And then, The Mawn said, very cautiously, that he doesn't think it's funny when people say they're serious when they're joking.

[QUOTE=chalsall;369434]If you two don't stop fighting I'm going to shut this server down.

I'm serious. I have better things to do with my time.[/QUOTE]

Oh come on... I think you need to upgrade your joke meter/detector. :smile:

[QUOTE=kracker;369437]Oh come on... I think you need to upgrade your joke meter/detector. :smile:[/QUOTE]

And you yours. :wink:

[QUOTE=LaurV;369410]OTOH, [URL="http://www.mersenne.org/report_top_500_custom/?team_flag=0&type=1004&rank_lo=39&rank_hi=40&start_date=2000-01-01&end_date=&B1=Get+Report"]hey buddy, do you feel the heat[/URL]? :razz: (no, this is serious, it does not include the bogus credit from GPU72! I am right behind of you, coming fast!)[/QUOTE]

Did you happen to check about another 20 spots higher....?

[QUOTE=chalsall;369443]And you yours. :wink:[/QUOTE]

I didn't want to set mine too high or my life would be a joke :wink:

[QUOTE=petrw1;369444]Did you happen to check about another 20 spots higher....?[/QUOTE]
I don't talk to you anymore! You are bad guy. Bad petrw1! Bad! Last time when we competed, you flied away and let me in the dust! :yucky: :yucky:

[QUOTE=kracker;369433]Nope, you mean chalsall passing me, not you. Next!:razz:[/QUOTE]
That's the funny part, I will overtake you without lifting a finger :paul:
We could keep it secret, but then, where the fun would be, hehe...
Anyhow, man, don't be upset, that's life, you win few, you lose few :razz:

[QUOTE=LaurV;369469]That's the funny part, I will overtake you without lifting a finger :paul:
We could keep it secret, but then, where the fun would be, hehe...
Anyhow, man, don't be upset, that's life, you win few, you lose few :razz:[/QUOTE]

Change that will to may :razz: I don' t believe anything I actually see in firepower... I guess that goes for me too... Electricity is such a b!tch

[QUOTE=LaurV;369469]That's the funny part, I will overtake you without lifting a finger :paul:
We could keep it secret, but then, where the fun would be, hehe...
Anyhow, man, don't be upset, that's life, you win few, you lose few :razz:[/QUOTE][URL="http://oppositelock.jalopnik.com/favorite-sr-71-story-1079127041"]Amusing tale[/URL] from [STRIKE]Curtis Cooper[/STRIKE] a SR-71 pilot.

We knew the story (although a slightly different).
Thanks for sharing it!

I've noticed many exponents in the high 69M range are not going through GPU to 72 before being assigned for LL testing: [url]http://mersenne.org/report_exponent/default.php?exp_lo=69700000&exp_hi=69701000&full=1[/url]

Any reason for this?

[QUOTE=ixfd64;369560]I've noticed many exponents in the high 69M range are not going through GPU to 72 before being assigned for LL testing: [url]http://mersenne.org/report_exponent/default.php?exp_lo=69700000&exp_hi=69701000&full=1[/url]

Any reason for this?[/QUOTE]

Yes, unfortunately a side effect of the new assignment policy is that GPU-TF has not kept pace with class 4 assignments. I've been doing 71->72 in that range for a while but more help is welcome.

[QUOTE=garo;369569]Yes, unfortunately a side effect of the new assignment policy is that GPU-TF has not kept pace with class 4 assignments. I've been doing 71->72 in that range for a while but more help is welcome.[/QUOTE]

I assume you would prefer TF to 74 (or even 75) but given the lack of firepower 72 is better than nothing at all.

It might be wise to lower the Let GPU72 Decide assignments to 73 bits instead of 74. I could push through a few dozen 72's per day and I will start working on that when I finish my current batch of work.

chalsall has been busy with real life so I don't want to burden him too much. So I've been getting my exponents straight off Primenet. It seems to work fine. Here is an example string
[url]http://v5www.mersenne.org/report_factoring_effort/default.php?exp_lo=69850000&exp_hi=69860000&bits_lo=1&bits_hi=71&txt=1&exassigned=1&B1=Get+Data[/url]

A bit of text massaging to create worktodo.txt and then stick it into Prime95 to reserve the exponents and then put them in the mfaktc worktodo. A bit manual yes but the best option at the moment.

Something similar can be done with GPU72. It takes some playing around with the range, factoring target level, and highest/lowest TF level drop-down.

[QUOTE=TheMawn;369581]It might be wise to lower the Let GPU72 Decide assignments to 73 bits instead of 74. [/QUOTE]


??

Let GPU72 Decide has not given me a 74 bit assignment for several weeks. They have all been 72—>73 in the 66M range.

I didn't mean to sound pushy. I know Chris has been busy but it was just a thought. And I know that I can pretty much choose my own exponents with the available options. I just assumed that most people just Let GPU72 Decide.

What is strange is that Chuck is getting a bunch of 72 -> 73 but I've been getting literally nothing but 73 -> 74. I'll change the bit level to 73 manually but that has been the default since I joined GPU72 last summer. I'll fiddle with the ranges too and see if I can get some Class 4 stuff.

[QUOTE=TheMawn;369601]What is strange is that Chuck is getting a bunch of 72 -> 73 but I've been getting literally nothing but 73 -> 74. I'll change the bit level to 73 manually but that has been the default since I joined GPU72 last summer. I'll fiddle with the ranges too and see if I can get some Class 4 stuff.[/QUOTE]

Sorry guys... Busy doesn't even being to describe it...

"Let GPU72 Decide" via MISFIT has been to 73 for about a month.

"What Makes Sense" still honors what the user has pledged.

