mersenneforum.org Raman's plans and questions
 Register FAQ Search Today's Posts Mark Forums Read

 2009-08-07, 17:43 #1 Raman Noodles     "Mr. Tuch" Dec 2007 Chennai, India 3·419 Posts Raman's plans and questions 6,343+ is about 90% sieved, and is expected to complete in about 20 days or so. I am thinking about 6,335+ for my next number. Is anybody planning for this number, or shall I take it up? If not, I will go to do so for either 7,393+ or 10,339+. Suppose that 6,335+ is easier by GNFS, then I will go after 6,340+ or something like that up. Last fiddled with by Batalov on 2009-08-23 at 21:21 Reason: Moved from 2+ thread (obviously, nothing here is about 2+)
2009-08-07, 21:21   #2
R.D. Silverman

Nov 2003

22·5·373 Posts

Quote:
 Originally Posted by Raman 6,343+ is about 90% sieved, and is expected to complete in about 20 days or so. I am thinking about 6,335+ for my next number. Is anybody planning for this number, or shall I take it up? If not, I will go to do so for either 7,393+ or 10,339+. Suppose that 6,335+ is easier by GNFS, then I will go after 6,340+ or something like that up.
Given your resources, 6,335+ will be much more difficult for you than 7,393+
,7,396+, or 10,339+.

 2009-08-08, 07:06 #3 Raman Noodles     "Mr. Tuch" Dec 2007 Chennai, India 23518 Posts I will give a few more days to decide about it up. I am more likely to have more resources since this month onwards. Right now, I have joined up IIT Madras, to study up for my Master's degree. I think that I should have taken up the tests such as GRE or TOEFL to pursue up my higher studies at the USA. By the way, thus can I know about the fact that how old you are so? And then what type of work you are doing so at? RSA security? Last fiddled with by Raman on 2009-08-08 at 07:08
 2009-08-21, 07:12 #4 Raman Noodles     "Mr. Tuch" Dec 2007 Chennai, India 3·419 Posts It does not matter whether I do an easy number, or a slightly harder number, I will be always be able to do it up. 6,343+ is 93% sieved, and sieving is expected to complete up in another 15 days or so. I have got another cluster of systems at my school, most of the systems are idle most of the time, and that we can always be able to switch up users, in case, one has to work with a particular system, running the sieving program within the background. Each of the systems use Linux, and I have downloaded up the GGNFS binaries for Linux. I can also use up gnfs-lasieve4I15 too. But, there is a problem with the GGNFS binaries for Linux, unable to execute them up. ../gnfs-lasieve4I14e -k -o spairs.out -v -n0 -a task.job ../gnfs-lasieve4I14e: error while loading shared libraries: libgmp.so.3: cannot open shared object file: No such file or directory This error occurs only in Core 2 Duo systems, and not in Pentium Dual Core or the other systems. I assume that Core 2 Duo is the most powerful than all the other older processors, except for Core 2 Quad, and then that Core i7. Thus, what should be done up so to get rid of this error, please?
 2009-08-21, 11:46 #5 jasonp Tribal Bullet     Oct 2004 3·1,181 Posts This is a library load error; you compiled the siever on a machine whose runtime libraries are different than the machines you want to run the siever on. This is the big danger of distributing precompiled unix binaries. The easiest solution is to compile the code while logged into one of those machines, but that requires the development tools to be available there. Also, make sure to get permission from the systems group to run your binary in the background, at low priority or possibly only at night. People have gotten in big trouble by not doing so, and if they give you permission then they can help you get the most computing cycles out of your job.
2009-08-22, 03:50   #6
Raman
Noodles

"Mr. Tuch"
Dec 2007
Chennai, India

3×419 Posts

Quote:
 Originally Posted by jasonp The easiest solution is to compile the code while logged into one of those machines, but that requires the development tools to be available there.
I compiled the code only on the Linux System, in a Pentium Dual core processor, and the siever works fine with that. But, the same siever does not work up in a Core 2 Duo at all.

Code:
    pentium2                   Intel Pentium 2
pentium3                   Intel Pentium 3
pentium4                   Intel Pentium 4
prescott                 Intel Pentium 4 with SSE3
pentium-m                   Intel Pentium M
athlon             AMD Athlon (k7)
athlon-xp         AMD Athlon XP (k7) with SSE
x86_64                   AMD Opteron/Athlon64 (k8)
nocona                   Intel 64-bit-capable Xeon/Pentium
ppc_970                  PowerPC 970
ppc_7450                  PowerPC 7450
doc             Documentation
snapshot                Sources snapshot
install             Installation
clean                  Clean up
I used up "make pentium4". What target should I use up for a Core 2 Duo processor?

