[QUOTE=sonjohan;354265]I estimate that we would've kept up the TF from 71 to 74 if the big guns were still there.[/QUOTE]

I appreciate your message, but please let me be clear...

GPU to 74 [B][I][U]IS[/U][/I][/B] currently keeping up. All the way from 71. Even with the ebb and flow of participants.

And, there are automatic "safety valves" in place to release candidates already at 73 for LL assignment if we ever fall behind. Again, automatic; while I monitor things closely, I'm not needed in the loop for this to happen iff needed.

To cut to the chase, David has admitted in the past that his agenda in all this nagging is for him, personally, to be assigned low candidates because they take less time, and there's a (slightly) greater probably that the lower candidates are prime.

David could, of course, simply sign up to GPU72 and have access to the (small number of) low candidates we hold for LL assignments. Ironically, David was actually the very first beta tester of this service...

But, at the end of the day, the truth is the next Mersenne Prime is probably somewhere up in the high 60Ms or even low 70Ms range; all the work below has to be done by someone(s). Those of us who TF are simply helping those who LL to not waste their valuable time if it can be avoided.

[QUOTE=davieddy;354199]You still don't get it, do you?[/QUOTE]

I can't say I exactly appreciate this as a rebuttal to what I'm trying to say. To reiterate:

We can get every exponent to 100M up to 73 before ever even looking at getting anything to 74. In the meantime, exponents in the upper 60M with only 73 bits done on them are being LL'ed because the TF effort is too focused on 2018's work instead of 2013's work.

Alternatively, we can focus on having a smaller buffer of more deeply factored exponents, which means that no work today is being done without the optimal amount of trial factoring.


I've been getting 30 GHz-Days of credit for every trial factoring job I do from 73 to 74 via my GPUs. Also, I get roughly 150 GHz-Days on a LL test in the same range via my CPU.

Trial factoring TO 73 would also take 30 GHz days. By the looks of things, it takes about two-fifths the effort to TF to 74 as it does to LL. Chris seems to believe that for every 150 GHz-Days that LL is getting, there IS 60 GHz-Days of TF. If there is, then we CAN keep up TF to 74 before releasing for further work.


David: How much of a 73-bits buffer do you even want? I found 384 composite results in the report recent cleared in the last 24 hours. If you're asking for a buffer of 50,000 then you've just gone completely <censored> mad. FOUR MONTHS? Explain to me how even a week's worth is going to suddenly evaporate.

The 50,000 exponents to 73 buffer you're asking for requires the same amount of work as a 16,666 exponent to 74 buffer. How is THAT not enough? It's like you think we're choosing between doing a few hundred to 74 versus doing tens of thousands to 73. If we HAD the capacity to make this buffer for ourselves then we would HAVE the capacity to be so far ahead of the LL curve with 74 bits that we'd be looking at doing 75 bits a few millions early.

I don't see the relevance of this discussion, since we are ahead of the LL and DC wavefronts. If you don't like factoring to 74 bit, nobody is forcing you to do so. You can take expo's to 73 and somebody else might do 73 to 74. Besides, the TF GPUto72 is doing is 'extra', in other words, higher than CPUs efficiently can do.

[QUOTE=VictordeHolland;354287]I don't see the relevance of this discussion.[/QUOTE]

Please understand this.

It is ill-advised to step in-between giants when they're being stupid. :wink:

[QUOTE=VictordeHolland;354287]I don't see the relevance of this discussion, since we are ahead of the LL and DC wavefronts. If you don't like factoring to 74 bit, nobody is forcing you to do so. You can take expo's to 73 and somebody else might do 73 to 74. Besides, the TF GPUto72 is doing is 'extra', in other words, higher than CPUs efficiently can do.[/QUOTE]

This discussion mostly seems relevant to David's need to feel that he has some form of influence. If he cannot obtain this in the form of others following his advice, he gets it by provoking turmoil. He revels in gaining the attention of others, whether that attention is negative or positive.

He has been predicting a disastrous shortfall in factoring for as long as I have hung out here. It seems that he has probably done so for longer, but I feel no real need to review the history. In any case, it does not seem that these predictions are accurate.

Since the justifications, rationales, and proposed remedies for these predicted disasters have run a wide, and sometimes contradictory gamut, I am inclined to see the whole thing as a displacement of hostility and resentment arising from other causes. It seems that the community respect, [I]which Chalsall has earned through much work,[/I] is probably one of these causes.

However, I rather think that there may well be multiple burrs under David's blanket, or bees in his bonnet if you prefer.

EDIT:
Chalsall said: [QUOTE]Please understand this.

It is ill-advised to step in-between giants when they're being stupid.[/QUOTE]
I suppose that I understand the peril, and can take the heat.

[QUOTE=kladner;354293]I suppose that I understand the peril, and can take the heat.[/QUOTE]

Good on you.

Some think they can cause you discomfort with no cost.

Some might be wrong....

[QUOTE=davieddy;354199]
To cut to the chase, David has admitted in the past that his agenda in all this nagging is for him, personally, to be assigned low candidates because they take less time, and there's a (slightly) greater probably that the lower candidates are prime.

David could, of course, simply sign up to GPU72 and have access to the (small number of) low candidates we hold for LL assignments. Ironically, David was actually the very first beta tester of this service...

But, at the end of the day, the truth is the next Mersenne Prime is probably somewhere up in the high 60Ms or even low 70Ms range; all the work below has to be done by someone(s). Those of us who TF are simply helping those who LL to not waste their valuable time if it can be avoided.[/QUOTE]
Those were the good ol' days.

I reserved a 45M exponent (TFed to 68) from Primenet, and put out a request for a GPU owner to TF it a few more bits with mfaktc (which Oliver had just announced). Among several offers, Eric Christenson (remember him?) promptly TFed it to 75.
I thanked him suitably and assured him that I would acknowledge his contribution should it turn out to be prime.
We engaged in much mutually instructive discussion of what the appropriate bit level might be.

I posted this anedote and requested a similar exponent. As Chris recalls correctly, he duly presented me with a 45M exponent TFed to 72 and P-1ed.

This simply goes to show that I am a veteran of this topic of discussion, and Chris ought to know better than he sees fit to let on.

David

[QUOTE=davieddy;354298]This simply goes to show that I am a veteran of this topic of discussion, and Chris ought to know better than he sees fit to let on.[/QUOTE]

But, by deduction, does that not also mean that I am also knowledgeable?