I just updated the GPU72 Manual Assignment page to default to a pledge of 73 rather than 74. Those who use MISFIT and use option "0" (WMS) rather than option "9" (LGD) will need to change your pledge level, since WMS honors the pledge level, while LGD sets it.

The good news is the starvation for Cat4 assignments should only last for another month or so, and we will be able reclaim about 80% of the assignments sub-optimally TFed over time.

Also, so everyone knows, if "Spidy" detects that a 69M candidate not yet P-1'ed is about to be released at less than 73 "bits", it pulls its "rip-cord", and releases lower candidates at the highest bit levels not yet P-1'ed, sorted by TF level descending.

GPU72 also now releases automatically any candidates TF'ed to at least 73, and P-1'ed. Oliver should release about a thousand such candidates tomorrow.

Are we having fun yet??? :smile:

[QUOTE=chalsall;369603]
Are we having fun yet??? :smile:[/QUOTE]

It has been an interesting diversion. I threw some 69M's in the hopper. It was also interesting to experiment with GPU72 Assignments responses to different settings. Before I started limiting the upper end of the range, it once delivered a block of 69M's, filled out with some 332M's.

[QUOTE=kladner;369612]It has been an interesting diversion. I threw some 69M's in the hopper. It was also interesting to experiment with GPU72 Assignments responses to different settings. Before I started limiting the upper end of the range, it once delivered a block of 69M's, filled out with some 332M's.[/QUOTE]

Just wondering...

Is this my mistake, or yours?

[CODE]mysql> select User,Status,Trust,DisplayName,Created from Users where UN like "klad%";
+----------------------------------+--------+-------+-------------+---------------------+
| User | Status | Trust | DisplayName | Created |
+----------------------------------+--------+-------+-------------+---------------------+
| 8ad778c87c3d88f688fef4ebad36994c | 1 | 1 | ktony | 2011-11-01 14:54:55 |
| 0a68f617e684700767e03f40c1fda177 | 0 | 0 | | 2013-03-09 20:33:31 |
| 6cac8167336e8ff4ca04b4f50a011882 | 0 | 0 | | 2013-03-09 20:39:14 |
+----------------------------------+--------+-------+-------------+---------------------+
[/CODE]

I confess a lack of perception as to what you're asking. :unsure:
On further study, I can offer these observation.
kladner is my forum name, and actual GIMPS User Name. ktony is the GIMPS Display name. The 2011 date is perhaps when I started in GPU72. The other two with creations minutes apart look like some artifact, perhaps of rebuilding hardware and OS on this end.

Another thought comes to me which I will communicate privately. :smile:

I can now explain why I've been so busy...
 

Hey all.

The project which has been consuming so much of my time has now "gone public": [url]http://www.hivgateway.com/[/url]

I'm particularity proud of the use of "info-graphics" -- pretty and interactive graphics which draw the user in, but also communicates useful information for those who know how to read them.

[url]http://www.hivgateway.com/author/0472ab58d779522125ac5eec361b40d9/[/url]

and

[url]http://www.hivgateway.com/category_bubble/[/url]

The first is, in my mind, the most interesting. It shows the entity relationships between Authors, with a "weight". You can then hover over the outer bar to dynamically drill down on who that single author has co-authored with. You can click on the outer bar or the author's name to drill down on that author.

For the geeks reading this, both charts are implemented using the [URL="http://d3js.org/"]Data-Driven Documents[/URL] JavaScript library. A remarkably powerful and sophisticated library.

We are to present this on Monday to many "Big-ups". If anyone who tries this out sees any SPEs, please let me know.

[QUOTE=chalsall;369960]Hey all.

The project which has been consuming so much of my time has now "gone public": [url]http://www.hivgateway.com/[/url]

I'm particularity proud of the use of "info-graphics" -- pretty and interactive graphics which draw the user in, but also communicates useful information for those who know how to read them.

[url]http://www.hivgateway.com/author/0472ab58d779522125ac5eec361b40d9/[/url]

and

[url]http://www.hivgateway.com/category_bubble/[/url]

The first is, in my mind, the most interesting. It shows the entity relationships between Authors, with a "weight". You can then hover over the outer bar to dynamically drill down on who that single author has co-authored with. You can click on the outer bar or the author's name to drill down on that author.

For the geeks reading this, both charts are implemented using the [URL="http://d3js.org/"]Data-Driven Documents[/URL] JavaScript library. A remarkably powerful and sophisticated library.

We are to present this on Monday to many "Big-ups". If anyone who tries this out sees any SPEs, please let me know.[/QUOTE]

Well done, Chris, congratulations! :smile: :flex:

Luigi

I am delighted to see that your work has addressed such a deep need. That is an amazing interactive graphic.

Kudos, Thanks, and Praises! :party:

[QUOTE=kladner;370000]I am delighted to see that your work has addressed such a deep need.[/QUOTE]

Thanks guys. I am quite proud of this work.

I'm working with some really great people here, and Barbados is actually leading the Caribbean in the fight against HIV/AIDs.

Particular thanks to Xyzzy who privately pointed out a subtle issue which I never would have seen because I have looked at the pages far too often, and so it was simply filtered out. Interesting how human perception works.... :smile:

Hi Chris I am back. Very nice that work. Congrats. That is something I will definitively read! (as much as I will be able too, I am kinda interesting in the subject).

mis post, heh.

I grabbed some LL TF to 73 bits (everything default, asked for 100 assignments) and the preview showed 70,002,XXX from 72 to 73, but I got 66M 71 to 73. I don't really care what kind of work I get as far as TF is concerned but because this has been brought up earlier, I thought I'd mention it now.

I also grabbed a hundred DC-TF since I haven't done any in a while. How are we doing there?

