sr2sieve client - Page 2

engracio · 2007-07-08, 12:43

After about 8 hrs of crunching, here is the current status. Jjsieve only found 1 factor and sr2sieve found 2. Both found the first factor, jjsieve is still trying to get to the the 2nd factor. Sobistrator say jjsieve is 12% done. Sr2sieve say 15.52% done. I think I will let it run for a while and see any other major difference.
PS jjsieve also found the second factor just now.

Jjsieve at pmin=631411945377727 @ 417kp/s
Sr2sieve at p=631415516700253, 577177 p/sec

D:\Psr2sieve>sr2sieve-intel -s -v
sr2sieve 1.5.13 -- A sieve for multiple sequences k*b^n+/-1.
Using SSE2 code path, L1 data cache 32Kb (detected), L2 cache 2048Kb (detected).
Read 2320262 terms for 18 sequences from dat format file `SoB.dat'.
Split 18 base 2 sequences into 178 base 2^360 subsequences.
Using 32 Kb for the baby-steps giant-steps hashtable, maximum density 0.30.
Using baby step method sse2/16, giant step method sse2/8, ladder method sse2/16.
Using 1024Kb for the Sieve of Eratosthenes bitmap.
Expecting to find factors for about 10.78 terms in this range.
sr2sieve started: 991 <= n <= 49999997, 631400000000000 <= p <= 631500000000000
631403118735439 | 33661*2^39142248+1
631413371167559 | 24737*2^41431303+1
p=631415516700253, 577177 p/sec, 2 factors, 15.52% done, ETA 10 Jul 02:34

pmin=631411475615741 @ 418kp/s
631411517236739 | 55459*2^8475790+1 (duplicate)
pmin=631411542724607 @ 418kp/s
pmin=631411609833469 @ 417kp/s
pmin=631411676942327 @ 418kp/s
pmin=631411744051199 @ 418kp/s
pmin=631411811159999 @ 418kp/s
pmin=631411878268741 @ 416kp/s
pmin=631411945377727 @ 417kp/s

engracio · 2007-07-08, 12:45

Quote:

Originally Posted by mgpower0

i have found sr2 sieve to be quicker on my intels by quite a bit, but on the amd there seems to be little if any difference over jjsieve

I have to agree with you. I think I will run a comparison on my X2 4600+ like on the c2d.

engracio · 2007-07-08, 14:10

Here is the setup on my X2 4600@2640 running sr2sieve and jjsievesse2. First blush is that c2d 6300@2100 is a little faster. We shall see.

Found SoBStatus.dat with unfinished work.
Continuing from last save point.
Setting priority to Process 0 Thread 3
Running in Sierpinski mode
Using CMOV optimizations
Using SSE optimizations
Using SSE2 optimizations
Expected 18 k values.
Found 18 k values.
Setup took 2.866812 seconds.
P Range 633400000000000 <= p <= 633500000000000
N Range 991 <= n <= 49999997
pmin=633400024999961 @ 520kp/s
pmin=633400049999971 @ 523kp/s
pmin=633400074999997 @ 520kp/s
pmin=633400099999991 @ 523kp/s
pmin=633400124999969 @ 524kp/s
pmin=633400149999989 @ 523kp/s