[QUOTE=VictordeHolland;354287]I don't see the relevance of this discussion, since we are ahead of the LL and DC wavefronts. If you don't like factoring to 74 bit, nobody is forcing you to do so. You can take expo's to 73 and somebody else might do 73 to 74. Besides, the TF GPUto72 is doing is 'extra', in other words, higher than CPUs efficiently can do.[/QUOTE]

This isn't so much a "GPU72 is forcing me to do something" as much as a "it's preventing me from doing something" issue. As I understand it, we've upped the amount of TF we are doing on the lower 60M's (the limit was 73 and is now 74) or some such and the result of all that is the exponents in the low 60M's still require some trial factoring work done on them.

David believes that there has been sufficient work done on them and he is bitter about the fact that he is now "forced" to work on larger exponents because the lower ones are being "held up" by GPU72 until the trial factoring has been done on them. It seems to personally hurt him that people are getting an exponent in the 66 million when they could be getting one in the 62 million. Note that all of my exponents right now (four of them) are in the ranges "being held up" so I'm not too sure what that's about.

[QUOTE=TheMawn;354311]
David believes that there has been sufficient work done on them and he is bitter about the fact that he is now "forced" to work on larger exponents because the lower ones are being "held up" by GPU72 until the trial factoring has been done on them.[/QUOTE]

I respect how keen and quick you are to learn the ropes around here.
(BTW here's a toast to Sir Ben Ainsley for rescuing the Yanks in the America's Cup. I suspect he might have given me a run for my money in my sailing prime.)

I find the whole GPUto72 ethos of personalizing everything [url=http://www.youtube.com/watch?v=IkEylc1-18M]tasteless[/url].

If and when I say "I" would prefer to do something or other, I usually mean the man on the Clapham omnibus.

Keep up the good work!

D

[QUOTE=kladner;354293]This discussion mostly [/QUOTE]
+1 :goodposting:

We however like David, we don't know why, but we in the same time are somehow tired of him requesting other people to work for him (like in "I only do LL, fcuk you TF-ers who do not TF my expos to 85 bits and P-1 them well, before they are assigned to me! (therefore increasing my chance to find a prime)"). (for newcomers, he used to sing this song some time ago, but he changed since he started not to like chalsall :smile:).

I never liked this kind of people, and that is the reason I don't like curtisc to much too. But don't take it personally, I am just an old and envious guy.

David is upset that he can't get lower expos to LL (and increasing his chance to find a prime even more), but in the same time he is ranting about gpu72 hoarding low exponents and assign them preferentially. Who stops him to became a part of gpu72 team and get low exponents from the pool? He even won't need to be a part of the team, just adjust the proxy in p95. I tell you what stops him, it is his personal self-pride.

Statistically, I personally don't see any proof of the fact that gpu72 is slowing GIMPS down. Contrarily. But what David doesn't understand, the world we used to live in is changing, we are too old for it now. Years ago we did not have cars, mobile phones, strong computers. We used to walk more, talk to persons face to face, compute with the pencil, and LL exponents of 20M. Now, the last is not possible anymore. Somebody else did it, we have to live harder now, driving to our jobs, having hard time to learn how to use iphones and gps-es, and LL-ing 60M exponents...

@chalsall: For this graph [URL]http://www.gpu72.com/reports/overall/graph/assigned/[/URL]

Can you make it 3D so we can see the exponent ranges by age also? Kinda like your range weekly progress videos are setup?

[QUOTE=flashjh;354812]@chalsall: For this graph [URL]http://www.gpu72.com/reports/overall/graph/assigned/[/URL]

Can you make it 3D so we can see the exponent ranges by age also? Kinda like your range weekly progress videos are setup?[/QUOTE]

I could.

But, can't you do that in your head? :smile:

Well sure, but then I don't have a picture to look at :) Plus, as I get older I get distracted easily. I need something to help me focus better.

[QUOTE=flashjh;354823]Well sure, but then I don't have a picture to look at :)[/QUOTE]

LOL... OK, I gave it a try, and I think you'll agree it makes the information harder (and sometimes impossible) to interpret than a standard stacked 2D bar graph....

[QUOTE=chalsall;354979]LOL... OK, I gave it a try, and I think you'll agree it makes the information harder (and sometimes impossible) to interpret than a standard stacked 2D bar graph....[/QUOTE]
Personally like it, but either way. If you can make it a 2D/3D option, cool. Otherwise, thanks for trying :smile:

The Makes Sense[SUP]TM[/SUP] assignments I've been grabbing for TF via GPU72.com have been exclusively 73 to 74 in the low 60's range (consistently going up as we do them; in the upper 63M right now). How long should it be before it turns back into 71 (or something) to 74?

[QUOTE=TheMawn;356179]The Makes Sense[SUP]TM[/SUP] assignments I've been grabbing for TF via GPU72.com have been exclusively 73 to 74 in the low 60's range (consistently going up as we do them; in the upper 63M right now). How long should it be before it turns back into 71 (or something) to 74?[/QUOTE]
According to [URL="https://www.gpu72.com/reports/current_level/"]this[/URL] in ~24,000 exponents.

[QUOTE=TheMawn;356179]How long should it be before it turns back into 71 (or something) to 74?[/QUOTE]

It is impossible to say exactly, but given our current 30 day average of about 344 candidates a day to 74, and that there are currently ~2,500 candidates ready for assignment already at 73 below 65M but without P-1 done (a criteria for WMS), I would estimate in about a week or so.

Keep in mind you can always set your work preference to be "Lowest TF Level", which sorts by lowest current factor depth (71) followed by lowest candidate (currently ~64.576M).

[QUOTE=Uncwilly;356191]According to [URL="https://www.gpu72.com/reports/current_level/"]this[/URL] in ~24,000 exponents.[/QUOTE]

[URL="https://www.gpu72.com/reports/available/nop-1/"]This report[/URL] (showing available candidates without P-1 done) is a better data source to use for this estimation. Note that GPU72 doesn't own all the candidates listed in the above report.

And then there are people like me doing almost exclusively 71->72 64M because I want to find more factors faster, for fun. About 5% of the time I get a 70->72. Since starting GPU72 14 days ago I've found 8 factors.