[QUOTE=TheMawn;370524]I don't really care what kind of work I get as far as TF is concerned but because this has been brought up earlier, I thought I'd mention it now.[/QUOTE]

With George's knowledge, I brought in everything in the 70M range not yet assigned and not yet TFed to at least 72 bits. This was to prevent anything below 72 bits from being assigned for P-1 work.

Not likely to occur -- we're two days away from having the Category 4 candidates start to be recycled (several hundred a day), so we should be able to start bringing back in candidates not yet TFed to 74 bits, and return to that TF level as the default release level shortly.

[QUOTE=TheMawn;370524]I also grabbed a hundred DC-TF since I haven't done any in a while. How are we doing there?[/QUOTE]

We're looking fine; LaurV, I and others (and occasionally Jerry) have build up a comfortable buffer, and again, several hundred will be recycled per day starting Thursday. But, such work is still needed; we're just not under a "crunch" situation.

[QUOTE=chalsall;370551]Not likely to occur -- we're two days away from having the Category 4 candidates start to be recycled (several hundred a day), so we should be able to start bringing back in candidates not yet TFed to 74 bits, and return to that TF level as the default release level shortly.[/QUOTE]

Where do you get that information from? The Assignmentrules on the threshold page would only say Cat4 candidates to be recycled after 180 respectively 360 days. Or did I miss something else in the forum according to that?

[QUOTE=manfred4;370561]Where do you get that information from?[/QUOTE]

I have Spiders... :wink:

[QUOTE=manfred4;370561]Or did I miss something else in the forum according to that?[/QUOTE]

You missed that the original recycling rules would still be in place. As in, a candidate claimed but not reported on for 60 days would be recycled (with the exception of Manual Assignments, which are allowed 180 days if not manually extended).

In two days at least 354 candidates will be recycled under this rule, and the number increases significantly over time.

Ah I see. The new assignment rules and recycling rules are listed on the site, but the recycling is still noch implemented as stated there.
For GPUto72 that's a good thing because the buffer at that range is not built well enough yet.

But are the Rules getting implemented anytime soon then? Because that would only lead to more candidates having to be TF'd over the triple amount of time those C4 assignments would stay stuck.

[QUOTE=manfred4;370582]Ah I see. The new assignment rules and recycling rules are listed on the site, but the recycling is still noch implemented as stated there.
For GPUto72 that's a good thing because the buffer at that range is not built well enough yet.[/QUOTE]

Take a few more sips from the flask. And then watch how it's done.

[QUOTE=chalsall;370551]With George's knowledge, I brought in everything in the 70M range not yet assigned and not yet TFed to at least 72 bits. This was to prevent anything below 72 bits from being assigned for P-1 work.[/QUOTE]
That is a very good idea. In this case, would you please extend [URL="http://www.gpu72.com/reports/current_level/"]these tables[/URL] vertically? (few rows more). And for estimated completion too. If (and only if) any space constrains, the first two rows in both tables can be dropped). I don't know others, but I am watching those tables at least 3-4 times per week, to see how we progress.

[QUOTE=TheMawn;370524]I grabbed some LL TF to 73 bits (everything default, asked for 100 assignments) and the preview showed 70,002,XXX from 72 to 73, but I got 66M 71 to 73. [/QUOTE]

[QUOTE=chalsall;370551]With George's knowledge, I brought in everything in the 70M range not yet assigned and not yet TFed to at least 72 bits. This was to prevent anything below 72 bits from being assigned for P-1 work.[/QUOTE]

Chris, I think you missed the part where the preview window showed one thing, but the actual assignment ended up being different.

[QUOTE=LaurV;370608]In this case, would you please extend [URL="http://www.gpu72.com/reports/current_level/"]these tables[/URL] vertically?[/QUOTE]

Done.

[QUOTE=axn;370611]Chris, I think you missed the part where the preview window showed one thing, but the actual assignment ended up being different.[/QUOTE]

I didn't miss it, but forgot to speak to it.

The 70M range was brought in, but not made available for assignment (with one user's exception, which George, Garo and Jerry was aware of). I made a SPE, and didn't change the preview code.

Sorry about that -- this was a precautionary action I undertook. Based on the recycling rules clearly documented at [URL="http://mersenne.org/thresholds/"]the Mersenne.org Assignment Rules[/URL] page, the "starvation" of Category 4 candidates is about to end.

[QUOTE].....the "starvation" of Category 4 candidates is about to end.[/QUOTE]

Does this mean that we will return to trial factoring to 74 where appropriate?

[QUOTE=kladner;370692]Does this mean that we will return to trial factoring to 74 where appropriate?[/QUOTE]

Yes.

Probably in about two weeks time, as resources allow.

I had my gall bladder removed a couple of weeks ago, and was in the hospital for a week before that getting tests done, and so I missed some things regarding what was being factored, and now I see my computer started pulling numbers in the 70M range because they were lowest factored. Sigh. Since we seem to be having some sort of crisis I don't honestly understand about numbers in the 'high 69M range' I grabbed 500 assignments > 69.5M to put to 73. 500 because that is about a week and a half worth of work, and am releasing any other assignments I have.

I am having trouble understanding the exact nature of the category 4 problem, but I hope this helps. When I tried to reserve 'let gpu72 decide' on the web site it gave me 66M numbers to factor from 73 to 74, and that was not the behavior I expected from the above posts.

[QUOTE=Aramis Wyler;371002]I am having trouble understanding the exact nature of the category 4 problem, but I hope this helps.[/QUOTE]

It will indeed! Thanks!

The "Cat 4" problem is simply that I didn't appropriately appreciate just how much "churn" occurred for a large amount of participants. I thought we had enough of a buffer for George to activate the new assignment rules, not realizing that for the first two months there would be ~1,000 assignments requested per day for "Cat 4", rather than 300 (the number of assignments actually completed).

