Distributed sieving discussions - Page 4

axn · 2006-07-07, 20:27

I've tried to fix it so it is clearer.

rogue · 2006-07-07, 21:28

I would like to see a thread on sieving rates. Thus I could compare the speed of PPC to other platforms easily. Also, could someone post a input file and factors found file for a small range extracted from the current zip file so that those of us building our own version could verify the results?

Many thanks if this could be done.

(edit)
Actually, if someone can sieve a range from 10e9-11e9 and send me the factors they find, that would be good enough for me. I can't speak for anyone else.

BTW, OS X does not support nested functions, as found in srfile.c. I assume this will not be needed to sieve.

(edit2)
When using ARCH=ppc64, the function pre2_mulmod64_fini() is not defined. I modified arithmetic.h to use a empty function for ppc64 since AFAICT it is only necessary for x86.

axn · 2006-07-07, 22:09

Quote:

Originally Posted by rogue

I would like to see a thread on sieving rates. Thus I could compare the speed of PPC to other platforms easily.

Done

Quote:

Originally Posted by rogue

Actually, if someone can sieve a range from 10e9-11e9 and send me the factors they find, that would be good enough for me. I can't speak for anyone else.

YGM

Quote:

Originally Posted by rogue

BTW, OS X does not support nested functions, as found in srfile.c. I assume this will not be needed to sieve.

That's correct -- only sr5sieve is needed for _this_ project. srfile is a utility meant to work with the full build (srsieve) -- not relevant for our purpose.

Quote:

Originally Posted by rogue

When using ARCH=ppc64, the function pre2_mulmod64_fini() is not defined. I modified arithmetic.h to use a empty function for ppc64 since AFAICT it is only necessary for x86.

Obviously, geoff will be able to speak authoritatively on this one, but I believe you are essentially correct.

rogue · 2006-07-07, 23:43

Stupid me. Since sieving is complete to 25e9 those k/n pairs are already removed. How about 50e9-51e9? I'm seeing factors in that range. Once the factors are verified, I will reserve a range.

geoff · 2006-07-08, 02:40

Quote:

Originally Posted by rogue

BTW, OS X does not support nested functions, as found in srfile.c. I assume this will not be needed to sieve.

When using ARCH=ppc64, the function pre2_mulmod64_fini() is not defined.

Both fixed in 0.3.4.

rogue · 2006-07-08, 15:35

I've verified a large range of factors sent by Anand, so I'm good to go.

I will reserve 100e9 - 200e9. I have four CPUs and should complete in about three weeks.

axn · 2006-07-08, 16:05

I have moved out the discussions from the reservation thread to here.

Greenbank · 2006-07-11, 09:23

I have 20GB diskspace and unmetered bandwidth hosting. I already pay the bill for this so it doesn't matter how much I throw on top of it.

If you want to use this then please let me know and I'll create an account for you. (This is for sieving related files only.)

michaf · 2006-07-12, 20:02

I've noticed that my latest reservation for prp-ing in the range
91.8-92.0k is already sieved to 104e9.

Are we doing double work in the distributed sieving, or is there a separate sieve running someplace for numbers upto 100k?

If so, we need to get the distributed sieve going faster :>

Oh, axn1, could you add a column to the results:
factors per 10^9 sieved. It would be nice to see how many you can expect once we get more results.

axn · 2006-07-12, 21:24

Quote:

Originally Posted by michaf

I've noticed that my latest reservation for prp-ing in the range
91.8-92.0k is already sieved to 104e9.

Are we doing double work in the distributed sieving, or is there a separate sieve running someplace for numbers upto 100k?

The current PRP range was sieved separately till 104e9 (for about 170 k's with 60k <= n <= 100k) at which point I stopped. The distributed sieve is for the complete k's for 0 < n <= 2,000,000. This one will be sieved to much higher levels (currently srsieve is limited to 4T - we'll easily cross that). We are "kind of" doing double work, since the PRP range is getting resieved as part of the main sieve effort. However, that should not be a big deal.

Quote:

Originally Posted by michaf

Oh, axn1, could you add a column to the results:
factors per 10^9 sieved. It would be nice to see how many you can expect once we get more results.

Are you looking for expected factors per G or actual factors per G ?

geoff · 2006-07-13, 02:44

Quote:

Originally Posted by axn1

(currently srsieve is limited to 4T - we'll easily cross that).

The limit is actually 2^62 for i386 machines and 2^52 for others. 4e12 was the limit of the prime sieve, beyond that some composites would be sieved, but that only affected the efficiency not the results. In the current version the prime sieve will generate all primes up to 2^62, although it takes a lot of memory (> 1GB).

2006-07-07, 21:28	#35
rogue "Mark" Apr 2003 Between here and the 18CB₁₆ Posts	I would like to see a thread on sieving rates. Thus I could compare the speed of PPC to other platforms easily. Also, could someone post a input file and factors found file for a small range extracted from the current zip file so that those of us building our own version could verify the results? Many thanks if this could be done. (edit) Actually, if someone can sieve a range from 10e9-11e9 and send me the factors they find, that would be good enough for me. I can't speak for anyone else. BTW, OS X does not support nested functions, as found in srfile.c. I assume this will not be needed to sieve. (edit2) When using ARCH=ppc64, the function pre2_mulmod64_fini() is not defined. I modified arithmetic.h to use a empty function for ppc64 since AFAICT it is only necessary for x86. Last fiddled with by rogue on 2006-07-07 at 21:43

2006-07-07, 23:43	#37
rogue "Mark" Apr 2003 Between here and the 11·577 Posts	Stupid me. Since sieving is complete to 25e9 those k/n pairs are already removed. How about 50e9-51e9? I'm seeing factors in that range. Once the factors are verified, I will reserve a range. Last fiddled with by rogue on 2006-07-07 at 23:43

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Soapbox Discussions	only_human	Soap Box	41	2019-11-16 15:46
Our discussions here: how can we improve things?	Brian-E	Soap Box	105	2013-11-10 12:26
n=390000 discussions and old reservations	Mini-Geek	Twin Prime Search	31	2010-05-22 23:13
Primegrid discussions	pacionet	Twin Prime Search	17	2007-01-20 11:22
Automated PRP discussions	ltd	Sierpinski/Riesel Base 5	20	2006-09-02 22:19

2006-07-07, 20:27	#34
axn Jun 2003 5053₁₀ Posts	I've tried to fix it so it is clearer.

2006-07-08, 15:35	#39
rogue "Mark" Apr 2003 Between here and the 11·577 Posts	I've verified a large range of factors sent by Anand, so I'm good to go. I will reserve 100e9 - 200e9. I have four CPUs and should complete in about three weeks.

2006-07-08, 16:05	#40
axn Jun 2003 5053₁₀ Posts	I have moved out the discussions from the reservation thread to here.

2006-07-11, 09:23	#41
Greenbank Jul 2005 2×193 Posts	I have 20GB diskspace and unmetered bandwidth hosting. I already pay the bill for this so it doesn't matter how much I throw on top of it. If you want to use this then please let me know and I'll create an account for you. (This is for sieving related files only.)

2006-07-12, 20:02	#42
michaf Jan 2005 479 Posts	I've noticed that my latest reservation for prp-ing in the range 91.8-92.0k is already sieved to 104e9. Are we doing double work in the distributed sieving, or is there a separate sieve running someplace for numbers upto 100k? If so, we need to get the distributed sieve going faster :> Oh, axn1, could you add a column to the results: factors per 10^9 sieved. It would be nice to see how many you can expect once we get more results.