We'll see if I'll continue throwing 320 GHz days/day at GPU72 once my next electricity bill arrives lol

Just pay it[B]™[/B] (or have someone else look at it and pay it for you). It's the only way :)

[QUOTE=flashjh;356226]Just pay it[B]™[/B] (or have someone else look at it and pay it for you). It's the only way :)[/QUOTE]

Haha. It's bad enough that I want to buy more GPU power to throw at trial-factoring Just Because[B]™[/B].

[QUOTE=Mark Rose;356228]Haha. It's bad enough that I want to buy more GPU power to throw at trial-factoring Just Because[B]™[/B].[/QUOTE]

Trust me. You're not the only one. There isn;t even a plausible excuse for getting a GTX 660 Ti whilst already owning a GTX 670 (can't even SLI different cards) besides sending ASUS an I-Love-You note with a big fat cheque for $250.

I occasionally wander over to Newegg and try to make the cheapest prime machine I can think of. I'll probably be one of those people with four or five computers humming in some cold room, and about thirty years too young for the part :razz: Speaking of...

I took a gander over to mersenne.ca's throughput calculator. I compared an i5-3570k to an fx-8350 or whatever it was. Are those benchmarks legit? I'm seeing similar-release-date processors from AMD getting absolutely crushed by Intel. I've noticed a serious lack of talk regarding AMD processors, so it doesn't surprise me that Intel is the stronger hitter, but holy crap!

The reason I ask is AMD always ends up powering the hypothetical budget systems I put together in the cart. I was thinking to myself that AMD is so far away from being able to have the memory bandwidth to support eight cores at 5 GHz (with even 1866 being barely achievable in a lot of cases) but it looks like the CPU is awful too.

Does AMD just fall to the side because of lack of interest? Or have people been trying to make it work in vain?

AMDs work well for general purpose office/web stuff, IMHO.
AMD's implementation of AVX (and FMA) is a failure, at least for Primenet purposes. SSE2 code runs faster than AVX code on AMD.

[QUOTE=TheMawn;356268]Trust me. You're not the only one. There isn;t even a plausible excuse for getting a GTX 660 Ti whilst already owning a GTX 670 (can't even SLI different cards) besides sending ASUS an I-Love-You note with a big fat cheque for $250.[/QUOTE]

Oh I bought a GTX 760 three weeks ago to game. How much time have I gamed? 15 minutes. How much time have I trial factored? 21000 minutes lol.

[QUOTE]I occasionally wander over to Newegg and try to make the cheapest prime machine I can think of. I'll probably be one of those people with four or five computers humming in some cold room, and about thirty years too young for the part :razz: Speaking of...[/QUOTE]

So I see I'm not the only one with that affliction! Normally I'm putting in ~450 GHz towards Seventeen or Bust. But recently [url=http://www.seventeenorbust.com/stats/users/user.mhtml?userID=18096]this guy[/url] came along with the equivalent of 100 4-core Haswells. He nearly doubled the project throughput. Makes me feel insignificant.

[QUOTE]I took a gander over to mersenne.ca's throughput calculator. I compared an i5-3570k to an fx-8350 or whatever it was. Are those benchmarks legit? I'm seeing similar-release-date processors from AMD getting absolutely crushed by Intel. I've noticed a serious lack of talk regarding AMD processors, so it doesn't surprise me that Intel is the stronger hitter, but holy crap!

The reason I ask is AMD always ends up powering the hypothetical budget systems I put together in the cart. I was thinking to myself that AMD is so far away from being able to have the memory bandwidth to support eight cores at 5 GHz (with even 1866 being barely achievable in a lot of cases) but it looks like the CPU is awful too.

Does AMD just fall to the side because of lack of interest? Or have people been trying to make it work in vain?[/QUOTE]

AMD is only competitive on a capital basis. I used to buy all AMD but the last two systems I got have been Intel. Intel chips are faster per clock and faster per watt. AMD's Bulldozer architecture sucks, and they're still using it to make new chips. AMD is also a full process shrink behind Intel, which makes it difficult to produce competitive products.

AMD chased all the smart engineers out of the company a few years back when AMD started having financial troubles. Now they're really suffering. Keeping AMD alive is their APU, as well as being the chip in the new version of all three main consoles being released this and next year.

[QUOTE=TheMawn;356268]There isn;t even a plausible excuse for getting a GTX 660 Ti whilst already owning a GTX 670[/QUOTE]

You can use the 660 as a physx card for the 670, it actually helps some games. :max:

Lol well now that you mention it, I already have set the GTX 660 Ti to dedicated PhysX. Now, I just have to remember to turn mfaktc off or slow it down whenever I play on playing a PhysX enabled game.


Well, AMD definitely is dominating the budget market. Intel is making some serious project with hyper-efficient processors but I think they're scared of having multiple platforms, and chips like the i7-4770k with integrated graphics are all-around good instead of absolutely stupendous in targeted sectors. People with an unlocked i7 aren't usually going to be wanting integrated graphics.

I think AMD are doing well for the position they're in. If I understand correctly AMD and Intel were once upon a time in reversed positions. You just have to know how to play from behind, and making super cheap platforms is a good way to go. Making the processors for the next generation of consoles is absolutely massive. If they're using a chip that would be available for PCs, they'll have a good supply of CPUs which exceed the design specs for the console and they can sell them for PCs instead. You're going to have a number of absolutely sublime chips if you're manufacturing hundreds of millions for consoles.

Athlons, and especially Athlon64s were highly competitive when they came out. I should have switched to Intel, though, with my second-last rebuild. It was long past time. Mind, the FX 8350 isn't a bad chip, but the reduced FPU count was just plain stupid. Even my Athlon II hex core was not match for an i5 quad of its day, though it came closer, and the six full cores were useful.

EDIT: I knew AMD was really screwing up (and screwed) when they started laying off engineers at their Austin campus. Even before that, I think the ATI purchase was a mistake. They tied up too much capital, and maybe bit off more than they could chew on the technical front.

[QUOTE=kladner;356408]Athlons, and especially Athlon64s were highly competitive when they came out. I should have switched to Intel, though, with my second-last rebuild. It was long past time. Mind, the FX 8350 isn't a bad chip, but the reduced FPU count was just plain stupid. Even my Athlon II hex core was not match for an i5 quad of its day, though it came closer, and the six full cores were useful.

