[QUOTE=Prime95;258773]select @base_fftlen = max(fftlen), @base_timing = max(timing) from t_gimps_credit_timings where fftlen < @fftlen
select @next_fftlen = min(fftlen), @next_timing = min(timing) from t_gimps_credit_timings where fftlen > @fftlen[/QUOTE]Thanks... but that doesn't help me much if I don't have access to the `t_gimps_credit_timings` table. If you could give me a [COLOR="Blue"]SELECT * FROM `t_gimps_credit_timings`[/COLOR] (or an export from phpMyAdmin or mysqldump or whatever is easy) then I can reconstruct the data.

32 8.7840000000000003E-7 0 743
48 1.1399999999999999E-6 743 1099
64 1.4087999999999998E-6 1099 1469
80 0.0000017592 1469 1827
96 0.000002052 1827 2179
112 2.4623999999999999E-6 2179 2539
128 2.5271999999999997E-6 2539 2905
160 3.5639999999999997E-6 2905 3613
192 4.2935999999999999E-6 3613 4311
224 5.2511999999999998E-6 4311 5029
256 5.4743999999999996E-6 5029 5755
320 7.3439999999999995E-6 5755 7149
384 8.7839999999999992E-6 7149 8527
448 0.000010536 8527 9933
512 1.1327999999999999E-5 9933 11359
640 1.7039999999999999E-5 11359 14119
768 2.0591999999999999E-5 14119 16839
896 2.5415999999999999E-5 16839 19639
1024 2.7647999999999997E-5 19639 22477
1280 3.9119999999999998E-5 22477 27899
1536 4.7592000000000001E-5 27899 33289
1792 5.6879999999999998E-5 33289 38799
2048 6.1680000000000006E-5 38799 44339
2560 8.5199999999999997E-5 44339 55099
3072 1.0247999999999999E-4 55099 65729
3584 0.00012648 65729 76559
4096 1.2959999999999998E-4 76559 87549
5120 0.000192 87549 108800
6144 2.2559999999999998E-4 108800 129900
7168 2.7599999999999999E-4 129900 151300
8192 0.0002856 151300 172700
10240 3.9839999999999998E-4 172700 214400
12288 4.8479999999999997E-4 214400 255300
14336 5.9040000000000004E-4 255300 297300
16384 6.1439999999999997E-4 297300 340400
20480 8.3760000000000008E-4 340400 423300
24576 0.0010176 423300 504600
28672 1.2599999999999998E-3 504600 587500
32768 1.3127999999999998E-3 587500 671400
40960 1.7256000000000001E-3 671400 835200
49152 2.0951999999999998E-3 835200 995500
57344 2.5607999999999998E-3 995500 1158000
65536 2.7384000000000002E-3 1158000 1325000
81920 3.8592000000000001E-3 1325000 1648000
98304 4.6871999999999999E-3 1648000 1966000
114688 5.6808000000000006E-3 1966000 2287000
131072 5.9928000000000004E-3 2287000 2614000
163840 7.3535999999999992E-3 2614000 3251000
196608 9.0743999999999998E-3 3251000 3875000
229376 1.0775999999999999E-2 3875000 4512000
262144 1.1975999999999999E-2 4512000 5158000
327680 1.5311999999999999E-2 5158000 6421000
393216 1.8839999999999999E-2 6421000 7651000
458752 0.02256 7651000 8908000
524288 0.0252 8908000 10180000
655360 3.3264000000000002E-2 10180000 12650000
786432 4.1375999999999996E-2 12650000 15070000
917504 4.9200000000000001E-2 15070000 17550000
1048576 5.5920000000000004E-2 17550000 20050000
1310720 7.0080000000000003E-2 20050000 24930000
1572864 8.5919999999999996E-2 24930000 29690000
1835008 0.10224 29690000 34560000
2097152 0.11375999999999999 34560000 39500000
2621440 0.14999999999999999 39500000 49100000
3145728 0.18312 49100000 58520000
3670016 0.21839999999999998 58520000 68130000
4194304 0.24360000000000001 68130000 77910000
5242880 0.31295999999999996 77910000 96830000
6291456 0.37919999999999998 96830000 115300000
7340032 0.45839999999999997 115300000 134200000
8388608 0.504 134200000 153400000
10485760 0.67200000000000004 153400000 190700000
12582912 0.81120000000000003 190700000 227300000
14680064 0.98399999999999987 227300000 264600000
16777216 1.0800000000000001 264600000 302600000
20971520 1.4903999999999999 302600000 376100000
25165824 1.8215999999999999 376100000 448000000
29360128 2.1791999999999998 448000000 521500000
33554432 2.3832 521500000 596000000

Thanks. Unfortunately, that data is identical to what I have already, so that's not the problem. The credit formula I have for P-1 is the one [url=http://www.mersenneforum.org/showpost.php?p=160774&postcount=14]you posted 27-Jan-2009[/url]:[code]if ( $B1 >= $B2 )
return ( $timing * ( 1.45 * $B1 ) / 86400.0 );
else
return ( $timing * ( 1.45 * $B1 + 0.079 * ($B2 - $B1) ) / 86400.0 );[/code]Has it been modified since then?

