Glucas/Mlucas errors...
 

My iBook can pass all 31 Glucas self tests, but it fails immediately (bus error) when I try to run a real exponent... Mlucas fails on every self test I try... The real exponent I'm trying with Glucas is 9167489...

Glucas... http://sourceforge.net/project/showfiles.php?group_id=24518
Mlucas... http://gravity.psu.edu/~khanna/hpc.html

Here is the output from the Glucas self test... http://www.teamprimerib.com/txt/selftest.res

Here is the output from Mlucas...

[code:1][Xyzzy:~/Desktop/Mlucas] mvang% ./Mlucas
looking for worktodo.ini file...
no worktodo.ini file found...switching to interactive mode.
Enter exponent, FFT length in K (set K = 0 for default FFT length) >9490001,480
Enter 'y' to run a self-test, <return> for a full LL test >y
Enter number of iterations for timing test>
100
Enter index of radix set to be used for the FFT:
(See file fft_radix.txt for a list of available choices; enter -1 to get the default) >-1
Enter 'y' to enable per-iteration error checking, <return> for no error checking >y
p is prime...proceeding with Lucas-Lehmer test...
M( 9490001 ): using FFT length 480K = 491520 8-byte floats.
this gives an average 19.3074564615885 bits per digit
INFO: Using real* 8 for FFT sincos and DWT weights tables inits.
WARN: radix set -1 not available - using defaults.
Using complex FFT radices 15 16 32 32
M 9490001 Roundoff warning on iteration 24 maxerr = 0.406250000000
M 9490001 Roundoff warning on iteration 25 maxerr = 0.500000000000
FATAL ERROR...Halting execution.
STOP
[/code:1]
[code:1][Xyzzy:~/Desktop/Mlucas] mvang% ./Mlucas
looking for worktodo.ini file...
no worktodo.ini file found...switching to interactive mode.
Enter exponent, FFT length in K (set K = 0 for default FFT length) >87700003,4608
Enter 'y' to run a self-test, <return> for a full LL test >y
Enter number of iterations for timing test>
100
Enter index of radix set to be used for the FFT:
(See file fft_radix.txt for a list of available choices; enter -1 to get the default) >-1
Enter 'y' to enable per-iteration error checking, <return> for no error checking >y
p is prime...proceeding with Lucas-Lehmer test...
M( 87700003 ): using FFT length 4608K = 4718592 8-byte floats.
this gives an average 18.5860534244114 bits per digit
INFO: Using real* 8 for FFT sincos and DWT weights tables inits.
WARN: radix set -1 not available - using defaults.
Using complex FFT radices 9 16 16 32 32
M87700003 Roundoff warning on iteration 27 maxerr = 0.500000000000
FATAL ERROR...Halting execution.
STOP
[/code:1]
The hardware is a 800MHz PowerPC 750FX with 640MB of memory and 512KB L2...

What am I doing wrong?

Re: Glucas/Mlucas errors...
 

[quote="Xyzzy"]My iBook can pass all 31 Glucas self tests, but it fails immediately (bus error) when I try to run a real exponent...[/quote]
That's strange.
Glucas selftests use the same codepath as the regular tests, but use a different set of ini/que-files.
How did you setup Glucas for testing M9167489?
Please post the contents of glucas.{ini,que}.

Re: Glucas/Mlucas errors...
 

[quote="Xyzzy"]
Here is the output from Mlucas...

[code:1][Xyzzy:~/Desktop/Mlucas] mvang% ./Mlucas
looking for worktodo.ini file...
no worktodo.ini file found...switching to interactive mode.
Enter exponent, FFT length in K (set K = 0 for default FFT length) >9490001,480
Enter 'y' to run a self-test, <return> for a full LL test >y
Enter number of iterations for timing test>
100
Enter index of radix set to be used for the FFT:
(See file fft_radix.txt for a list of available choices; enter -1 to get the default) >-1
Enter 'y' to enable per-iteration error checking, <return> for no error checking >y
p is prime...proceeding with Lucas-Lehmer test...
M( 9490001 ): using FFT length 480K = 491520 8-byte floats.
this gives an average 19.3074564615885 bits per digit
INFO: Using real* 8 for FFT sincos and DWT weights tables inits.
WARN: radix set -1 not available - using defaults.
Using complex FFT radices 15 16 32 32
M 9490001 Roundoff warning on iteration 24 maxerr = 0.406250000000
M 9490001 Roundoff warning on iteration 25 maxerr = 0.500000000000
FATAL ERROR...Halting execution.
STOP
[/code:1]

Apparently I was overly aggressive in setting the maximum exponent for the 480K FFT length - try using 512K instead. If you're planning a full test of an exponent in this range, you can override the program's default FFT length by starting the run in interactive mode as above and specifying the appropriate FFT length, then after the program gets to the first checkpoint, kill the run and restart from the worktodo.ini file - the run will continue with the manually set FFT length, but will write outputs to the pXXX.stat file, as it normally does for batch mode runs.