The good news is as of tonight, we're now two months into the new assignment rules. Thus, the original recycling rule of "60 days without report, no assignment" will kick in, and a great many candidates will be recycled, and within a week or two the pressure will be off.

[QUOTE=Aramis Wyler;371002]When I tried to reserve 'let gpu72 decide' on the web site it gave me 66M numbers to factor from 73 to 74, and that was not the behavior I expected from the above posts.[/QUOTE]

You should have only seen that if you changed the default pledge level from 63 to 64. And, by the way, "Let GPU72 Decide" is only available through MISFIT -- on the GPU72 manual assignment web page the option is "What Makes Sense" -- subtly different.

Lastly, the reason you were given some 70Ms to process is because you've configured MISFIT to use option #1 -- Lowest TF level. The 70M range was not actually intended to be exposed, but I made a mistake in not changing the preview code, which someone pointed out.

Sorry for the f' ups on my end. (And, as an aside, this is a "tough and exacting crowd"! Just the way I like it!!! :smile:)

In fact, I do know the difference between the 'what makes sense' and 'Let Gpu72 decide' options are, and [I]both are available on the get work form for lltf[/I], and I stand by the statement you quoted.

[QUOTE=Aramis Wyler;371008]In fact, I do know the difference between the 'what makes sense' and 'Let Gpu72 decide' options are, and [I]both are available on the get work form for lltf[/I], and I stand by the statement you quoted.[/QUOTE]

Ah hell... Sorry; you are correct and I was wrong. It is sobering to be to told how your own software works... :smile:

Please let me know if you see any other issues.

lol! fwiw it's pretty cool how the let gpu decide options so clearly disables the irrelevant field, nice work on that. :smile:

New info though, purely coincidental. While I picked up the new assignments and put them in the queue, exactly 1 70M number finished on each video card. The spider does not appear to have picked up these results, which makes sense if they weren't supposed to be assigned.

I'm not having a hate on challsall day, though, it's just I'm finally off all the narcotics and so I'm going through about a month's worth of data on my end.

[QUOTE=Aramis Wyler;371016]I'm not having a hate on challsall day, though, it's just I'm finally off all the narcotics and so I'm going through about a month's worth of data on my end.[/QUOTE]

Not a problem. I know I'm "Only Human", and so often make stupid mistakes. Feedback is always welcome and appreciated. :smile:

So you know, the GPU72 side "Observing Spider" *IS* watching the 70M range, so it should have picked up the submission to Primenet within 30 minutes. It does a "full scan" at 10 and 40 minutes after each hour.

Please PM me the candidates in question, and I'll drill down.

Interesting, must have been luck of the draw. The 70M numbers ([URL="http://www.mersenne.org/report_exponent/default.php?exp_lo=70765103&full=1"][B][COLOR=#000080]70765103[/COLOR][/B][/URL], [URL="http://www.mersenne.org/report_exponent/default.php?exp_lo=70765853&full=1"][B][COLOR=#000080]70765853[/COLOR][/B][/URL]) show on gpu72 at 2014-04-12 21:15:53 whereas the other 13 numbers I submitted all showed at 2014-04-12 20:53:38. It's interesting because it was 2 cards, 2 distinct sequential submissions each with 1 70M number at the end, so the first 70M number submitted before 6 of the 69M numbers and the other 70M number was at the end, and yet the first scan picked up all of the 69M numbers and neither of the 70M numbers. It's not important - I'm glad it's working. It must have been some unusual timing with the submissions and the scan.

[QUOTE=Aramis Wyler;371024]It's not important - I'm glad it's working. It must have been some unusual timing with the submissions and the scan.[/QUOTE]

Thanks for the update. And, for the record, I didn't change anything for the "observations" to succeed.

This was probably a timing issue. What I refer to as a "harmonic situation" between autonomous actors; sometimes related to a "race condition", but often different.

But, please, bring any future perceived issues forward publicly. This is a very complicated machine we've communally built -- we all want to find any bugs that exist.

Last night during maintenance on my main system the power supply blew out (including a nice smoke display).


Unfortunately, I don't have the money to continue replacing parts so this system will stay offline. That's ~1400 Ghz/days down.

@Chris: Since I don't have access to the drive now, it will take a few days to know which assignments are stuck on the system. I'll move them to a working system when I can regenerate the work file.

[QUOTE=flashjh;371059]Last night during maintenance on my main system the power supply blew out (including a nice smoke display).[/QUOTE]

Ah, heck... Sorry to hear of your loss -- it's always sad when the "Magic Blue Smoke" leaks out....

[QUOTE=flashjh;371059]@Chris: Since I don't have access to the drive now, it will take a few days to know which assignments are stuck on the system. I'll move them to a working system when I can regenerate the work file.[/QUOTE]

OK, thanks for the heads up for resource planning.

Thankfully the Cat4 recycling is now happening in earnest (about 600 or so candidates a day on average), so we should still be able to pull ahead again. But if anyone has any additional firepower they can bring to bear in the LL TF range, it would be very appreciated....

[QUOTE=chalsall;371062]But if anyone has any additional firepower they can bring to bear in the LL TF range, it would be very appreciated....[/QUOTE]In celebration of (unexpectedly) getting a second GTX 670 today, I'll take a break from my usual project and pick up 200 candidates (100 for each 670). Assuming I can run them 24/7 that would be 500GHd/d, but I don't think I can promise that I'll be running 24h/d. Still, it's something.

[QUOTE=chalsall;371025]But, please, bring any future perceived issues forward publicly. This is a very complicated machine we've communally built -- we all want to find any bugs that exist.[/QUOTE]Minor issue:
When asking for LL-TF work, if I enter a starting number, like say 332190000, GPU72 does not automatically update the previews. It used to. It takes selecting Lowest Exponent (or the like) to get it to update. Don't know if this is feature or not.