D:\Psr2sieve>sr2sieve-amd -s -vv
sr2sieve 1.5.13 -- A sieve for multiple sequences k*b^n+/-1.
Using SSE2 code path, L1 data cache 64Kb (detected), L2 cache 512Kb (detected).
Read 2320262 terms for 18 sequences from dat format file `SoB.dat'.
Split 18 base 2 sequences into 178 base 2^360 subsequences.
BSGS range: 373*373 - 4961*28.
Using 64 Kb for the baby-steps giant-steps hashtable, maximum density 0.15.
Best time for baby step method sse2/2: 162900.
Best time for baby step method sse2/4: 136545.
Best time for baby step method sse2/8: 129193.
Best time for baby step method sse2/16: 128886.
Best time for baby step method gen/2: 178689.
Best time for baby step method gen/4: 144727.
Best time for baby step method gen/8: 125004.
Best time for baby step method gen/1: 188706.
Best time for giant step method sse2/2: 129972.
Best time for giant step method sse2/4: 123224.
Best time for giant step method sse2/8: 125234.
Best time for giant step method sse2/16: 124713.
Best time for giant step method gen/2: 120865.
Best time for giant step method gen/4: 114396.
Best time for giant step method gen/8: 117523.
Best time for giant step method gen/1: 161642.
Best time for ladder method sse2/2: 10436.
Best time for ladder method sse2/4: 7887.
Best time for ladder method sse2/8: 6994.
Best time for ladder method sse2/16: 7161.
Best time for ladder method gen/2: 9790.
Best time for ladder method gen/4: 7803.
Best time for ladder method gen/8: 6342.
Best time for ladder method gen/1: 16651.
Best time for ladder method add/1: 9215.
Using baby step method gen/8, giant step method gen/4, ladder method gen/8.
Using 256Kb for the Sieve of Eratosthenes bitmap.
Expecting to find factors for about 10.75 terms in this range.
sr2sieve started: 991 <= n <= 49999997, 633400000000000 <= p <= 633500000000000
p=633400226231197, 539878 p/sec, 0 factors, 0.23% done, ETA 10 Jul 12:45

engracio · 2007-07-08, 19:55

I think I'll stop now. I'm pretty happy how much faster the sr2sieve compare to jjsieve. Unless you guys think otherwise. Sr2sieve picked the 3rd factor jjsieve is still trying to get there. This is on the c2d 6300. Ill post the x2 4600 stuff on that box.

pmin=631422817013741 @ 419kp/s
pmin=631422884122609 @ 418kp/s
631422924150659 | 55459*2^48884686+1 (duplicate)
pmin=631422951231473 @ 417kp/s
pmin=631423018340351 @ 418kp/s
pmin=631423085449187 @ 417kp/s

Using 1024Kb for the Sieve of Eratosthenes bitmap.
Expecting to find factors for about 10.78 terms in this range.
sr2sieve started: 991 <= n <= 49999997, 631400000000000 <= p <= 631500000000000
631403118735439 | 33661*2^39142248+1
631413371167559 | 24737*2^41431303+1
631426452846979 | 168451*2^37488996+1
p=631430985686789, 561293 p/sec, 3 factors, 30.99% done, ETA 10 Jul 01:06

engracio · 2007-07-08, 20:02

This is on the X2 4600 box. Sobistrator say it is 11.3% completed. Sr2sieve 11.66%. Sr2sieve is still little bit faster on AMD X2 but definitely a lot faster on c2d.

pmin=633411249999929 @ 524kp/s
pmin=633411274999961 @ 523kp/s
pmin=633411299999963 @ 523kp/s
pmin=633411324999983 @ 525kp/s
pmin=633411349999997 @ 524kp/s
pmin=633411374999947 @ 518kp/s

Best time for ladder method add/1: 9215.
Using baby step method gen/8, giant step method gen/4, ladder method gen/8.
Using 256Kb for the Sieve of Eratosthenes bitmap.
Expecting to find factors for about 10.75 terms in this range.
sr2sieve started: 991 <= n <= 49999997, 633400000000000 <= p <= 633500000000000
633402880943863 | 152267*2^27483099+1
p=633411662000609, 533153 p/sec, 1 factor, 11.66% done, ETA 10 Jul 12:27

mgpower0 · 2007-07-08, 23:51

pretty much the conclusion I have come too. I have decided to run sr2sieve for all my rigs i like the fact it gives you an ETA on the range

hhh · 2007-09-24, 14:18

So, should I post a link in the reservation list post? Is there a configuration where jjsieve is better? I didn't do much sieving lately, so I'm not up to date.