EDIT: I knew AMD was really screwing up (and screwed) when they started laying off engineers at their Austin campus. Even before that, I think the ATI purchase was a mistake. They tied up too much capital, and maybe bit off more than they could chew on the technical front.[/QUOTE]

They did it their way... :smile:

Luigi

[QUOTE=kladner;356408]Athlons, and especially Athlon64s were highly competitive when they came out. I should have switched to Intel, though, with my second-last rebuild. It was long past time. Mind, the FX 8350 isn't a bad chip, but the reduced FPU count was just plain stupid. Even my Athlon II hex core was not match for an i5 quad of its day, though it came closer, and the six full cores were useful.

EDIT: I knew AMD was really screwing up (and screwed) when they started laying off engineers at their Austin campus. Even before that, I think the ATI purchase was a mistake. They tied up too much capital, and maybe bit off more than they could chew on the technical front.[/QUOTE]
The Athlon IIs are roughly comparable to Core 2s.

[QUOTE=henryzz;356769]The Athlon IIs are roughly comparable to Core 2s.[/QUOTE]

OK, thanks. I had trouble either with the Athlon II chip, or the motherboard it was on. That's when I ended up with the FX. I should test it again. It really was a solid performer, even if it's a generation behind in comparison with Intel.

Ahh... gpu72 you thief what are you doing with some of my credit?? :razz:

(Just kidding.)

[QUOTE=kracker;356874]Ahh... gpu72 you thief what are you doing with some of my credit?? :razz:[/QUOTE][URL="http://www.mersenne.ca/credit.php?worktype=LL&exponent=38103407"]49.216 is correct credit amount[/URL]. Or at least it should be... :smile:

Hmm... Do I have a few stuck assignments or something preventing me?

[code]
Sorry kracker, but you already have too many assignments. Since you joined the project you have done on average 7.62 GHz Days of P-1 Factoring work per day.
You currently have 119 assignments totalling 368.56 GHz Days of work assigned, or 48 days worth based on your history. The oldest is 7 days old.
In order to have access to additional assignments, please complete some more of the work already assigned, or [URL="https://www.gpu72.com/contact/"]Contact Us[/URL] and explain why you should be exempted from this automated policy.

[/code]


I can do much, [I]much [/I]more than 7.62 GHz days...

No stuck assignments (in 7 days? c'mon! stuck is after 60 days or so!) and nobody has nothing against you. The server computes a 30 days moving average with each of your work type. If you don't do a type of work for a while, you lose the "score". This way gpu72 guards itself against idiots who reserve work "just because is fun to reserve work" and never do any real crunching. Which, of course, it is not your case, but the server doesn't know "yet". Some guy can't reserve more work than he can digest for a delimited period.

One solution is that you report your results as soon as they come out from the pipe for few days (i.e. don't delay the reports!), which will increase your recorded "average ability" to do P-1. Once the numbers raise, they stay high, the server "learns" that you [U]can do[/U] work, and will let you reserve more and more work. I think you are already doing this, if I look to your stats, your limits are already increasing, I can see them growing, waaaa....hehe...

Second solution is that you contact Chris and convince him to put you in the (short) list with exceptions. I am on that list, because Chris knows the hardware I have, and the fact that I don't produce "continuously". I have "periods" when the hardware gets free and I get the mood, and push it high, lots of reservations, and periods when I am busy with other things and the numbers decrease. I used to get this message often when I tried to reserve work after periods of "hibernation". I will tell you how I convinced Chris to add me to the exception list: I told him that if I try to reserve work and I can't, I turn my attention to something else, ignoring gpu72 for a while :razz:, so he got scared... :razz: Maybe you can tell him the same...

A minor nitpick about the GPU to 72 website: on the "Completed Assignments" page, the count should say "1000[COLOR="Red"]+[/COLOR] (or [COLOR="Red"]>[/COLOR]1000) Assignments completed" if the user has submitted more than 1,000 results.

[QUOTE=LaurV;358955].[/QUOTE]

Well the point is, not all my machines are in continuous work/submitting. I don't like submitting manually every day for those computers which I don't have access to 24/7...So I give them a "list" :smile:

The "Work Done By <Username>" graph X axis labels are running over each other:
[url]http://www.gpu72.com/graphs/worker/56f1b7572536a14513b08c88b2ba9578/[/url]

M64515779 has a factor: 6189543690695864260687 [TF:72:73:mfaktc 0.20 barrett76_mul32_gs]
M64515779 has a factor: 6273948586500036869753 [TF:72:73:mfaktc 0.20 barrett76_mul32_gs]
found 2 factors for M64515779 from 2^72 to 2^73 [mfaktc 0.20 barrett76_mul32_gs]

Not something I recall ever seeing. Bound to happen eventually I suppose

Nice! You won't see this, for the most of us, as we use the "when a factor is found, exit after current class". Your factors are from two different classes (1977 and respective 3964). So you have set the mfaktc to "when a factor is found, exit after current bitlevel", or not set to exit at all. Some people (me included) believe that is waste of time, in the most of the cases, after a factor is found, you don't need to "go higher". Some other people will blame me, and say that you can miss the smaller factor (not in your case, the smaller factor is in the smaller class) if you don't finish the bitlevel, etc.

Opinions vary, and finding factors is fun!

All in all, this duet of factors is very rare, a very nice find, congratulations.

I didn't know this was a thing. I suppose saving on average half the time on one exponent every time a factor is found isn't a huge deal but still.

I've gone and changed that. Neat feature.

[QUOTE=LaurV;359166]All in all, this duet of factors is very rare, a very nice find, congratulations.[/QUOTE]It's mostly rare because of the size of the factors (since as LaurV said it's common to stop immediately upon finding a factor), and that the two factors are in the same bit level. There's no shortage of exponents with multiple known factors, on my [URL="http://www.mersenne.ca/manyfactors.php"]Many Factors page[/URL] I don't list anything that has less than 4 known factors since there's far too many of them. But that's not grouped by size of factor, nor limited to factors in the same bit level. Running a quick database query on exponents below 10M I see 5959 (of 444942 = 1.3%) exponents have at least 2 known factors in the same bitlevel. Looking in the 50M range it's 3855/359501 = 1.1%

[QUOTE=TheMawn;359188]I suppose saving on average half the time on one exponent every time a factor is found isn't a huge deal[/QUOTE]
You are totally right here.

To make a bit of history talk, here people are split in two teams: some believe that is no reason to continue testing once a factor is found, and some other believe that letting the bitlevel unfinished is trouble, and not quite ethical... Well...in the future, our children will have faster hardware (and nothing better to do with it! :razz:) and will wish to continue to push the factoring limits higher, they may start with the next biltlevel and lose factors if they lay behind. Or some motivation like that. Because PrimeNet stores only the bitlevel which was done, it does not store the fact that such bitlevel was only "half done". Here again, we are victims of our imperfect tools: the right way should be to add the right field to the database.

