and it didn't work for that so i tried it on my hard disk and still had the same issue. Sunix tell me its an issue with both my OS and motherboard so no refund. Funny thing is it has the issue with both my Gigabyte and Intel motherboards. I do not recomend these things unless its a really good hard disk you are really desperate to put onto SATA.
Does E@H *force* disk accesses? That would be a feature. My G4 eMac almost always has a couple hundred MB of RAM (real, not virtual) sitting unused while E@H is running; why should it ever go to disk?
If you have BOINC or E@H set standard, without using a RAMDISK, or alike, yes BOINC will write the interval results to the harddisk. But that is a good idea. Maybe someone should post that on the 'Wish List' for BOINC or E@H. I am unsure, if that would be a E@H wish only.
If you don't write to disk and keep everything in RAM then when you shutdown or have a power failure all crunching is lost.
A few projects don't have checkpoints (disk writes) and those ones have nothing but complaints from the people who run them due to lost crunching and credits.
If you don't write to disk and keep everything in RAM then when you shutdown or have a power failure all crunching is lost.
A few projects don't have checkpoints (disk writes) and those ones have nothing but complaints from the people who run them due to lost crunching and credits.
Not to brag or anything :-> but using a Mac and a UPS I go months between shutdowns. E@H can probably live with that.
This discussion has prompted me to change the "Write to disk at most every" general preferences value from the default six seconds to 600 seconds. Maybe the default should be a lot larger than six seconds?
No need to call it bragging lol. I run Linux machines and generally the only time i reboot is a power failure or a boinc problem so for me no checkpoints is never a worry.
I did an RCN monster WU without checkpoints for 16 days before i pulled the plug as i needed to take some hardware out for another computer. I promptly restarted it on another machine (faster) and finished it. after that my record for a non-checkpoint WU is 202 hours.
But we have to remember there are many windoze users out there and a week is close to a record for most of them without something failing or needing a reboot.
Edit:Im also considering a UPS myself as i wanna see how long it is between BOINC problems lol. That and i dont want my newer rigs getting spiked by power surges/failures.
I found curiousity in many of your posts. So, instead of replying to each and every one, I will give a general idea of what I am thinking, and curious about... I am unsure the model of the HDD I am currently using for Virtual Memory, Firewall, and BOINC on my desktop. But it is a older HDD. About 2GB, so that gives a rough idea on the age....
Alinator:
Am I correct in assuming, that you believe, or know, that it is more effiecient, and less damage inflicting to keep a HDD spinning at all times, in opposition to having it power down, and back up?
__________
Before setting up the secondary drive, I noticed that even with BOINC being the only user-loaded program running at night, my drive would turn off, and turn on, much more than every 3600 seconds (My current WTD interval).. I didn't do much research on it, but from what I did attempt to learn, I was unable to find what was causing my disk to be writing much of the time... I am thinking about the fact, that the WTD interval, is only MAX WTD. But, I do not believe that BOINC would write to disk and save data ever 10 minutes or so, when a short WU takes more than 4 hours on my 448MHZ desktop... So, any suggestions?
NOTE: When I use a program called Metriscope to monitor overall computer systems. It tells me that my cache is writing much. But, I assumed that was not on my actual drive, but something else. Maybe someone here has an idea about that.
LOL, yes like most fora, conversations tend to wander considerably with the parameters of the thread theme! ;-)
As a general rule of thumb when it comes to drives, I try to avoid spinup/down cycles as much as I can. The reason is the spindle motor electronics are probably the most stressed subsystem in the drive, and experience has shown me they tend to fail when they are drawing their maximum current while spinning the drive up.
So in the case of where you have a separate physical drive for the swap file, I would tend to keep that one spinning all the time. Likewise, for a system which runs 24/7 I keep the system drive spun up all the time.
The only case I can think of where I would routinely let the drive spindown is large drive used for archival/backup storage which is infrequently used.
