mersenneforum.org (https://www.mersenneforum.org/index.php)
-   GPU to 72 (https://www.mersenneforum.org/forumdisplay.php?f=95)

 James Heinrich 2015-09-17 02:21

[QUOTE=LaurV;410573]Ha! You did it for him and didn't do it for me! :tantrum:

 lycorn 2015-09-17 08:46

[QUOTE=James Heinrich;410556]Easily done.[/QUOTE]

Gosh, you were fast. Thanks :bow::bow:

 LaurV 2015-09-17 12:02

Which is actually not the same, my proposition was gaining us one more column :razz:
edit: joking apart, it looks much better now, thanks.

 James Heinrich 2015-09-17 12:11

[QUOTE=LaurV;410610]Which is actually not the same, my proposition was gaining us one more column :razz:
edit: joking apart, it looks much better now, thanks.[/QUOTE]But since your mystery proposal was made in a thread you know I don't read I'm very unlikely to fill it unless you re-post your request here. You know, this thread is called "Features request?" :wink:

 LaurV 2015-09-17 13:38

For the records, [URL="http://www.mersenneforum.org/showthread.php?p=410460"]here it was[/URL]. It is not important anymore. I see now a different problem, the links to your tables. They are not valid anymore the next day, because the things moved. Some "time absolute" links, like "from date, to date" would have to be used in posts like the one I put there, otherwise the links I give don't have any connection anymore with what I was saying :whistle:
For example my "two days ago" link clicked today does not show what it was showing when I clicked yesterday... hehe... well, you know what I mean... :razz:

 James Heinrich 2015-09-17 14:27

[QUOTE=LaurV;410616]Some "time absolute" links, like "from date, to date" would have to be used in posts like the one I put there, otherwise the links I give don't have any connection anymore with what I was saying :whistle:[/QUOTE]Something I'm aware of and I'm working to correct. Of course, it will most likely break existing links :redface:
But that's why this is still considered "beta" :smile:

 James Heinrich 2015-09-17 15:44

[QUOTE=LaurV;410616]I see now a different problem, the links to your tables. They are not valid anymore the next day, because the things moved. Some "time absolute" links, like "from date, to date" would have to be used in posts like the one I put there, otherwise the links I give don't have any connection anymore with what I was saying[/QUOTE]That problem is now solved, at the expense of breaking any existing links :whistle:

Some examples:[list][*][url]http://www.mersenne.ca/status/tf[/url] = most recent available single-day status[*][url]http://www.mersenne.ca/status/tf/20150913/0/1/0[/url] = single-day status of a particular date (13-Sep-2015)[*][url]http://www.mersenne.ca/status/tf/0/2/1/0[/url] = compare most recent data to data of 2 days before[*][url]http://www.mersenne.ca/status/tf/20150913/2/1/0[/url] = compare specific date (13-Sep-2015) to data of 2 days prior[*][url]http://www.mersenne.ca/status/tf/20150916/20150913/1/0[/url] = compare specific date (16-Sep-2015) to data of another specific date (13-Sep-2015)[/list]

 chalsall 2015-09-17 16:17

[QUOTE=James Heinrich;410630]That problem is now solved, at the expense of breaking any existing links :whistle:[/QUOTE]

Thanks James. I've updated the links on GPU72.

However, there's something not quite right with your GUI... For example, if you go to [URL="http://www.mersenne.ca/status/tf/0/7/3/4000"]http://www.mersenne.ca/status/tf/0/7/3/4000[/URL] things are as expected. But if you then change the "Days Prior" field and Submit, things are not as expected.

 James Heinrich 2015-09-17 16:25

It seems to be working for me.

It's possible your browser is caching the .js file -- do a Ctrl-F5 to force-refresh the linked javascript file and I think it should fix your problem?

 chalsall 2015-09-17 16:28

[QUOTE=James Heinrich;410637]It's possible your browser is caching the .js file -- do a Ctrl-F5 to force-refresh the linked javascript file and I think it should fix your problem?[/QUOTE]

Thanks.... :redface:

 James Heinrich 2015-09-17 17:29

Another forced refresh will bring some changes that fixes some issues with the days-back part of the GUI.

Did I ever mention how much I hate working with Date objects in Javascript?
And then just when you think you got it working you try a different browser and it does something totally different. :bangheadonwall:

[QUOTE=James Heinrich;410637]It seems to be working for me.

It's possible your browser is caching the .js file -- do a Ctrl-F5 to force-refresh the linked javascript file and I think it should fix your problem?[/QUOTE]

FWIW, whenever I update a JS or CSS or something like that, I bump a query parameter on it which will usually cache bust it.

Like /blah.js?v=20150917.1

Of course how cacheable it is depends on the server and it's expiration time on objects in the first place, but oh well.

 chalsall 2015-09-17 18:49

[QUOTE=Madpoo;410657]FWIW, whenever I update a JS or CSS or something like that, I bump a query parameter on it which will usually cache bust it.[/QUOTE]

Yes... This is a common "trick" used with AJAX fetches et al.

 James Heinrich 2015-09-17 19:12

[QUOTE=Madpoo;410657]FWIW, whenever I update a JS or CSS or something like that, I bump a query parameter on it which will usually cache bust it.[/QUOTE]I usually do as well, I was lazy. I have added that now.

 LaurV 2015-09-17 19:42

Brilliant James! It is working perfect for me, and it seems the old links are redirected too (i.e. it won't let me with some "error 404" or so). Few more clicks and I am in the range and date I want. Perfect. Thanks a million. That is what I call "technical support in real time". You and Scott rock at it!

 snme2pm1 2015-09-20 02:59

Something doesn't feel right in the propagation of the date specific qualification.
The algorithm seems too quick to introduce an absolute date.
I maintain tabs that I would refresh on new days.
In recent days, I tend to find that they have become date specific rather than "today".
Thus when refreshed they present unchanged information.

P.S. What would be the proper place for future discussion about mersenne.ca. facilities?
Chris might become annoyed by the continued use of this thread for matters quite unrelated to his GPU72 project.
Whilst there was a historic connection, that has been broken.