[QUOTE=Uncwilly;371095]Minor issue:
When asking for LL-TF work, if I enter a starting number, like say 332190000, GPU72 does not automatically update the previews. It used to. It takes selecting Lowest Exponent (or the like) to get it to update. Don't know if this is feature or not.[/QUOTE]

I have usually found it necessary to click in any other field, after entering data in a particular one, to get the entry to be included. It amounts to a refresh.

[QUOTE=kladner;371096]I have usually found it necessary to click in any other field, after entering data in a particular one, to get the entry to be included. It amounts to a refresh.[/QUOTE]That used to work. I have to click on one of the block of four choices in the drop down.

[QUOTE=James Heinrich;371094]In celebration of (unexpectedly) getting a second GTX 670 today, I'll take a break from my usual project and pick up 200 candidates (100 for each 670). Assuming I can run them 24/7 that would be 500GHd/d, but I don't think I can promise that I'll be running 24h/d. Still, it's something.[/QUOTE]

Nice! :smile:

[QUOTE=Uncwilly;371097]That used to work. I have to click on one of the block of four choices in the drop down.[/QUOTE]

Still seems to work as it was for me. The exception is that, if the drop-down is in the default "WMS" choice, no entry makes any difference except number of assignments.

[QUOTE=James Heinrich;371094]Still, it's something.[/QUOTE]

Thanks! Every bit helps.... :smile:

[QUOTE=Uncwilly;371095]It used to. It takes selecting Lowest Exponent (or the like) to get it to update. Don't know if this is feature or not.[/QUOTE]

Thanks for pointing that out... I added a $Low = 66000000 for WMS and LGD, so we would focus on the Cat4 range for a while. I didn't realize this would screw up the WMS for the 332M range. This has been fixed.

[QUOTE=James Heinrich;359028]The "Work Done By <Username>" graph X axis labels are running over each other:
[url]http://www.gpu72.com/graphs/worker/56f1b7572536a14513b08c88b2ba9578/[/url][/QUOTE]Dredging this up from 5 months ago because it's still not fixed :smile:

Currently the smallest available Cat4 assignment is in the 73M - I think you should be releasing some more exponents / reserve further ranges for TF.

[QUOTE=manfred4;371367]Currently the smallest available Cat4 assignment is in the 73M - I think you should be releasing some more exponents / reserve further ranges for TF.[/QUOTE]

Not true. The highest which appears on Primenet's summary page is in the 71M range, not 73M.

And few (if any) are actually being assigned there, and none not TFed below 73 "bits"

"Spidy" is doing its job, and pulling its "rip-cord" as needed. Please see the below as an example of what Spidy does every five minutes.