H.

mgpower0 · 2007-09-25, 09:10

sr2 seem to be quicker on all cpu's, only a little bit quicker on amd, quite a lot on intels
Tell you what is quick is sr2sieve for 64bit Linux. Currently running a range on my E6400 C2D 2.13ghz. On 32 bit xp was getting around 720kp/sec with windows sr2 around 670kp/sec with jjsieve and now 935kp/sec with sr2sieve for 64bit Linux
All are available here if people want to try them out, Make sure to read the read me file as you need to make a txt doc with your range in it called sr2work.txt, and needs to be run with -s or -r to run with Sob/psp comb dat or riesel dat. READ the readme file
Linky, Latest releases at the bottom
http://www.geocities.com/g_w_reynolds/sr2sieve/

Brucifer · 2007-09-25, 17:33

I was wondering how long before the 64-bit linux client would get a little more recognition. :)

That's all I use now for sieving. No reason at all for a dedicated 64-bit capable cruncher to not be running linux, and using the 64-bit sr2 client.

mgpower0 · 2007-09-26, 08:00

Yep run all my DC projects on Linux now. If it won't work on 64 bit Linux I won't run it

Gomeler · 2007-09-26, 21:02

No way.. just fired up an instance of JJSieveSSE2 and 1.5.19 sr2sieve x86 and the speedup is huge.
Core 2 Duo Allendale - 460x7, 3220 MHz
2x1GB 460MHz 5-5-5-23
Bloated Windows XP
JJSieveSSE2 : 620kp/s
sr2sieve: 819kp/s

Going to let sr2sieve run through the night and just make sure both programs are picking up the same factors and if so then I'll be switching over. Uses 20mb of ram vs 50mb also, great stuff.

2007-07-08, 12:43	#12
engracio May 2007 11² Posts	After about 8 hrs of crunching, here is the current status. Jjsieve only found 1 factor and sr2sieve found 2. Both found the first factor, jjsieve is still trying to get to the the 2nd factor. Sobistrator say jjsieve is 12% done. Sr2sieve say 15.52% done. I think I will let it run for a while and see any other major difference. PS jjsieve also found the second factor just now. Jjsieve at pmin=631411945377727 @ 417kp/s Sr2sieve at p=631415516700253, 577177 p/sec D:\Psr2sieve>sr2sieve-intel -s -v sr2sieve 1.5.13 -- A sieve for multiple sequences kb^n+/-1. Using SSE2 code path, L1 data cache 32Kb (detected), L2 cache 2048Kb (detected). Read 2320262 terms for 18 sequences from dat format file `SoB.dat'. Split 18 base 2 sequences into 178 base 2^360 subsequences. Using 32 Kb for the baby-steps giant-steps hashtable, maximum density 0.30. Using baby step method sse2/16, giant step method sse2/8, ladder method sse2/16. Using 1024Kb for the Sieve of Eratosthenes bitmap. Expecting to find factors for about 10.78 terms in this range. sr2sieve started: 991 <= n <= 49999997, 631400000000000 <= p <= 631500000000000 631403118735439 \| 336612^39142248+1 631413371167559 \| 247372^41431303+1 p=631415516700253, 577177 p/sec, 2 factors, 15.52% done, ETA 10 Jul 02:34 pmin=631411475615741 @ 418kp/s 631411517236739 \| 554592^8475790+1 (duplicate) pmin=631411542724607 @ 418kp/s pmin=631411609833469 @ 417kp/s pmin=631411676942327 @ 418kp/s pmin=631411744051199 @ 418kp/s pmin=631411811159999 @ 418kp/s pmin=631411878268741 @ 416kp/s pmin=631411945377727 @ 417kp/s Last fiddled with by engracio on 2007-07-08 at 13:22