 LaurV 2015-09-20 07:13

The non-date-specific links still work for me, you can bookmark them and click the bookmark instead of clicking refresh tab - tht is also a single mouse click.

Edit: I don't think Chris is upset about this discussion at all, it is connected to GPu72 and Gimps Visu anyhow. And because you reminded me, one thing which should be extremely useful to be saved from the old side, for "historical" reasons, is the collection of the movies with moving candle bars... Few of the movies I really love, I was part of them, and it should be a pity to be lost... Of course, an update of some of them would not hurt either, a lot o' progress was done since that time and a lot of other people contributed to it much more than myself, but I know that updating them is a freaking time consuming activity (I mean not only to put the data together, but rendering and compressing the movie too)

 alpertron 2015-09-20 16:49

There is no need for movies. From the data, a JavaScript application can show the same animation by using canvas. Since the application can be cached in the Web browser, only the data will need to be downloaded, so the bandwidth to reproduce the animation will be small.

 chalsall 2015-09-20 20:32

[QUOTE=LaurV;410884]Edit: I don't think Chris is upset about this discussion at all, it is connected to GPu72 and Gimps Visu anyhow.[/QUOTE]

Absolutely. To the contrary, I'm very thankful to James for undertaking this.

I had hoped to have had the cycles to fix this myself, but I've recently learnt (the hard way) that "snagging" (read: finishing to completion) two houses' construction takes way more time that you might expect! (Damn Humans! Damn Atoms! :wink:)

 chalsall 2015-09-20 20:35

[QUOTE=alpertron;410902]There is no need for movies. From the data, a JavaScript application can show the same animation by using canvas. Since the application can be cached in the Web browser, only the data will need to be downloaded, so the bandwidth to reproduce the animation will be small.[/QUOTE]

An excellent idea!

I had used the videos as a "test bed" for HTML5 video, but what you suggest would eliminate the heavy server-side processing, and pass the load onto the client. Perfect for data-driven videos.

 James Heinrich 2015-09-20 22:16

[QUOTE=snme2pm1;410877]Something doesn't feel right in the propagation of the date specific qualification.
The algorithm seems too quick to introduce an absolute date.[/QUOTE]I have spent the afternoon reworking the GUI. Hopefully it now makes use of more sensible URLs with relative or absolute date comparisons for your convenience.

On the flip side, it's pretty likely I've broken something and not noticed it, so please let me know what it is and I'll try and fix it :smile:

PS: as for graphs, they're on my to-do-soon list. I'll try and include an option for animation as well (guaranteed not to work for [i]retina[/i] :whistle:)

 petrw1 2015-09-21 15:27

[QUOTE=James Heinrich;410917]I have spent the afternoon reworking the GUI. Hopefully it now makes use of more sensible URLs with relative or absolute date comparisons for your convenience.

On the flip side, it's pretty likely I've broken something and not noticed it, so please let me know what it is and I'll try and fix it :smile:

PS: as for graphs, they're on my to-do-soon list. I'll try and include an option for animation as well (guaranteed not to work for [i]retina[/i] :whistle:)[/QUOTE]

I see the new GUI menu and I like it....however now when I click on a row in the table to zoom it does NOT zoom in fact it goes back to 0 - 900M.

 James Heinrich 2015-09-21 15:35

[QUOTE=petrw1;410967]I see the new GUI menu and I like it....however now when I click on a row in the table to zoom it does NOT zoom in fact it goes back to 0 - 900M.[/QUOTE]Indeed. Thanks. Fixed.
I also uploaded the missing .js file for the calendar for picking specific dates.

 CuriousKit 2015-09-21 16:56

[Feature Suggestion] "Cyclic Cooldown"

I have a suggestion for a feature for a future iteration of Prime95: "Cyclic Cooldown".

Basically, an option to, say, tell Prime95 to periodically stop the threads to give time for the machine to cool down. I'm not sure how practical it is to obtain the CPU temperature using assembly language, especially something that doesn't require high privileges, but the idea is that the feature will still allow you to run Prime95 on machines that are prone to overheating.

My own case in point: I once acquired a netbook that I tried to run Prime95 on, but after around 10 to 15 minutes, the machine forcefully turned itself off because it overheated, and the only way I could get around it was to play with the system settings to slow the CPU down. But would it be feasible to have a feature where you can have the worker threads running for 5 to 10 minutes (configurable), and then do a full stop for another 5 to 10 minutes (also configurable, so you can tailor the times if you know your machine well) so the machine can cool down, before starting again. Obviously it means the tests can take twice as long, but at least it can do [I]something[/I] without shutting off without having a chance to save etc, and can be left unattended.

Of course, the best solution would be periodic monitoring of the CPU core temperature and shutting down the worker threads if it reaches an upper threshold, and not starting up again until it cools below a lower threshold, but I'm not sure how practical that is.

 petrw1 2015-09-21 17:07

[QUOTE=CuriousKit;410973]I have a suggestion for a feature for a future iteration of Prime95: "Cyclic Cooldown".

Basically, an option to, say, tell Prime95 to periodically stop the threads to give time for the machine to cool down. I'm not sure how practical it is to obtain the CPU temperature using assembly language, especially something that doesn't require high privileges, but the idea is that the feature will still allow you to run Prime95 on machines that are prone to overheating.

My own case in point: I once acquired a netbook that I tried to run Prime95 on, but after around 10 to 15 minutes, the machine forcefully turned itself off because it overheated, and the only way I could get around it was to play with the system settings to slow the CPU down. But would it be feasible to have a feature where you can have the worker threads running for 5 to 10 minutes (configurable), and then do a full stop for another 5 to 10 minutes (also configurable, so you can tailor the times if you know your machine well) so the machine can cool down, before starting again. Obviously it means the tests can take twice as long, but at least it can do [I]something[/I] without shutting off without having a chance to save etc, and can be left unattended.

Of course, the best solution would be periodic monitoring of the CPU core temperature and shutting down the worker threads if it reaches an upper threshold, and not starting up again until it cools below a lower threshold, but I'm not sure how practical that is.[/QUOTE]

This is in a way already there though not with any automatic sensors....

