Rieselator

[vaughan]

I will be finishing a range in about 1 day 19 hours.

What do I do next?

How do I submit the factors?

How do I either stop Riesel Sieving or grab another range and put it in Rieselator so it can keep going on the "next range" when the current one is finished. I don't want to lose more than a week's work.

[carlos]

What do I do next?

Add another range to sieve or let the client stop sieving.

How do I submit the factors?

Go here:

http://www.rieselsieve.com/sieve/

And add your nick, paste factors and send them.

How do I either stop Riesel Sieving or grab another range and put it in Rieselator so it can keep going on the "next range" when the current one is finished. I don't want to lose more than a week's work.

To add another range click on ranges button, then select your machine then click on new button. The client will process the next range only if you started the client with -w flag. If you don't use the -w flag the client will stop automatically.

Carlos

[bryanRS]

Vaughan / Carlos,

The -w flag will keep it waiting if the client reaches the end of its work AND the nextrange.txt file is empty. If the file is not empty, it will pick up the work and keep going.

You can also submit the factors right from rieselator, by clicking the "Submit" button.

Reserve a new range from the stats pages, then click "Ranges". Select a machinge, choose "new" and enter in the ragne information. Click "Equalize" and it will spread the new range across all of your machines equally.

Wash, rinse, and repeat as desired. ;)

Bryan

[Anonymous]

I get the jump on this by dividing the range into units of 1 and feeding them into my clients as I go along.

ie (using non project numbers for this example)
105000 - 110000 becomes
105000 - 106000
106000 - 107000
107000 - 108000
108000 - 109000
109000 - 110000

note that 105000 - 106000 then 107000 - 108000 will miss a range.
I hope no-one splitting ranges is making this simple mistake and the projects have some way of guarding against this.

It seems that people have done this in the past and wondered why their points were so low. It is the only reason, other than the natural factor distribution, that I can think of.

This way I can farm out batches of single unit ranges to different machines, measure their progress over several days and then farm out the rest relative to speed. I found this manual method to be much, much more accurate than the "Equalize" function which I found to be very poor at re-distribution relative to completion speed.

Another benefit is it gives you a much finer grain of re-start points if the system fails and reboots and the measurement point is corrupted. During the first range set I allocated to one of my machines the power failed and this happened. Reisel Seive did not find the last save point and I had to start it again. That was when I broke up the entire range to the lowest significant digit. I was very pleased with the resulting control over my allocations and output levels.

I imagine (and hope) that the awaited automated versions of the Riesel and PSP seivers will allocate single unit stages like this automatically by dealing out ranges in "signifiacant" digit increments to better spread the result return rate.

[AMDave]

Looks like I lost my cookie again !
That was me BTW :D

[vaughan]

I agree Dave. In a new automated Siever I'd like to be able to see if the range is completed yet or not - a progress report. Sort of like SOB does or Rieselator. Sometimes a machine reboots for a reason - like Windowsupdates - and I have forgotten to restart PSP-Sieve. A few days go by with it running some other project and I check the PSP-S reservation page and discover that I haven't finished the range. I then hunt through my farm trying to find the computer that had the range allocated to it (I check the SoBstatus.dat file).

Why is it called SoBstatus.dat? Is it related to Seventeen or Bust?

[vaughan]

Wash, rinse, and repeat as desired.

Yep it sure did, when I checked it was at 0.1% and had started the 100G range again. I confirmed that the next range was associated with that computer but it didn't switchover to it. I clicked submit for the 97 (I think it was) factors.

Now how do I make it stop running the same work again and start the new 5G range I reserved?

I've switched to LLRnet tonight will check in the morning.

[bryanRS]

It started the same range again? Or did multiple boxes start the same range? It shouldn't restart it. But, to stop it from running a range, stop the client as you have, delete the RieselStatus.dat file, and then start the client again. It will ask you to manually enter a range, enter it, and things will be back to normal.

Bryan

[vaughan]

Just the one box running.

[AMDave]

Submit daily.
It's more fun. :D
(It also means I can see your progress better :P )

...one more thing about slicing up the 500G range I did.
To start off with I fed the whole lot into all of my clients.
Then I hit the Equalize button to test how it would re-allocate.
The "Out of memory error" was not what I was expecting.
It can only handle about 50 or so range entries.
500 range entries across 5 clients was too much for it.

Also, testing on 20 ranges, I observed that it "coalesced" the ranges back together instead of just reallocating the individual ranges to each of the clients which ticked me off. After that I handled the whole lot from a spreadsheet. It was so easy. I was able to top them up as they went along. If I over allocated to any client, I just edited the tail of the client's file "cut" from there and pasted into another client. As they were done, I marked them off as completed by that machine which gave me an exact measurement of the rate at which they were getting done.

This resulted in perfect control to even out the work across 5 clients all working at different rates. Too easy.

One surprise I discovered was that the Centrino Laptop fared very well indeed at seiving. So much so I got suspicious. It took me a couple of days of monitoring it and the results before I was convinced that it was doing the job properly.