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
srsieve/sr2sieve enhancements	rogue	Software	300	2021-03-18 20:31
32-bit of sr1sieve and sr2sieve for Win	pepi37	Software	5	2013-08-09 22:31
sr2sieve question	SaneMur	Information & Answers	2	2011-08-21 22:04
How to use sr2sieve	nuggetprime	Riesel Prime Search	40	2007-12-03 06:01
Sr2sieve on PPC/Linux	BlisteringSheep	Sierpinski/Riesel Base 5	172	2007-07-06 21:29

2007-07-08, 14:10	#14
engracio May 2007 121₁₀ Posts	Here is the setup on my X2 4600@2640 running sr2sieve and jjsievesse2. First blush is that c2d 6300@2100 is a little faster. We shall see. Found SoBStatus.dat with unfinished work. Continuing from last save point. Setting priority to Process 0 Thread 3 Running in Sierpinski mode Using CMOV optimizations Using SSE optimizations Using SSE2 optimizations Expected 18 k values. Found 18 k values. Setup took 2.866812 seconds. P Range 633400000000000 <= p <= 633500000000000 N Range 991 <= n <= 49999997 pmin=633400024999961 @ 520kp/s pmin=633400049999971 @ 523kp/s pmin=633400074999997 @ 520kp/s pmin=633400099999991 @ 523kp/s pmin=633400124999969 @ 524kp/s pmin=633400149999989 @ 523kp/s D:\Psr2sieve>sr2sieve-amd -s -vv sr2sieve 1.5.13 -- A sieve for multiple sequences kb^n+/-1. Using SSE2 code path, L1 data cache 64Kb (detected), L2 cache 512Kb (detected). Read 2320262 terms for 18 sequences from dat format file `SoB.dat'. Split 18 base 2 sequences into 178 base 2^360 subsequences. BSGS range: 373373 - 496128. Using 64 Kb for the baby-steps giant-steps hashtable, maximum density 0.15. Best time for baby step method sse2/2: 162900. Best time for baby step method sse2/4: 136545. Best time for baby step method sse2/8: 129193. Best time for baby step method sse2/16: 128886. Best time for baby step method gen/2: 178689. Best time for baby step method gen/4: 144727. Best time for baby step method gen/8: 125004. Best time for baby step method gen/1: 188706. Best time for giant step method sse2/2: 129972. Best time for giant step method sse2/4: 123224. Best time for giant step method sse2/8: 125234. Best time for giant step method sse2/16: 124713. Best time for giant step method gen/2: 120865. Best time for giant step method gen/4: 114396. Best time for giant step method gen/8: 117523. Best time for giant step method gen/1: 161642. Best time for ladder method sse2/2: 10436. Best time for ladder method sse2/4: 7887. Best time for ladder method sse2/8: 6994. Best time for ladder method sse2/16: 7161. Best time for ladder method gen/2: 9790. Best time for ladder method gen/4: 7803. Best time for ladder method gen/8: 6342. Best time for ladder method gen/1: 16651. Best time for ladder method add/1: 9215. Using baby step method gen/8, giant step method gen/4, ladder method gen/8. Using 256Kb for the Sieve of Eratosthenes bitmap. Expecting to find factors for about 10.75 terms in this range. sr2sieve started: 991 <= n <= 49999997, 633400000000000 <= p <= 633500000000000 p=633400226231197, 539878 p/sec, 0 factors, 0.23% done, ETA 10 Jul 12:45*

