Future requests?
 

I see we don't have a thread for requesting new features :razz:

Well... my request is not exactly about GPU-2-72, but about the visualization page. When we go to the[URL="http://www.mersenne.info/trial_factored_tabular_data/0/0/"] max zoom out on the table[/URL], there is a lot of information lost on the right, because the smaller expos (under 14M) are factored not too high, and they have to be on the table. This is valid not only for the status (tabular) view, but for [URL="http://www.mersenne.info/trial_factored_tabular_delta_7/0/0/"]changes table[/URL] too. And it may affect lower levels of zoom too.

So, my suggestion would be to implement either a possibility to "move" left-right with the columns (pan, or select from-to bits), or either implement a larger table (that would go out of screen, but then use firefox's "pan" functions), or either have a checkbox on that page, to "ignore first 10M exponents", or better 15M, 20M, or either to have an "intermediary" zoom level which ignores first row. I keep in mind that the things on the upper ranges of exponents are still evolving (lowest bound is raising) and also less and less people are interested on the lower expos which are already "cleared", "factored", etc.

[QUOTE=LaurV;296490]So, my suggestion would be to implement either a possibility to "move" left-right with the columns (pan, or select from-to bits), or either implement a larger table (that would go out of screen, but then use firefox's "pan" functions), or either have a checkbox on that page, to "ignore first 10M exponents", or better 15M, 20M, or either to have an "intermediary" zoom level which ignores first row. I keep in mind that the things on the upper ranges of exponents are still evolving (lowest bound is raising) and also less and less people are interested on the lower expos which are already "cleared", "factored", etc.[/QUOTE]

A reasonable and logical request.

In all honesty, I haven't spent much time with the Mersenne.info site/code for several months. But I should do so.

Let me get back to you.... :smile:

I've been thinking it would nice to be able to select ranges other than those with boundaries/multiples at powers of 10; that is to say, be able to view 36-46M instead of just 30-40M and 40M-50M; that is to say (again) something like what you did with the "Workers's Progress" graphs on GPU272.

Ya know, not cuz you need more work or anything :razz:

[QUOTE=Dubslow;296522]I've been thinking it would nice to be able to select ranges other than those with boundaries/multiples at powers of 10; that is to say, be able to view 36-46M instead of just 30-40M and 40M-50M; that is to say (again) something like what you did with the "Workers's Progress" graphs on GPU272.[/QUOTE]

Yes... That request was also noted, and added to the todo list. It is also reasonable and logical. :smile:

Is it possible to generate a downloadable worktodo.txt file instead of having to cut and paste the individual lines generated on the 'get assignments' page?

[QUOTE=mognuts;296537]Is it possible to generate a downloadable worktodo.txt file instead of having to cut and paste the individual lines generated on the 'get assignments' page?[/QUOTE]

If this was mersenne.facebook I'd be looking for the "Like" button.

[QUOTE=LaurV;296490]When we go to the[URL="http://www.mersenne.info/trial_factored_tabular_data/0/0/"] max zoom out on the table[/URL], there is a lot of information lost on the right, because the smaller expos (under 14M) are factored not too high, and ...[/QUOTE]

gee for a split second I thought you were going to end your above sentence with:

[QUOTE]"so I'm going to GPU them all to 65 bits just so you don't have to bother changing the report."[/QUOTE]

oh well wishful thinking.
:rofl:

[QUOTE=petrw1;296558]gee for a split second I thought you were going to end your above sentence with:

oh well wishful thinking.
:rofl:[/QUOTE]

What is even funnier (or, perhaps, sadder) is I and a few others compiled our own versions of mprime to remove the sanity check, and took the last few hundred low candidates up to 60 bits using CPUs just for the hell of it a couple of years ago.... :smile:

[QUOTE=petrw1;296557][QUOTE=mognuts]Is it possible to generate a downloadable worktodo.txt file instead of having to cut and paste the individual lines generated on the 'get assignments' page?[/QUOTE]If this was mersenne.facebook I'd be looking for the "Like" button.[/QUOTE]

OK... I'm hearing that this is a desired feature... :wink:

Would it be OK if this was a function from the View Assignments page? It would be [U][I]much[/I][/U] easier to implement there.

[QUOTE=petrw1;296558]gee for a split second I thought you were going to end your above sentence with:

Quote:
"so I'm going to GPU them all to 65 bits just so you don't have to bother changing the report."

oh well wishful thinking.
:rofl:[/QUOTE]

[QUOTE=chalsall;296562]What is even funnier (or, perhaps, sadder) is I and a few others compiled our own versions of mprime to remove the sanity check, and took the last few hundred low candidates up to 60 bits using CPUs just for the hell of it a couple of years ago.... :smile:[/QUOTE]

Yeah, I would do it with pleasure! I am the crazy farang who took M1061 from 62 to 63 bits(**see Notes below) and the same guy who P-1 this exponent to those astronomical limits, [URL="http://www.mersenneforum.org/showpost.php?p=279785&postcount=121"]see here[/URL] (it took few months for P-1 with p95). I did it (the TF) with my own (slow multiplication) tool, who checks all 8kp+1 and 8kp+6p+1 numbers (prime or not), for natural k and p=1061, and the 63-to-64 is on the way too, but interrupted at about 15% some time ago. The same slow tool I use it for the stuff in [URL="http://www.mersenneforum.org/showpost.php?p=294009&postcount=1214"]this post[/URL] (trying P-1 on GPU). I would do "lower expos to 64 and higher" with pleasure if I would have a FAST software tool. I HAVE THE HARDWARE! That is why I am "hitting" other forum members in posts like [URL="http://www.mersenneforum.org/showpost.php?p=296501&postcount=1754"]this one[/URL], and keep my eyes on the range.

Notes:
** that job I reported as "mfaktc-faked-by-LaurV", the "faked" part does not refers to the job (the job was done last year!) but it refers to the report itself, as Primenet doesn't know my tool, and I had to report it as mfaktc. And with this opportunity I found out that the formula to compute the credit has a "gap". (read as "has an error, George, are you there?"). Generally the credit doubles with every bit-step for a given exponent, with the exception of exponents lower then 2000, from 62 to 63 bits, where it has a "times 4" jump. So, the credit it gives for a 62-63 bit on M1061 is about ~900GHzDays, but it should be only ~450GD. The credit for 61 to 62 is correctly given as ~230GD, and so far correct for all lower or higher. The "gap" is only 62 to 63 bits. The same error appears in the [URL="http://mersenne-aries.sili.net/credit.php?worktype=TF&exponent=1061&f_exponent=&b1=&b2=&numcurves=&factor=&frombits=62&tobits=63&submitbutton=Calculate"]mersenne-aries page[/URL], I believe James uses the same formulas. This should be a two-days job on a gtx580 with a "good mfaktc" program, but it took me over two weeks.