[QUOTE=James Heinrich;258828]Has it been modified since then?[/QUOTE]

Just the interpolation for FFT lengths not in the timings table. There is no way to predict which FFT size prime95 will use when you are near an FFT boundary. The xjmptable table in mult.asm lists all the available FFT sizes and approximate maximum exponent that can be handled.

I believe mprime used an 18M FFT on the test in question.

Well, that does reveal some discrepancies. The exponent ranges (the last two columns from your data in post #519 above) are what I'm using in my lookup table, but they don't match at all what's in mult.asm (if I'm reading it right). Using for example M332205149, the `t_gimps_credit_timings` data seems to indicate it should use a 20480K FFT. But in mult.asm it seems to say it should be a 17920K FFT. Did I read that correctly? If so, that would explain the ~15% credit difference issue I raised in post #514.

Yes, the server's timing table is based on FFT lengths available in prime95 version 25. Version 26 introduced many more FFT lengths and faster timings.

[QUOTE=James Heinrich;258747][i]KingKurly[/i] [url=http://www.mersenneforum.org/showpost.php?p=258746]noted in this post[/url] that P-1 on [url=http://mersenne.org/report_exponent/?exp_lo=332205149]M332205149[/url] (B1=3365000, B2=95902500) he got 181.3220 GHz-days of credit; but this doesn't match at all with [url=http://mersenne-aries.sili.net/prob.php?exponent=332205149&b1=3365000&b2=95902500&factorbits=76]my calculation[/url] of 210.272541 GHz-days.[/QUOTE]Good news -- after interpolating my own timings table for all the missing new FFT sizes, as well as figuring out which FFT range the exponents now fall under, I'm getting "[url=http://mersenne-aries.sili.net/prob.php?exponent=332205149&b1=3365000&b2=95902500&factorbits=76]181.321973 GHz-days[/url]" as the estimate credit, which looks pretty close. :smile:

