Primenet web design

[snme2pm1]

Originally Posted by Madpoo

It's just a nicer presentation and can give some sorting options other than the default "date".

I didn't want to muddy the waters by mentioning tables.
I guess we'll see what you choose to stitch together readily.
One recurring theme is to duck the noise from the outer reaches of TF exploration!
There would be several ways for that to happen of course.

[snme2pm1]

Originally Posted by Madpoo

It's just a nicer presentation and can give some sorting options other than the default "date".

Indeed it is, and various sorting can quickly shift unwanted noise out of sight.

The top of page update notice on revised Recent Results (unlike Recently Cleared) tends to be too long to render on my chrome browser.

Don't you have a day job or something?
Looking vastly better now.
I suppose one of the few remaining text tables is the Exponent Status Distribution, not that I'm complaining.

[Madpoo]

Originally Posted by snme2pm1

Indeed it is, and various sorting can quickly shift unwanted noise out of sight.

The top of page update notice on revised Recent Results (unlike Recently Cleared) tends to be too long to render on my chrome browser.

The recent results page shows the last 3000 entries logged in. That last column with the result text can be very long, especially with...interesting... entries like this:

no factor for M93928907 from 2^70 to 2^71 [mfaktc 0.20 barrett76_mul32]no factor for M93928943 from 2^70 to 2^71 [mfaktc 0.20 barrett76_mul32]

As in, why did that client feel the need to include *2* results in one line of text? Bug in mfaktc where it forgot to add a line break?

So yeah, the results can sometimes be really long. I should look at having text in that last column wrap, and I actually added a class to that text "wrapit" that just tells it to break a word if needed, since some LONG factors cause wrapping issues. But it's not wrapping the text normally right now, apparently.

Scroll horizontally in that table though to see everything for now. If I can make it wrap okay without looking funny, I'll do that. That's actually what I *meant* to do but kind of ran out of time last night.

Originally Posted by snme2pm1

Don't you have a day job or something?
Looking vastly better now.
I suppose one of the few remaining text tables is the Exponent Status Distribution, not that I'm complaining.

Indeed, but nights and weekends are a good time for tinkering. If I can do some website updates while watching TV, I can feel better about myself.

[TheMawn]

Originally Posted by Madpoo

Indeed, but nights and weekends are a good time for tinkering. If I can do some website updates while watching TV, I can feel better about myself.

I couldn't possibly do productive work while watching TV.

[James Heinrich]

Originally Posted by Madpoo

As in, why did that client feel the need to include *2* results in one line of text? Bug in mfaktc where it forgot to add a line break?

I'm curious how that result got in. The new manual-results parser is pretty strict about submitted data formatting, so it wouldn't (or at least shouldn't) have come in that way. Perhaps it came in via misfit which I believe is still using proprietary submission code which is less discerning when parsing results.

[Madpoo]

Originally Posted by TheMawn

I couldn't possibly do productive work while watching TV.

I'll confess that during most of last night's season premiere of "The Walking Dead", I closed the laptop lid and set GIMPS aside. I have my priorities after all.

[Madpoo]

Here's a not-entirely off-topic question:

I was exploring the whole benchmarking thing, starting from the client side all the way up.

But what I could NOT figure out was, how do I benchmark a machine and have it upload those results to my account *without* actually getting that machine started on some exponent testing?

Let me explain what I was trying to do... My own account on the server only has one machine doing any work, and I've already run benchmarks on it, no problem.

But as part of my research into making the benchmark page on the site look better, I wanted to add more of my own benchmarks so I had a better set of data to work on.

I have a lot of machines I work with so I unzipped the Prime95 software on there, but the only way I could make it upload the benchmark results was to add my account and computer name, and also check the "use primenet" box. Which of course then makes it request work to do, which I didn't want. This was just for benchmarking purposes.

I at least set it to only use 1 worker thread on my 40-core box (I love that system), and once the benchmark was done and uploaded, I just had to "quit gimps" from that system to unreserve the LL test it checked out.

I wanted to benchmark a few more like my laptop, etc. to add more data points, but it's kind of a hassle to go through that.

Maybe there isn't a good way, but if anyone knows, I'm all ears.

[Xyzzy]

Before you start the client, create a prime.txt file that contains "NoMoreWork=1".

[Main thread Oct 13 13:47] Starting worker.
[Work thread Oct 13 13:47] Worker starting
[Work thread Oct 13 13:47] Setting affinity to run worker on any logical CPU.
[Work thread Oct 13 13:47] No work to do at the present time.  Waiting.
[Comm thread Oct 13 13:47] Exchanging program options with server
[Comm thread Oct 13 13:47] Successfully quit GIMPS.
[Comm thread Oct 13 13:47] Done communicating with server.

[chalsall]

Hey Madpoo et al. Sorry for being "late to the party" on this; generally love the new "look and feel" of Primenet.

However, this morning I noticed a change which bothered me a bit. Please see the attached screenshot.

1. Note that the inner table is being rendered outside of the "wrapper" table.

1.1. Personally I don't like it when a site constrains the width of the rendering area. If my window is, for example, 1920 pixels wide (or, for that matter, 5760 pixels) the site should render into the space given (IMO).

2. Note that some lines in this report are rendered in approximately twice the vertial space as others. This appears to be because of the "Stage %" column being rendered without a "nowrap".

While I'm typing:

3. Appreciate the new color coding; would be good if an explination of said colors was given somewhere (I can infer what they approximately mean, but it would be good for a definitiatve definition).

4. I miss the UTC timestamp being rendered in the header of each page.

Very much appreciate the work being done.

[Madpoo]

Originally Posted by chalsall

Hey Madpoo et al. Sorry for being "late to the party" on this; generally love the new "look and feel" of Primenet.

However, this morning I noticed a change which bothered me a bit. Please see the attached screenshot.

1. Note that the inner table is being rendered outside of the "wrapper" table.

1.1. Personally I don't like it when a site constrains the width of the rendering area. If my window is, for example, 1920 pixels wide (or, for that matter, 5760 pixels) the site should render into the space given (IMO).

2. Note that some lines in this report are rendered in approximately twice the vertial space as others. This appears to be because of the "Stage %" column being rendered without a "nowrap".

While I'm typing:

3. Appreciate the new color coding; would be good if an explination of said colors was given somewhere (I can infer what they approximately mean, but it would be good for a definitiatve definition).

Very much appreciate the work being done.

Thanks darling... or is it "sir"?

The /assignments/ report is "new", in the sense that it's been around for a while, but unless you knew it was there, you wouldn't have known it was there. :)

In other words, it was not actively linked to from any other page on the site until just last night. I guess on the old server that report wasn't performing that well so it wasn't referenced anywhere. You're probably one of the folks that new that page existed beforehand.

As for the look of it right now, yeah, it's a work in progress, sorry for the mess. I had converted it to the new format but since it wasn't being linked to, I didn't really test it out that much. I did some updates on it last night (when I wasn't watching "walkers" being destroyed), but clearly I have some more work to do.

A couple things I *might* do is change the word "Exponents" to a shorter M_N, and instead of the "Work Types" header opening a new page with a table of what each work type means, do some kind of hover tip
instead. Just removing the (days) from 2 of those columns would tighten it up a bit as well, as long as people can figure out those #'s are in days? I could shrink the "(days)" text I guess...it's there but not as "there".

The coloring was a first stab I took at highlighting assignments that were overdue to varying degrees.
0-15 days past due = blue text
16-45 days past due = red text
46-90 days past due = black text on yellow
90+ days overdue = black text on red

A thought I had was to color code them to correlate in some way with the recycling rules, so as they approached being recycled, the colors would be more intense, but honestly the recycling rules scare me. :) There's a lot of "if then, but not, yet, plus, or else, random dice roll" stuff. Okay, it's not that bad, but I wasn't going to tackle the logic just yet.

In the end I may just opt for red text for anything past due and keep it simple. I just thought it would be cool to see at a glance what stuff out there was getting pretty stale and/or moldy.

What that page needs is text output like some others have (or maybe XML). It's another table suffering from potential information overload. It's good info but perhaps some is redundant or is best left out of the overview. Like the age and ETA could be figured out from the date assigned/due date, but seeing it in # of days is easier. Maybe omitting the raw dates from the table itself and have them show when hovering over the days/eta #'s would be good.

I'm not entirely sure the computer name should be there either... maybe another thing to show up when hovering over the account name perhaps. It could be useful in some ways, but not as an overall thing.

The report itself isn't meant to be exhaustive anyway since the underlying SQL query is limited to the first 1000 rows, but if you're looking at small, specific ranges it's handy.

That's why I set the default min/max exponents the way I did... you may notice it defaults to show the remaining assignments left to reach the milestone of proving "M(32582657) is the 44th Mersenne Prime". Just 29 DC's left. (the project needs more DC'ers, FYI)

I think once some of the column stuff gets cleaned up it will look better.

4. I miss the UTC timestamp being rendered in the header of each page.

You may notice that I've added that to the footer now. It's there if people want it, but not in-your-face for people who don't care. I even toyed with using some Javascript to have an updated time on the page but that mostly depends on visitors having their own clock and timezone set correctly, so I abandoned that.

[Madpoo]

Originally Posted by Xyzzy

Before you start the client, create a prime.txt file that contains "NoMoreWork=1".

[Main thread Oct 13 13:47] Starting worker.
[Work thread Oct 13 13:47] Worker starting
[Work thread Oct 13 13:47] Setting affinity to run worker on any logical CPU.
[Work thread Oct 13 13:47] No work to do at the present time.  Waiting.
[Comm thread Oct 13 13:47] Exchanging program options with server
[Comm thread Oct 13 13:47] Successfully quit GIMPS.
[Comm thread Oct 13 13:47] Done communicating with server.

Ah, perfect! I'll have to give that a try later.