In a word, finishing the bitlevel is not optimal for time, but is ethical, and it is what you pledged for, when you took the assignment.

Exiting after class is not ethical, it lets some work unfinished, but is "better" for the timing, assuming you do high bitlevels (to be clear: I am in this group of people, falling in the fallacy, see below).

Please remark the quotes of "better". Actually, here is where the argument came from, if you crunch at 70 bits, 5-6 minutes per exponent, you may find a factor every 70 assignments, which would take about 350 minutes, ~6 hours, so, every 6 hours you may lose 3 minutes (in average) to continue to check the rest of the bitlevel where you found the factor, and let the exponent "neat". If you TF at 75, and each exponent takes 5 hours, you may not like to lose 2.5-3 hours to finish a bitlevel, in spite of that on the long run, [U]the lost time is the same, or even lower, for higher TF levels[/U], i.e. you may find a factor every 75 assignments, which take 375 hours, from which you lose 2.5-3 hours to check the rest of the bitlevel.

So, theoretically, checking the whole bitlevel gets better as the bitlevel raises, because the factors are rarer there, and each time you find a factor, you "waste" in average half for the time per exponent per that bitlevel. The "fallacy" come from the fact that the "2-3 hours" are put together, seems to be a long time for "waste". At lower bitlevels you actually waste more time, but they are split in one minute here, one minute there...

As I said, opinions vary... You do your own [URL="http://en.wikipedia.org/wiki/Due_diligence"]due diligence[/URL] and set the parameters as you like.

[QUOTE=LaurV;359230]In a word, finishing the bitlevel is not optimal for time, but is ethical, and it is what you pledged for, when you took the assignment.[/QUOTE]

Thank you LaurV for phrasing it this way. An interesting interpretation and perspective.

I could very well be wrong (and would be happy to be corrected), but my understanding is it's a little more complicated than what you've outlined.

Based on something James told me quite some time ago when I sought his advise on how to fairly calculate "credits" (and which might have changed), Prime95 searches "upwards" when it TFs through a bit-level, while mfakt* searches downwards.

So it really comes down to those who want to know the lowest factor of all the MP candidates which have been factored cannot be sure they are all in the Primenet (and, separately, James') database for most of those candidates where a factor was found by mfakt*.

Some care; some don't.

Isn't it the case that mfaktc Factored results state "Partially tested," if 'StopAfterFactor=2'? Is this not carried over in PrimeNet reports on exponent status?

LaurV- Are you saying that only 'StopAfterFactor=0' is ethical?

On the other hand, by the time the hardware exists that people are trial factoring more or less for fun or to try to completely factor Mersenne numbers, they'll want to be double-checking the TF work to ensure nothing was missed in the first place. The likelihood of a factor being missed due to a GPU error is low, but for completeness they may want to TF every range twice to decrease the risk of anything being missed.


I too like the "What you promised to do" vs "What you set out to do" comparison. Quite subtle. Very interesting.

I'm trial factoring to cheaply eliminate candidates. As far as I am concerned, when I find a factor, I've eliminated that candidate and need not do any more work. If ever someone finds a serious flaw with this judgement, I'll be the first to switch back.

Of course, if someone did want to trial factor to higher limits for no better reason than finding as many factors as possible for known composites, they could just start over from the beginning of the bit level which contains the known factor. For the 1%-2% of exponents which DO have factors, re-doing the shortest bit level is really not a huge deal. If I found 2[SUP]70.01[/SUP] was a factor but it wasn't known if I factored all the way to 2[SUP]71[/SUP], redoing 2[SUP]70[/SUP] would be completely trivial in comparison to even 2[SUP]75[/SUP] (<2% of the total time [B]possibly[/B] wasted re-doing work on <2% of all exponents)

There's no "upwards or downwards" in currently implemented trial factoring in either of the programs. There are modulo classes that are turned into passes for efficiency. P95 and mfaktc have some differences in how these are used.

This misunderstanding probably stems from misreading George's remark in the code:
[CODE]/-- commonb.c around line 3950 --/

/* Set flag indicating a factor has been found! */

factor_found = TRUE;

/* We used to [B][COLOR="DarkRed"]continue factoring to find a smaller factor in a later pass[/COLOR][/B]. */
/* We'll continue to do this if the found factor is really small (less than */
/* 2^56) or if the user sets FindSmallestFactor in prime.ini. */

find_smaller_factor =
(end_bits <= (unsigned int) IniGetInt (INI_FILE, "FindSmallestFactor", 56));

/* Format and output a message */

makestr (facdata.asm_data->FACHSW,
facdata.asm_data->FACMSW,
facdata.asm_data->FACLSW, str);
sprintf (buf, "M%ld has a factor: %s\n", p, str);
OutputStr (thread_num, buf);
...[/CODE]
Well, this doesn't mean that P95 goes "downwards". It means that a smaller factor [I]may[/I] be found later in the remaining passes, in a different modulo class. [I]May[/I], not [I]will[/I].

And some (probably very few) people would like to know the smallest factor. Most won't care, only would care that this Mp is composite for the purposes of the project. Some would say that "it's not ethical to abandon the work that you signed up for; continue digging even though you already found water!", while others will retort, "No, I signed up to find water [u]or else[/u] until the end of the square plot. In this order. Instead of finishing digging this plot, I can do more use finding water on the next plot; well, or digging that next plot to its end if it will be barren."

[QUOTE=chalsall;359232]
So it really comes down to those who want to know the lowest factor of all the MP candidates which have been factored cannot be sure they are all in the Primenet (and, separately, James') database for most of those candidates where a factor was found by mfakt*.
[/QUOTE]
Just to pick nit: they can't do this anyways due to P-1 or ECM efforts.
E.g. I recently found a 94 bit factor for a M2,***,*** exponent (IIRC)(it's first known factor). It had been TF'd to 62 (again, running off memory). Who's to say there isn't a 62.1 bit factor for it?

[QUOTE=kladner;359237]LaurV- Are you saying that only 'StopAfterFactor=0' is ethical?[/QUOTE]
Not me! :smile:, remember, I am in the "fxuck the ethics!" team. :razz:
I just related the facts. There are advantages and disadvantages in both paths. You just chose yours.