As far as your 4608K run is concerned, I'm not sure how I settled on 87.7M as the max exponent for that FFT length, but it's much too large - my current version of the code lists 83.76M as pmax for 4608K.

Cheers,
-Ernst

Re: Glucas/Mlucas errors...
 

[quote="kiste"]How did you setup Glucas for testing M9167489?
Please post the contents of glucas.{ini,que}.[/quote]
glucas.que
[code:1]9167489,1,0,0[/code:1]
glucas.ini
[code:1]alternative_output_flag=2
iteration_output=1000
last_error_flag=1
qa_interim_file=100000
roundoff_check=1
save_iterations=10000
time_flag=1
verbose_flag=1[/code:1]
I also tried it with no glucas.ini and it had the same error...

Re: Glucas/Mlucas errors...
 

[quote="ewmayer"]Apparently I was overly aggressive in setting the maximum exponent for the 480K FFT length - try using 512K instead.[/quote]
It looks like that fixed it...

[code:1][Xyzzy:~/Desktop/Mlucas] mvang% ./Mlucas
looking for worktodo.ini file...
no worktodo.ini file found...switching to interactive mode.
Enter exponent, FFT length in K (set K = 0 for default FFT length) >9490001,512
Enter 'y' to run a self-test, <return> for a full LL test >y
Enter number of iterations for timing test>
100
Enter index of radix set to be used for the FFT:
(See file fft_radix.txt for a list of available choices; enter -1 to get the default) >-1
Enter 'y' to enable per-iteration error checking, <return> for no error checking >y
p is prime...proceeding with Lucas-Lehmer test...
M( 9490001 ): using FFT length 512K = 524288 8-byte floats.
this gives an average 18.1007404327393 bits per digit
INFO: Using real* 8 for FFT sincos and DWT weights tables inits.
WARN: radix set -1 not available - using defaults.
Using complex FFT radices 8 32 32 32
100 iterations of M 9490001 with FFT length 524288
Res64: F9C0781482193C32. Program: E2.7b
Clocks = 00:01:44.646
STOP[/code:1]
[quote="ewmayer"]As far as your 4608K run is concerned, I'm not sure how I settled on 87.7M as the max exponent for that FFT length, but it's much too large - my current version of the code lists 83.76M as pmax for 4608K.[/quote]
Do you have a known self-test result I can use to test this range?

Re: Glucas/Mlucas errors...
 

[quote="Xyzzy"]
Do you have a known self-test result I can use to test this range?[/quote]

Try 84000013, which should yield a 100-iteration Res64 = C7323B8F92DCB618. Some notes I mad to myself abot this self-test seem to indicate that (based on the max. average RO error I saw for this test) 84M may actually be overconservative for 4608K, but in any event it shouldn't cause you any problems running it.

Re: Glucas/Mlucas errors...
 

[quote="Xyzzy"]
glucas.que
[code:1]9167489,1,0,0[/code:1]
[/quote]
Apparently the docs aren't clear enough. The fields in glucas.que are supposed to be seperated by whitespace.
Try if the following works:
[code:1]9167489 1 0 0[/code:1]
Looks like we need to add improved input validation here.

As of Glucas 2.9 it's better to import PrimeNet assignments, instead of editing glucas.que directly.

Example: DoubleCheck=9167489,60

Save the PrimeNet assignment in a file i.e. workotodo.ini
Start Glucas with the filename as argument:
./Glucas wortodo.ini

It looks like I can get away with a 480K FFT... If my calculations are correct, it looks like it will take ~70 days... This is with error checking on, and the default radix... I'll try Glucas next to compare the speed...
[code:1][Xyzzy:~/Desktop/mlucas] mvang% nice -19 ./mlucas
looking for worktodo.ini file...
no worktodo.ini file found...switching to interactive mode.
Enter exponent, FFT length in K (set K = 0 for default FFT length) >9167489, 480
Enter 'y' to run a self-test, <return> for a full LL test >
Enter index of radix set to be used for the FFT:
(See file fft_radix.txt for a list of available choices; enter -1 to get the default) >-1
Enter 'y' to enable per-iteration error checking, <return> for no error checking >y
p is prime...proceeding with Lucas-Lehmer test...
M( 9167489 ): using FFT length 480K = 491520 8-byte floats.
this gives an average 18.6513041178385 bits per digit
INFO: Using real* 8 for FFT sincos and DWT weights tables inits.
WARN: radix set -1 not available - using defaults.
Using complex FFT radices 15 16 32 32
M( 9167489 ) iteration = 2000 clocks = 00:39:09.309. Res64: 313B5721DFABE7CD
WARN: radix set -1 not available - using defaults.
Using complex FFT radices 15 16 32 32
M( 9167489 ) iteration = 4000 clocks = 00:21:57.501. Res64: D9783AAA0FF5EF2D
WARN: radix set -1 not available - using defaults.
Using complex FFT radices 15 16 32 32
^C[/code:1]
21 [ENTER] 60 × 57.501 + 9167489 [ENTER] 2000 ÷ × 86400 ÷

