PrimeGrid is pleased to announce that manual sieving will no longer require establishing an account on another site, or need a separate upload/download server. On the left of every page, there's now a link for Manual Sieving. There you'll see a list of projects where sieving is available. They currently include two different GFN's, Factorial, Primorial and PPR12M. PPS Sieve in BOINC is PPR9M, PPR12M is the next higher n range.

Manual sieving is not for everyone. You select a project and reserve a range of p values. On the confirmation screen you can download the sieving program used along with the current sieve file, if any. The suggested command line is also listed. You run the program on your computer until it finishes, then upload your completed factor files to the server. If you don't feel comfortable with command line programs, please don't participate.

On the My Account page, there's now a link to your sieving reservations. You can always get to your own reservations with the url http://www.primegrid.com/my_reservations.php. Once you reserve a range, that's considered an "open" reservation. When you upload your factor files it changes to "pending" - or as it says onscreen "awaiting admin action". Once your files are verified to make certain the sieving was done properly, an administrator will change your reservation to "completed" and you'll be awarded PSA credit. PSA credit doesn't go into your account all at once, doing so would skew the BOINC statistics and make it look like you did an incredible amount of work in a single day. Some PSA credit is added every day, based on your Recent Average Credit number. Pending PSA credit can be seen on your badges page.

Please note this is sieving - removing candidates when we can find a number that divides them (meaning they're most definitely not prime). You can't find a prime with sieving, you can only prove that a number is not prime. We continue sieving until it takes less CPU time to run an LLR primality test than to remove a candidate by sieving. That's called the optimal sieving point. At that point no new sieving reservations will be allowed on that particular project.

We're not desperate for everyone to start sieving. This system was created because previously sieving involved using two different outside servers belonging to Lennart. With his absence from PrimeGrid we needed to come up with a new procedure for sieving before those two servers were taken down. This is the result.

Sieving takes a while. Depending on how large a range you select and how many cores you're running on, it could take anywhere from a few hours to a month. Please start out slowly, selecting the smallest range (1 of whatever the unit is) to make sure you can handle it. There are no hard deadlines, but you're expected to finish whatever you reserve. Anything more than two weeks' worth of work is probably excessive. Repeatedly making and cancelling reservations will lock you out of the system, so don't play around with it. The more sieving you do, the larger the range you can select. First-time sievers are limited to two open reservations at a time and each one no larger than 5 of whatever the minimum sieving size is. Every successfully completed reservation bumps up both those values, though relatively few people run more than one reservation at a time.

If you want to try finding a Factorial or Primorial prime, they're available in PRPNet.

Questions and comments can be posted here. This is a brand-new system written entirely from scratch. Though it's been tested, I'm sure a bug or two will turn up. Feel free to PM me if anything urgent turns up.

This is quite cool! I was always tempted to try the manual sieving but there were quite a lot of hoops for a casual user, this is greatly simplified and really easy to use, thank you all who created it!

Only comment (and it's rather silly) is that I wish you had waited for a non-challenge time to roll it out as they tend to stretch my power budget and all my CPUs are just now getting engaged in some primorial ooze action.

Trying out my first range now on a GPU, and will think about some tweaking of the speed variables on the next one.
____________
Eating more cheese on Thursdays.

This looks really nice - let's start some sieving :)
____________

Only comment (and it's rather silly) is that I wish you had waited...

I just found that I'd forgotten to put the factorial sieve file on the server. It's now on there. Either everyone is using the GPU program in spite of the pitfalls, or somehow nobody noticed.

A quick question for a beginner: what's the point of manual sieve VS just turning that into regular Boinc tasks? Wouldn't the later be more systematic, with results being available as soon as possible and without much risk for a user to start and never report back?

A quick question for a beginner: what's the point of manual sieve VS just turning that into regular Boinc tasks? Wouldn't the later be more systematic, with results being available as soon as possible and without much risk for a user to start and never report back?

A quick question for a beginner: what's the point of manual sieve VS just turning that into regular Boinc tasks? Wouldn't the later be more systematic, with results being available as soon as possible and without much risk for a user to start and never report back?

Getting new apps to work under BOINC isn't so easy.

Also, we like to have the BOINC projects available on most platforms. People would be rightly upset, for example, if a particular sieving badge was only available to people running Windows, or only on Nvida GPUs but not on CPUs or ATI GPUs, etc. You'll notice that a lot of sieves are only available for specific platforms.