Facts:

1. - GIMPS cares only about finding primes. There is where the glory and the money is. Once a factor is found (or a double LL is done), the exponent is scrapped. (don't argue with me here, the intention is not to argue with anyone).

2. - therefore once a factor is found, the factor is stored in the DB, and the "how far factored" is lost for the "public".

3. - other sites (mersenne.ca, etc) still keep the "how far factored", but some of them do not keep the "total" or "partial" for the last bit level, in case a factor was found there, and they do not keep the program used, the parameters (the number of classes), etc.

4. - some people care of finding factors more than finding primes, because finding factors is fun and easier to achieve, and factors may be useful for this or that math results, cryptography, whatever. Such people will continue to TF or P-1 mersenne numbers with known factors, higher and higher.

5. - once a factor is found, there is no way to know (beside doing your own recordings, homeworks, due diligence, etc) [B][U]how the factor was found[/U][/B]. This is important, because (as already pointed) different programs split the ranges differently (in bitlevels or in multiples of 60k candidates, etc) and for each range, do the sieving differently. Therefore, the guy who wants to "extend" the factorization, must [B][U]re-do the last bitlevel completely[/U][/B] i.e. to do [B][U]again[/U][/B] the work you already did. From this point of view, if you don't complete the range (bitlevel, or "up to 10^x, whatever you call it), is a bit unethical, like you asking/forcing the other guy to do part of the work you already did, to waste his resources to double part of your work.

To give you an example of calculus, assume one could use mfaktc to factor 2^71-1, with 420 classes. The factors are 228479, 48544121, and 212885833, but you don't know that yet. If you write them as 2*k*p+1, they are: 2*1609*71+1, 2*341860*71+1, 2*1499196*71+1.

Now, we have the modularity classes for k:

1609 = [B]349[/B] (mod 420)
341860 = [B]400 [/B](mod 420)
1499196 = [B]216 [/B](mod 420)

So, in this case, because mfaktc splits the range in classes, and it sieves and test the classes in increasing order, the biggest factor will be found first!!! Then the smallest factor will be found second, and only at the end mfaktc will find the middle factor (well, assuming we don't divide the other two factors out, which we can not do for high exponents).

If 4620 classes would be used ( bit too much for this exponent, bu we are talking here about exponents of millions), then

1609 = [B]1609 [/B](mod 4620)
341860 = [B]4600 [/B](mod 4620)
1499196 = [B]2316 [/B](mod 4620)

So the first factor is found first, then the biggest, then the middle one still found last.

If P95 is used, the factors will be found in order of the size: first, second, third.

If 2^191 is to be factored with mfakt*-like software, and considering we know the first (trivial) factor, depending of the number of classes, we can have either the third or the forth factor found first (here it makes more sense, as we still have a "cofactor" after dividing them out - think about exponents of millions.

[CODE]
gp > p=191; v=factorint(1<<p-1); print("\n"v[,1]); for(i=1,#v~,print(a=(v[i,1]-1)/(2*p)" : "a%4620" : "a%420" : "a%60))

[383, 7068569257, 39940132241, 332584516519201, 87274497124602996457]~
1 : 1 : 1 : 1
18504108 : 1008 : 168 : 48
104555320 : 100 : 100 : 40
870640095600 : 660 : 240 : 0
228467269959693708 : 3408 : 48 : 48
[/CODE]

Assuming that for some "exotic" exponent, there are 3 factors on the same testing range (bitlevel, etc), and the middle one in size is in the smallest class of k. That middle factor will be first found by mfakt* (which exits after the class), and stored in the DB.

If a guy wants to check for more factors, and he has no additional info (only the fact that the range was tested partially), then he [B][U]must recheck the last bitlevel, completely![/U][/B] Up, and down! Assuming he knows mfaktc was used, and he knows the number of classes, then he only need to check for the rest of the classes. Assuming he knows that P95 was used, he has to check all factor candidates bigger than the factor (due to the sieving style of p95). And so on.

(@chalsall: mfakt* does not test the candidates in "reversed" order, but in the incrementing order of the modularity classes, as in the example I just gave).

Of course, in the future the hardware will be more performant, and nitpickers will waste just few seconds to test our years-of-work, everything from scratch :razz:, but that is a different story.

This is where, actually, the dropped assignments matter... :razz:

Hey buddy, do you feel the heat? :razz:
[ATTACH]10451[/ATTACH]

hehehe...

[QUOTE=LaurV;359271]Hey buddy, do you feel the heat? :razz:
[ATTACH]10451[/ATTACH]

hehehe...[/QUOTE]

I just switched some to P-1 a week ago. Nope, I don't. Still chilly.

EDIT: I may get [I]quite a bit[/I] more cores soon. :razz:

[QUOTE=Batalov;359239]There's no "upwards or downwards" in currently implemented trial factoring in either of the programs.
This misunderstanding probably stems from misreading George's remark in the code[/QUOTE]My statement comes from George's code for calculating TF credit: there's one codeblock for TF done using Prime95, and one for "other". I don't have the code in front of me, but if it was something like this: If you have a factor of 70.9 bits, for Prime95 it would give you credit for 70 bits, plus 10% of 70-71 bits. For any other TF program, it would give you credit for 70 bits + 90% of 70-71.

For mesenne.ca I have added another credit-calculating function for mfakt* that determines which class the factor would be found in and grants credit accordingly since bitsize of factor (within a bitlevel) and runtime are unrelated.

[QUOTE=kracker;359278].....

EDIT: I may get [I]quite a bit[/I] more cores soon. :razz:[/QUOTE]

Is there a new toy on your horizon? :cool:

EDIT: BTW, I sure feel Uncwilly breathing down my neck in P-1. :ermm:

[QUOTE=kladner;359283]Is there a new toy on your horizon? :cool:[/QUOTE]

[SIZE="1"]Shhh.....[/SIZE]

[QUOTE=James Heinrich;359282]My statement comes from George's code for calculating TF credit: there's one codeblock for TF done using Prime95, and one for "other". I don't have the code in front of me, but if it was something like this: If you have a factor of 70.9 bits, for Prime95 it would give you credit for 70 bits, plus 10% of 70-71 bits. For any other TF program, it would give you credit for 70 bits + 90% of 70-71.

For mesenne.ca I have added another credit-calculating function for mfakt* that determines which class the factor would be found in and grants credit accordingly since bitsize of factor (within a bitlevel) and runtime are unrelated.[/QUOTE]
Your function makes much more sense!