[QUOTE=LaurV;296593]And with this opportunity I found out that the formula to compute the credit has a "gap". (read as "has an error, George, are you there?"). Generally the credit doubles with every bit-step for a given exponent, with the exception of exponents lower then 2000, from 62 to 63 bits, where it has a "times 4" jump. So, the credit it gives for a 62-63 bit on M1061 is about ~900GHzDays, but it should be only ~450GD. The credit for 61 to 62 is correctly given as ~230GD, and so far correct for all lower or higher. The "gap" is only 62 to 63 bits. The same error appears in the [URL="http://mersenne-aries.sili.net/credit.php?worktype=TF&exponent=1061&f_exponent=&b1=&b2=&numcurves=&factor=&frombits=62&tobits=63&submitbutton=Calculate"]mersenne-aries page[/URL], I believe James uses the same formulas.[/QUOTE]You can call it a "gap" if you like, but it's not an error. It's designed to reflect [url=http://mersenne-aries.sili.net/throughput.php?cpu1=Intel%28R%29+Core%28TM%292+CPU+6300+%40+1.86GHz|2048|0&mhz1=1000]actual CPU TF performance[/url] (example showing theoretical 1GHz Core2 Duo, basis of GHz-days metric) at around the 62-64 transition. If you look at [url=http://mersenne-aries.sili.net/bench.php]CPU TF benchmarks[/url] you'll notice a large jump in TF times around 62-63 bits (11ms -> 17ms in the Core2 example), but you'll also notice the GHz-days/day TF graph stays flatline at 1.0GHz-days/day despite the slowdown. GHz-days are architechture-specific performance and may not make sense in other contexts, such as using mfaktc which doesn't perform the same (but is analogous to the mfakto slowdown above 2^70).

The actual formula for George's code, which is also used on my site, includes this comment:[code]// bits tf_timing
// ---- ---------
// <=62 0.00465
// 63-64 0.00743
// >=65 0.00707[/code]

Thanks James, meantime I clarified it by myself, as I edited my previous post to put the link to your page in the "notes", then I saw the "show source" button on your page (I am not using it very often). Looking into the source I saw the "$scaling" variable. The time limit forbade me to edit the post.

[QUOTE=chalsall;296562]What is even funnier (or, perhaps, sadder) is I and a few others compiled our own versions of mprime to remove the sanity check, and took the last few hundred low candidates up to 60 bits using CPUs just for the hell of it a couple of years ago.... :smile:[/QUOTE]

Or you could have checked for a v24 client. It did not have the sanity check.

[QUOTE=chalsall;296563]Would it be OK if this was a function from the View Assignments page? It would be [U][I]much[/I][/U] easier to implement there.[/QUOTE]

OK, this has now been implemented on the View Assignments page after regenerating one or more assignments. Below the text to cut and paste there appears a "Download WORKTODO.TXT" button.

Could I ask people check that this works across all browsers and OSs? It works fine under Firefox on Linux, but I'm not sure the HTTP Header MIME type ("application/force-download") will work under Internet Exploder.

I'll also add this abiliity to each of the worktype reservation pages, so it will be available after someone reserves work initially. This is obviously so they don't have to go through the extra steps of Viewing Assignments, selecting the new ones, regenerating and then downloading.

[QUOTE=chalsall;297054]Could I ask people check that this works across all browsers and OSs? It works fine under Firefox on Linux, but I'm not sure the HTTP Header MIME type ("application/force-download") will work under Internet Exploder.[/QUOTE]Does anyone actually [i]use[/i] IE? :surprised

Unsurprisingly, it works perfectly under Windows on Firefox 3, Opera, and Chrome. IE fails to download (just displays as HTML, as in line breaks ignored).

It would probably be better to do this:[code]header('Content-type: text/plain');
header('Content-Disposition: attachment; filename="worktodo.txt"');
echo $worktodo;[/code]

[QUOTE=James Heinrich;297062]It would probably be better to do this:[code]header('Content-type: text/plain');
header('Content-Disposition: attachment; filename="worktodo.txt"');
echo $worktodo;[/code][/QUOTE]

Cool. Thanks James.

Of course, I had to modify the PHP to be Perl.... :wink:

Can you test it again, and make sure it works under Internet Exploder?

[QUOTE=chalsall;297063]Can you test it again, and make sure it works under Internet Exploder?[/QUOTE]Amazing, it does! :smile:
I'm sure I'll be the last person ever to know that. :razz:

@James Heinrich && chalsall :
Good job! :tu:

p.s. still waiting for my visualization-related request, see OP.... :razz:

edit: didn't realize this is first-in-last-served list (just kidding).

[QUOTE=LaurV;296593]Yeah, I would do it with pleasure! I am the crazy farang who took M1061 from 62 to 63 bits... [/QUOTE]
This number was ECM'd to ~200-bit (t60, i.e. 60-digits). It is virtually impossible for it to have 62, 63, 64-bit factors, or even 80-bit or 100-bit factors (if you are thinking about going forward, forward and beyond in your endevour). If you are thinking that "ECM misses some tough factors", then you are thinking about P-1. Estimate the corresponding probabilities using the known formulae.

[QUOTE=LaurV;296593]I HAVE THE HARDWARE![/QUOTE]
That is a fairly poor excuse to effectively waste it. Instead, it is a good idea to let the hardware run something useful and mainstream software -- and in parallel let your brain "run" some new and exciting ideas and possibly find something for the hardware. That's my two cents, but of course you can do whatever you desire if your h/w.

[QUOTE=Batalov;297077]...[/QUOTE]
I generally agree with everything you said. Did you click the link which follows after "I have the hardware" in my post? I am talking about ECM too, in almost the same words.

[QUOTE=James Heinrich;297067]I'm sure I'll be the last person ever to know that. :razz:[/QUOTE]

Indeed. You certainly saved me a great deal of Googling (and head bashing). It is annoying how Micro$oft still delude themselves that they're so dominate that their Internet browser does just about everything differently than everyone else... :cry:

To hopefully help others who have a need for the cross-browser ability to force a download rather than rendering the returned data, the Perl needed (using the CGI Perl module) is:

[CODE]$q = new CGI;
print $q->header( -type => "text/plain",
-Content_Disposition => "attachment; filename=\"worktodo.txt\"");[/CODE]

And, so everyone knows, I've added the "Download WORKTODO.TXT" functionality to all the "Get Assignment" pages.

Plus, I took the opportunity to implement a request from Dubslow from quite some time ago -- the P-1 assignment page now offers Number of CPUs / Number per CPU / [Worker #n] functionality.

[QUOTE=LaurV;297070]p.s. still waiting for my visualization-related request, see OP.... :razz:

edit: didn't realize this is first-in-last-served list (just kidding).[/QUOTE]

I know you were kidding, but so you know... What you've requested is a bit more "tricky" -- the download request was trivial.

Work on your and Dubslow's requests are already underway in the Mersenne.Info development and testing environment. Plus a couple of other things I've been wanting to implement for quite some time.... :smile:

[QUOTE=chalsall;297100]
Plus, I took the opportunity to implement a request from Dubslow from quite some time ago -- the P-1 assignment page now offers Number of CPUs / Number per CPU / [Worker #n] functionality.[/QUOTE]

Ironically, I no longer do work for GPU272 :razz: (Though hopefully I'll return at some point)

[QUOTE=chalsall;297102]I know you were kidding, but so you know... What you've requested is a bit more "tricky" -- the download request was trivial.

Work on your and Dubslow's requests are already underway in the Mersenne.Info development and testing environment. Plus a couple of other things I've been wanting to implement for quite some time.... :smile:[/QUOTE]

Overtake predictions based on work-per-day averages, like from the old Team Prime Rib web site? Not really a request of mine, just that it might be an interesting diversion since most of the parts already seem to be there.