In the case of the PPR sieve (which is very similar to BOINC PPS Sieve), running it on BOINC wouldn't be a technical problem, but it's the sieve we'll use after the current BOINC sieve. The current BOINC sieve, which we need sooner, is a higher priority. If both were on BOINC, we would lose some PPS sieve participation, which is not what we want.

There is this "B value desired", which is at 7 default. Should I care to ever change it (and why)?

There is this "B value desired", which is at 7 default. Should I care to ever change it (and why)?

It's a tuning parameter. Play around with different settings and see which runs fastest. It varies depending on your hardware.

[Geeky performance tuning stuff follows. Read only if this whole manual sieving thing sounds easy to do and isn't at all confusing.]

Also, with the GFN sieves, I've personally found that running TWO sieves on one GPU produces more total throughput than running just one sieve. You can tell the new setup page you have two GPUs and it will give you command lines for both, but then run both commands without the '-d1' so both apps run on a single GPU. Again, you can experiment with this to see what works best on your hardware. Depending on your particular hardware, three or more sieves running simultaneously could be fastest.

I have 1 small request. Once you decide what and how many T's or G's to reserve and hit Reserve you are taken to a screen where you are given the switches and other information. Embedded is the range you need to sieve but I was wondering if you can add a small block that just shows the actual range as well as the other useful information you show.

One person has now been burned by using the GPU program for Factorial sieving. He handed in four factors when there should have been eleven (because I ran it myself as his upload did not look right). You CANNOT INTERRUPT the fsievecl program. Factors it's already found will be missing from the ouput file. If your GPU has been flaky in the past, don't use fsievecl.

If people can't use the fslevecl program properly, I'll be removing it and not accepting anything created with it (yes, I can tell). Please, accuracy is important and all I'm asking is that you do things properly.

I can't seem to upload the factor file for my reservation...

I tried 2 separate times and both times (after a minute or two) it said "All uploads failed - reservation is unchanged."

I'm curious if you have a timeout for how long a file upload can take? Centurylink upload from my house is painfully slow at 0.5Mbps to somewhere only a few miles away (that's right, megabits. not even megabytes), so that may be the issue. That speed on my factor file would be over a minute.

Thanks,
-Golfer

EDIT: Compressed w/ 7zip and uploaded, worked fine. Probably was the file size/slow upload issue.
____________

275*2^3585539+1 is prime!!! (1079358 digits)
Proud member of Aggie the Pew

One person has now been burned by using the GPU program for Factorial sieving. He handed in four factors when there should have been eleven (because I ran it myself as his upload did not look right). You CANNOT INTERRUPT the fsievecl program. Factors it's already found will be missing from the ouput file. If your GPU has been flaky in the past, don't use fsievecl.

If people can't use the fslevecl program properly, I'll be removing it and not accepting anything created with it (yes, I can tell). Please, accuracy is important and all I'm asking is that you do things properly.

That's why the warning is there. Many experienced sievers have used it in the past, though. Rather than have them complaining about how they're not allowed to use it, I reluctantly put it in there. I could easily hide it from new users, though. This code is still evolving and changing.

I can't seem to upload the factor file for my reservation...

I tried 2 separate times and both times (after a minute or two) it said "All uploads failed - reservation is unchanged."

EDIT: Compressed w/ 7zip and uploaded, worked fine. Probably was the file size/slow upload issue.

Just by the way, if I were to have to name a sieving priority right this minute, it's GFN N=2097152. Those will become the new short GFN workunits within a few weeks at most. Any sieving done there eliminates having to test those candidates with genefer. You can see the results of recent sieving here. And just so everyone knows, that URL will be changing soon as Comcast is taking away their free hosting as of 8 October. The new home will be somewhere on this site.

Yes, I'd been wondering about that. I've had problems myself with large files, but that was on my local server. I have to check what apache and php are both allowing - the limit is going to be the smaller of whatever is configured. We don't pay for incoming bandwidth so I have no reason to have low limits.

Later edit: You definitely ran into a file size limit (now changed). There was also a 1-minute limit on how long an upload could take. That's now changed to 2 minutes but I would recommend putting any PPR12M factor file into .zip, .7z or any other common compressed format to make things faster on your end. Your uncompressed file should have taken somewhere in the neighborhood of 45 seconds to upload.

Just by the way, if I were to have to name a sieving priority right this minute, it's GFN N=2097152. Those will become the new short GFN workunits within a few weeks at most. Any sieving done there eliminates having to test those candidates with genefer. You can see the results of recent sieving here. And just so everyone knows, that URL will be changing soon as Comcast is taking away their free hosting as of 8 October. The new home will be somewhere on this site.