[QUOTE]If you are running on a laptop or in a hot room and you want to slow the
program down to reduce heat, then you can add this line to prime.txt:
Throttle=n
where n is the percentage of time the program should run. For example,
Throttle=40 means the program will run 40% and be idle 60% of the time.[/QUOTE]

In my own experiment, if my memory is somewhat intact: If you use Throttle=90 for example it will run for 90 seconds then stop for 10 etc....

In my humble opinion: I believe that the core continually getting hot/cool/hot/cool is harder on it than a constant temp even if it is on the high side....

 James Heinrich 2015-09-21 17:21

[QUOTE=CuriousKit;410973]I have a suggestion for a feature for a future iteration of Prime95[/QUOTE]Note that this thread is for feature requests on [url]www.gpu72.com[/url] (and by oblique extension [url]www.mersenne.ca[/url]) -- feature suggestions for Prime95 itself should probably go in the [URL="http://mersenneforum.org/forumdisplay.php?f=10"]Software[/URL] forum.

 CuriousKit 2015-09-21 17:26

Erg, sorry. While trying to avoid making a new topic, I did a flat search for "Suggestion" without really looking properly at the forum board. Thanks for the correction.

[QUOTE=CuriousKit;410976]Erg, sorry. While trying to avoid making a new topic, I did a flat search for "Suggestion" without really looking properly at the forum board. Thanks for the correction.[/QUOTE]

And even though we're pretty off-topic, for future reference, on a Windows system it's available through a simple WMI call.

For example, run this from a command line (might need to run as admin):
[CODE]wmic /namespace:\\root\wmi PATH MSAcpi_ThermalZoneTemperature get CurrentTemperature[/CODE]

I think it outputs as Celsius * 100, e.g. I get "2815" when I run it which is 28.15 C ( ~ 82.7 F). But on this system I think that may be giving me a temp zone reading, not necessarily the CPU. But you get the idea.

Problem is, you'd want to measure it over time... it'll probably reach a high temp in seconds after starting, so you'd be more interested in throttling it to keep it from hitting the max temp for too long a period? Or introduce some throttling internally to try and keep the temp in a sweet spot.

Either way, a laptop prone to overheating probably isn't the best place to run something like this.

 Mark Rose 2015-09-21 19:51

[QUOTE=James Heinrich;410975]Note that this thread is for feature requests on [url]www.gpu72.com[/url] (and by oblique extension [url]www.mersenne.ca[/url]) -- feature suggestions for Prime95 itself should probably go in the [URL="http://mersenneforum.org/forumdisplay.php?f=10"]Software[/URL] forum.[/QUOTE]

What is needed is a Feature Request thread for each.

 Xyzzy 2015-09-21 22:35

[QUOTE=Mark Rose;410989]What is needed is a Feature Request thread for each.[/QUOTE]Maybe post a request in a "Request for Feature Request Threads" thread?

:max:

 VBCurtis 2015-09-22 03:33

[QUOTE=Xyzzy;410999]Maybe post a request in a "Request for Feature Request Threads" thread?

:max:[/QUOTE]

Gold. :tu:

[QUOTE=Xyzzy;410999]Maybe post a request in a "Request for Feature Request Threads" thread?

:max:[/QUOTE]

:bow: :grin:

 James Heinrich 2015-09-22 13:36

[QUOTE=chalsall;408174]I would be happy to provide you with the code which generates the tables, graphs etc. It's in Perl though...[/QUOTE]Is your graph-generating code home-rolled, or are you using a chart library of some kind?
I'm not sure if I should write my own chart generator to make something along the lines of the existing mersenne.info graphs (i.e. PNG output) or something fancy and JS-based like [url=http://www.amcharts.com/demos/3d-stacked-column-chart/]amCharts[/url].

 chalsall 2015-09-22 14:31

[QUOTE=James Heinrich;411042]Is your graph-generating code home-rolled, or are you using a chart library of some kind?[/QUOTE]

On Mersenne.info and GPU72, all graphs are generated using [URL="http://search.cpan.org/dist/GDGraph/Graph.pm"]GD:Graph[/URL]. Remarkably fast.

However, you might want to consider a client-side solution; I've had great success with the [URL="http://d3js.org/"]Data-Driven Documents[/URL] Javascript library. Absolutely amazing what you can do with it!

[QUOTE=James Heinrich;411042]Is your graph-generating code home-rolled, or are you using a chart library of some kind?
I'm not sure if I should write my own chart generator to make something along the lines of the existing mersenne.info graphs (i.e. PNG output) or something fancy and JS-based like [url=http://www.amcharts.com/demos/3d-stacked-column-chart/]amCharts[/url].[/QUOTE]

*GASP* Javascript? Oh noes... that would mean Retina couldn't view pretty graphs. Oh well... he'd just say that graphs are useless and prefers a table of numbers. :smile: (I had to get that in... sorry Retina!)

I looked at Google Charts for some mersenne.org things but didn't get too deep into them:

I may explore it more down the road, but just wanted to point it out.

 alpertron 2015-09-22 16:04

I think it is OK to use JavaScript, which is a Web standard and supported by all browsers. Of course if someone wants to disable JavaScript, he will not be able to see the graph, but this is a problem of the person who does not want to see Web pages as they were designed. In the same way, if you want to disable CSS, you will see Web pages incorrectly.

 James Heinrich 2015-09-22 16:08

[QUOTE=Madpoo;411055]*GASP* Javascript? Oh noes... that would mean Retina couldn't view pretty graphs.[/QUOTE]Absolutely. I [url=http://www.mersenneforum.org/showpost.php?p=410917&postcount=312]already promised[/url] that he wouldn't be able to :smile:

[QUOTE=Madpoo;411055]I looked at Google Charts ... just wanted to point it out.[/QUOTE]Yes, that would have been my first choice since I'm passingly familiar with it, unfortunately there's no option for stacked columns or other 3-dimension charts (I don't mean the columns look 3D, I mean 3 data dimensions).