Re: Glucas/Mlucas errors...
 

[quote="ewmayer"]Try 84000013, which should yield a 100-iteration Res64 = C7323B8F92DCB618.[/quote]
I get a different Res64...

[code:1][Xyzzy:~/Desktop/mlucas] mvang% ./mlucas
looking for worktodo.ini file...
no worktodo.ini file found...switching to interactive mode.
Enter exponent, FFT length in K (set K = 0 for default FFT length) >84000013,4608
Enter 'y' to run a self-test, <return> for a full LL test >y
Enter number of iterations for timing test>
100
Enter index of radix set to be used for the FFT:
(See file fft_radix.txt for a list of available choices; enter -1 to get the default) >-1
Enter 'y' to enable per-iteration error checking, <return> for no error checking >y
p is prime...proceeding with Lucas-Lehmer test...
M( 84000013 ): using FFT length 4608K = 4718592 8-byte floats.
this gives an average 17.8019233279758 bits per digit
INFO: Using real* 8 for FFT sincos and DWT weights tables inits.
WARN: radix set -1 not available - using defaults.
Using complex FFT radices 9 16 16 32 32
100 iterations of M84000013 with FFT length 4718592
Res64: 6EA52961F9CAD418. Program: E2.7b
Clocks = 00:19:38.994
STOP[/code:1]

Re: Glucas/Mlucas errors...
 

[quote="kiste"]Apparently the docs aren't clear enough. The fields in glucas.que are supposed to be seperated by whitespace.
Try if the following works:
[code:1]9167489 1 0 0[/code:1]
Looks like we need to add improved input validation here.[/quote]
The corrected glucas.que works...

One other glitch I found is the contents of the glucas.ini file are case sensitive... I used all lowercase and none of the lines got recognized... I had to capitalize the first word in each line... :(

I'll post timings in a few minutes...

Looks like ~32 days...

[code:1][Xyzzy:~/Desktop/Glucas-2.9.0.ppc750.MacOSX-darwin_55] mvang% nice -19 ./glucas Appended ComputerID=408208707 in file glucas.ini.
[Fri Mar 7 15:02:58 2003]
Going to work with exponent 9167489
Starting from iteration 1. Exponent 9167489.
Iter. 1000 ( 0.01%), Err= 0.074, 280.94 user 92% CPU (0.304 sec/iter).
Iter. 2000 ( 0.02%), Err= 0.074, 281.95 user 93% CPU (0.302 sec/iter).
Iter. 3000 ( 0.03%), Err= 0.066, 281.46 user 94% CPU (0.301 sec/iter).
Iter. 4000 ( 0.04%), Err= 0.082, 281.49 user 93% CPU (0.302 sec/iter).
Iter. 5000 ( 0.05%), Err= 0.062, 282.02 user 94% CPU (0.301 sec/iter).
Iter. 6000 ( 0.07%), Err= 0.074, 281.92 user 94% CPU (0.301 sec/iter).
Iter. 7000 ( 0.08%), Err= 0.070, 281.73 user 94% CPU (0.300 sec/iter).[/code:1]
9167489 [ENTER] .300 × 86400 ÷

Re: Glucas/Mlucas errors...
 

[quote="Xyzzy"][quote="ewmayer"]Try 84000013, which should yield a 100-iteration Res64 = C7323B8F92DCB618.[/quote]
I get a different Res64...