Just by the way, if I were to have to name a sieving priority right this minute, it's GFN N=2097152. Those will become the new short GFN workunits within a few weeks at most. Any sieving done there eliminates having to test those candidates with genefer. You can see the results of recent sieving here.

Sieving by range for GFN N=2097152 Factors Found Removed from sieve
Range Start 0
Range End 53893T
User Name Honza
Factors Found 26556851
Factor Density 492770.0
Removed from sieve 26556851
Removed Density 492770.0
Usable by GeneferCUDA 149413
GPU Time saved 1442.09 years
Usable by Genefer64 CPU 5228799
Time saved 1101832.40 years

It's nice that the manual sieving have been somewhat "simplified" to use for average users.

Always wanted to try it, did it now for some days and can say that i kind of like the way it's done (well, after i realised how to start it, probably some beginner's mistake :) ).

All in all, good job with that :)

A small typo in user interface. When reserving Primorial Prime Sieving, the form displays: "Files to upload when finished: fpfactors_8226G-8230G.txt". The leading "f" is erroneous, file will be named "pfactors_...". Maybe it's just a copy-n-paste error from part of command line: "-fpfactors_8226G-8230G.txt", where "-f" is an option and "pfactors_..." - file name.

A small typo in user interface. When reserving Primorial Prime Sieving, the form displays: "Files to upload when finished: fpfactors_8226G-8230G.txt". The leading "f" is erroneous, file will be named "pfactors_...". Maybe it's just a copy-n-paste error from part of command line: "-fpfactors_8226G-8230G.txt", where "-f" is an option and "pfactors_..." - file name.

I'm now showing links for statistics screens for three of the sieving projects. Stats for factorial and primorial exist and they'll be shown shortly. Please note that these stats predate the sieving system by several years and used to live on my own hosting space. That free hosting is being discontinued by my ISP, so now they'll be hosted on PrimeGrid. That's why the sieving stats pages don't show the PrimeGrid logo or menus and will open in a new window by default. It's not a high priority at the moment for me to brand those pages as belonging to PrimeGrid.

Only the GFN pages are personalized with user names. That's a manual process for me (entering those reservations into the stats system on my computer) and I don't intend to personalize any of the other project stats like that. GFN stats can also be sorted on any column by clicking on the column header. Click again to reverse the sort order.

And as long as I'm posting, please note that it takes me exactly the same amount of time to verify a 1P sieve upload as a 100P sieve upload. For every range successfully sieved, you gain the ability to reserve larger ranges. Please feel free to reserve larger ranges as I won't consider them late until at least two weeks have gone by. They certainly don't have to be finished in a day. On the other hand, do what you're comfortable with. If that means small ranges, fine.

And while I'm thinking about it: PSA credit is automatically applied to BOINC credit once a day at 0400 UTC. I may not validate manual sieving ranges constantly throughout the day, but I'll make every effort to validate all uploads before 0400. How much of that resulting credit goes into BOINC every day depends on your Recent Average Credit - the higher it is the more credit is moved. The minimum moved is 100 credits per day and there is no predefined maximum.

Can we just make a bunch of reservations and run the sieve as just one huge command line? Say, I reserve range 10-20, then 20-30, then 30-40, and just upload a 10-40 file in one of the tasks.

I'm asking because I have to leave my PC away for a few days, and I can't just keep adding the inputs for new tasks. That would allow me to keep up with the sieve.

Can we just make a bunch of reservations and run the sieve as just one huge command line? Say, I reserve range 10-20, then 20-30, then 30-40, and just upload a 10-40 file in one of the tasks.

I'm asking because I have to leave my PC away for a few days, and I can't just keep adding the inputs for new tasks. That would allow me to keep up with the sieve.

Can we just make a bunch of reservations and run the sieve as just one huge command line? Say, I reserve range 10-20, then 20-30, then 30-40, and just upload a 10-40 file in one of the tasks.

I'm asking because I have to leave my PC away for a few days, and I can't just keep adding the inputs for new tasks. That would allow me to keep up with the sieve.

(Note that Jim has to do X amount of work for every reservation, regardless of size, so it reduces his work load if you make one large reservation.)

Jim will give you a definitive answer, but if the answer is "no", you *can* do this:

1) Cancel any reservations you have that you haven't started. There's no penalty for canceling them, and it doesn't break anything.

2) Make one big reservation.

3) Sieve that large reservation while you're away.