 James Heinrich 2015-09-22 17:48

So I've tossed up some basic graphs by amCharts. It's slightly heavy with so many data points loading, but it's very easy to use (for me :smile:)
I'll still need to tweak a number of things, like axis labels and so forth. But it's a start.

 alpertron 2015-09-22 18:01

[QUOTE=James Heinrich;411066]So I've tossed up some basic graphs by amCharts. It's slightly heavy with so many data points loading, but it's very easy to use (for me :smile:)
I'll still need to tweak a number of things, like axis labels and so forth. But it's a start.[/QUOTE]

Very good. But I think that the bars should be thinner, so we can see the bars behind them.

 chalsall 2015-09-22 18:06

[QUOTE=James Heinrich;411066]I'll still need to tweak a number of things, like axis labels and so forth. But it's a start.[/QUOTE]

Looking great James!

One thing I found I had to do was make the bars relatively thin (on both axis), so you could see all of the ones behind. Also, I had the higher factored bars in front as there are often fewer of those than lower factored.

 alpertron 2015-09-22 18:34

Another suggestion would be to add the bit size of the exponents at the right of the image so the meaning is more clear.

For example: for [url]http://www.mersenne.ca/status/tf/0/0/3/0[/url] you should write 62, 63, 64, ... at the right.

 LaurV 2015-09-23 05:31

[QUOTE=chalsall;411068]Looking great James![/QUOTE]
+1! B-e-a-utiful! We love the on/off checkboxes on the right!
Still waiting for the movies :razz: (just kidding, man, you did a wonderful job!)

 Dubslow 2015-09-23 08:04

[QUOTE=alpertron;411060]I think it is OK to use JavaScript, which is a Web standard and supported by all browsers. Of course if someone wants to disable JavaScript, he will not be able to see the graph, but this is a problem of the person who does not want to see Web pages as they were designed. In the same way, if you want to disable CSS, you will see Web pages incorrectly.[/QUOTE]

CSS can't (at least not that I know of) be used against the person viewing the website. JS very easily can, retina has nothing against the authors or the websites themselves, he is only in the self protection game.

Edit: The imgur zero-day exploit yesterday was an excellent example.

 alpertron 2015-09-23 12:51

It always depends on the type of Web sites you visit. In general you will not find these problems on mathematical or other "boring" (for most people) stuff. Hackers like to attack sites that are visited by most people.

I have visited Web sites for 18 years with JavaScript enabled, and I have never had any incident.

[QUOTE=Dubslow;411090]CSS can't (at least not that I know of) be used against the person viewing the website. JS very easily can, retina has nothing against the authors or the websites themselves, he is only in the self protection game.

Edit: The imgur zero-day exploit yesterday was an excellent example.[/QUOTE]

I thought there was some exploit a while back that used some CSS maneuvering to put itself over a logon box (for example) and make it seem like you're typing in there, but it's going into an iframe.

Of course there was more to it than that, like the iframe had to be able to interact with the main site, so it really only affected certain sites that were setup poorly, but still...

I may have the details on that wrong, but I recall something like that affecting Facebook?

Okay, I admit, I skimmed the article. But I figured since CSS can position things and a poor iframe implementation could let outside code do something like interact with the parent frame, then maybe, just maybe?

 chris2be8 2015-09-24 15:55

[QUOTE=alpertron;411100] I have visited Web sites for 18 years with JavaScript enabled, and I have never had any incident.[/QUOTE]

That's any incident you know of. Security specialists tend to be slightly paranoid, which is generally considered a virtue in the field.

And avoiding "dangerous" web sites is harder than you think. At least skip any site with advertising on it (you can't rely on AdBlock+ to stop all advertising).

Chris

 alpertron 2015-09-24 18:44

[QUOTE=chris2be8;411189]That's any incident you know of. Security specialists tend to be slightly paranoid, which is generally considered a virtue in the field.

And avoiding "dangerous" web sites is harder than you think. At least skip any site with advertising on it (you can't rely on AdBlock+ to stop all advertising).

Chris[/QUOTE]

There is no problem with ads. With time people acquire "ad blindness", so ads are automatically ignored by the brain, and you can concentrate in the Web page subject without even knowing what is the advertisement about.

 chalsall 2015-09-24 19:21

[QUOTE=alpertron;411195]There is no problem with ads.[/QUOTE]

Incorrect. Ads can come with malicious Javascript.

[QUOTE=chalsall;411198]Incorrect. Ads can come with malicious Javascript.[/QUOTE]

And often do.

The websites I manage run ads. Some come from Google Adsense, some are direct buys, and they all get checked to make sure they're cool.

The problem comes about when an advertiser buys a placement using some benign looking stuff, and then once it's approved, they change up what they serve from their own system. The old bait and switch.

End result is someone visits our site and gets one of those bogus "you've got a virus!" things that takes over the window and can't be closed using conventional means. It's annoying, we hate it, our customers hate it, Google hates it, and they do end up getting banned for life, but these are sock puppets and show up again somewhere else.

I just got my first Chrome on Android "you've got a virus" pop-up... super annoying. I normally root my phones and install an ad-blocker, but in this case I haven't done that yet, and sure enough, the darn thing pops up and even made my phone vibrate at max intensity... super annoying.

Of course the idiot who gets these and then says "oh noes, I better click on that shiny flashy button and type in my credit card #" really only has himself to blame at that point. Worst case is when the malicious ad has zero-day stuff or redirects you somewhere that does.

 LaurV 2015-09-25 03:49

[QUOTE=Madpoo;411208]and can't be closed using conventional means[/QUOTE]
One more reasons to love Firefox. We used to go endless roundabouts to avoid clicking on popping stuff like (when you click the red x) "this page asks you to confirm that you want to close it". One can do strange stuff with those popups, if you allow it (by clicking on it, because you don't know what you are clicking on, they can overpose pictures over the real popups)... Sometime using escape key closes the popup but you still could not close the page, because when clicking the red x, you get the popup again. With the old FF we could drag the tab on the desktop, where another instance of FF would open (move the current tab to new browser window) and we closed the other instance from the task manager, this way avoiding to close all open tabs. Since people claimed to the company and suggested patches - advantages of having an open source browser - they changed last year: now you just click once more on the red x and the bad tab, together with all malefic popups, is gone. We hate when a site asks us "are you sure you want to close me?" after we click the red x, forcing us to click an "yes" on an uncertain popup window.