For anyone interested, this is my version of the timing table and FFT sizes, derived from George's data above, plus the contents of gwnum/mult.asm:[code]static $fft_timing = array(
33554432 => 2.3832E+0,
33030144 => 2.3577E+0,
32768000 => 2.3449E+0,
31457280 => 2.2812E+0,
29360128 => 2.1792E+0,
28311552 => 2.0898E+0,
27525120 => 2.0227E+0,
26214400 => 1.9110E+0,
25165824 => 1.8216E+0,
23592960 => 1.6974E+0,
22937600 => 1.6456E+0,
22020096 => 1.5732E+0,
20971520 => 1.4904E+0,
19660800 => 1.3621E+0,
18874368 => 1.2852E+0,
18350080 => 1.2339E+0,
16777216 => 1.0800E+0,
16515072 => 1.0680E+0,
16384000 => 1.0620E+0,
15728640 => 1.0320E+0,
14680064 => 9.8400E-1,
14155776 => 9.4080E-1,
13762560 => 9.0840E-1,
13107200 => 8.5440E-1,
12582912 => 8.1120E-1,
11796480 => 7.5900E-1,
11468800 => 7.3725E-1,
11010048 => 7.0680E-1,
10485760 => 6.7200E-1,
9830400 => 6.1950E-1,
9437184 => 5.8800E-1,
9175040 => 5.6700E-1,
8388608 => 5.0400E-1,
8257536 => 4.9830E-1,
8192000 => 4.9545E-1,
7864320 => 4.8120E-1,
7340032 => 4.5840E-1,
7077888 => 4.3860E-1,
6881280 => 4.2375E-1,
6553600 => 3.9900E-1,
6291456 => 3.7920E-1,
5898240 => 3.5436E-1,
5734400 => 3.4401E-1,
5505024 => 3.2952E-1,
5242880 => 3.1296E-1,
4915200 => 2.9128E-1,
4718592 => 2.7828E-1,
4587520 => 2.6961E-1,
4194304 => 2.4360E-1,
4128768 => 2.4045E-1,
4096000 => 2.3887E-1,
3932160 => 2.3100E-1,
3670016 => 2.1840E-1,
3538944 => 2.0958E-1,
3440640 => 2.0296E-1,
3276800 => 1.9194E-1,
3145728 => 1.8312E-1,
2949120 => 1.7070E-1,
2867200 => 1.6552E-1,
2752512 => 1.5828E-1,
2621440 => 1.5000E-1,
2457600 => 1.3867E-1,
2359296 => 1.3188E-1,
2293760 => 1.2735E-1,
2097152 => 1.1376E-1,
2064384 => 1.1232E-1,
2048000 => 1.1160E-1,
1966080 => 1.0800E-1,
1835008 => 1.0224E-1,
1769472 => 9.8160E-2,
1720320 => 9.5100E-2,
1638400 => 9.0000E-2,
1572864 => 8.5920E-2,
1474560 => 7.9980E-2,
1376256 => 7.4040E-2,
1310720 => 7.0080E-2,
1228800 => 6.5655E-2,
1179648 => 6.3000E-2,
1146880 => 6.1230E-2,
1048576 => 5.5920E-2,
1032192 => 5.5080E-2,
983040 => 5.2560E-2,
917504 => 4.9200E-2,
884736 => 4.7244E-2,
819200 => 4.3332E-2,
786432 => 4.1376E-2,
737280 => 3.8334E-2,
688128 => 3.5292E-2,
655360 => 3.3264E-2,
589824 => 2.9232E-2,
573440 => 2.8224E-2,
524288 => 2.5200E-2,
491520 => 2.3880E-2,
458752 => 2.2560E-2,
409600 => 1.9770E-2,
393216 => 1.8840E-2,
344064 => 1.6194E-2,
327680 => 1.5312E-2,
294912 => 1.3644E-2,
262144 => 1.1976E-2,
245760 => 1.1376E-2,
229376 => 1.0776E-2,
196608 => 9.0744E-3,
163840 => 7.3536E-3,
147456 => 6.6732E-3,
131072 => 5.9928E-3,
122880 => 5.8368E-3,
114688 => 5.6808E-3,
98304 => 4.6872E-3,
86016 => 4.0662E-3,
81920 => 3.8592E-3,
73728 => 3.2988E-3,
65536 => 2.7384E-3,
61440 => 2.6496E-3,
57344 => 2.5608E-3,
49152 => 2.0952E-3,
40960 => 1.7256E-3,
32768 => 1.3128E-3,
28672 => 1.2600E-3,
24576 => 1.0176E-3,
20480 => 8.3760E-4,
16384 => 6.1440E-4,
14336 => 5.9040E-4,
12288 => 4.8480E-4,
10240 => 3.9840E-4,
8192 => 2.8560E-4,
7168 => 2.7600E-4,
6144 => 2.2560E-4,
5120 => 1.9200E-4,
4096 => 1.2960E-4,
3584 => 1.2648E-4,
3072 => 1.0248E-4,
2560 => 8.5200E-5,
2048 => 6.1680E-5,
1792 => 5.6880E-5,
1536 => 4.7592E-5,
1280 => 3.9120E-5,
1024 => 2.7648E-5,
896 => 2.5416E-5,
768 => 2.0592E-5,
640 => 1.7040E-5,
512 => 1.1328E-5,
448 => 1.0536E-5,
384 => 8.7840E-6,
320 => 7.3440E-6,
256 => 5.4744E-6,
224 => 5.2512E-6,
192 => 4.2936E-6,
160 => 3.5640E-6,
128 => 2.5272E-6,
112 => 2.4624E-6,
96 => 2.0520E-6,
80 => 1.7592E-6,
64 => 1.4088E-6,
48 => 1.1400E-6,
32 => 8.7840E-7,
);