[QUOTE=chalsall;297063]Can you test it again, and make sure it works under Internet Exploder?[/QUOTE]

It works fine with IE9. Thank you for this new feature, I will find it very useful.

Show what exponents would be assigned...
 

Perhaps this is a silly request, it's just something that came to mind. Feel free to dismiss it.

Would it be of interest to people to see what assignments would be made (given the optional range, will factor to, and options settings selected) before actually receiving the assignments? Sort of a "preview" option?

Chuck

[QUOTE=Chuck;297897]Perhaps this is a silly request[/QUOTE]No, it would serve a reasonable purpose.

[quote]Would it be of interest to people to see what assignments would be made (given the optional range, will factor to, and options settings selected) before actually receiving the assignments? Sort of a "preview" option?[/quote]Suppose exponent XXXXXXXX is on the preview list. Between the time you see the list and the time you indicate that you want to go ahead with the assignment, exponent XXXXXXXX might be assigned to someone else. Indeed, it's possible that the none of the assignments you actually receive will be on the preview list because while you were pondering the preview list [strike]PrimeNet[/strike] the relevant central assignment authority[sup]*[/sup] assigned all those exponents to other non-previewing users.

One fix for that would be to place a "hold" on the exponents shown to you in the preview -- but ... There's already a workaround that has long been in use: just go ahead with your assignment request, then promptly review the list of exponents you received, then release the assignments that you don't want. You're welcome to do that (of course, you're expected to release unwanted assignments in not-too-long a time).

- - -

[sup]*[/sup] Tip of the hat to Dubslow :-)

[QUOTE=cheesehead;297902] PrimeNet [/QUOTE]
Umm... GPU to 72 forum...
:smile:
(Good post otherwise :razz:)

[QUOTE=cheesehead;297902]One fix for that would be to place a "hold" on the exponents shown to you in the preview[/QUOTE]

More code for me, plus an additional field in the Assignments table... :cry:

[QUOTE=cheesehead;297902] -- but ... There's already a workaround that has long been in use: just go ahead with your assignment request, then promptly review the list of exponents you received, then release the assignments that you don't want. You're welcome to do that (of course, you're expected to release unwanted assignments in not-too-long a time).[/QUOTE]

Indeed. This technique is already used by several people. They ask for ten or so, and if they don't like them they unreserve them. If they do, then they reserve some more using the same settings.

If there was enough demand, I could add a button to the LLTF, DCTF and P-1 assignments to unreserve everything just assigned.

As an aside, please don't do this for true LL and DC assignments. They are always assigned in order of Exponent (read: lowest first), and because of the AnonSpidy trick we use any LL and DC assignments which are unreserved have to go through the un-register/re-register cycle in order to get a new AID.

I should have added a note:\

[QUOTE=cheesehead;297902]There's already a workaround that has long been in use: just go ahead with your assignment request, then promptly review the list of exponents you received, then release the assignments that you don't want.[/QUOTE]Note that this workaround does, in effect, accomplish the same thing as the review/hold procedure, with no added code.

Next and Previous on "Completed Assignments"
 

I think a next and prev. page would work here... after a while the list kinda gets... Unless you want to see everything you've done on one page... that might have been the intent.

[URL]http://gpu72.com/account/completed/[/URL]

[QUOTE=kracker;302222]I think a next and prev. page would work here... after a while the list kinda gets... Unless you want to see everything you've done on one page... that might have been the intent.[/QUOTE]

Currently that report is limited to 1000 entries. Is that too much?

I could shorten that down, and add a "next" function if desired. Please let me know.

[QUOTE=chalsall;302287]Currently that report is limited to 1000 entries. Is that too much?

I could shorten that down, and add a "next" function if desired. Please let me know.[/QUOTE]

Ah, I didn't know it was set to 1000, any way is fine :smile:

@chalsall
Man, we have a saying: the dogs are barking, the bear is passing. Instead of wasting your time arguing with Mr.D. in the parallel thread (thanks god some mod decided to close it!), better invest that time in modifying those tables I am talking about since ages... and let the dogs bark.

To remind you, [U]first case[/U] was about the "view assignments" page, first table (I think is called production heuristics), where the LL and DC are together in the same line, there is no line for "total" or "overall" like in individual reports tables, and this really started to piss me off since we have more/different work types (like LLP-1 and DCP-1), they may need separate rows too. That is minor, but counting how many of my assignments are LL and how many are LLDC makes me crazy. If I leave to look in the other table where they are separate, when I come back I need to input the password again (I don't store passwords, except for my computer at home).

The [U]second case[/U] (less but still important) is the visualization tool table, cutting first line, or having custom ranges (the post which started this very current "[URL="http://www.mersenneforum.org/showthread.php?t=16727"]feature request[/URL]" thread, still so long time ignored....)

[QUOTE=LaurV;303484]@chalsall Man, we have a saying: the dogs are barking, the bear is passing.[/QUOTE]

LaurV... Please appreciate that GPU72 and Mersenne.info are a hobby. Although, admittedly, one that consumes quite a bit of my time...

Everything (or, at least, just about everything) that anyone asks for is placed on a todo list. Some are easy to do and are done quickly. Others are a bit more involved, and take some time.

Your request for breaking out the heuristics information is more involved that it might appear, because that data is used by the system for more than just presenting the information to the user.

The changes on Mersenne.info you suggested are underway in my development environment. I'm taking the opportunity to doing a fairly major upgrade to that system. The videos I just exposed were something I had almost competed months ago, and finally got back to.

Patience grasshopper....

I have a suggestion for a feature: I'd like to know whether my DCs came back that my result was 1. good, (matched a previous result) 2. unverified, (didn't match any previous result) or 3. bad (other results matched). My reason for this is that monitoring DC results is a good indicator that my machine is, without a doubt, stable. Mersenne.org doesn't already show you this information on your [URL="http://www.mersenne.org/results/"]results[/URL] page, it just shows that the LL returned as "C" (composite) and shows the result with the last two bytes masked, regardless of whether your result was good or bad.

when manually reporting, PrimeNet is saying if the residues matched or not, indirectly, by saying "this clears the exponent" bla bla. This can also be implemented in the spiders (or maybe is already, I don't use spiders).

[QUOTE=LaurV;303646]when manually reporting, PrimeNet is saying if the residues matched or not, indirectly, by saying "this clears the exponent" bla bla. This can also be implemented in the spiders (or maybe is already, I don't use spiders).[/QUOTE]

Thanks for the info. I've been using Prime95 to work and report the work. I haven't checked the logs or comm. window to see if it reports this info to the user (I can't check it at the moment). If it reports this, I for one am OK with how it is now (though the spiders reporting it as well - if they don't already - would be a good addition IMHO). :smile:

In Prime95 it will look like this:
[code][Work thread Jun 27 07:59:22] Iteration: 25020000 / 25024693 [99.9812%]. Per iteration time: 8.304 ms.
[Work thread Jun 27 08:00:01] M25024693 is not prime. Res64: 505B65B8808CE497. We8: 9C81285E,11843949,00000000
[Work thread Jun 27 08:00:01] No work to do at the present time. Waiting.
[Comm thread Jun 27 08:00:01] Sending result to server: UID: Dubslow/Gravemind, M25024693 is not prime. Res64: 505B65B8808CE497. We8: 9C81285E,11843949,00000000, AID: 153DF6EA5F0CC33CB045339C4DDE22D0
[Comm thread Jun 27 08:00:01]
[Comm thread Jun 27 08:00:02] PrimeNet success code with additional info:
[B][Comm thread Jun 27 08:00:02] LL test successfully completes double-check of M25024693[/B]
[Comm thread Jun 27 08:00:02] CPU credit is 21.4448 GHz-days.[/code]
If you get a mismatch, that line will be missing.