[QUOTE=LaurV;411221]One more reasons to love Firefox. We used to go endless roundabouts to avoid clicking on popping stuff like (when you click the red x) "this page asks you to confirm that you want to close it"...[/QUOTE]

That used to drive me nuts, but at the time I was using IE most of the time (corporate websites depended on it, bleagh).

Now I've been using the latest versions of Chrome for years and I don't know if it's just me, or the fact that I run the AdBlocker Plus plugin, but I haven't seen one of those annoying "you can't close me!" things in years. Good riddance.

I used to resort to killing the "iexplore.exe" process entirely... sometimes that was the only "safe" way to handle it without winding up in a click-hell cycle.

 James Heinrich 2015-10-03 17:54

[url=http://www.mersenne.ca/status/tf]TF visualization[/url] graph now supports [url=http://www.mersenne.ca/status/tf/0/7/2/0]delta graphs[/url] (change between two dates).

Also I added some controls so you can play manipulate the graph in pseudo-3D (in my experience it's smooth on Chrome and choppy in Firefox, and I don't care if it works in IE or not :smile:)

 chalsall 2015-10-03 18:06

[QUOTE=James Heinrich;411909][url=http://www.mersenne.ca/status/tf]TF visualization[/url] graph now supports [url=http://www.mersenne.ca/status/tf/0/7/2/0]delta graphs[/url] (change between two dates).[/QUOTE]

Super cool James!!! :smile:

Any progress on the P-1 / LL / DC deltas? We're flying a little blind at the moment...

[QUOTE=James Heinrich;411909]...and I don't care if it works in IE or not :smile:)[/QUOTE]

Does anyone still use Internet Exploder?

 chalsall 2015-10-03 18:34

[QUOTE=chalsall;411913]Super cool James!!! :smile:[/QUOTE]

Just looking at this a little deeper... Examining the [URL="http://www.mersenne.ca/status/tf/0/7/3/4000"]delta graph for the 40M to 50M range over the last week[/URL], for example, we clearly see the huge dumps by Anonymous.

Two points...

First, the right hand key column on the graph shows TF levels between 70 bits and 78, and yet there were no deltas beyond 73 bits in this range. This appears also in the main graph, where half the graph is transparent.

Secondly, is amCharts able to show stacked 3D bar charts? It might be cool to show where factors were found, so the visual columns are aligned.

Please don't take this the wrong way; I am very appreciative of what you've done here. And, everyone is a critic, few are artists.... :smile:

 James Heinrich 2015-10-03 18:53

1 Attachment(s)
[QUOTE=chalsall;411913]Any progress on the P-1 / LL / DC deltas? We're flying a little blind at the moment...[/QUOTE]Not currently. I need to pull in that data from somewhere. It's on my to-think-about list. I just spent the week moving (again) so my house is a pile of boxes right now...

[QUOTE=chalsall;411913]Does anyone still use Internet Exploder?[/QUOTE]Apparently ~8% of site visitors, for whatever that's worth. That number supposedly excludes spiders and bots, but who knows.

[QUOTE=chalsall;411914]First, the right hand key column on the graph shows TF levels between 70 bits and 78, and yet there were no deltas beyond 73 bits in this range.[/QUOTE]I had to suppress some scaling code for delta graphs where I ran into divide-by-zero errors. I didn't have time to examine the problem so I hid it :blush: I'll have to go back and fix it properly sometime.

[QUOTE=chalsall;411914]Secondly, is amCharts able to show stacked 3D bar charts? It might be cool to show where factors were found, so the visual columns are aligned.[/QUOTE]I [url=http://www.amcharts.com/demos/]think so[/url], but perhaps you could give me a more explicit example of what you're thinking of in this case.

 chalsall 2015-10-03 19:11

[QUOTE=James Heinrich;411915]I just spent the week moving (again) so my house is a pile of boxes right now...[/QUOTE]

Been there, done that. Sucks big time.

[QUOTE=James Heinrich;411915]I had to suppress some scaling code for delta graphs where I ran into divide-by-zero errors. I didn't have time to examine the problem so I hid it :blush: I'll have to go back and fix it properly sometime.[/QUOTE]

Yeah. Most languages don't handle division by zero well. You have to introduce a special case conditional to avoid an exception.

 James Heinrich 2015-10-03 19:17

[QUOTE=chalsall;411917]Most languages don't handle division by zero well.[/QUOTE]Obviously it's not supposed to be dividing by zero. :smile:
My code logic was predicated on a single-day data set, I kinda just shoved the delta data into it and hoped it worked, and it mostly did. I need to poke around some more when I have some more time and see what it's trying to do and fix it properly.

 chalsall 2015-10-03 22:38

[QUOTE=James Heinrich;411918]Obviously it's not supposed to be dividing by zero. :smile:[/QUOTE]

There can be cases where dividing by zero is valid.

This simply results in infinity.

 science_man_88 2015-10-03 22:44

[QUOTE=chalsall;411930]There can be cases where dividing by zero is valid.

This simply results in infinity.[/QUOTE]

this depends on direction see:

 chalsall 2015-10-03 22:55

[QUOTE=science_man_88;411931]this depends on direction see:[/QUOTE]

In my opinion, only an idiot would link to a YT video to enter an argument.

 science_man_88 2015-10-03 22:57

[QUOTE=chalsall;411932]In my opinion, only an idiot would link to a YT video to enter an argument.[/QUOTE]

sorry I thought I was the village idiot for some reason

 chalsall 2015-10-03 23:03

[QUOTE=science_man_88;411933]sorry I thought I was the village idiot for some reason[/QUOTE]

You are, we suspect.

A serious question... Are you related to Rick Mercer?

 science_man_88 2015-10-03 23:09

[QUOTE=chalsall;411934]You are, we suspect.

A serious question... Are you related to Rick Mercer?[/QUOTE]

no not to my knowledge and our family genealogist died within the last month so I can't ask them. Also you also tend to get repetitive because that's the 4th time you've replied to a post I made that way in under 3 years.

 James Heinrich 2015-10-03 23:11

The ranges should be correctly set for the delta graphs now.

 LaurV 2015-10-04 03:27

[QUOTE=science_man_88;411935]no not to my knowledge and our family genealogist died within the last month so I can't ask them. Also you also tend to get repetitive because that's the 4th time you've replied to a post I made that way in under 3 years.[/QUOTE]
You know what? I would take that as a compliment. When he said it first time I googled for the guy and I really like that guy and what he is doing for the people with disabilities. I would like him to be my relative.

Now back on topic, @James
[QUOTE=James Heinrich;411936]The ranges should be correctly set for the delta graphs now.[/QUOTE]
Man I like the delta graphs! I can't understand how do you find the free time to do all that stuff. I wanted just to add that the delta graphs are better viewed from the angle of zero (due to the negative bars are difficult to see in 3D) so you could make a asterisk note under the graphs (better viewed from a zero angle, or so). Another idea, but that is already on the nitpicking side, is to make the negatives darker or change their border. But the zero angle note would be enough.

 Uncwilly 2015-10-04 03:32

[QUOTE=James Heinrich;411915]I had to suppress some scaling code for delta graphs where I ran into divide-by-zero errors. I didn't have time to examine the problem so I hid it :blush: I'll have to go back and fix it properly sometime.

I [url=http://www.amcharts.com/demos/]think so[/url], but perhaps you could give me a more explicit example of what you're thinking of in this case.[/QUOTE]
This seems to break things:
[url]http://www.mersenne.ca/status/tf/0/7/4/33200[/url]

 James Heinrich 2015-10-04 03:52

[QUOTE=Uncwilly;411946]This seems to break things: [url]http://www.mersenne.ca/status/tf/0/7/4/33200[/url][/QUOTE]Thanks, fixed. It broke because I was lazy :redface: with the code and it caught me out by having no results in the first line of the table. I've fixed the code to be saner. :smile:

 Dubslow 2015-10-04 05:31

(Off topic post)

[QUOTE=chalsall;411930]There can be cases where dividing by zero is valid.

This simply results in infinity.[/QUOTE]
That is of debatable merit in the computer realm. Perhaps in some symbolic logic packages, but almost never in PHP or JS or similar. Even including the symbolic logic packages, overall the cases where it is not an error are outweighed by the cases where it is by a couple orders of magnitude, so making them an error by default is a very sensible thing.
[QUOTE=science_man_88;411931]this depends on direction see:

[QUOTE=chalsall;411932]In my opinion, only an idiot would link to a YT video to enter an argument.[/QUOTE]

Why?

There's nothing different between the content of the YouTube video and the content of a couple of topical Wikipedia articles, and sm_88 is correct.

In the majority of cases, linking a YouTube video in an argument might be considered dubious, but in this case it certainly wasn't, and... well, your resulting ad hominem certainly doesn't improve your own image.

Edit: Unless you consider the point specifically about sign to be irrelevant?

 chalsall 2015-10-06 21:47

[QUOTE=Dubslow;411955]There's nothing different between the content of the YouTube video and the content of a couple of topical Wikipedia articles, and sm_88 is correct.[/QUOTE]

I fundamentally disagree.

Videos can't be "parced" by many, including often myself.

This isn't always because of bandwidth or hardware limitations, but sometimes because of the unwillingness to have Javascript running on their machines where you don't really know what the code is doing to present the video.

Further, if one can't write what they mean to say, why should I listen to them? At least link to text (HTML et al) which I can read. I tend to read far faster than most can speak.

 science_man_88 2015-10-06 22:09

[QUOTE=chalsall;412102]Further, if one can't write what they mean to say, why should I listen to them? At least link to text (HTML et al) which I can read. I tend to read far faster than most can speak.[/QUOTE]

it depends on the circumstances is what I meant because one thing pointed out is you can define division by 0 on the real number line if you say which direction you are going because the limits going both ways are only the same under absolute value as they are [TEX]\pm \infty[/TEX]

 chalsall 2015-10-06 23:38

[QUOTE=science_man_88;412105]it depends on the circumstances is what I meant...[/QUOTE]

If it cost you something to do something, and you hadn't succeeded yet, would the cost to successes radio not be infinite at that point in time per attempt?

 James Heinrich 2015-10-07 01:29

[QUOTE=chalsall;411913]Any progress on the P-1 / LL / DC deltas?[/QUOTE]First version of LL data is up now:
[url]http://www.mersenne.ca/status/ll[/url]

Resolution is limited to 1M because I'm (currently) grabbing data from [url]http://www.mersenne.org/primenet/[/url] which is trivial to parse for the data I need.

Data in this report goes back to 01-Feb-2012.

 James Heinrich 2015-10-07 19:07

Now with graphs.

 chalsall 2015-10-07 20:19

[QUOTE=James Heinrich;412169]Now with graphs.[/QUOTE]

Unbelievably cool!!!

Thanks James. :smile:

P.S. Sorry for this, but how's the P-1 query going? :wink:

 James Heinrich 2015-10-07 20:51

[QUOTE=chalsall;412174]how's the P-1 query going? :wink:[/QUOTE]:max:

More seriously, perhaps you can break down more specifically what you'd like to see on the P-1 data. Mersenne.info doesn't have any P-1 data that I saw so I can't copy your good ideas from there.
If you just want to know how many exponents in each range have had P-1 done I should be able to extract that data reasonably easily. If you want something more than that, please explain to me [i]before[/i] I start working on it :smile:

 chalsall 2015-10-07 21:14

[QUOTE=James Heinrich;412177]More seriously, perhaps you can break down more specifically what you'd like to see on the P-1 data. Mersenne.info doesn't have any P-1 data that I saw so I can't copy your good ideas from there.
If you just want to know how many exponents in each range have had P-1 done I should be able to extract that data reasonably easily. If you want something more than that, please explain to me [i]before[/i] I start working on it :smile:[/QUOTE]

Mersenne.info never tracked P-1 progress nor deltas. In all honesty, when I started mersenne.info P-1 was almost non existent.

The issue we are now facing is we are (as a community) seriously overpowered with P-1, so tracking candidates without P-1 compared to those with P-1 done is important for our trending and resource management.

To answer your question... Very similar to your TF'ing reports; how many candidates were TFed, how many candidates were P-1'ed (at what bit level), how many candidates were DC'ed (at what bit level), how many candidates were LL'ed (at what bit level).