[code:1][Xyzzy:~/Desktop/mlucas] mvang% ./mlucas
looking for worktodo.ini file...
no worktodo.ini file found...switching to interactive mode.
Enter exponent, FFT length in K (set K = 0 for default FFT length) >84000013,4608
Enter 'y' to run a self-test, <return> for a full LL test >y
Enter number of iterations for timing test>
100
Enter index of radix set to be used for the FFT:
(See file fft_radix.txt for a list of available choices; enter -1 to get the default) >-1
Enter 'y' to enable per-iteration error checking, <return> for no error checking >y
p is prime...proceeding with Lucas-Lehmer test...
M( 84000013 ): using FFT length 4608K = 4718592 8-byte floats.
this gives an average 17.8019233279758 bits per digit
INFO: Using real* 8 for FFT sincos and DWT weights tables inits.
WARN: radix set -1 not available - using defaults.
Using complex FFT radices 9 16 16 32 32
100 iterations of M84000013 with FFT length 4718592
Res64: 6EA52961F9CAD418. Program: E2.7b
Clocks = 00:19:38.994
STOP[/code:1][/quote]

Ah, I didn't look closely enough at your output previously. You're using v2.7b for these - that's the (relatively ancient) Fortran-90 version - no wonder it's less accurate and a heck of a lot slower than expected. I thought you had built the C version of 2.7c at some point?

Using the C version of 2.7c, I have no trouble running the self-test for your 9.49M exponent at 480K FFT length:

[code:1]
M9490001: using FFT length 480K = 491520 8-byte floats.
this gives an average 19.307456461588540 bits per digit
INFO: Using real*16 for FFT sincos and DWT weights tables inits.
WARN: radix set -1 not available - using defaults.
Using complex FFT radices 15 16 32 32
100 iterations of M9490001 with FFT length 491520
Res64: 01A4E738255C522B. AvgMaxErr = 0.215053013. Program: E2.7c
Clocks = 00:00:05.533
[/code:1]

Also, to compare timing-test results between the f90 and C code, you need to run the f90 code for one more iteration than the C, or the C code for one less iteration than the f90. When I run the C/v2.7c code for 99 iterations, I get the same as your "100-iteration" result for v2.7b:

[code:1]
M84000013: using FFT length 4608K = 4718592 8-byte floats.
this gives an average 17.801923327975803 bits per digit
INFO: Using real*16 for FFT sincos and DWT weights tables inits.
WARN: radix set -1 not available - using defaults.
Using complex FFT radices 9 16 16 32 32
99 iterations of M84000013 with FFT length 4718592
Res64: 6EA52961F9CAD418. AvgMaxErr = 0.107991536. Program: E2.7c
Clocks = 00:01:09.213
[/code:1]

Using the more-accurate C version, I can also run all the way up to 87.7M at 4608K (though the RO warnings and the AvgMaxErr > 0.3 indicate this is too large an exponent for a full-length run at this FFT length - p ~= 87.0M is probably the maximum one could safely use):
[code:1]
M87700003: using FFT length 4608K = 4718592 8-byte floats.
this gives an average 18.586053424411350 bits per digit
INFO: Using real*16 for FFT sincos and DWT weights tables inits.
WARN: radix set -1 not available - using defaults.
Using complex FFT radices 9 16 16 32 32
M 87700003 Roundoff warning on iteration 77 maxerr = 0.437500000000
M 87700003 Roundoff warning on iteration 91 maxerr = 0.406250000000
100 iterations of M87700003 with FFT length 4718592
Res64: 3C41DA97B5B87409. AvgMaxErr = 0.323716518. Program: E2.7c
Clocks = 00:01:09.997
[/code:1]

Re: Glucas/Mlucas errors...
 

[quote="ewmayer"]Ah, I didn't look closely enough at your output previously. You're using v2.7b for these - that's the (relatively ancient) Fortran-90 version - no wonder it's less accurate and a heck of a lot slower than expected. I thought you had built the C version of 2.7c at some point?[/quote]
That was on my Sun... This is my new [url=http://www.teamprimerib.com/jpg/ibook.jpg]iBook[/url]... I got the binary from that link I posted earlier... (I'd love to build a new binary, but I wouldn't know where to start!)

If you want a SSH account on my box to work on a Mac client let me know... :)

Hello. Oups. What happened with Mlucas? :sad:

I got this on IBM Power8 and Red Hat 7.2.

Mlucas installed without problem ... I hope. Because now not sure.

[CODE][Mar 02 21:59:06] M43892801 Iter# = 200000 clocks = 00:16:24.308 [ 0.0984 sec/iter] Res64: A82FA0F649C8F274. AvgMaxErr = 0.170602632. MaxErr = 0.250000000
[Mar 02 22:20:42] M43892801 Iter# = 210000 clocks = 00:21:35.630 [ 0.1296 sec/iter] Res64: BAFC8160194E1EFA. [COLOR="Red"]AvgMaxErr = 0.000000000. MaxErr = 0.000000000[/COLOR]
[/CODE]

[QUOTE=Lorenzo;430176]Hello. Oups. What happened with Mlucas? :sad:

I got this on IBM Power8 and Red Hat 7.2.