As far as I know, the only spider that submits LL results is the one I made for CUDALucas (based on chalsall's mfakt* spider). That one checks for a mismatch before submitting, so that if you didn't get a match, you retain the assignment so you can ask somebody else here to run their own quick TC for you.

Request: Overtake dates?

[SIZE=1](Newegg should sponsor that column!)[/SIZE]

So since I figured that we would be taking all of the remaining exponents to 73 since we seem to be safely over 58M, I decided to put in that I wanted to go to 73, and asked it to grab what makes sense.

It gave me exponents from 72->73 from 52M to 55M.

This is not what I wanted so I threw them all back and just went back to doing it to 72.

[QUOTE=KyleAskine;304109]So since I figured that we would be taking all of the remaining exponents to 73 since we seem to be safely over 58M, I decided to put in that I wanted to go to 73, and asked it to grab what makes sense.

It gave me exponents from 72->73 from 52M to 55M.

This is not what I wanted so I threw them all back and just went back to doing it to 72.[/QUOTE]

The problem was you changed the highest range to be 60,000,000, rather than leaving it as the default of 100,000,000.

"What makes sense" falls back to exactly what you ask for if you change the ranges.

Sorry -- that probably doesn't make sense....

[QUOTE=chalsall;304117]The problem was you changed the highest range to be 60,000,000, rather than leaving it as the default of 100,000,000.

"What makes sense" falls back to exactly what you ask for if you change the ranges.

Sorry -- that probably doesn't make sense....[/QUOTE]

That is because I saw there was a lone 58M exponent sitting at 70 bits. I wanted to grab it so I clicked that I wanted to take the lowest TF to 72.

This gave me only the numbers > 60M since they were all at 69 bits. :(

I can just try to set better ranges and do it myself. I mean, the tools are there to grab everything myself, but I am lazy and stupid.

Also... after you grab some exponents, you go to a screen telling you what they are and giving you an option to download a worktodo.txt. Can we also have a throw these back button? I mess up quite often (like last night), and have to go back and click the radio boxes 100 times to throw them all back.

I am the idiot user you need to idiot proof for!

[QUOTE=KyleAskine;304173]Can we also have a throw these back button?[/QUOTE]I would also like to see that, I assume it should be a fairly simple implementation.

[QUOTE=James Heinrich;304175]I would also like to see that, I assume it should be a fairly simple implementation.[/QUOTE]

Yeah, quite easy.

There is now a "Unreserve all of these" button on the LL TF, DC TF, LL P-1 and DC P-1 assignment pages.

Can [STRIKE]we[/STRIKE]you rename the buttons to "Go Fishing" and "Throw them back"?

:popcorn:

[QUOTE=Dubslow;304184]Can [STRIKE]we[/STRIKE]you rename the buttons to "Go Fishing" and "Throw them back"?

:popcorn:[/QUOTE]Or "Catch" and "Release"?

Would it be possible to make an option on the overall statistics to set the max # of days shown? I have 8 months of data and it has shrunk the graph down to where it is impossible to see anythong other than a broad generalization. Maybe leave that one there and add a 'past 100 days' graph so it is easier to see?

Feature request: Ability to optionally tag an assignment with a machine name. Perhaps allow for sorting and/or stats based on machine names.

Use case: I have two main machines doing P-1 work. The larger of the two, I load up with more assignments and do it less often. My laptop, I load up usually with only one or two assignments at a time and I add assignments as needed. The machines are different speeds. It would be nice if there were some way to differentiate between the two in the stats, or at least on my assignments page. There have been times in the past where because the larger of the two machines had such a deep stack of assignments, I had trouble getting new single assignments for the laptop.

This is not a fully baked feature request I guess, but I wanted to write it down somewhere before I forgot about it.

--

Side question: Is there a way to request a small LL with a given TF level? Last night there was one LL in the 45M range that was TF'd in excess of 73 bits. I kinda want to grab it, but I kept getting offered one that was "only" TF'd to 72.

chalsall: On our 'Factors found by' page, can you add a sortable date column?

[QUOTE=flashjh;304347]chalsall: On our 'Factors found by' page, can you add a sortable date column?[/QUOTE]

Easy. Done.

[QUOTE=KingKurly;304312]Feature request: Ability to optionally tag an assignment with a machine name. Perhaps allow for sorting and/or stats based on machine names.[/QUOTE]

Yeah. Bdot asked for something similar a while ago. It's on my todo list.

[QUOTE=KingKurly;304312]Side question: Is there a way to request a small LL with a given TF level? Last night there was one LL in the 45M range that was TF'd in excess of 73 bits. I kinda want to grab it, but I kept getting offered one that was "only" TF'd to 72.[/QUOTE]

While it wouldn't be difficult to do, there isn't currently any way to do that, other than ask for a bunch, and throw back what you don't like. But [B][I][U]PLEASE[/U][/I][/B] don't do that -- the way LL and DC assignments are handled (AnonSpidy trick), any assignments which are unreserved have to go through a process of being passed back to PrimeNet and then recaptured.

Is this something people actually want? As it is the system gives the lowest available candidate, which in my mind makes the most sense as the lower the candidate, the faster it is to process. Being TFed to a slightly higher level only increases its likelihood of being Prime by a tiny ([I][U]tiny[/U][/I]!!!) amount.

[QUOTE=bcp19;304254]Would it be possible to make an option on the overall statistics to set the max # of days shown? I have 8 months of data and it has shrunk the graph down to where it is impossible to see anythong other than a broad generalization. Maybe leave that one there and add a 'past 100 days' graph so it is easier to see?[/QUOTE]

Yeah, I have to climb into that section of code, and handle those who have been with the project since the beginning. Been thinking about how best to handle this.

It's high on my list....

[QUOTE=chalsall;304351]While it wouldn't be difficult to do, there isn't currently any way to do that, other than ask for a bunch, and throw back what you don't like. But [B][I][U]PLEASE[/U][/I][/B] don't do that -- the way LL and DC assignments are handled (AnonSpidy trick), any assignments which are unreserved have to go through a process of being passed back to PrimeNet and then recaptured.[/QUOTE]Hmm, even if we never actually claim the assignment with mprime (completing the transfer)? I would've thought that assignment would stay with you. (No worries, it's not like I put a lot of effort into trying to get the test, I think I got 2-3 assignments and threw them back without ever 'claiming' them. And I won't do it again.) ...Oh. It just clicked as to why it matters. You need a new AID so that I/others don't know the old one. Makes sense now. Okay. Won't do that again.

[QUOTE=chalsall;304351]
Is this something people actually want?[/QUOTE]Apparently. :smile: But not something I feel strongly about.
[QUOTE=chalsall;304351]As it is the system gives the lowest available candidate, which in my mind makes the most sense as the lower the candidate, the faster it is to process. Being TFed to a slightly higher level only increases its likelihood of being Prime by a tiny ([I][U]tiny[/U][/I]!!!) amount.[/QUOTE]Fair enough. It was more a "hey it's there, why not take the 'nicest' one available?"

[QUOTE=chalsall;304350]Easy. Done.[/QUOTE]
Thank you (and to whomever moved my request to the appropriate thread)

Assignments preview...
 

Hey all.

Since several people are using the "catch and release" technique to see what assignments are available, I figured I'd make it easier for everyone...

The LL and DC Trial Factoring assignment pages now have an AJAX preview showing what assignments you'd be given based on the parameters entered into the form. The last three digits of each candidate are hidden, and the GHz Days of work being pledged is shown.

Obviously this is not guaranteed -- someone else might grab the assignments you're viewing in between the time you see the preview and you actually click the "Get Assignments" button. Or Spidy might get something new from PrimeNet in the interim. But it should help lower the releasing cycling....

That's pretty cool.

-- Craig

"Obviously this is not guaranteed -- someone else might grab the assignments..."

Disclaimer acknowledged, but this is a welcome development. Available assignments evaporating twixt the cup and the lip is hardly a new phenomenon. Thanks for this and all your work, Chris.

[QUOTE=bcp19;304254]Would it be possible to make an option on the overall statistics to set the max # of days shown? I have 8 months of data and it has shrunk the graph down to where it is impossible to see anythong other than a broad generalization. Maybe leave that one there and add a 'past 100 days' graph so it is easier to see?[/QUOTE]

After playing around with this for a bit, I've come to the conclusion (and implemented) showing the latest 236 days worth of results on the "GHz Days Done" and "Saved" graphs for those who have been with the project for a long time is the best solution.

The cumulative Work Done graph (being line rather than bar) did not need this -- it continues to show the full history.

Great work! (Referring to the status thread)

I've had an idea: for the [URL="http://www.gpu72.com/reports/estimated_completion/"]Estimated Completion[/URL] chart, could you add a third column after PrimeNet that shows how many expos are assigned for LL, and therefore not accessible to us? It shouldn't be too hard to spider that information from PrimeNet's [URL="http://www.mersenne.org/primenet/"]status page[/URL] (e.g. [code]/5[0-9]{1}000000 5[0-9]{4} | [0-9]{5}( [0-9 ]{5}){4} |( [0-9 ]{5}){2} ([0-9 ]{5}) [0-9 ]{5} |/ # Group 3 is the target[/code]

The change that shows you a sample of the assignments you'll get on the get assignments page is really cool!

Request: Email notification of soon-to-be-expired exponents?

When one of my DC finishes, I always check if the residue matches the one from the original test by clicking the exponent on the 'Completed Assignments' page. Maybe this can be displayed on that page as well? Displaying the exponent row in green or red or some other way?

We have now so many beautiful new columns in the [URL="http://www.gpu72.com/account/assignments/"]"assignment" report table[/URL]... Percent... Age... Last Updated... Estimated Completion... Days To Go... what a pity we can't SORT the table by clicking those columns... And generally, what a pity that for existent sortable columns the sorting direction does not change with alternative clicks...

[QUOTE=LaurV;307252]We have now so many beautiful new columns in the [URL="http://www.gpu72.com/account/assignments/"]"assignment" report table[/URL]... Percent... Age... Last Updated... Estimated Completion... Days To Go... what a pity we can't SORT the table by clicking those columns... And generally, what a pity that for existent sortable columns the sorting direction does not change with alternative clicks...[/QUOTE]

Sigh... :wink:

Give me some time LaurV. I first need to build a new "Account Settings" page so everyone can enjoy this new ability...

Oh, completely unrelated but I'd might as well mention it now... I've installed a SSL cert on the GPU72.com server. This means you can now access the system using [url]https://www.gpu72.com/[/url]. Not that there's any reason to do so -- it's not like it's an e-commerce site or anything, but I like to experiment.

Please do note that because of the way SSL works, this will only work cleanly with the [url]www.gpu72.com[/url] domain; the language sub-domains will warn that the cert is not valid. If you proceed anyway the traffic will still be encrypted but you'll probably see a [COLOR="Red"][strike]https:[/strike][/COLOR] or similar indication in your browser's URL display.

[QUOTE=BigBrother;307039]When one of my DC finishes, I always check if the residue matches the one from the original test by clicking the exponent on the 'Completed Assignments' page. Maybe this can be displayed on that page as well? Displaying the exponent row in green or red or some other way?[/QUOTE]

That's an excellent suggestion. Added to my "Todo" list.

[QUOTE=chalsall;307255]Sigh... :wink:[/QUOTE]

PrimeNet uses [URL="http://www.kryogenix.org/code/browser/sorttable/"]this thingy[/URL] to sort its tables, it's stupidly easy to use... (but works, doesn't require a whole other page, etc...)

[QUOTE=Dubslow;307263]PrimeNet uses [URL="http://www.kryogenix.org/code/browser/sorttable/"]this thingy[/URL] to sort its tables, it's stupidly easy to use... (but works, doesn't require a whole other page, etc...)[/QUOTE]

I'd thought of that. Certainly lighter than jQuery in cases where you're not already needing it.

Ability to "drop" computers? (like clones)
[URL]http://www.gpu72.com/account/computers/[/URL]

[QUOTE=kracker;308609]Ability to "drop" computers? (like clones)
[URL]http://www.gpu72.com/account/computers/[/URL][/QUOTE]
or merge like primenet.:smile:

Is there a way to create a new worktype that allows the user to choose a ratio of first time checks to double checks?

For example, say we want 5 first time checks and then a double check, repeatedly.

We thing (?) this would allow your system to flag computers that do not have matching double check results as suspect.

:chalsall:

@chalsall: On the Workers' Overall Progress report, could you insert the heading columns (no need links, same as in the bottom of the table) somewhere in the middle of the table? Say after position 50, and try to keep it inserted every 50 lines if the table grow? (like there will be one thousand users in the future, don't ask me to scroll up and down to see which column is P-1 results and which one is TF to 69 bits?) (well, by that time we may have more columns too, like TF to 147 bits, and it would be even more difficult to sail through so many numbers/columns/lines :razz:)

Hmm, it already may be implemented, but is there a way to show work done in GPU72 daily/weekly/monthly etc. for a specific category or all?

[QUOTE=kracker;312664]Hmm, it already may be implemented, but is there a way to show work done in GPU72 daily/weekly/monthly etc. for a specific category or all?[/QUOTE]

There are the [URL="https://www.gpu72.com/reports/overall/graph/"]weekly[/URL], [URL="https://www.gpu72.com/reports/overall/graph/month/"]monthly[/URL], [URL="https://www.gpu72.com/reports/overall/graph/quarter/"]quarterly[/URL] and [URL="https://www.gpu72.com/reports/overall/graph/halfyear/"]half-yearly[/URL] graph reports. These are accessible from a sub-menu which should appear when you "hover" over the "Overall System Progress" report menu item. Or, alternatively, click on the sole graph under the [URL="https://www.gpu72.com/reports/overall/"]Overall System Progress[/URL] report, then click on any of the graphs to show the next level.

Note that the data on the farthest right will usually show a marked decline, as many of our workers only submit results once every week or so. When they do, the data is "smoothed" across the period of time that the assignments have been out.

Does this provide what you're after?

[QUOTE=chalsall;312666]There are the [URL="https://www.gpu72.com/reports/overall/graph/"]weekly[/URL], [URL="https://www.gpu72.com/reports/overall/graph/month/"]monthly[/URL], [URL="https://www.gpu72.com/reports/overall/graph/quarter/"]quarterly[/URL] and [URL="https://www.gpu72.com/reports/overall/graph/halfyear/"]half-yearly[/URL] graph reports. These are accessible from a sub-menu which should appear when you "hover" over the "Overall System Progress" report menu item. Or, alternatively, click on the sole graph under the [URL="https://www.gpu72.com/reports/overall/"]Overall System Progress[/URL] report, then click on any of the graphs to show the next level.

Note that the data on the farthest right will usually show a marked decline, as many of our workers only submit results once every week or so. When they do, the data is "smoothed" across the period of time that the assignments have been out.

Does this provide what you're after?[/QUOTE]

Oh, sorry, I meant it for the "specific" user, something *like* [URL="https://www.gpu72.com/reports/workers/lltf/"]this[/URL], except it would show work done for that user month/day/week etc

On the page for getting assignments, in the box that says 'Sample assignments based on the above settings...', can you add a line there that gives a good estimate of the total days it will take to do the work selected based on the current average and the number of assignments to reserve? Since you already put 'xx.xxx GHz Days; x.xxx of your days.' there, I can use the # of assignments and multiply, but I'm being lazy ;)

[QUOTE=flashjh;314512]On the page for getting assignments, in the box that says 'Sample assignments based on the above settings...', can you add a line there that gives a good estimate of the total days it will take to do the work selected based on the current average and the number of assignments to reserve? Since you already put 'xx.xxx GHz Days; x.xxx of your days.' there, I can use the # of assignments and multiply, but I'm being lazy ;)[/QUOTE]
That wouldn't be practical for anyone who has more than 1 GPU.

[QUOTE=bcp19;314583]That wouldn't be practical for anyone who has more than 1 GPU.[/QUOTE]
I guess I don't see why? What do you mean because in my assignments page I can see my average per day and when I select assignments it tells me x.xxx of your days. I'm just asking for a general number so if I select 10 or 100, etc., assignments I can see about what it will take me to finish them. It's my job to divide them as necessary over the GPUs.

In either case, it's just a nicety for when I'm selecting exponents.

[QUOTE=flashjh;314585]I guess I don't see why? What do you mean because in my assignments page I can see my average per day and when I select assignments it tells me x.xxx of your days. I'm just asking for a general number so if I select 10 or 100, etc., assignments I can see about what it will take me to finish them. It's my job to divide them as necessary over the GPUs.

In either case, it's just a nicety for when I'm selecting exponents.[/QUOTE]
I have 7 GPUs running 18 instances of mfaktc/o and average over 1100 GHzD/Day. I add assignments on a per instance, not a per GPU case. 2 of the GPUs I request 25 assignments at a time (for 5 total instances) which would be roughly 135 GHzD work or ~3 hours when compared to my average. 1 GPU (running 2 instances) I get 100 assignments at a time, ~1654 GHzD or 1.5 days, the other 4 get 50 at a time and combined have 11 instances, but get an added 180-800GHzD work due to assignment type.

The GPUs I get 25 assignments for are kinda slow, doing 5-6 assignments per day, so this adds 4-5 days of work. The GPU I get 100 for completes 11 per day, so this adds ~9 days of work and the ones with 50 complete between 6 and 9 per day which adds between 5 and 9 days of work. I check file size daily to determine if more assignments need to be added, so some days I have 0-3 files I add work to, some days 8-10.

[QUOTE=bcp19;314699]I have 7 GPUs running 18 instances of mfaktc/o and average over 1100 GHzD/Day. I add assignments on a per instance, not a per GPU case. 2 of the GPUs I request 25 assignments at a time (for 5 total instances) which would be roughly 135 GHzD work or ~3 hours when compared to my average. 1 GPU (running 2 instances) I get 100 assignments at a time, ~1654 GHzD or 1.5 days, the other 4 get 50 at a time and combined have 11 instances, but get an added 180-800GHzD work due to assignment type.

The GPUs I get 25 assignments for are kinda slow, doing 5-6 assignments per day, so this adds 4-5 days of work. The GPU I get 100 for completes 11 per day, so this adds ~9 days of work and the ones with 50 complete between 6 and 9 per day which adds between 5 and 9 days of work. I check file size daily to determine if more assignments need to be added, so some days I have 0-3 files I add work to, some days 8-10.[/QUOTE]

That makes sense. I have 5 and I fill them all at the same time. Sometimes I have to balance a few workers, but since the cards are all the same it works out pretty well for me. I just wish I had a program to take my assignments and automatically distribute them to the workers.

[QUOTE=bcp19;314583]That wouldn't be practical for anyone who has more than 1 GPU.[/QUOTE]

This is the issue I have to deal with.

The estimate is based on the individual user's overall performance -- not the individual Computer, nor the individual GPU instance (since even one GPU needs more than one instance to be efficient).

A non-trivial, but not intractable, problem.

Unfortunately I've been a little busy lately on other matters.

[QUOTE=flashjh;314701]That makes sense. I have 5 and I fill them all at the same time. Sometimes I have to balance a few workers, but since the cards are all the same it works out pretty well for me. I just wish I had a program to take my assignments and automatically distribute them to the workers.[/QUOTE]
I thought the mfaktXapp did just that.

[QUOTE=bcp19;315045]I thought the mfaktXapp did just that.[/QUOTE]

I haven't seen it yet. I will check it out. Thanks for the tip.

On the page that gives you all the assignments there is a button to make a worktodo.txt file. Can you add a button to copy all the newly assigned work to the clipboard?

[QUOTE=flashjh;322421]On the page that gives you all the assignments there is a button to make a worktodo.txt file. Can you add a button to copy all the newly assigned work to the clipboard?[/QUOTE]

I can look into that. However, can you not simply highlight the assignments and press Ctrl-C?

[QUOTE=chalsall;322426]I can look into that. However, can you not simply highlight the assignments and press Ctrl-C?[/QUOTE]
Of course I can, and that's what I do. Just asking ;)

[QUOTE=flashjh;322433]Of course I can, and that's what I do. Just asking ;)[/QUOTE]

This is non trivial, since it is dangerous.

Naturally M$'s Internet Exploder allows it, but all other browers (correctly) make it difficult.

Since all the alternative solutions come down to pre-selecting the text which the user then has to Cntl-C, I'm just going to leave the system as it is.

This is a non issue in my opinion but I have a suggestion on how it could almost be done.

How about putting the text in an FORM text edit element, that way you only need to click the text and press ctrl-A and ctrl-C to mark the text.
I usually makes the mistake and presses those keys before thinking and ends up marking all of the page instead of just the assignment. :grin:

/Göran

[QUOTE=Axelsson;322442]How about putting the text in an FORM text edit element, that way you only need to click the text and press ctrl-A and ctrl-C to mark the text.[/QUOTE]Focusing the element and automatically selecting the text on page load with element.select() is trivial and takes care of "ctrl-A". It's the "ctrl-C" part that no longer works in modern browsers (it used to be a simple part of browser Javascript implementations, but that disappeared 5+ years ago for security reasons.