I may have to rethink my system for increasing the size of a range you can reserve as you gain experience.

You, Rafael, can now reserve a range of 200P at a time with GFN2097152 (or 200T of PPR12M). 200 of whatever the minimum unit is. That's the maximum I've configured the project for. I don't see a reason to increase that project max reservation size at this time.

How do you run a bunch of smaller reservations while you have to be away? You put the individual command lines into a batch file. Then when one reservation finishes the next one starts immediately. A batch file is just a text file with a .bat extension. Create it in a plain text editor like notepad. Each line contains the command for a different reservation.

Anyone else who'd like to reserve ranges larger than they can now right away should PM me. People who'd sieved for us in the past started with a limit of 50, everyone else started with 5 and it grows by 1 for every range completed successfully. Please note that's not a total limit, it's per reservation.

It seems I was mistaken about the penalty for cancelling reservations, so you shouldn't wantonly cancel them after all.
____________
My lucky number is 75898⁵²⁴²⁸⁸+1

But i have one question for that: do i have to put all of them before i start sieving or can i edit the .bat while its running, so that i can add reservations later on?

Possibly because of a minor change I made, there are some overlapping reservations. While I investigate this phenomenon, new reservations are suspended so that the situation doesn't get worse.

Later edit: There was a bug that was there from the beginning, just needed the right circumstances to trigger it. Other code perpetuated the overlap once it started. I'm now waiting to grab today's database backup so I can see how the changed code handles the already-screwed-up reservations. Everyone who has reserved a range, overlapping or not, will get credit for that range. In the case of the GFN sieving stats, because I'm counting factors removed, overlaps will not show up there.

Final edit: New code has been tested and this should not happen again. New reservations are reenabled.

It's tme to talk about overdue reservations. On my admin screen, anything older than two weeks shows up as overdue. It's not that there's a hard deadline, but it lets me stay aware of what's happening. At the moment there are three "overdue" reservations. They're all for minimum sieving ranges - all three are each for 1G of Factorial sieving. Even using the CPU program on a single core, that takes about 30 hours for my slowest computer. Two weeks seems like more than enough time for a 1.25-day runtime. In the case of the two older reservations, I've written to the user asking him/her to please either cancel the reservation (in which case it'll get handed out to the next user asking for a 1G range) or to do the work. Neither user has responded. Once seven additional days have gone by since the message, I'll cancel the reservation and block that user from making new ones. In the case of the two users I've already written to, it will then be 23 days after the reservation was made. If a user can't bother to either respond to my PM or cancel a reservation him/her self then that user has no business doing manual sieving.

For anyone who has a history of actually finishing sieving and especially for people reserving large ranges, I will not get upset about it taking more than two weeks. In fact I like large reservations much more than small ones. The amount of work on my side is the same per reservation no matter how large or small it is.

I knew that by opening manual sieving up to everyone at PrimeGrid we'd get users who wouldn't follow through on their reservations. It's really simple - if you can't do the work within the time allotted, either contact me or cancel the reservation. There's a send message button on every post I make and you can always find me by clicking on the contacts link at the bottom of every PrimeGrid page. I'll always try to work with you.

Probably a silly question but, playing with a new GPU and trying to get the over clock set to a stable level, if I observe a line that says computation error not factors found or xxxxxx is it safe to assume the entire factor file is trashed and I need to start it all over again?

Cheers

Thoughts and observations on GFN sieving.

* This applies to all manual sieving, not just GFN. If you've done sieving before and are confident you're going to complete the sieve, I think it's easier to grab large reservations. Less work for you, less work for Jim. I personally like to reserve about a week's worth of sieving at once. I just reserved 800P worth of GFN 131072 sieving as 4 200P chunks. It will take about 10 days, and I don't need to worry about it again until then. If I have to stop the sieving for some reason (rebooting, games, whatever), that's easy and restarting it is no problem.

* All four of those 200P reservations are running simultaneously on my single GPU. With a lot of sieve programs, it's beneficial to run multiple instances of the sieve on the GPU because the program doesn't use the full GPU. You need to experiment with this to see what works best on your hardware, since your results *will* vary based on the GPU, the CPU, and which GFN range you're sieving. Higher ranges like 4194304 use the GPU a lot more efficiently than low ranges like 131072. That's why you get more credit per P for lower ranges -- the idea is the same credit per hour, regardless of how fast the sieve is.