[CODE]20140416_184001 INFO: Get Manual Work spider starting...
20140416_184002 INFO: PM-1...
20140416_184003 INFO: 100: 66293263,73,0 (7ED910084013041CAE74C7518DAA8E83) -- Keep: 0
20140416_184003 INFO: 100: 66293267,73,0 (9C8839A5E63E3550F667035741F4A717) -- Keep: 0
20140416_184003 INFO: 100: 66293323,73,0 (C22548FD2A14050435B61B40CEA04EA4) -- Keep: 0
20140416_184003 INFO: 100: 66293327,73,0 (20FC45A3EE5609E52401239D2AAC214C) -- Keep: 0
20140416_184003 INFO: 100: 66293329,73,0 (4D58086875138668AD2C7A0EF83F4D0D) -- Keep: 0
20140416_184003 INFO: 100: 66293443,73,0 (B2A29909FAB5BB3A5D1200CDF305FCB8) -- Keep: 0
20140416_184003 INFO: 100: 66293497,73,0 (618AC794092FF9FE434FE034DF2529BC) -- Keep: 0
20140416_184003 INFO: 100: 66293509,73,0 (DFAC94852BF160DAF526EA8079671A50) -- Keep: 0
20140416_184003 INFO: 100: 66293533,73,0 (01F6E1BAB473B01E4C41687F8B93F5C7) -- Keep: 0
20140416_184003 INFO: 100: 66293587,73,0 (106CC4D73A00FD4A91C4589C4DFFBC01) -- Keep: 0
20140416_184003 INFO: Category 3...
20140416_184004 INFO: 100: 61576939,74,1 (67323B64EA5E7E47CF54EDC1CB2B53D3) -- Keep: 0
20140416_184004 INFO: 100: 61577057,74,1 (8332D7FF0625880002427E3ED023C127) -- Keep: 0
20140416_184004 INFO: 100: 61581893,74,1 (104128B156C06CFE01A26C27267DAA7E) -- Keep: 0
20140416_184004 INFO: 100: 61582271,74,1 (0EB8B8B2CE656A79A3F7C92DBD52059F) -- Keep: 0
20140416_184005 INFO: 100: 61638259,74,1 (7BA3C053E3EA8FA3FE161BC04B36B8EF) -- Keep: 0
20140416_184005 INFO: 100: 61650319,74,1 (D8BB15005C2969027EC88848D9018B7F) -- Keep: 0
20140416_184006 INFO: 100: 61659019,74,1 (1D1D33F34D2DEAF99EBD86BD1D373E70) -- Keep: 0
20140416_184006 INFO: 100: 61659469,74,1 (E14E9CEFDCDFFF0EE0CCB77E38B19474) -- Keep: 0
20140416_184006 INFO: 100: 61662421,74,1 (182FCFDC3A27C93FCD4A603191F08A14) -- Keep: 0
20140416_184006 INFO: 100: 61662551,74,1 (85BB5E16395746BB324782EAB588E209) -- Keep: 0
20140416_184007 INFO: 100: 61665839,74,1 (6757E159A94367BE29C8A6CA5E0C7584) -- Keep: 0
20140416_184007 INFO: 100: 61695509,74,1 (77E235F1FF1C07EB765A8043EEECE1CC) -- Keep: 0
20140416_184007 INFO: Category 4...
20140416_184008 INFO: 100: 66293701,73,0 (72D077F239B35613078036CFA12720B6) -- Keep: 0
20140416_184008 INFO: 100: 66293737,73,0 (2AB4B2CC3E128E0CA9F7E3431AAC0FD1) -- Keep: 0
20140416_184008 INFO: 100: 66293743,73,0 (8331B79691403033B461618BEF239222) -- Keep: 0
20140416_184008 INFO: 100: 66293789,73,0 (05E55AB2AB6158AA26FFE10416B3F033) -- Keep: 0
20140416_184009 INFO: 100: 66293803,73,0 (04C1C6E7B04EA6C8B87B47BF2667B116) -- Keep: 0
20140416_184009 INFO: 100: 66293809,73,0 (1C6C4EEA62B196A9A1D5BE3321C2E750) -- Keep: 0
20140416_184009 INFO: 100: 66293947,73,0 (29DB3381848947A04F1CEC13FD13B617) -- Keep: 0
20140416_184009 INFO: 100: 66294023,73,0 (CEB4376D23C7506AA51B3AABB9E404E9) -- Keep: 0
20140416_184010 INFO: 100: 66294061,73,0 (BE158DF1FE1BE89CF882D02314AB463C) -- Keep: 0
20140416_184010 INFO: 100: 66294091,73,0 (CF90A493FB8E1FD89A019D1190F0338F) -- Keep: 0
20140416_184011 INFO: 100: 66294113,73,0 (A831071BA7BCD24F11C9CDBE8111EA91) -- Keep: 0
20140416_184011 INFO: 100: 66294161,73,0 (CC6D0868671E3C6E642AD9B856018745) -- Keep: 0
20140416_184011 INFO: 100: 66294187,73,0 (2A25229404CD57BF435518ACF2157729) -- Keep: 0
20140416_184011 INFO: 100: 66294313,73,0 (A2773DFD40CE733B0FC04912B9987B32) -- Keep: 0
20140416_184012 INFO: 100: 66294367,73,0 (2752CB3FE05BE3DA15B224D2385A777C) -- Keep: 0
20140416_184012 INFO: 100: 69117583,73,1 (A61AAA863D6024B38479CD4243871734) -- Keep: 0
20140416_184013 INFO: 100: 69171917,73,1 (34259FD203FD6B771F65CF677AA65817) -- Keep: 0
20140416_184013 INFO: 100: 69306617,73,1 (7B208DB82B95A5C46A8EEF75F20B4CD0) -- Keep: 0
20140416_184014 INFO: 100: 69703369,74,1 (5C4446C9D837C71C7BD2A2050D64B82F) -- Keep: 0
20140416_184014 INFO: 100: 69882523,73,1 (D2B7EA0277947F42BF53A30531B847E7) -- Keep: 0
20140416_184014 INFO: 100: 69889487,73,1 (2DB7B30446101929A81DB6A0CC152CB7) -- Keep: 0
20140416_184014 INFO: 100: 69894533,74,1 (54D41F8F384E69D859BAAC0738C6C880) -- Keep: 0[/CODE]

Edit: And just so you all know, the "Keep: 0" at the end of each line means the assignment(s) will be returned to Primenet within 60 seconds.

Just noticed a tiny thing on /account/getassignments/lltf/ (not sure if applies to other assignment types). On pageload the Will Factor To box is set to "73". If you change to "Let GPU72 Decide" it changes to 74, if you change to "LMH Depth-first" it changes to 78. But if you change back to "What makes sense" it doesn't revert to 73.

Would appear to be controlled in /js/assign_preview.js::DoOption()

Just an early heads up, but I'll remind you later...

From Friday May 16 to Monday May 26, I will be on another continent so my CPU and more importantly both GPUs will be down for 10 days @ 600 GHz-Days per day.

[QUOTE=TheMawn;371683]Just an early heads up, but I'll remind you later...

From Friday May 16 to Monday May 26, I will be on another continent so my CPU and more importantly both GPUs will be down for 10 days @ 600 GHz-Days per day.[/QUOTE]You could run a few handfuls of 332M numbers up to 82 or 83 during that time. Would not make a big results file. (Just take any early ones, or ones that are already at 76 or above.):bow:

[QUOTE=Uncwilly;371688]You could run a few handfuls of 332M numbers up to 82 or 83 during that time. Would not make a big results file. (Just take any early ones, or ones that are already at 76 or above.):bow:[/QUOTE]I think the issue was more that his system would be off, rather than suffering from lack of maintenance for 10 days. For 10 days I can even pre-feed my systems doing >1G TF where exponents are completed (quite literally) every 2 seconds.

[QUOTE=James Heinrich;371094]In celebration of (unexpectedly) getting a second GTX 670 today, I'll take a break from my usual project and pick up 200 candidates (100 for each 670).[/QUOTE]This is getting weird. I've completed about 90 assignments (67M, 71-73) on each 670, and my old 670 has found 6 factors and the new (to me) 670 has found none. I [i]think[/i] it's just random chance that put all the factorable assignments on the first GPU; the second one passes the self-test and did indeed find one of the factors I found on the other GPU when I was getting suspicious. But odd.

[QUOTE=James Heinrich;371696] the second one passes the self-test and did indeed find one of the factors I found on the other GPU when I was getting suspicious.[/QUOTE]

Then you're probably fine. I get streaks like that too and they freak me out a bit also. It definitely isn't worth double-checking even if the GPU is known to miss the odd factor.

If I'm ever that bored, I might see if I could learn how to make mfaktx run a few thousand exponents with factors on only the class that contains that factor. It would be fast as hell and a person could get a meaningful success rate.