Largest 100 factors chart
 

I'd like to see a sortable "Date found" column added to the [B]Largest 100 Factors Found[/B] chart. When I get bumped down the list due to someone finding a larger factor, it would be nice to be able to see who that was.

This is really getting to be a rarified list — you don't even make the list now unless your factor is over 104 bits.

Hey Chris, can you pull in some 37M-38M exponents for LLDC? (and assign them to me, hehe)

Kracker talked me into it, and he is perfectly right, thinking about efficiency of LL-ing with clLucas, for which the 2097152 FFT is much faster comparing with other sizes near by ([U]double speed[/U] on our cards, compared to 1M9 FFT which would be the default for 30M expos). Therefore, a 30M test and a 37M test takes about the same time with clLucas, using this FFT (doing a 30M with the default FFT take double time!). So why not use clLucas at its max efficiency? As ~38M is the limit of the expos which can use this FFT, why not?

Now, I could pull the expos by myself from PrimeNet, but that way I won't get credit on GPU72, and this guy is out for my neck, you know... :razz:

[QUOTE=LaurV;355232]Now, I could pull the expos by myself from PrimeNet, but that way I won't get credit on GPU72, and this guy is out for my neck, you know... :razz:[/QUOTE]

LOL...

Certainly -- trivial to facilitate.

Could you give me an exact range you're interested in? Or, alternatively, the highest exponent you're interested in?

I would point out that many of these won't be TFed as far as is now appropriate. I'll bring in several of the very highest TFed I can find, plus several more at lower TF levels. Hopefully some of the DC TFer's will contribute some cycles to manage and "feed" this new "micro-wave". :smile:

You are my man, I knew it! I can do the proper TF by myself.

From here, the max exponent I would be interested would be around 38900000. Say that anything between 37M5 and 38M85 would be perfect, going higher the error is a bit extreme and I fear "resumes" with "increased" (and much slower) FFT.

[CODE]Microsoft Windows [Version 6.1.7601]
Copyright (c) 2009 Microsoft Corporation. All rights reserved.

e:\-99-Prime\clLucas>cll 37000000
Platform :Advanced Micro Devices, Inc.
Device 0 : Tahiti


start M37000000 fft length = 1966080
^C caught. Writing checkpoint.

e:\-99-Prime\clLucas>cll -c 100 -t 37000000
Platform :Advanced Micro Devices, Inc.
Device 0 : Tahiti


continuing work from a partial result M37000000 fft length = 1966080 iteration = 498
Iteration 500 M( 37000000 )C, 0x4b6379ed9a066ad9, n = 1966080, clLucas v1.01 err = 0.2344 (0:01 real, 1.6293 ms/iter, ETA 16:44:42)
Iteration 600 M( 37000000 )C, 0xde63e79aa67e8f91, n = 1966080, clLucas v1.01 err = 0.3125 (0:00 real, 8.9109 ms/iter, ETA 91:34:58)
Iteration 700 M( 37000000 )C, 0xac10c1595c4679d8, n = 1966080, clLucas v1.01 err = 0.3125 (0:00 real, 9.0032 ms/iter, ETA 92:31:49)
Iteration 800 M( 37000000 )C, 0x4c27ce026b9ae5da, n = 1966080, clLucas v1.01 err = 0.3438 (0:00 real, 8.8773 ms/iter, ETA 91:14:12)
err = 0.375, increasing n from 1966080

start M37000000 fft length = 2097152
Iteration 100 M( 37000000 )C, 0xe5d22ad7ff931756, n = 2097152, clLucas v1.01 err = 0.0625 (0:00 real, 5.9706 ms/iter, ETA 61:21:51)
Iteration 200 M( 37000000 )C, 0x3c3b3e3d7920401d, n = 2097152, clLucas v1.01 err = 0.0625 (0:01 real, 5.5693 ms/iter, ETA 57:14:24)
Iteration 300 M( 37000000 )C, 0x7d64ce0b69e906dc, n = 2097152, clLucas v1.01 err = 0.0625 (0:00 real, 5.7091 ms/iter, ETA 58:40:35)
Iteration 400 M( 37000000 )C, 0x1cd77766d6c06764, n = 2097152, clLucas v1.01 err = 0.0625 (0:01 real, 5.6013 ms/iter, ETA 57:34:05)
Iteration 500 M( 37000000 )C, 0x4b6379ed9a066ad9, n = 2097152, clLucas v1.01 err = 0.0625 (0:01 real, 5.5778 ms/iter, ETA 57:19:33)
^C caught. Writing checkpoint.

e:\-99-Prime\clLucas>cll -c 100 -t 39000000
Platform :Advanced Micro Devices, Inc.
Device 0 : Tahiti

start M39000000 fft length = 1966080
err = 0.46875, increasing n from 1966080

start M39000000 fft length = 2097152
Iteration 100 M( 39000000 )C, 0x9428aa08031269b9, n = 2097152, clLucas v1.01 err = 0.2422 (0:01 real, 5.8975 ms/iter, ETA 63:53:20)
Iteration 200 M( 39000000 )C, 0x554f712ca7404cb8, n = 2097152, clLucas v1.01 err = 0.2422 (0:00 real, 5.6646 ms/iter, ETA 61:21:57)
Iteration 300 M( 39000000 )C, 0x840a71e3cd509484, n = 2097152, clLucas ..........
Iteration 700 M( 39000000 )C, 0x50c87b4c1545a09c, n = 2097152, clLucas v1.01 err = 0.25 (0:00 real, 5.6653 ms/iter, ETA 61:22:23)
Iteration 800 M( 39000000 )C, 0xbf1da98c8579a5b9, n = 2097152, clLucas v1.01 err = 0.25 (0:01 real, 5.7615 ms/iter, ETA 62:24:54)
.................
Iteration 2300 M( 39000000 )C, 0xfe5b6160e4d53b21, n = 2097152, clLucas v1.01 err = 0.25 (0:01 real, 5.6537 ms/iter, ETA 61:14:41)
^C caught. Writing checkpoint.

e:\-99-Prime\clLucas>cll -c 100 -t 38500000
Platform :Advanced Micro Devices, Inc.
Device 0 : Tahiti

start M38500000 fft length = 1966080
err = 0.5, increasing n from 1966080

start M38500000 fft length = 2097152
Iteration 100 M( 38500000 )C, 0xf417504c714b0041, n = 2097152, clLucas v1.01 err = 0.1563 (0:00 real, 6.0322 ms/iter, ETA 64:30:37)
Iteration 200 M( 38500000 )C, 0x671e287d0342b172, n = 2097152, clLucas v1.01 err = 0.1563 (0:01 real, 5.6750 ms/iter, ETA 60:41:27)
^C caught. Writing checkpoint.