static $fft_size_lookup_table = array(
0 => 32,
743 => 32,
1099 => 48,
1469 => 64,
1827 => 80,
2179 => 96,
2539 => 112,
2905 => 128,
3613 => 160,
4311 => 192,
5029 => 224,
5755 => 256,
7149 => 320,
8527 => 384,
9933 => 448,
11359 => 512,
14119 => 640,
16839 => 768,
19639 => 896,
22477 => 1024,
27899 => 1280,
33289 => 1536,
38799 => 1792,
44339 => 2048,
55099 => 2560,
65729 => 3072,
76559 => 3584,
87549 => 4096,
108800 => 5120,
129900 => 6144,
151300 => 7168,
172700 => 8192,
214400 => 10240,
255300 => 12288,
297300 => 14336,
340400 => 16384,
423300 => 20480,
504600 => 24576,
587500 => 28672,
671400 => 32768,
835200 => 40960,
995500 => 49152,
1158000 => 57344,
1243000 => 61440,
1325000 => 65536,
1485000 => 73728,
1648000 => 81920,
1725000 => 86016,
1966000 => 98304,
2287000 => 114688,
2452000 => 122880,
2614000 => 131072,
2929000 => 147456,
3251000 => 163840,
3875000 => 196608,
4512000 => 229376,
4837000 => 245760,
5158000 => 262144,
5781000 => 294912,
6421000 => 327680,
6716000 => 344064,
7651000 => 393216,
7967000 => 409600,
8908000 => 458752,
9547000 => 491520,
10180000 => 524288,
11100000 => 573440,
11410000 => 589824,
12650000 => 655360,
13250000 => 688128,
14160000 => 737280,
15070000 => 786432,
15690000 => 819200,
16930000 => 884736,
17550000 => 917504,
18800000 => 983040,
19740000 => 1032192,
20050000 => 1048576,
21850000 => 1146880,
22490000 => 1179648,
23360000 => 1228800,
24930000 => 1310720,
26120000 => 1376256,
27900000 => 1474560,
29690000 => 1572864,
30900000 => 1638400,
32420000 => 1720320,
33370000 => 1769472,
34560000 => 1835008,
37030000 => 1966080,
38570000 => 2048000,
38880000 => 2064384,
39500000 => 2097152,
43060000 => 2293760,
44250000 => 2359296,
46030000 => 2457600,
49100000 => 2621440,
51450000 => 2752512,
53460000 => 2867200,
54950000 => 2949120,
58520000 => 3145728,
60940000 => 3276800,
63970000 => 3440640,
65790000 => 3538944,
68130000 => 3670016,
73060000 => 3932160,
76090000 => 4096000,
76680000 => 4128768,
77910000 => 4194304,
84920000 => 4587520,
87250000 => 4718592,
90760000 => 4915200,
96830000 => 5242880,
101200000 => 5505024,
105300000 => 5734400,
108200000 => 5898240,
115300000 => 6291456,
120000000 => 6553600,
126000000 => 6881280,
129500000 => 7077888,
134200000 => 7340032,
143800000 => 7864320,
149800000 => 8192000,
151000000 => 8257536,
153400000 => 8388608,
167200000 => 9175040,
172000000 => 9437184,
178300000 => 9830400,
190700000 => 10485760,
199500000 => 11010048,
207600000 => 11468800,
213400000 => 11796480,
227300000 => 12582912,
236700000 => 13107200,
248400000 => 13762560,
255500000 => 14155776,
264600000 => 14680064,
283900000 => 15728640,
295500000 => 16384000,
297800000 => 16515072,
302600000 => 16777216,
329800000 => 18350080,
339300000 => 18874368,
352500000 => 19660800,
376100000 => 20971520,
393800000 => 22020096,
409300000 => 22937600,
420700000 => 23592960,
448000000 => 25165824,
467600000 => 26214400,
488400000 => 27525120,
503500000 => 28311552,
521500000 => 29360128,
559300000 => 31457280,
582200000 => 32768000,
585300000 => 33030144,
596000000 => 33554432,
);[/code]

[QUOTE=James Heinrich;258851]Good news -- after interpolating my own timings table for all the missing new FFT sizes, as well as figuring out which FFT range the exponents now fall under, I'm getting "[url=http://mersenne-aries.sili.net/prob.php?exponent=332205149&b1=3365000&b2=95902500&factorbits=76]181.321973 GHz-days[/url]" as the estimate credit, which looks pretty close. :smile:[/QUOTE]Ok, so see if I have this right.

I can take the exponent and match it to the entry in the second table to find FFT size. Then take the FFT size and look-up the number in the first table.
Then take [U]that[/U] number and input it into the formula above as $timing. And that will give a GHz-days value.

[QUOTE=Uncwilly;258858]Ok, so see if I have this right.... And that will give a GHz-days value.[/QUOTE]Yes.
Same for L-L, except the formula is simpler ($timing * $exponent / 86400).

[QUOTE=James Heinrich;258878]Yes.[/QUOTE]:tu:So I will now have a way to figure P-1 credit for the 100M digit range. Will try to get this implemented in my spreadsheet before vacation. :juggle: