New GIMPS Visualisation Tool report types...
 

Hey Uncwilly et al...

Just to let you all know, I've created a new class of report types over on the GIMPS Visualisation Tool which you all will probably find interesting. Please see:

[URL="http://www.mersenne.info/trial_factored_tabular_delta_180/3/332000000/"]Trial Factored Depth Change for Range 332,000,000 to 333,000,000 between 2011.08.05 and 2011.02.06[/URL]

It shows you guys are doing some serious work in this range, and have eliminated 324 exponents in the last six months.

[QUOTE=chalsall;268446]It shows you guys are doing some serious work in this range, and have eliminated 324 exponents in the last six months.[/QUOTE]
In that range the GHz-days is near 250,000. We have lots more to eliminate, maybe around 2000 in the range that you mention. It would take us from 57% factored out to 61%.

Here is a status report for the range from 332192831 to 332399999:

[code]Date of data 8/20/2011
Average bit depth for first 100 expos 78.34
Average bit depth for first 1000 expos 77.00
100th active expo (no factor found) 332198357
1000th active expo (no factor found) 332246009
Unitless total effort number 163,982,336
Number of first 100 expos to 2^76 95
Number of first 1000 expos to 2^75 902
Number left in range 4241

Estimated expos in range to be removed 448
(by taking all expos to 2^82)[/code]
[code]Bit # at bit level
71 1753
72 739
73 235
74 156
75 204
76 437
77 262
78 31
79 398
80 23
81 3
P-1 128 ≈11,200,000 effort[/code]It seems that several of the heavy hitters have left for the standard TF ranges. Graph of current exponent statuses attached.

Here is the monthly status report for the range from 332192831 to 332399999:

[code]Date of data 9/20/2011
Average bit depth for first 100 expos 78.35
Average bit depth for first 1000 expos 77.03
100th active expo (no factor found) 332198357
1000th active expo (no factor found) 332246009
Unitless total effort number 165,528,576
Number of first 100 expos to 2^76 96
Number of first 1000 expos to 2^75 906
Number left in range 4237

Estimated expos in range to be removed 444
(by taking all expos to 2^82)[/code]
[code]Bit # at bit level
71 1579
72 835
73 278
74 179
75 181
76 464
77 266
78 30
79 398
80 24
81 3
P-1 137 ≈12,000,000 effort[/code]Nifty graphic attached.

Here is the monthly status report for the range from 332192831 to 332399999:

[code]Date of data 10/20/2011
Average bit depth for first 100 expos 78.36
Average bit depth for first 1000 expos 77.07
100th active expo (no factor found) 332198357
1000th active expo (no factor found) 332246023
Unitless total effort number 166,515,712
Number of first 100 expos to 2^76 97
Number of first 1000 expos to 2^75 913
Number left in range 4233

Estimated expos in range to be removed 440
(by taking all expos to 2^82)[/code]
[code]Bit # at bit level
71 1459
72 895
73 313
74 192
75 152
76 501
77 266
78 30
79 398
80 24
81 3
P-1 147 ≈13,000,000 effort[/code]

Here is the monthly status report for the range from 332192831 to 332399999:

[code]Date of data 11/20/2011
Average bit depth for first 100 expos 78.36
Average bit depth for first 1000 expos 77.09
100th active expo (no factor found) 332198357
1000th active expo (no factor found) 332246107
Unitless total effort number 167,133,184
Number of first 100 expos to 2^76 97
Number of first 1000 expos to 2^75 915
Number left in range 4230

Estimated expos in range to be removed 294
(by taking all expos to 2^79)[/code]
[code]Bit # at bit level
71 1372
72 938
73 324
74 208
75 159
76 505
77 269
78 30
79 398
80 24
81 3
P-1 151 ≈13,400,000 effort[/code]
I have been dedicating 3 CPU's to this effort as detailed below:
2 cores: lowest available exponents up to 74 bits (typically those already at 73), 2 hyperthreading CPU's with each virtual core (4 total) moving lowest available exponents to 73 bits (typically those that are now already at 72).
As of Nov 23, I have started changing 1 CPU to: 1 v-core moving the lowest available to 72 bits and 1 v-core moving the lowest available to 74.

Can mfaktc handle these exponents? If so, how long (estimate) it takes to go ONE bit-level higher (as from 77 to 78, these being the numbers in your report, if I understood it right). Can you send me an assignment file with few exponents to try? (1-5, and if it works ok, I may request more). Never tried my gpu for this high range, and I am curious..

[QUOTE=LaurV;279953]Can mfaktc handle these exponents?
[/QUOTE]

Yes, sure!


Why don't get your assignments here?

[URL]http://www.mersenne.org/manual_assignment/[/URL]

[SIZE=1][B]Number of CPUs (cores) you need assignments for (maximum 12): 1[/B][/SIZE]
[SIZE=1][B]Number of assignments you want for each core:[/B] [B]5[/B][/SIZE]
[B][SIZE=1]Preferred work type: Trial factoring
[/SIZE][SIZE=1]Optional exponent range: 332192831 to 332399999[/SIZE][/B]

I think I am a bit idiot, sorry! Of course it works! I am the one who took 400001233 to 77 and 400009999 to 75 bits in the first days when I started playing with mfaktc weeks ago, to check its possibilities! (the neighborhood of these expos was TF-ed to 65 bits only, and I thing there is not much progress since).

I have read the Uncwilli's post quite superficially and I (wrongly) assumed we are talking about OBD exponents here (3.32G, and not 332M). That is why I did not know how to get assignments. Primenet handles only expos under 1G. The question remains for that range of 3G++, and also the request for few expos in THAT range, if mfaktc can handle them. I only want to test my GPU, so not being really interested in 100M digits numbers for now. Sorry for making waves. Do you know who is handling that range, so I could ask for few assignments?

[QUOTE=LaurV;279964] I only Do you know who is handling that range, so I could ask for few assignments?[/QUOTE]

Try here: Operation Billion Digits

[url]http://mersenneforum.org/forumdisplay.php?f=50[/url]

@Laurv: :smile:
You just look over the OBD web page, and post what you will do in the reservation thread, after making sure that noone has also reserved it in the last week...it's that simple.

Question for the peanut gallery: Is the TF effort for the 100M digit range worth keeping in the GPU to 72 tool? I like getting the exponents that have the least TF on them for my GPU.