* On the GFN sieve, try playing with the B (blocking) parameter. It can have a significant effect on the amount of lag you get and the speed of the sieve. Note that a larger B parameter is NOT always faster. Both Jim and I have observed that having a B parameter that's too high will slow the sieve down. On my system when running the GFN 131072 range, B9 seems to work best. It's expected that lower B numbers will run slower, but counterintuitively I also found that higher B numbers run slower -- and also have more screen lag.

* I use a simple "go.bat" file to start the sieve. It looks like this:

Even if I would take "big" reservation I cannot make it over 7P since it is limit ( on my case).
Ok I can make few 7P reservation but still it is nothing compared others 200P I sow...
Am I punished for some kind ( all my reservation is done under 16 hours from reservation and results are goos since I got credits)
App says that I can make 16.1P per day.
____________
92*10^^1585996-1 NEAR-REPDIGIT PRIME :) :) :)
4 * 650^⁴⁹⁸¹⁰¹-1 CRUS PRIME
2022202116^¹³¹⁰⁷²+1 GENERALIZED FERMAT
Proud member of team Aggie The Pew. Go Aggie!

Even if I would take "big" reservation I cannot make it over 7P since it is limit ( on my case).
Ok I can make few 7P reservation but still it is nothing compared others 200P I sow...
Am I punished for some kind ( all my reservation is done under 16 hours from reservation and results are goos since I got credits)
App says that I can make 16.1P per day.

Thanks Michael!
Jim, please arise my limit to 16P, it is one day for my GPU :)

____________
92*10^^1585996-1 NEAR-REPDIGIT PRIME :) :) :)
4 * 650^⁴⁹⁸¹⁰¹-1 CRUS PRIME
2022202116^¹³¹⁰⁷²+1 GENERALIZED FERMAT
Proud member of team Aggie The Pew. Go Aggie!

Quick question: is there a priority between GFN 65536 VS 131072, and between GFN 2097152 VS PPR12M?

I'm currently running 1 instance of the 1st 2 options (because we need them) and one instance of the second (because the other 2 just don't use the GPU at all). I don't really have a preference for either, so if you guys would like to choose what I'm going to work on, feel free.....



.... until you release GFN WR on OCL3, that is. I'll switch to that once it comes out :)

Crun-chi

You could make a few reservations then put them in a batchfile so they process sequentially.

Jim, please arise my limit to 16P, it is one day for my GPU :)

Jim, please arise my limit to 16P, it is one day for my GPU :)

Done.

I've modified the upload routines to help eliminate a lot of common mistakes. For example a lot of people doing GFN sieving have uploaded the checkpoint file (starts with a c) rather than the factor file (starts with an f). Some uploads have been for the wrong range. All those should now be caught by the system. Please let me know by private mesasge if you're not able to upload a file you know is correct. I'll see those messages a lot faster than forum posts.

Edit: There was a problem accepting PPR12M uploads, now fixed.

More changes to the sieving system. I've reduced the number of upload dialogs to 4 because now you can select multiple files in each file upload dialog. So if all the factor files are in a single directory, you can select them all at once. There's more checking on the file uploaded.

The text stating what range you're uploading to is now a lot larger and easier to spot. As with the last changes I made, if there's a problem, use that "Send message" on the left to sent me a private message. That triggers an email and I'll know immediately about the problem. I try not to read the forums more than twice a day, which is why posts here are not the best way to get my attention.

... PSA credit doesn't go into your account all at once, doing so would skew the BOINC statistics and make it look like you did an incredible amount of work in a single day. Some PSA credit is added every day, based on your Recent Average Credit number. Pending PSA credit can be seen on your badges page.

These rules predate my membership at PrimeGrid. From what I was given to understand, they were negotiated with the BOINC developers. They date from a time when work was credited at most once a week but could have been once a month. Imagine a month's worth of GPU sieving work (which was done outside of BOINC) all hitting your credit in one transaction. I have no intention of changing the way this works. We promised to distribute credit in relationship to the user's RAC, so that's what we'll keep doing.

To explain, PSA credit goes into its own queue (visible in My Badges and every day at 0400 GMT an amount equal to 80% of the user's Recent Average Credit is moved from the queue into BOINC total credit. Doing so increases the RAC so that more credit will be applied the next day. As of the time I'm writing this, I have more than 3,000,000 credits waiting to move across. According to my stats page on Free-DC (link available on the Your account page) I'm getting somewhere in the neighborhood of 400,000 per day and that's been increasing daily. If I keep up this same level of sieving, I'll eventually reach the point where my entire sieving output is credited the same day. There is no built-in maximum on the system, it depends entirely on RAC (with a 100-credit minimum per day in case the user has a zero RAC).