This will help us all balance our collective resources to obtain optimization.

I hope that makes sense; please let me know if it doesn't.

 James Heinrich 2015-10-07 21:22

Your example seems to have many dimensions. Perhaps you could help me visualize that better with a sample table of data you'd like to see? Do you want to see all that data in one table? Or is that 4 separate reports?
I'm not sure how easy it is parse my data for what bitlevel the exponent was at when it was P-1'd (or LL'd), I'll have to investigate.

 Mark Rose 2015-10-07 22:25

As I understand it, the problem at hand is knowing how many unassigned exponents have no P-1, and what TF bit-level those exponents are at, and how fast those assignments are being consumed.

So it could be as simple as adding an option at the top to show only unassigned assignments without P-1.

The table and graph could show changes exactly like they currently do for TF, with and without deltas.

 gjmccrac 2015-10-07 22:41

Another way of thinking of it is a flag you can toggle to view the TF/LL reports for exponents that have been P-1'd, not P-1'd or the whole set.

[QUOTE=Mark Rose;412192]As I understand it, the problem at hand is knowing how many unassigned exponents have no P-1, and what TF bit-level those exponents are at...[/QUOTE]

Do you guys need a list of suboptimally TF'd stuff that hasn't had any P-1 work (and is available)? Is that the main thing needed? I can't really help much with the rate at which things are being processed, but I can generate lists like this (just the top 50 here):
[CODE]exponent TFBits
80010001 73
80012903 71
80012917 71
80012923 71
80012957 71
80013163 72
80013253 72
80013257 72
80013347 71
80013397 72
80013749 71
80013827 71
80013877 71
80013979 71
80014027 71
80014537 71
80014601 71
80014663 71
80014673 71
80014703 72
80016107 71
80016121 71
80016199 71
80016281 71
80016359 71
80016613 71
80016623 71
80016647 71
80016667 71
80017087 71
80017097 71
80017099 71
80017187 71
80017243 71
80017627 71
80017649 71
80017739 71
80017837 71
80017871 71
80018023 72
80018041 72
80018327 72
80018333 72
80018377 72
80018381 72
80018767 71
80018857 72
80018933 72
80018963 72
80019227 72[/CODE]

PS - That's using this little bit of clause for TF bit levels... I didn't know what the 80M+ stuff would be, ideally, thus the 75 bit for everything 80M+

[CODE] ((exponent between 0 and 40e6 and no_factor_to_bits<71)
OR (exponent between 40e6 and 50e6 and no_factor_to_bits<72)
OR (exponent between 50e6 and 65e6 and no_factor_to_bits<73)
OR (exponent between 65e6 and 80e6 and no_factor_to_bits<74)
OR (exponent between 80e6 and 999e6 and no_factor_to_bits<75))[/CODE]

 James Heinrich 2015-10-08 04:40

I don't have easy access to which exponents are assigned or not, so if it's important to you guys to have the P-1 report based on that then I'll need to enlist [i]Madpoo[/i]'s help to generate some data for me.
It shouldn't be filtered by "suboptimal" TF, I'd want data on TF levels of all exponents, and not a list of exponents but just a count per range. Something like[code]SELECT COUNT(*) AS howmany, (FLOOR(exponent / 10000) * 10000) AS 10k_range
FROM table
GROUP BY 10k_range, tf_bits, pm1_is_done, pm1_is_available, pm1_is_assigned[/code]The last 3 boolean columns should always have one and only one of the 3 set to true -- it's either done, or it's currently assigned, or it's not-done and not-assigned therefore available. Could just as well be represented by a single field (e.g. 0=done, 1=assigned, 2=available)
@Aaron: how heavy a query is something like that?

 Mark Rose 2015-10-08 04:52

[QUOTE=Madpoo;412209]Do you guys need a list of suboptimally TF'd stuff that hasn't had any P-1 work (and is available)? Is that the main thing needed? I can't really help much with the rate at which things are being processed, but I can generate lists like this (just the top 50 here):[/QUOTE]

No, it's more estimating how much TF can be done before PrimeNet runs out of higher-TFed exponents to hand out for P-1. For instance, recently GPU72 was holding on to a bunch of exponents at 73 bits, but PrimeNet started handing out exponents at 71 bits. It would have been better to release back some of those 73 bit TFed exponents. Ideally we want to TF them to 75 bits before releasing the exponent back to PrimeNet for P-1, but we don't have the throughput for that. The problem is knowing how big of a buffer of less than 75 but better than 71 TF level exponents to leave at PrimeNet. If we know we need 300 exponents per day, Chris can program GPU72 to make sure an appropriate amount are left with PrimeNet at say 74 bits or even 73 bits, based on the available TF capacity. So having the deltas is the most important thing.

 chalsall 2015-10-08 14:38

[QUOTE=Mark Rose;412213]For instance, recently GPU72 was holding on to a bunch of exponents at 73 bits, but PrimeNet started handing out exponents at 71 bits. It would have been better to release back some of those 73 bit TFed exponents.[/QUOTE]

Actually, what happened there is GPU72 /was/ releasing some at 73 bits when it "observed" that candidates were about to be assigned for P-1'ing at below that (up in the 80M range). But because it only "looks" at the situation every five minutes and Mr. P-1, and then later aurashift, were reserving candidates in such large batches, it wasn't keeping up with the releasing.

[QUOTE=Mark Rose;412213]If we know we need 300 exponents per day, Chris can program GPU72 to make sure an appropriate amount are left with PrimeNet at say 74 bits or even 73 bits, based on the available TF capacity. So having the deltas is the most important thing.[/QUOTE]

It's actually a bit more complicated than that, because of the non-linear nature of the assignment requests. Normally it's pretty steady, but sometimes large batches are requested. The ideal is to pull far enough ahead that we've got a comfortable buffer of candidates at 75 bits for both the P-1'ers, and separately, the LL'ers.

Having the deltas allow us to make an informed decision as to how many we can take to 75, and how many we should only take to 74. My goal is to not have assigned for LL'ing anything not at at least 75 bits above 65M. GPU72 will "recapture" anything released for P-1 at below 75 bits for final TF'ing, assuming a factor isn't found by the P-1 run of course.