However, the whole concept of credit, you know, is like this --

[QUOTE=Batalov;359287]the whole concept of credit, you know, is like this[/QUOTE]Certainly :smile:
Most of the TF work I do these days is beyond 1000M and therefore by definition gets no PrimeNet credit.

[QUOTE=kladner;359283]EDIT: BTW, I sure feel Uncwilly breathing down my neck in P-1. :ermm:[/QUOTE]
See attached to see how hot the breath is.
BTW, PrimeNet says that i should overtake you (and 2 or 3 more folks) on the lifetime list in 8 days

[QUOTE=Uncwilly;359320]See attached to see how hot the breath is.
BTW, PrimeNet says that i should overtake you (and 2 or 3 more folks) on the lifetime list in 8 days[/QUOTE]

Yep, and in GPU72 ranking as well.

[QUOTE=kladner;359321]Yep, and in GPU72 ranking as well.[/QUOTE]

And I'm 90% sure I can pass you in the near future as well :razz:

EDIT: That is, if I keep P-1 up instead of DC.

[QUOTE=kracker;359325]And I'm 90% sure I can pass you in the near future as well :razz:

EDIT: That is, if I keep P-1 up instead of DC.[/QUOTE]

Yeah. I kinda thought so, too. :cool:

I was never able to figure out how to access the all-time leaderboards. Can anyone enlighten me (as opposed to just giving me the link?)

[QUOTE=TheMawn;359335]I was never able to figure out how to access the all-time leaderboards. Can anyone enlighten me (as opposed to just giving me the link?)[/QUOTE]For PrimeNet, just go to Top Producers of what ever type of work you want to know about, then hit customize, set the date to something like 1990-01-01 (i.e. before the start of PrimeNet).
[url]http://www.mersenne.org/report_top_500_custom/?team_flag=0&type=1002&rank_lo=1&rank_hi=1000&start_date=1990-01-01&end_date=&B1=Get+Report[/url]
Look at the address bar after you click the link, it shows the start date.

Just fill in a date which is not in the last year (last 365 days).

Some time ago it was possible to get timed reports, like for the last 2 years or so, or for the last 3 months of 2011. This however was consuming the server, because there are lots of strange people out there (me is one of them) and they all have their own idea about the most important date they need reports for, and they used to change their mind every 5 minutes. :smile:

Therefore, after short discussions here on the forum, everybody agreed with the conclusion that we only need a 365 days report, and a lifetime report. Those reports can be pre-generated every hour, as opposed to generating them for different dates, different worktypes, etc, [U]every time [/U]when a user requested it. This lets the DB and the server free to do other things. When you request a report, the pre-generated report is printed on screen, therefore it can be one hour old. For sure this doesn't bother anyone. The 365 days report if your date is in the last year, or the lifetime if your date is older.

[QUOTE=kladner;359329]Yeah. I kinda thought so, too. :cool:[/QUOTE]

You can check weekly specific stats. Example,

[URL="https://www.gpu72.com/reports/workers/dc/week/"]Workers' Double Check Testing Progress for the [B][I][B][B]last Week[/B][/B][/I][/B][/URL]

[QUOTE=kracker;359637]You can check weekly specific stats. Example,

[URL="https://www.gpu72.com/reports/workers/dc/week/"]Workers' Double Check Testing Progress for the [B][I][B][B]last Week[/B][/B][/I][/B][/URL][/QUOTE]

Ah! Like this:
[url]https://www.gpu72.com/reports/workers/p-1/week[/url]

The OP was asking for top lists (leaderboards) on PrimeNet (maybe not in the right subforum, but that the question was about, therefore my answer). Your clarification is however, welcomed.

Lost the link to the thread with DC mismatches, most probably it was deleted or moved to archive. Anyhow:

@chalsall: May I have 30680549 back? (i.e. don't assign it to another user, as long as it was not assigned already)?

I see a mismatch was reported, by one of my GPUs which never gave failures in the past (i.e. when that occasionally happened, the original residues were wrong, not mine). I always triple-check the mismatches, usually before reporting, but this time I missed it! The triple check found a mismatch with the checkpoint file at iteration ~19M, (the checkpoint had all bytes zero inside (!!?!?)), after which obviously the next checkpoints had different residues. Still about 8M to go, so it will be ready in few hours, and obviously my residue is the one which is wrong this time. So, if not already assigned, keep it unassigned, or assign it back to me please.

[SIZE=2]
[/SIZE]

[QUOTE=LaurV;359972]may I have 30680549 back?[/QUOTE]
I already DC-ed, finished, matched. Still hanging assigned to GPU72, most probably it was assingned to a different user in the meantime, which user is not using P95, as the key is not claimed yet, ho he will waste ~20 hours of his GPU time, for a triple check which will be not accepted bt PrimeNet (same shift!). Well done, I hope that user is kracker :razz:

[QUOTE=LaurV;359997]Well done, I hope that user is kracker :razz:[/QUOTE]

Hey LaurV.

I'm a little too busy right now to "reach into the stream" and pull out a specific candidate just as it floats by just for you...

Let the systems do their job....

[QUOTE=LaurV;359997]I already DC-ed, finished, matched. Still hanging assigned to GPU72, most probably it was assingned to a different user in the meantime, which user is not using P95, as the key is not claimed yet, ho he will waste ~20 hours of his GPU time, for a triple check which will be not accepted bt PrimeNet (same shift!). Well done, I hope that user is kracker :razz:[/QUOTE]
Already? 20 hours?? You guys with these modern machines make me nervous. I do DCs one at a time on my single 6-year-old PC and spend 2-3 months on each one!

[QUOTE=Brian-E;360137]Already? 20 hours?? You guys with these modern machines make me nervous. I do DCs one at a time on my single 6-year-old PC and spend 2-3 months on each one![/QUOTE]
If you read carefully, it was already triple checked to 19M, so only about 11M iterations left, which is about 8 hours for this range, the whole run is like 23-30 hours or so, depending of the exponent (29-37M).

When I reported it, I didn't see it is a mismatch, I sent it with few other results together and only checked partially, but after I saw the mismatch I started a triple-check. I always do that, usually [U]before[/U] reporting the DC (to keep the exponent in the gpu72 pool, otherwise it may get assigned to somebody else). This time I started the triple check and I expected/hoped my result to be correct, and the initial residue to be wrong, this happened in the past more that the other way around. Well, this time I was wrong. And yes, the request to Chris was a bit childish too... As it turned out, the exponent is now DC-ed, but still hanging on Gimps Visualization. Some user might get is assigned, and waste GPU time with it. The comment about Kracker was a joke, and I know he is taking it as a joke.