Your current Recent Average Credit is visible on the Your account page.

Thanks for explanation Jim

____________
92*10^^1585996-1 NEAR-REPDIGIT PRIME :) :) :)
4 * 650^⁴⁹⁸¹⁰¹-1 CRUS PRIME
2022202116^¹³¹⁰⁷²+1 GENERALIZED FERMAT
Proud member of team Aggie The Pew. Go Aggie!

I've been doing a few Primorial PSA units, but I haven't been seeing any PSA credit showing on my account.

Thanks for the info. I didn't know manual sieving credit was handled differently than other PSA projects.

Hey there, long time no see. I'm quite impressed by the new manual sieving system. Good work!

Just two questions:
What's my limit on reservation of work?
And why the heck is fpsieve still bound to 32 threads or less? It's 2015, d'oh! ;)
____________

Hey there, long time no see. I'm quite impressed by the new manual sieving system. Good work!

Just two questions:
What's my limit on reservation of work?
And why the heck is fpsieve still bound to 32 threads or less? It's 2015, d'oh! ;)

What's my limit on reservation of work?

Back on PrimeSearch forum, we used to give general note to a reservation like ETA on when we can expect it to finish, "PC failed; need to rerun my reservation" and similar.

A more general note not limited to computer identification could be used for other purposes and of benefit to both participants and project staff.
____________
My stats

Bad idea, actually. Why have yet another method of communicating, which is one more place we need to check regularly? There's already forums and private messages. Why clutter up web pages with useless fields to pass notes to admins? You already have the ability to communicate with us. You're doing so right now, no?

As for keeping track of one's sieving ranges, there's post-it notes, text files, notebooks, crayons, and, my personal favorite, writing messages in blood. On the walls. Don't forget to write the 'R's backwards.

::removes tongue from cheek::

Seriously, it's not really necessary, and a major concern is not wanting to make the web pages more cluttered than they need to be. Plus Jim's busy on a lot of other stuff.
____________
My lucky number is 75898⁵²⁴²⁸⁸+1

Currently you are allowed 10 simultaneous sieving jobs. That include in-progress work and ranges waiting for admin approval. At this time you cannot create new reservations.

I need 15-16 jobs now, can someone change it now?

"Someone" also needs to learn some manners.

What is the maximum time allowed for a reservation before it gets cancelled?

The instructions include a suggested upload name. For factorial that takes the form of ffactors_{start}-{end}.txt. Here are a few recent uploads:

ffactors_7673G-7688G.txt
ffactors_7703G-7707G.txt
ffactors_7707G-7708G.txt
ffactors_7718G-7722G.txt
ffactors_7722G-7725G.txt

"fsieve" sounds like a sieve file and the code is trying to prevent people uploading the wrong file. For example on GFN sieving there are two files, one starts with a "c" and one starts with an "f". The one starting with "f" is the one I need, the other is just a checkpoint file. If you try to upload the one starting with "c" you get a very specific error message telling you that.

If anyone uploads a file with the wrong name, I wind up manually renaming it. I quickly got tired of renaming factorial factor files named "fsieve.log" (the hardcoded name used by the GPU sieving program) and so the upload code now does it for me. If a factorial upload is not named "fsieve.log" then the code expects the filename to refer to the range sieved like the filenames above. It used to be very common for people to upload the wrong factor file to the wrong reservation. Now they have to work harder to screw it up.

File naming standards are not enforced for .zip, .7z or .rar files. I expect the files inside such archival formats to have standard names, otherwise I have to take time to open them and then name them properly. I tend to try to get through all the day's sieving on a rather tight deadline, so I don't look kindly on misnamed files, especially when every line in such a file generates an error message on my side. That happens if for example the range the filename says it is doesn't match what's in the file.

Manual sieving is just as manual on my side as on yours. Please help me out by naming your upload files the way I'm expecting.

JimB

Had you restarted it, it would have continued from where it left off with at most five minutes wasted. I find that GFN sieving takes very little CPU and by using the W1 parameter on the higher n's it takes almost nothing. Setting too high a B value on a slower GPU can cause display driver crashes, which also crashes the sieving program.

@Jim - so I am back on the factorial sieving and you are correct in that it does give you the file name to send up. If you change the "Sieve using" to GPU and update the file name instructed to send up is "fsieve.log" which has been the file I've been uploading.

I'm now wondering if the naming process for the gpu would work like the instructions for the cpu. If so I can make that happen on my end.

Cheers Rick