[QUOTE=chalsall;412237]Having the deltas allow us to make an informed decision as to how many we can take to 75, and how many we should only take to 74. My goal is to not have assigned for LL'ing anything not at at least 75 bits above 65M. GPU72 will "recapture" anything released for P-1 at below 75 bits for final TF'ing, assuming a factor isn't found by the P-1 run of course.[/QUOTE]

I know this is going to show my ignorance of the setup, but when GPU72 reserves exponents to take up to 75 (in that range, for example), does it actually hold on to them until it reaches 75 then? So if someone gets that work and only takes it up to 74 bits but doesn't want to go to 75, that exponent will sit in the GPU72 reservations unless action is taken to unreserve it to feed the P-1 beasts?

I don't know if this would be a better or worse situation, but maybe GPU72 could reserve exponents to be done a bit level at a time instead of (optimistically?) taking a bunch and going all the way from like 72 to 75 in one go. Would it be better for this situation to take a wide range and first get them up to 73, then 74, then 75, sending them back to Primenet between each bit level being complete?

In a perfect world it wouldn't matter if a bunch of workers took a batch and went through them one exponent at a time, taking each one all the way to 75, then returning it. But given the circumstances, it seems like maybe the better option is to take at least the exponents in the leading P-1 edge and only do a single bit more at a time rather than going for the full 75 at once.

Again, it may be my total lack of comprehension on what's happening under the hood, so please disregard if I'm way off base. It wouldn't be the first time.

 chalsall 2015-10-08 19:27

[QUOTE=Madpoo;412249]Again, it may be my total lack of comprehension on what's happening under the hood, so please disregard if I'm way off base. It wouldn't be the first time.[/QUOTE]

:smile:

Primenet currently hands out candidates for LL'ing based on the P-1'ed status first. P-1 done is assigned before no P-1 done, then the lowest candidates without P-1 done, grouped by the status %1M.

This is further complicated by the four different Categories for DC and LL.

It can be argued that GPU72 is actually "starving" Primenet such that only those candidates appropriately TF'ed / P-M'ed are available.

 storm5510 2019-04-16 23:36

[QUOTE=chalsall;412252]...It can be argued that GPU72 is actually "starving" Primenet such that only those candidates appropriately TF'ed / P-M'ed are available.[/QUOTE]

This could be the reason why my assignments are one bit higher on each side than what I have in the config file. I [U]do not[/U] have a problem with this at all. Just a curiosity .:smile:

 chalsall 2019-04-17 14:39

[QUOTE=storm5510;513893]Just a curiosity .:smile:[/QUOTE]

I'm not clear what you're asking / stating?

 petrw1 2021-09-23 00:25

Can we get a Data_As_Of timestamp on this report.

[url]https://www.gpu72.com/account/twok/[/url]

Thanks

 LaurV 2021-09-24 04:51

+1. :iws:
(albeit the update time is on the upper left corner of that page :razz:)

 axn 2021-09-24 06:31

[QUOTE=LaurV;588532](albeit the update time is on the upper left corner of that page :razz:)[/QUOTE]
which is useless as not all the results reported to mersenne.org by that time is reflected in the update.

 chalsall 2021-09-24 16:50

[QUOTE=axn;588536]which is useless as not all the results reported to mersenne.org by that time is reflected in the update.[/QUOTE]

I don't really understand the feature request/bug report.

Because this report is so computationally expensive, it is only generated once an hour (at the top of the hour, as indicated in the top-left hand corner of every page). Yes, Wayne... You used to have access to the "real-time" report, but then you leaked the "secret" URL so I made it cached as well.

With regards to the report being out of sync with Primenet, that will (almost) *always* be the case. GPU72 only "spiders" Primenet twice an hour to get updates wrt factors found and P-1'ing/TF'ing completed.

 petrw1 2021-09-24 16:52

[QUOTE=LaurV;588532]+1. :iws:
(albeit the update time is on the upper left corner of that page :razz:)[/QUOTE]

Am I blind/short sighted/tunnel vision? :blush:

In any case .... cool!

 petrw1 2021-09-24 16:54

[QUOTE=chalsall;588607]Yes, Wayne... You used to have access to the "real-time" report, but then you leaked the "secret" URL so I made it cached as well.[/QUOTE]

But that explains why it refreshes so fast now.

 chalsall 2021-09-24 16:55

Been there for as long as GPU72 has existed. (Not meaning to be mean or anything, but I believe it was actually you who asked it be rendered as UTC rather than AST many years ago...)

 petrw1 2021-09-24 17:30

[QUOTE=chalsall;588610]Been there for as long as GPU72 has existed. (Not meaning to be mean or anything, but I believe it was actually you who asked it be rendered as UTC rather than AST many years ago...)[/QUOTE]

Not the first time; won't be the last time I've had a senior moment.

 axn 2021-09-25 01:48

[QUOTE=chalsall;588607]I don't really understand the feature request/bug report.
<snip>
With regards to the report being out of sync with Primenet, that will (almost) *always* be the case. GPU72 only "spiders" Primenet twice an hour to get updates wrt factors found and P-1'ing/TF'ing completed.[/QUOTE]

When the report says "Updated: xxxx UTC", I expect all the results reported to primenet by xxxx UTC to be there. If that is not the case can you please give a "Data as of" time as well which represents the last time primenet was spidered?

I don't really care what time report was generated. I only care how recent the data actually is.

 chalsall 2021-09-25 23:52

[QUOTE=axn;588643]I don't really care what time report was generated. I only care how recent the data actually is.[/QUOTE]

I hear your reasoning for the change request. Having thought about it, I'm going to remain with the current code paths.

There are many clients (read: workers) who interact with GPU72 by way of direct IPC. The Colab TF'ers and P-1'ers for example.

The timestamp on the report at [URL="https://www.gpu72.com/twok/"]https://www.gpu72.com/twok/[/URL] should perhaps be interpreted as the knowledge GPU72 had at that moment in time. One hour resolution is the same as provided by Primenet for many of its reports.

I hope and trust this is acceptable, taking into consideration the computational and bandwidth constraints of the IPC between these two systems.

All times are UTC. The time now is 15:17.