[QUOTE=LaurV;360148]...the comment about Kracker was a joke, and I know he is taking it as a joke.[/QUOTE]


LaurV----->:ouch:

a joke as well.... :razz:

[QUOTE=LaurV;360148]If you read carefully, it was already triple checked to 19M, so only about 11M iterations left, which is about 8 hours for this range, the whole run is like 23-30 hours or so, depending of the exponent (29-37M).[/QUOTE]
I went through your post with a fine tooth comb, mate.:smile:
And you were attributing the 20 hours' unnecessary work to Kracker who, I had assumed, would not have thought to contact you for any save file you may have produced at 19M. Then again, I'm always open to surprises...:rolleyes:

[QUOTE=Brian-E;360169]I went through your post with a fine tooth comb, mate[/QUOTE]
Maybe through the wrong post. See one post before that one in which I was trying to fry the poor man.. There were two posts in a row (and if you want to suggest that I am replying to myself, well,... you may be totally right, I am getting a little bit dumber with the age, but don't tell anyone :smile:)

Anyhow, we move on now...

Shameless self-promotion plug...
 

Hey All.

Please forgive me for going completely tangential, and a bit of self-promotion...

But I thought some of you might be interested in knowing why I've been so busy lately... a new company I'm helping to form here in Barbados: [URL="http://www.forresearch.com/"]For Research Inc.[/URL].

And if anyone wants us to help design and analyse a medical study, please let me know.... :wink:

[QUOTE=chalsall;360738]Hey All.

Please forgive me for going completely tangential, and a bit of self-promotion...

But I thought some of you might be interested in knowing why I've been so busy lately... a new company I'm helping to form here in Barbados: [URL="http://www.forresearch.com/"]For Research Inc.[/URL].

And if anyone wants us to help design and analyse a medical study, please let me know.... :wink:[/QUOTE]

Congrats!

I would think twice before publishing this:

[code]
We're passionate about data.
About knowing what data are out there.
About understanding the quality of these data.
[/code]

:smile:

Luigi

[QUOTE=chalsall;360738]And if anyone wants us to help design and analyse a medical study, please let me know.... :wink:[/QUOTE]

What can we do to help?

[QUOTE=chalsall;360738]... a new company I'm helping to form here in Barbados: [URL="http://www.forresearch.com/"]For Research Inc.[/URL].

And if anyone wants us to help design and analyse a medical study, please let me know.... :wink:[/QUOTE]
All the best with the new venture!

[QUOTE=chalsall;360738]
And if anyone wants us to help design and analyse a medical study, please let me know.... :wink:[/QUOTE]
What are the areas of interest and what types of data would the group prefer?

Just lost my watercooled system due to a leak. Have to wait until Saturday to see if it can be salvaged. If anything is broken, it will take time to get fixed, so -~1400 GHzDays/Day until it comes back up (if ever) :no:

[QUOTE=flashjh;361218]Just lost my watercooled system due to a leak. Have to wait until Saturday to see if it can be salvaged. If anything is broken, it will take time to get fixed, so -~1400 GHzDays/Day until it comes back up (if ever) :no:[/QUOTE]

Oh, heck!

Sorry to hear about this Jerry, and thanks for the heads-up.

That kind of firepower is not easily compensated for.

[QUOTE=flashjh;361218]Just lost my watercooled system due to a leak. Have to wait until Saturday to see if it can be salvaged. If anything is broken, it will take time to get fixed, so -~1400 GHzDays/Day until it comes back up (if ever) :no:[/QUOTE]

Your water broke?

[QUOTE=swl551;361228]Your water broke?[/QUOTE]GRR! :ick:
(:rofl:)

[QUOTE=swl551;361228]Your water broke?[/QUOTE]
:w00t:
hahaha, good one!

(hey, you better do something with that misfit for P-1, instead of wasting time here making jokes! :razz:)

[QUOTE=chalsall;361226]Oh, heck!

Sorry to hear about this Jerry, and thanks for the heads-up.

That kind of firepower is not easily compensated for.[/QUOTE]

Cue Daveiddy....

[QUOTE=swl551;361228]Your water broke?[/QUOTE]

Ha, yes, and it wasn't pretty.

System is back up, lost one 580, but it wasn't liquid that did the damage. The card that stopped working probably just failed on its own. So, down one 580, but that's better than the whole system. :smile:

Woo! Crisis averted! :)

[QUOTE=flashjh;361861]So, down one 580, but that's better than the whole system. :smile:[/QUOTE]

Thanks for the news Jerry. Also, much thanks to all those who stepped up to the plate while you were off-line. Not a single candidate was released at 73. :smile:

I am getting a new (used) 580 today!

Oh yeah!

[QUOTE=chalsall;361868]Thanks for the news Jerry. Also, much thanks to all those who stepped up to the plate while you were off-line. Not a single candidate was released at 73. :smile:[/QUOTE]
You're quite welcome. I appreciate what everyone does here! I have to seek out some replacement 580s, they have to MSI Lightnings because that's what I have waterblocks for. Eventually I'll get the system back up to full. Once I complete runs on some DC expos I have for CUDALucas testing, I'll move that card to TF to make up some ground.

good news ... bad news
 

We (I) missed 10,000,000 GPU GhzDays recently.

Davideddy close your ears but as of 5 PM today there are no LL assignments below 68M available.

Yeah. Will be a big jump from my 550Ti. 2 year old cards are finay becoming affordable for cheapskates like me.

[QUOTE=petrw1;361879]Davideddy close your ears but as of 5 PM today there are no LL assignments below 68M available.[/QUOTE]

Don't Panic. [SUP](TM)[/SUP]

Everything is under control. Nothing not TFed to at least 74 and P-1ed is being released for LLing.

With his permission, I sometimes have Jerry's cycles dedicated to taking 68M's up to 74, which are then released back to Primenet for P-1 assignments. I've "taught" spidy how to have Primenet release high candidates which are already fully and appropriately "cooked".

I'm doing this now, in preparation for a batch of candidates TFed to 74, but not yet P-1'ed, to be released this evening.

Please rest assured I never do this without being "at console".

I just went into both my systems and had them export the completed work so far today, anything else will post at the normal time.