2007-07-08, 19:55	#15
engracio May 2007 11² Posts	I think I'll stop now. I'm pretty happy how much faster the sr2sieve compare to jjsieve. Unless you guys think otherwise. Sr2sieve picked the 3rd factor jjsieve is still trying to get there. This is on the c2d 6300. Ill post the x2 4600 stuff on that box. pmin=631422817013741 @ 419kp/s pmin=631422884122609 @ 418kp/s 631422924150659 \| 554592^48884686+1 (duplicate) pmin=631422951231473 @ 417kp/s pmin=631423018340351 @ 418kp/s pmin=631423085449187* @ 417kp/s Using 1024Kb for the Sieve of Eratosthenes bitmap. Expecting to find factors for about 10.78 terms in this range. sr2sieve started: 991 <= n <= 49999997, 631400000000000 <= p <= 631500000000000 631403118735439 \| 336612^39142248+1 631413371167559 \| 247372^41431303+1 631426452846979 \| 1684512^37488996+1 p=631430985686789,* 561293 p/sec, 3 factors, 30.99% done, ETA 10 Jul 01:06

2007-07-08, 20:02	#16
engracio May 2007 121₁₀ Posts	This is on the X2 4600 box. Sobistrator say it is 11.3% completed. Sr2sieve 11.66%. Sr2sieve is still little bit faster on AMD X2 but definitely a lot faster on c2d. pmin=633411249999929 @ 524kp/s pmin=633411274999961 @ 523kp/s pmin=633411299999963 @ 523kp/s pmin=633411324999983 @ 525kp/s pmin=633411349999997 @ 524kp/s pmin=633411374999947 @ 518kp/s Best time for ladder method add/1: 9215. Using baby step method gen/8, giant step method gen/4, ladder method gen/8. Using 256Kb for the Sieve of Eratosthenes bitmap. Expecting to find factors for about 10.75 terms in this range. sr2sieve started: 991 <= n <= 49999997, 633400000000000 <= p <= 633500000000000 633402880943863 \| 1522672^27483099+1 p=633411662000609*, 533153 p/sec, 1 factor, 11.66% done, ETA 10 Jul 12:27

2007-07-08, 23:51	#17
mgpower0 Feb 2007 11 Posts	pretty much the conclusion I have come too. I have decided to run sr2sieve for all my rigs i like the fact it gives you an ETA on the range

2007-09-24, 14:18	#18
hhh Jun 2005 565₈ Posts	So, should I post a link in the reservation list post? Is there a configuration where jjsieve is better? I didn't do much sieving lately, so I'm not up to date. H.

2007-09-25, 09:10	#19
mgpower0 Feb 2007 11₁₀ Posts	sr2 seem to be quicker on all cpu's, only a little bit quicker on amd, quite a lot on intels Tell you what is quick is sr2sieve for 64bit Linux. Currently running a range on my E6400 C2D 2.13ghz. On 32 bit xp was getting around 720kp/sec with windows sr2 around 670kp/sec with jjsieve and now 935kp/sec with sr2sieve for 64bit Linux All are available here if people want to try them out, Make sure to read the read me file as you need to make a txt doc with your range in it called sr2work.txt, and needs to be run with -s or -r to run with Sob/psp comb dat or riesel dat. READ the readme file Linky, Latest releases at the bottom http://www.geocities.com/g_w_reynolds/sr2sieve/

2007-09-25, 17:33	#20
Brucifer Dec 2005 313 Posts	I was wondering how long before the 64-bit linux client would get a little more recognition. :) That's all I use now for sieving. No reason at all for a dedicated 64-bit capable cruncher to not be running linux, and using the 64-bit sr2 client.

2007-09-26, 08:00	#21
mgpower0 Feb 2007 11 Posts	Yep run all my DC projects on Linux now. If it won't work on 64 bit Linux I won't run it

2007-09-26, 21:02	#22
Gomeler Dec 2006 2³ Posts	No way.. just fired up an instance of JJSieveSSE2 and 1.5.19 sr2sieve x86 and the speedup is huge. Core 2 Duo Allendale - 460x7, 3220 MHz 2x1GB 460MHz 5-5-5-23 Bloated Windows XP JJSieveSSE2 : 620kp/s sr2sieve: 819kp/s Going to let sr2sieve run through the night and just make sure both programs are picking up the same factors and if so then I'll be switching over. Uses 20mb of ram vs 50mb also, great stuff.