In any event a drive is a man-made device, and therefore can fail at anytime for a variety of reasons, so the best strategy is to keep them cool, keep them clean, and keep them backed up. :-)
RE: RE: BTW: A RAMDISK
)
At the moment I'm running Knoppix, so I'm not running another programm but another OS. Sorry ;) It loads from CD into RAM. After loading, you're able to write to the RAMdisk it created. But it's generally easy to use.
Happy crunching,
Bert
Somnio ergo sum
Oh ok, Thanks for the reply
)
Oh ok, Thanks for the reply though Bert. I searched, with no success for a free good Ramdisk program on www.download.com...
d3xt3r.net
RE: Oh ok, Thanks for the
)
Perhaps this is useful:
-search for ramdisk.sys on support.microsoft.com.
-this one has a free basic version. It looks good on its screenshots.
HTH,
Bert
Somnio ergo sum
RE: RE: The hard disks i
)
Sadly i have one of these IDE - SATA converters. It set me back $40 and didn't do anything for me :(. It wouldn't talk to my BOIS or my OS it just timed out. So wasted money. It could be i got a bad one but i don't plan on buying more of them to find out lol. Im better off buying a new hard disk for $80 and having a faster one with a lot more space.
NB: I originally brought the adapter due to not being able to find a SATA DVD burner and it didn't work for that so i tried it on my hard disk and still had the same issue. Sunix tell me its an issue with both my OS and motherboard so no refund. Funny thing is it has the issue with both my Gigabyte and Intel motherboards. I do not recomend these things unless its a really good hard disk you are really desperate to put onto SATA.
Does E@H *force* disk
)
Does E@H *force* disk accesses? That would be a feature. My G4 eMac almost always has a couple hundred MB of RAM (real, not virtual) sitting unused while E@H is running; why should it ever go to disk?
If you have BOINC or E@H set
)
If you have BOINC or E@H set standard, without using a RAMDISK, or alike, yes BOINC will write the interval results to the harddisk. But that is a good idea. Maybe someone should post that on the 'Wish List' for BOINC or E@H. I am unsure, if that would be a E@H wish only.
d3xt3r.net
If you don't write to disk
)
If you don't write to disk and keep everything in RAM then when you shutdown or have a power failure all crunching is lost.
A few projects don't have checkpoints (disk writes) and those ones have nothing but complaints from the people who run them due to lost crunching and credits.
RE: If you don't write to
)
Not to brag or anything :-> but using a Mac and a UPS I go months between shutdowns. E@H can probably live with that.
This discussion has prompted me to change the "Write to disk at most every" general preferences value from the default six seconds to 600 seconds. Maybe the default should be a lot larger than six seconds?
No need to call it bragging
)
No need to call it bragging lol. I run Linux machines and generally the only time i reboot is a power failure or a boinc problem so for me no checkpoints is never a worry.
I did an RCN monster WU without checkpoints for 16 days before i pulled the plug as i needed to take some hardware out for another computer. I promptly restarted it on another machine (faster) and finished it. after that my record for a non-checkpoint WU is 202 hours.
But we have to remember there are many windoze users out there and a week is close to a record for most of them without something failing or needing a reboot.
Edit:Im also considering a UPS myself as i wanna see how long it is between BOINC problems lol. That and i dont want my newer rigs getting spiked by power surges/failures.
RE: I found curiousity in
)
LOL, yes like most fora, conversations tend to wander considerably with the parameters of the thread theme! ;-)
As a general rule of thumb when it comes to drives, I try to avoid spinup/down cycles as much as I can. The reason is the spindle motor electronics are probably the most stressed subsystem in the drive, and experience has shown me they tend to fail when they are drawing their maximum current while spinning the drive up.
So in the case of where you have a separate physical drive for the swap file, I would tend to keep that one spinning all the time. Likewise, for a system which runs 24/7 I keep the system drive spun up all the time.
The only case I can think of where I would routinely let the drive spindown is large drive used for archival/backup storage which is infrequently used.
In any event a drive is a man-made device, and therefore can fail at anytime for a variety of reasons, so the best strategy is to keep them cool, keep them clean, and keep them backed up. :-)
Alinator