[QUOTE=TheMawn;371700]Then you're probably fine. I get streaks like that too and they freak me out a bit also.[/quote]Normally I wouldn't think much of it. It's just that I just got this "new" video card from a friend who upgraded to a 780ti, and it hasn't (ever) found any previously-unknown factors, so I'm just cautious.

[QUOTE=TheMawn;371700]If I'm ever that bored, I might see if I could learn how to make mfaktx run a few thousand exponents with factors on only the class that contains that factor. It would be fast as hell and a person could get a meaningful success rate.[/QUOTE]That's what the self-test does. I believe the automatic selftest that runs every time you start mfaktc has 92 exponents with known factors, you can also run two levels of extended self-test, per the help:[code]-h display this help and exit
-d <device number> specify the device number used by this program
-tf <exp> <min> <max> trial factor M<exp> from 2^<min> to 2^<max> and exit
instead of parsing the worktodo file
-st run builtin selftest (1557 testcases) and exit
-st2 run builtin selftest (all testcases) and exit
-v <number> set verbosity (min = 0, default = 1, max = 2)[/code]

[QUOTE=TheMawn;371683]From Friday May 16 to Monday May 26, I will be on another continent so my CPU and more importantly both GPUs will be down for 10 days @ 600 GHz-Days per day.[/QUOTE]

OK, thanks for the heads up.

Geezzz... Jerry has a power supply failure resulting in significant loss of cycles. Then LaurV loses two of his four cards due to another PSU failure (he was able to bring one back online using a spare (but smaller) unit), then people dare to take vacations!!! :wink:

I had hoped to be able to start going to 74 again at the end of the month, but now it looks like that won't be possible for a while longer.

The good news is we're slowly pulling ahead again going to 73 in the Cat4 range, and reclaiming the 1,588 candidates released at 72 a while ago as they expire. It's amazing just how much "churn" occurs up there, but as of the last couple of weeks no candidates are being assigned by Primenet for either LL'ing or P-1'ing at less than 73.

As always, if anyone has any spare GPU cycles, LL TF'ing would be much appreciated!!! :smile:

I should be able to bring my 580 online as of May 2nd, which should add another ~430GHd/d

[QUOTE=James Heinrich;371710]I should be able to bring my 580 online as of May 2nd, which should add another ~430GHd/d[/QUOTE]

Coolness! Thanks. Good news! :smile:

Soon I'll be able to move my GPU space heater out of the bedrooms and down into the basement. Then they can run round the clock again.

I must say the proved to be EXCELLENT room heaters over the winter.

[QUOTE=swl551;371716]I must say the proved to be EXCELLENT room heaters over the winter.[/QUOTE]

:smile: BTW, did you ever get them controlled via a USB temperature sensor?

[QUOTE=chalsall;371718]:smile: BTW, did you ever get them controlled via a USB temperature sensor?[/QUOTE]

I got a cheap one from Amazon, but it proved unreliable. The professional sensors were too much $$. So I just used varying pre-defined duty cycles based on expected daily and overnight temperatures.

[QUOTE=swl551;371716]Soon I'll be able to move my GPU space heater out of the bedrooms and down into the basement. Then they can run round the clock again.

I must say the proved to be EXCELLENT room heaters over the winter.[/QUOTE]

It's not the heat the (now dead) 580 put out (97C!!) but the noise :no::no:

Now running dual 660's (overclocked) - getting 375 days/day and ...silence :smile:

[QUOTE=Gordon;371740]It's not the heat the (now dead) 580 put out (97C!!) but the noise :no::no:

Now running dual 660's (overclocked) - getting 375 days/day and ...silence :smile:[/QUOTE]

Hmm, so you're getting less out of dual 660's than a 580 with less sound and more power consumption.

EDIT: Is Kepler really more power efficient W/GHz? Hmm...

[QUOTE=Gordon;371740]Now running dual 660's (overclocked) - getting 375 days/day and ...silence :smile:[/QUOTE]

That doesn't sound right to me. Are you running two instances of mfactx? Or maybe you mean 375/card?

On my gtx480 + gtx580 I get 820 GHD/day in the 69M range with 2 instances of mfactc..

Sounds [url=http://www.mersenne.ca/mfaktc.php]right to me[/url].
GTX 660 (not Ti) at stock speed (no boost) should be about 179GHd/d, so two of them about 358, so 375 sounds about right.

For reference:
GTX 480 (368) + GTX 580 (433) = 801GHd/d at stock clocks (no boost)

When we ran our 660Ti it was good for ~275 a day.

With a bit of P-1 testing on the CPU we could just barely squeak by 2,000 a week.

Mine are not reference, they are the evga FTW Hydro Copper 1 and 2. But his also was overclocked. I'm surprised though that the 660 runs at that rate - I'm glad it's quiet.

[QUOTE=Xyzzy;371789]When we ran our 660Ti it was good for ~275 a day.[/QUOTE]Remember the 660 Ti has 30% more GFLOPS than the non-Ti version that [i]Gordon[/i] has.

Video card nomenclature is too confusing!

:edit:

[QUOTE=Xyzzy;371795]Video card nomenclature is too confusing![/QUOTE]

Please do try to keep up.

[QUOTE=chalsall;371798]Please do try to keep up.[/QUOTE]

:confused:

[COLOR="White"](A little unnecessary I think, but purely imo) [/COLOR]

Recent Ti-suffixed GPUs have roughly a 25-30% boost over their non-Ti counterparts:[quote]GeForce GTX 780/Ti +27%
GeForce GTX 750 Ti +25%
GeForce GTX 660 Ti +31%
GeForce GTX 560 Ti +16% (+20% for the 448-core version)[/quote]Where it gets more subtly confusing is when the same model number is used for dramatically different-performing GPUs. For example the GeForce GTX 460 (907 GFLOPS) should not be confused with the GeForce GTX 460 (1046 GFLOPS), nor with the GeForce GTX 460 SE (749) or the GeForce GTX 460M (518).