e:\-99-Prime\clLucas>cll -c 100 -t 38800000
Platform :Advanced Micro Devices, Inc.
Device 0 : Tahiti

start M38800000 fft length = 1966080
err = 0.5, increasing n from 1966080

start M38800000 fft length = 2097152
Iteration 100 M( 38800000 )C, 0x2ebd3da835145b3b, n = 2097152, clLucas v1.01 err = 0.2031 (0:01 real, 6.0696 ms/iter, ETA 65:24:58)
Iteration 200 M( 38800000 )C, 0x026e432efa7bf419, n = 2097152, clLucas v1.01 err = 0.2188 (0:00 real, 5.6810 ms/iter, ETA 61:13:41)
Iteration 300 M( 38800000 )C, 0x9d3433e16cdaf1f0, n = 2097152, clLucas v1.01 err = 0.2188 (0:00 real, 5.6626 ms/iter, ETA 61:01:47)
Iteration 400 M( 38800000 )C, 0x7cf749c542bddb06, n = 2097152, clLucas v1.01 err = 0.2188 (0:01 real, 5.9468 ms/iter, ETA 64:05:31)
^C caught. Writing checkpoint.

e:\-99-Prime\clLucas>cll -c 100 -t 38900000
Platform :Advanced Micro Devices, Inc.
Device 0 : Tahiti

start M38900000 fft length = 1966080
err = 0.5, increasing n from 1966080

start M38900000 fft length = 2097152
Iteration 100 M( 38900000 )C, 0x7814abe1baede2be, n = 2097152, clLucas v1.01 err = 0.2188 (0:01 real, 5.9974 ms/iter, ETA 64:48:18)
Iteration 200 M( 38900000 )C, 0xbafaf821a144eb63, n = 2097152, clLucas v1.01 err = 0.2344 (0:00 real, 5.6806 ms/iter, ETA 61:22:53)
Iteration 300 M( 38900000 )C, 0xea3715ed348d20bb, n = 2097152, clLucas v1.01 err = 0.25 (0:01 real, 5.7576 ms/iter, ETA 62:12:49)
..................
Iteration 1100 M( 38900000 )C, 0x42b675b01e56dfa1, n = 2097152, clLucas v1.01 err = 0.2813 (0:00 real, 5.7782 ms/iter, ETA 62:26:04)
Iteration 1200 M( 38900000 )C, 0xe232c10a5b0ac175, n = 2097152, clLucas v1.01 err = 0.2813 (0:01 real, 5.6548 ms/iter, ETA 61:06:05)
^C caught. Writing checkpoint.

e:\-99-Prime\clLucas>cll -c 100 -t 38850000
Platform :Advanced Micro Devices, Inc.
Device 0 : Tahiti

start M38850000 fft length = 1966080
err = 0.46875, increasing n from 1966080

start M38850000 fft length = 2097152
Iteration 100 M( 38850000 )C, 0x18cd7fe8785b025a, n = 2097152, clLucas v1.01 err = 0.1875 (0:01 real, 5.7970 ms/iter, ETA 62:33:33)
Iteration 200 M( 38850000 )C, 0x7321f59a23b709e2, n = 2097152, clLucas v1.01 err = 0.2188 (0:01 real, 5.7387 ms/iter, ETA 61:55:48)
Iteration 300 M( 38850000 )C, 0x953d6ed545c9f7d7, n = 2097152, clLucas v1.01 err = 0.2188 (0:00 real, 5.5454 ms/iter, ETA 59:50:35)
^C caught. Writing checkpoint.

e:\-99-Prime\clLucas>[/CODE]Kracker will be interested too, for sure (see the first part of the test above, for an idea of the speed difference)

edit: on the other hands, I just switched another computer (CPU work) to DC and I got a [URL="http://www.mersenne.org/report_exponent/?exp_lo=27645643&exp_hi=&B1=Get+status"]27M assignment[/URL]. I mean "legally", without any tricks nor any "preferential treatment", just pure luck, requested a DC and got a 27M. I thought the 27M range is already gone, for ages... Hehe, I did the honors and manually moved it into the worktodo.txt of a faster computer, but it will still take 5-6 days, because the queue in front of it.

[QUOTE=LaurV;355236]I can do the proper TF by myself.[/QUOTE]

OK. The work needed to facilitate this is underway.

Please feel free to do some TFing work in the 37.5M to 38.5M range (available from the standard DCTF form) in the precipitation of the DC work being exposed shortly... :smile:

An important question: based on [URL="http://www.mersenne.ca/cudalucas.php?model=13"]James' analysis[/URL], what in your mind is the proper TF level for this range (read: what are you comfortable doing?)? 71 or 72?

[QUOTE=LaurV;355236]edit: on the other hands, I just switched another computer (CPU work) to DC and I got a [URL="http://www.mersenne.org/report_exponent/?exp_lo=27645643&exp_hi=&B1=Get+status"]27M assignment[/URL]. I mean "legally", without any tricks nor any "preferential treatment", just pure luck, requested a DC and got a 27M. I thought the 27M range is already gone, for ages...[/QUOTE]

That still occasionally (although rarely) happens.

"Spidy" is at your service.... :smile:

Thanks chalsall, LaurV.

I can do TF as well, if needed.

@LaurV: greater -c will be faster, gpu usage goes down a bit on checkpoint.

EDIT: "officially" rank 2 on DC! :razz:

EDIT2;
[quote]
Please feel free to do some TFing work in the 37.5M to 38.5M range (available from the standard DCTF form) in the precipitation of the DC work being exposed shortly... :smile:
[/quote]
Started to 70-71 bits.

[QUOTE=chalsall;355238]An important question: based on [URL="http://www.mersenne.ca/cudalucas.php?model=13"]James' analysis[/URL], what in your mind is the proper TF level for this range (read: what are you comfortable doing?)? 71 or 72?[/QUOTE]

OK, the work on GPU72 to facilitate this has been (mostly) completed... The system now watches for any DC candidates between 37.5M and 38.5M TFed to at least 72 "bits" and transfers them to "Anon" for assignment. These are currently available from the [URL="https://www.gpu72.com/account/getassignments/dc/"]DC Assignment form.[/URL]. Just set the "Optional Range" low field to be at least 35M, and you should see them in the preview.

I will add an "Option Field" to the form for "GPU Optimized" which will do this automatically in a while (read: a day or so).

Also, I didn't hear back as to what the best TF level is: 71 or 72? Right now there are [URL="https://www.gpu72.com/reports/available/lldc/"]28 candidates[/URL] TFed to at least 72 bits ready for assignment.

And, If I may share... Sorry this took a bit longer than expected -- one of my business partners, my girlfriend, and I were "the hares" scheduled to set a [URL="http://www.barbadoshash.com/"]Barbados Hash House Harriers[/URL] run yesterday.

We set a wonderful 12 km run in St. Andrews (the hilliest part of Barbados) in the morning, and then the Universe decided to play a joke on us, and the sky opened up around noon with unbelievable amounts of rain (visibility dropped to less than 500 meters at one point).

Despite the fact that some parts of Barbados flooded (with vehicles literally being washed off the road and bridges overwhelmed), more than sixty people ventured out for the run. [URL="https://plus.google.com/photos/102753076471691256943/albums/5931681138724786321"]These pictures[/URL] might give you a bit of an idea. What isn't explicitly clear from the pictures is it continued to rain for the entire run and the "lime" afterwards -- Hashers, like GIMPers, can be strange people (see pic 80 for the three "hares", if you care)....