make ppc_7450
Code:
make[1]: Entering directory /home/ramanv/Factor/GGNFS/trunk'
echo "#define GGNFS_VERSION \"0.77.1-20060722-7450\"" > include/version.h
make[2]: Entering directory /home/ramanv/Factor/GGNFS/trunk/src'
gcc -I. -I.. -I../include -I/usr/local/include -DNDEBUG -O3 -funroll-loops -finline-functions -ftracer -fomit-frame-pointer -W -Wall -march=7450 -pipe -DGGNFS_HOST_GENERIC -DMALLOC_REPORTING -DGMP_BUG -L/usr/local/lib -o ../bin/sieve sieve.c clsieve.c getprimes.o fbmisc.o squfof.o rels.o blanczos64-no-mmx.o poly.o mpz_poly.o mpz_mat.o smintfact.o misc.o ecm4c.o nfmisc.o matsave.o montgomery_sqrt.o matstuff.o dickman.o fbgen.o llist.o if.o rellist.o intutils.o lasieve4/mpz-ull.o dlmalloc.o modinv32.o makefb.o -lgmp -lm
sieve.c:1: error: bad value (7450) for -march= switch
sieve.c:1: error: bad value (7450) for -mtune= switch
clsieve.c:1: error: bad value (7450) for -march= switch
clsieve.c:1: error: bad value (7450) for -mtune= switch
make[2]: *** [../bin/sieve] Error 1
make[2]: Leaving directory /home/ramanv/Factor/GGNFS/trunk/src'
make[1]: *** [common] Error 2
make[1]: Leaving directory /home/ramanv/Factor/GGNFS/trunk'
make: *** [ppc_7450] Error 2
Quote:
 Originally Posted by jasonp Also, make sure to get permission from the systems group to run your binary in the background, at low priority or possibly only at night. People have gotten in big trouble by not doing so, and if they give you permission then they can help you get the most computing cycles out of your job.
I don't understand. Thus, can you please elaborate more clearly what you are trying to say so?

Quote:
 Originally Posted by jasonp if they give you permission then they can help you get the most computing cycles out of your job.
What's the way to influence up the people? Please give me some of the tips to do so up!

For example, GIMPS is run by millions of users all around North America & Europe. GMP-ECM, GGNFS & msieve are being used up by thousands of users all around the world. What's the way to publish it up, or make it up to become popular? People who should give me up the permission, must understand the significance of these distributed computing projects.

Last fiddled with by Raman on 2009-08-22 at 03:56

2009-08-22, 03:51   #7
Raman
Noodles

"Mr. Tuch"
Dec 2007
Chennai, India

100111010012 Posts

make x86_64
Attached Files
 errors.zip (1.8 KB, 272 views)

Last fiddled with by jasonp on 2010-01-02 at 14:32 Reason: zipped output

 2009-08-22, 08:08 #8 henryzz Just call me Henry     "David" Sep 2007 Cambridge (GMT/BST) 3×1,979 Posts I think nocona is the best thing to use for core 2 in on a 64-bit operating system. To get a good 64-bit binary of the siever though you need to compile the experimental siever. I think i use the binaries in post #28 of http://www.mersenneforum.org/showthread.php?t=11660. Later on in that thread someone said that there isnt much difference between x86_64 compiled binaries and core 2 ones.
2009-08-22, 09:17   #9
fivemack
(loop (#_fork))

Feb 2006
Cambridge, England

192C16 Posts

Quote:
 Originally Posted by Raman make x86_64
Try 'make clean; make x86_64'; the messages from 'ld' indicate that you have some files compiled for 32-bit architecture being picked up by the linker.

Congratulations on getting to IIT!

 2009-08-22, 12:52 #10 jasonp Tribal Bullet     Oct 2004 3·1,181 Posts Any time you build a unix binary on one machine and move it to another unix machine, you are assuming the runtime libraries on the two machines are similar enough that the binary will run. This is not true on your systems, but the OS detects the incompatibility. People who are responsible for many computers like nothing less than to find out a user is acting as if he owns those machines. If the machines are running at 100% all the time and other users complain that they are slow and that they can't do their work, that's more work for the administrators. Just find someone in charge, ask nicely to use them during off-peak hours, add a link here, and ask for suggestions about running a binary in the background. If they don't want to bother, that's their right; just use one at a time. Here is a case of what could happen otherwise (and a more famous one). Perhaps the rules are different in India, but as a student at a very competitive university you cannot afford to risk getting into trouble. I wish I knew how to popularize the kinds of projects that we work on. NFSNET was a good way to do so, but it depends too much on the free time of the project founders. These guys have added a BOINC layer to the sieving tools, which is another nice step. Last fiddled with by jasonp on 2009-08-22 at 12:56
 2009-08-22, 18:45 #11 Raman Noodles     "Mr. Tuch" Dec 2007 Chennai, India 3·419 Posts This is my -1+2^7 th post I think that I will finish off with 6,343+ first. That sieving will take about 15 days or so. Followed up by the Linear Algebra. Based upon the observation of the laboratory during the last few weeks, the lab was only in peak occupation during the afternoons during the weekdays. The other times, the lab has been more than 50% vacant. I think that I will start up my project from the third week of September onwards? Only 15 hours out of 168 hours per week, the lab had been full. There are 80 systems. Of that only 10 systems are Pentium Dual Core, 40 are the slower Celeron D and Pentium 4 processors, and 30 are Core 2 Duo, in which binary is not yet able to be compiled as still. It is not feasible to manage executing the binary within all the systems, at once, but only 5 or so. I will ask for the permission, as early as possible, anyway, however, but, what do you mean by "ask them up nicely" & "the rules are different within India"? Last fiddled with by Raman on 2009-08-22 at 18:47 Reason: 127 = (71 * 2) - (17 - 2)

 Similar Threads Thread Thread Starter Forum Replies Last Post rogue Conjectures 'R Us 35 2013-11-09 09:03 Raman Chess 8 2013-04-16 20:52 Kosmaj Riesel Prime Search 6 2009-05-20 01:27 Kosmaj Riesel Prime Search 6 2006-09-29 22:32 pyrodave Software 17 2004-06-05 12:27

All times are UTC. The time now is 09:42.

Sun Nov 28 09:42:09 UTC 2021 up 128 days, 4:11, 0 users, load averages: 0.92, 1.05, 1.04