client suggestions

crash893 · 2002-09-16, 22:20

i dont know if there is already a post like this
if there is feel free to delete it.

some things i would like in the next client mabey

something to add to the prime.ini to get all the undoc.txt stuf to show up in the drop menus

or even mabey if you add soemthing to the prime.ini that it is in the menu.

i would also like to see something that would tell you your min iteration and the avarage iteration times.

at a nother forum i read at they made a bootable version of folding@home
so you dont need to buy a os and it also requires less system resorces.

i think that would be really helpfull to people seting up farms.

these are just suggestions and since i dont know much about the inter workings of the prime client ( yet) if these things are hard or imposable then forget they were ever said

anybody else have any suggestions.

greeners · 2002-09-17, 14:39

I would like ...

- to have a flag that would allow me to startup prime95.exe, let it upload a result and then quit ... then I could script the uploading rather than doing it by hand :D

Prime95 · 2002-09-17, 19:45

Quote:

Originally Posted by greeners

I would like ...

- to have a flag that would allow me to startup prime95.exe, let it upload a result and then quit ... then I could script the uploading rather than doing it by hand :D

Kludgy, but might work. Download winnt29.zip. Run ntprime -c. Cross your fingers. If you try it let us know if it works.

Prime Monster · 2002-09-22, 20:10

It would be nice to have links to this forum, the PrimeNet server and the Mersenne org home page on the Help Menu.

Alf

dswanson · 2002-12-27, 06:41

George,

I run a bunch of clients from a parallel set of subdirectories on a central server, all within the same allotment of disk space. I have each of these clients configured to save a checkpoint file every 1M iterations.

A couple of times now I've run into a problem when I run up against my storage allocation on the server. Each time the client tries to write a checkpoint file it reports an error because the disk is full, and continues to merrily chug along without writing the file. This is fine and dandy.

The real problem occurs when the client COMPLETES an exponent: It's unable to write the .spl file containing the result because the disk is full. So just starts working on the next exponent in the worktodo file AND FORGETS ALL ABOUT THE RESULT!!!! Once I clean up some disk space the client starts writing checkpoint and p/q files for the new exponent, but the old exponent's result remains lost. The best I can do is restart the lost exponent from the last available checkpoint file, which in some cases has been as much as two days old.

Would it be possible to tweak the client so that in the event of a .spl write error it retains the result(s) even while processing the next exponent(s), and writes it to the .spl file at the next successful write attempt?

S3SJK · 2002-12-28, 09:26

Hi,

I would like to see a way of manipulating the order that exponents are attempted via the software rather than have to go into the worktodo.ini every time.

Thanks

battlemaxx · 2002-12-28, 21:23

I'd like to have a way to "farm out" the clients, when installing to a fleet of boxen, perhaps have a setting or force a read from a custom .ini file that specifies some criteria to fill in the blanks that need to be initially set.

For example, e-mail, username, etc. are probably the same almost all of the time when deploying en masse. But, we'd need a different machine name or naming pattern. It'd also be nice to specify the work to do, or have some way to override the "default" "best fit" analysis.

Then there is a monitoring tool...which isn't a client piece...but still would be nice...or some way to run prime95 through a gimps queue like setiqueue handles seti@home.

2002-09-16, 22:20	#1
crash893 Sep 2002 450₈ Posts	client suggestions i dont know if there is already a post like this if there is feel free to delete it. some things i would like in the next client mabey something to add to the prime.ini to get all the undoc.txt stuf to show up in the drop menus or even mabey if you add soemthing to the prime.ini that it is in the menu. i would also like to see something that would tell you your min iteration and the avarage iteration times. at a nother forum i read at they made a bootable version of folding@home so you dont need to buy a os and it also requires less system resorces. i think that would be really helpfull to people seting up farms. these are just suggestions and since i dont know much about the inter workings of the prime client ( yet) if these things are hard or imposable then forget they were ever said anybody else have any suggestions.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Suggestions for Copenhagen?	Spherical Cow	Lounge	6	2012-04-24 18:36
two suggestions for PrimeNet	ixfd64	PrimeNet	1	2010-12-17 21:28
Suggestions and Errors	kar_bon	Riesel Prime Data Collecting (k*2^n-1)	8	2010-11-29 00:34
Any good suggestions please?	kakos22	Math	6	2010-07-09 10:40
Suggestions for a New Processor	Primeinator	Hardware	12	2009-06-10 15:01

2002-09-17, 14:39	#2
greeners Sep 2002 3 Posts	I would like ... - to have a flag that would allow me to startup prime95.exe, let it upload a result and then quit ... then I could script the uploading rather than doing it by hand :D

2002-09-22, 20:10	#4
Prime Monster Aug 2002 2²·5·13 Posts	It would be nice to have links to this forum, the PrimeNet server and the Mersenne org home page on the Help Menu. Alf

2002-12-27, 06:41	#5
dswanson Aug 2002 2³×5² Posts	George, I run a bunch of clients from a parallel set of subdirectories on a central server, all within the same allotment of disk space. I have each of these clients configured to save a checkpoint file every 1M iterations. A couple of times now I've run into a problem when I run up against my storage allocation on the server. Each time the client tries to write a checkpoint file it reports an error because the disk is full, and continues to merrily chug along without writing the file. This is fine and dandy. The real problem occurs when the client COMPLETES an exponent: It's unable to write the .spl file containing the result because the disk is full. So just starts working on the next exponent in the worktodo file AND FORGETS ALL ABOUT THE RESULT!!!! Once I clean up some disk space the client starts writing checkpoint and p/q files for the new exponent, but the old exponent's result remains lost. The best I can do is restart the lost exponent from the last available checkpoint file, which in some cases has been as much as two days old. Would it be possible to tweak the client so that in the event of a .spl write error it retains the result(s) even while processing the next exponent(s), and writes it to the .spl file at the next successful write attempt?

2002-12-28, 09:26	#6
S3SJK Dec 2002 3×5 Posts	Hi, I would like to see a way of manipulating the order that exponents are attempted via the software rather than have to go into the worktodo.ini every time. Thanks

2002-12-28, 21:23	#7
battlemaxx Nov 2002 2B₁₆ Posts	I'd like to have a way to "farm out" the clients, when installing to a fleet of boxen, perhaps have a setting or force a read from a custom .ini file that specifies some criteria to fill in the blanks that need to be initially set. For example, e-mail, username, etc. are probably the same almost all of the time when deploying en masse. But, we'd need a different machine name or naming pattern. It'd also be nice to specify the work to do, or have some way to override the "default" "best fit" analysis. Then there is a monitoring tool...which isn't a client piece...but still would be nice...or some way to run prime95 through a gimps queue like setiqueue handles seti@home.