View Single Post
Old 2006-05-02, 00:44   #1
ATH's Avatar
Dec 2003

24×32×23 Posts
Default Picturing progress

For quite awhile I have had this idea it would be nice to get an overview of the entire gimps progress like an image of all the exponents where you could see which stage they are on.

I finally got around to writing a little c++ program that reads the 5 data files "status.txt", "cleared.txt", "", "" and "" and makes an image from the data. Here is the result so far:
mersenne.rar (4.5 Mb).

It is 12 images of the progress from July 2005 to April 28th 2006, the time on the images from 2006 are the UTC time from status.txt/cleared.txt. The 3 from 2005 is not so accurate since the data files are not from the same date. The july one uses from July 17th while 4 others are from July 1st, in september is from september 22nd while rest is from september 3rd, and in november status.txt is from october 31st, from november 2nd and rest from november 1st. If anyone have the set of 5 files from dates in 2005 or earlier I'm interested.

Each image is 1000x2500 pixels bmp in 16 colors to keep them "small" and each pixel represent a prime exponent starting in bottom left corner. So a pixel in, say, the 1330th row from the bottom and 726th column from the left would be (1330-1)*1000 + 726 = 1,329,726th prime which using The Nth Prime Page we see is 20,996,011 so it is M40. Each image show from M(2) up to the 2,500,000th prime M(41,161,739).
The pixel colors represents different stages:
white : mersenne prime
black : double checked
green : one LL-check
blue : factor(s) known
yellow : ongoing LL-check in status.txt
orange : ongoing doublecheck in status.txt
red : ongoing factoring in status.txt
purple : status unknown

The colors are not so good, I found it hard to find colors that would be easy to seperate from seen far away so you are welcome to comment if you find better combinations. The colors in each bmp file are located from the 55th to the 86th byte in the format BGR 0 BGR 0 BGR.... (so blue,green,red bytes, reverse of RGB seperated with zeros):
55th to 86th byte: 255,255,255,0, 0,0,0,0, 0,130,0,0, 130,0,0,0, 0,220,220,0, 0,100,200,0, 0,0,225,0, 250,0,250,0
white is the 255,255,255, black is the 0,0,0, green 0,130,0, blue 130,0,0, yellow 0,220,220, orange 0,100,200, red 0,0,225, purple 250,0,250.

Like I posted here: " many exponents appear in several of the 5 data files, and there are many more different types of double apperances than I wrote in the post. So when the program sort through the files I needed a rule to determine the color of those conflicting pixels. I made the following "hierarchy" of what I thought was most important down to the least important for each exponent, so once one status is confirmed it ignores the rest of the list below:
1) The 43 mersenne primes
2) Factor(s) known
3) Doublechecked
4) Ongoing doublecheck in status.txt
5) One LL test done
6) Ongoing LL test in status.txt
7) Ongoing factoring in status.txt
8) status unknown

The 1000x2500 images are really very big so I cut them down to 2 images of 1000x1000 images: 500k-1500k.rar (1.75 Mb) and 1500k-2500k.rar (2 Mb).
First image is from 500,001st prime M(7,368,791) to 1,500,000th prime M(23,879,519), and 2nd is from 1,500,001st prime M(23,879,539) to 2,500,000th prime M(41,161,739). These 2 smaller images skip the first 500,000 primes but they are all factored and doublechecked so nothing new is going on.

I made a series of daily images from April 15th 2006 to May 1st 2006: daily status.rar (2.9 Mb). They are 1000x1000 images from 1,400,001st prime M(22,182,379) to 2,400,000th prime M(39,410,867) and all colors are black except yellow (ongoing LL test) and red (ongoing factoring) so they reflect the daily changes in status.txt. The orange (ongoing doublecheck) are not in them since its so far below the current LL/factoring ranges and I wanted them at 1000x1000.

Finally a single image 1000x4631 of the entire 79.3M range used on the status page on GIMPS homepage: 79301461.rar (675 Kb).
It ends with 4,631,000th prime M(79,301,461) and the data is from April 28th 2006 22:00 UTC.

Last fiddled with by ATH on 2006-05-02 at 00:50
ATH is online now   Reply With Quote