How long does factoring run on 2^33,xxx,xxx type numbers?

nukemyrman · 2003-03-03, 14:18

Hi,
I'm running my first 2^33M-1 number and my machine has been factoring for 2 days now. It is a 2.5GHz, 1Gbyte DRAM machine. Does this sound reasonable?
Thank you! Nick

QuintLeo · 2003-03-03, 14:36

It takes about a day for my P4 2.36 Ghz machine to factor a 20,xxx,xxx number - but those only factor to 2^67 depth.

33,xxx,xxx factor to 2^69 depth - so should take 4 days or a little less for your machine.

nukemyrman · 2003-03-03, 14:45

Thanx Q.
I think it is great we're getting some good factoring in for the Number Theory guys and gals.
Nuke

eepiccolo · 2003-03-03, 15:44

I recently factored a 2^333xxxxx-1 number to 2^68, and it took 42 hours running non stop with nothing else running on a P4 2.4Ghz. I think you would actually have to have an exponent greater than 3510000 to have Prime95 automatically trial factor to 2^69. I'm not sure, but I think the number I was working on had been previously TFed to 2^60.

Prime95 · 2003-03-03, 16:16

Quote:

Originally Posted by nukemyrman

I'm running my first 2^33M-1 number and my machine has been factoring for 2 days now. It is a 2.5GHz, 1Gbyte DRAM machine. Does this sound reasonable?

Trial factoring or P-1 factoring?

nukemyrman · 2003-03-03, 17:31

I don't mind running the factoring software a little longer than "Standard P95" is there soemhting I can put in the .int file to give it the OK to look a little further?
Is how much time longer would I consume, for example, if I checked just one more bit. Also, would I lose the work I've done up 'till now.

smh · 2003-03-03, 18:56

Factoring 1 bit deeper means you are factoring twice as far, so this takes at least double so long.

pakaran · 2003-03-03, 19:08

And a lot longer than twice in the case of going past certain numbers of bits, 62 and more so 65 come to mind.

Is the machine a P4 or an XP? The XP tends to factor pretty fast.

lycorn · 2003-03-06, 10:42

nukemyrman wrote:
I don't mind running the factoring software a little longer than "Standard P95"...

Don´t use a P4 for that! P4s perform like crap on Trial Factoring, so stick to the minimum TF defined by the client, and use the power of the SSE2 instruction set for LL. From some data I have collected, I assure you it is a waste to engage P4s on long factoring assignments.

eepiccolo · 2003-03-06, 13:29

I don't remember where I saw it, but I thought I saw somewhere that the P4 does really good with TFing once you get past 2^64, since than you have more than 64 bits. Is this true?

lycorn · 2003-03-06, 13:35

Shortly after having posted my advice to nukemyrman, I realized we were in fact talking about TFing beyond 2^64. For those ranges, the SSE2 instructions are used, so I think P4s do indeed a lot better than up to 2^63. I don´t have figures for those ranges, but I will at least give P4s the benefit of doubt...

:)

2003-03-03, 14:18	#1
nukemyrman Mar 2003 Yucaipa, CA, USA 8₁₀ Posts	How long does factoring run on 2^33,xxx,xxx type numbers? Hi, I'm running my first 2^33M-1 number and my machine has been factoring for 2 days now. It is a 2.5GHz, 1Gbyte DRAM machine. Does this sound reasonable? Thank you! Nick

2003-03-06, 13:29	#10
eepiccolo Dec 2002 Frederick County, MD 2×5×37 Posts	Trial Factoring over 2^64 I don't remember where I saw it, but I thought I saw somewhere that the P4 does really good with TFing once you get past 2^64, since than you have more than 64 bits. Is this true?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
How long it takes to factoring the 512-bit number?	Pepek	Msieve	5	2012-09-14 16:32
What is work type "LL Test with no factoring"?	Svenie25	PrimeNet	4	2011-02-22 20:08
how long it will take factoring a big number 512b	sinide	Factoring	8	2010-11-19 08:03
LL no factoring work type	edorajh	Information & Answers	1	2010-04-16 16:55
long numbers. Programming.	chrow	Factoring	3	2003-08-24 17:41

2003-03-03, 14:36	#2
QuintLeo Oct 2002 Lost in the hills of Iowa 700₈ Posts	It takes about a day for my P4 2.36 Ghz machine to factor a 20,xxx,xxx number - but those only factor to 2^67 depth. 33,xxx,xxx factor to 2^69 depth - so should take 4 days or a little less for your machine.

2003-03-03, 14:45	#3
nukemyrman Mar 2003 Yucaipa, CA, USA 8₁₀ Posts	Thanx Q. I think it is great we're getting some good factoring in for the Number Theory guys and gals. Nuke

2003-03-03, 15:44	#4
eepiccolo Dec 2002 Frederick County, MD 2×5×37 Posts	I recently factored a 2^333xxxxx-1 number to 2^68, and it took 42 hours running non stop with nothing else running on a P4 2.4Ghz. I think you would actually have to have an exponent greater than 3510000 to have Prime95 automatically trial factor to 2^69. I'm not sure, but I think the number I was working on had been previously TFed to 2^60.

2003-03-03, 17:31	#6
nukemyrman Mar 2003 Yucaipa, CA, USA 10₈ Posts	I don't mind running the factoring software a little longer than "Standard P95" is there soemhting I can put in the .int file to give it the OK to look a little further? Is how much time longer would I consume, for example, if I checked just one more bit. Also, would I lose the work I've done up 'till now.

2003-03-03, 18:56	#7
smh "Sander" Oct 2002 52.345322,5.52471 29×41 Posts	Factoring 1 bit deeper means you are factoring twice as far, so this takes at least double so long.

2003-03-03, 19:08	#8
pakaran Aug 2002 249₁₀ Posts	And a lot longer than twice in the case of going past certain numbers of bits, 62 and more so 65 come to mind. Is the machine a P4 or an XP? The XP tends to factor pretty fast.

2003-03-06, 10:42	#9
lycorn "GIMFS" Sep 2002 Oeiras, Portugal 1571₁₀ Posts	nukemyrman wrote: I don't mind running the factoring software a little longer than "Standard P95"... Don´t use a P4 for that! P4s perform like crap on Trial Factoring, so stick to the minimum TF defined by the client, and use the power of the SSE2 instruction set for LL. From some data I have collected, I assure you it is a waste to engage P4s on long factoring assignments.

2003-03-06, 13:35	#11
lycorn "GIMFS" Sep 2002 Oeiras, Portugal 1571₁₀ Posts	Shortly after having posted my advice to nukemyrman, I realized we were in fact talking about TFing beyond 2^64. For those ranges, the SSE2 instructions are used, so I think P4s do indeed a lot better than up to 2^63. I don´t have figures for those ranges, but I will at least give P4s the benefit of doubt... :)