Iffy GPU
 

I'm bringing up two second hand cards I just got my hands on, two Radeon 6950. One is running just fine but the other one have some random hardware errors and are failing circa 1% of the tests in the -st2 self test.
Is it okay to use it for TF? It still finds 99% of the factors.

/Göran

[QUOTE=kracker;371799]:confused:

[COLOR="White"](A little unnecessary I think, but purely imo) [/COLOR][/QUOTE]

I like to push the edge of humor. Mike understands....

[QUOTE=Axelsson;371801]I'm bringing up two second hand cards I just got my hands on, two Radeon 6950. One is running just fine but the other one have some random hardware errors and are failing circa 1% of the tests in the -st2 self test.
Is it okay to use it for TF? It still finds 99% of the factors.

/Göran[/QUOTE]

Is it overclocked?

[QUOTE=kracker;371755]Hmm, so you're getting less out of dual 660's than a 580 with less sound and more power consumption.

EDIT: Is Kepler really more power efficient W/GHz? Hmm...[/QUOTE]

It's the sound that's the killer for me, the 580 was like sitting next to an industrial blower unit :lol:

[QUOTE=Aramis Wyler;371785]That doesn't sound right to me. Are you running two instances of mfactx? Or maybe you mean 375/card?

On my gtx480 + gtx580 I get 820 GHD/day in the 69M range with 2 instances of mfactc..[/QUOTE]

Pailt GTX-660 - stock speed - 171 days/day

Gigabyte Windforce x2 - 200 Mhz overclock - 207 days/day

[QUOTE=Gordon;371805]It's the sound that's the killer for me, the 580 was like sitting next to an industrial blower unit :lol:[/QUOTE]
When the 570 fan was rattling I was going totally nuts. One minute it is a june bug's buzz, then it's some witch's cackling, then it starts, then it stops, then it's some toilet noises... It is impossible to work, and impossible to walk away from the computer because some work has to be done. Torture! :furious:

If you haven't seen it...
 

If you haven't seen it, the film "Man on the Moon" is worth buying, watching, and/or stealing.

I'm not a big fan of Jim Carrey, as he tends to star in very stupid stuff (IMO).

But his acting in this film, portraying the work and life of Andy Kaufman, is quite remarkable.

Edit: [url]http://en.wikipedia.org/wiki/Man_on_the_Moon_(film[/url])

(going on a tangent? where have you been all this time)
 

I never knew how well Kaufman treated the audience when annoyed with their requests to do his "old stuff" he read the [STRIKE]full[/STRIKE] [I]Great Gatsby[/I] to them, instead. (After watching [I]The Great Gatsby[/I], I can feel their pain.)
[QUOTE=Wiki]In reality, Kaufman was more jokey with his audiences: He would read a few pages, and then he would ask the audience if they wanted him to keep reading, or play a record. When the audience chose to hear the record, the record he cued up was a recording of him continuing to read The Great Gatsby from where he had left off. He never actually read (or played) the entirety of The Great Gatsby to an audience, but he sometimes liked to claim that he had.[/QUOTE]
He was a very [I]mathematical[/I] man.

[QUOTE=Batalov;371812]He was a very [I]mathematical[/I] man.[/QUOTE]

And a so-so wrestler....

[QUOTE=kracker;371803]Is it overclocked?[/QUOTE]
No, I even took it down from 800 MHz core to 500 MHz core clock and no difference.
Funny thing is that a friend played around with it bitcoin mining and it worked flawlessly over the weekend at stock speed.

Running under Linux 32-bit Debian.

/Göran

[QUOTE=Axelsson;371814]No, I even took it down from 800 MHz core to 500 MHz core clock and no difference.
Funny thing is that a friend played around with it bitcoin mining and it worked flawlessly over the weekend at stock speed.

Running under Linux 32-bit Debian.

/Göran[/QUOTE]

Well, if you are mining in a pool it is very hard if not impossible to tell a gpu error...

In mfakto, does it fail randomly or in a certain spot?

[QUOTE=kracker;371815]Well, if you are mining in a pool it is very hard if not impossible to tell a gpu error...

In mfakto, does it fail randomly or in a certain spot?[/QUOTE]
If I understood my friend correct there is a top-like display that records the accumulated number of errors locally on the miner. It is used for trimming the maximum amount of overclocking a GPU can sustain and that was what we were using.

The errors from mfakto appears more or less randomly, two consecutive runs gives different errors. I haven't found any pattern among the errors yet but the jury is still out.

[QUOTE=chalsall;371708]As always, if anyone has any spare GPU cycles, LL TF'ing would be much appreciated!!! :smile:[/QUOTE]

Thanks to everyone who's brought additional TF'ing resources to bear (and, of course, those who have been doing this work for quite some time).

Because of this, we can start to slowly migrate back to TF'ing to 74. Jerry and LaurV's GPUs have been tasked with going to 74 as of today. On May 1st or so MISFIT users who use the "Let GPU72 decide" option will be assigned work to 74. Then sometime later (as the situation allows) the manual assignment page will revert to default to 74.

As always, "Spidy" will be watching the situation, and react appropriately (read: not allow anything to be assigned for either LL'ing or P-1'ing at less than 73). And, also (as always), anyone who doesn't want to go to 74 can simply choose a lower "pledge level", and choose any other option (e.g. "What Makes Sense", "Lowest TF level", "Lowest Exponent", etc).

This doesn't mean the pressure is off -- please keep the cycles coming! :smile: I just wanted to give everyone an update on the situation so there's no confusion as the pledge level changes over the next few weeks.