Mlucas installed without problem ... I hope. Because now not sure.

[CODE][Mar 02 21:59:06] M43892801 Iter# = 200000 clocks = 00:16:24.308 [ 0.0984 sec/iter] Res64: A82FA0F649C8F274. AvgMaxErr = 0.170602632. MaxErr = 0.250000000
[Mar 02 22:20:42] M43892801 Iter# = 210000 clocks = 00:21:35.630 [ 0.1296 sec/iter] Res64: BAFC8160194E1EFA. [COLOR="Red"]AvgMaxErr = 0.000000000. MaxErr = 0.000000000[/COLOR]
[/CODE][/QUOTE]

Hi, Lorenzo - Through your v14.1, and *only* in scalar-double mode (x86 SIMD always leaves it on) Mlucas will turn off roundoff error checking of convolution outputs if the first 200k iterations run without any ROE warnings. When I first coded this the typical performance hit of ROE checking was 3-7% ... never seen an actual slowdown like yours, much less a huge slowdown.

This feature was one of the first things I ditched in my current v16.0 code-dev plan ... too dangerous, as ROE warnings can catch not just dangerously high errors but many forms of data corruption.

Your results should be fine, but I suggest you manually go into your Mlucas.c source file, change line 342 from
relax_err=TRUE;
to
relax_err=FALSE;
recompile that file, relink your binary and halt/restart your run.

[B]ewmayer[/B], Thank you very much! :smile:

Now it working as before. Very strange with speed. Hope i will verify LL-test.

[CODE][Mar 28 11:29:42] M43892801 Iter# = 17350000 clocks = 00:16:25.361 [ 0.0985 sec/iter] Res64: 9B98340B8DDCA51F. AvgMaxErr = 0.170527673. MaxErr = 0.226562500
[Mar 28 11:46:07] M43892801 Iter# = 17360000 clocks = 00:16:24.359 [ 0.0984 sec/iter] Res64: C4197B369FA08B12. AvgMaxErr = 0.170579383. MaxErr = 0.250000000
[Mar 28 12:02:32] M43892801 Iter# = 17370000 clocks = 00:16:25.449 [ 0.0985 sec/iter] Res64: B61947BA0900E712. AvgMaxErr = 0.170759450. MaxErr = 0.250000000
[Mar 28 12:18:57] M43892801 Iter# = 17380000 clocks = 00:16:24.465 [ 0.0984 sec/iter] Res64: 964B7F7B7BE64D97. AvgMaxErr = 0.170774248. MaxErr = 0.234375000
[Mar 28 12:35:23] M43892801 Iter# = 17390000 clocks = 00:16:25.531 [ 0.0986 sec/iter] Res64: 285FDC8DAC1094B6. AvgMaxErr = 0.170833351. MaxErr = 0.218750000
[Mar 28 12:51:48] M43892801 Iter# = 17400000 clocks = 00:16:24.742 [ 0.0985 sec/iter] Res64: B1C09B8BBF18AF65. AvgMaxErr = 0.170672593. MaxErr = 0.250000000[/CODE]

[QUOTE=Lorenzo;430232]Now it working as before. Very strange with speed. Hope i will verify LL-test.[/QUOTE]

We aim to please. :) When the current run finishes, let us know if the result matches that of the 1st-time test already on file.

Still can't believe removal of the extra operations used in the roundoff check caused such a huge slowdown - your compiler must be doing some really bizarre 'optimizations'.

[CODE][May 06 00:24:49] M43892801 Iter# = 41170000 clocks = 00:02:29.151 [ 0.0149 sec/iter] Res64: 853F7CB9D2160D0B. AvgMaxErr = 0.199261261. MaxErr = [COLOR="red"]0.281250000[/COLOR]

[May 06 00:27:19] M43892801 Iter# = 41180000 clocks = 00:02:29.457 [ 0.0149 sec/iter] Res64: 038A7FDF252DD6C4. AvgMaxErr = 0.199309877. MaxErr = [COLOR="Red"]0.281250000[/COLOR]
[/CODE]

Hello! Looks like i have some problem. :cry:
There are Ok? Or MaxErr shouldn't be greater than 0.25?

MaxErr cannot exceed 0.5. You are looking good.

[QUOTE=Prime95;433290]MaxErr cannot exceed 0.5. You are looking good.[/QUOTE]

Ahhh, yes!) :smile: Just got result. [URL="http://www.mersenne.org/report_exponent/?exp_lo=43892801&full=1"]43892801[/URL] was succesfully doublecheced.
Thank you!