1cpu vs 2cpu

[cardmaker]

hello, im testing new hardware factoring a 100digit and no much diference make between 1 cpu vs 2 cpu only 30sec does it normal?
is the same number and same bin(script 0.84 , msieve 1.52), only the change is the number os cores.
NUM_CORES = 12
THREADS_PER_CORE = 2
to
NUM_CORES = 24
THREADS_PER_CORE = 2

and the result whit 1 cpu:

Number: example
N = 1522605027922533360535618378132637429718068114961380688657908494580122963258952897654000350692006139 (100 digits)
Divisors found:
r1=37975227936943673922808872755445627854565536638199 (pp50)
r2=40094690950920881030683735292761468389214899724061 (pp50)
Version: Msieve v. 1.52 (SVN Nicht versioniertes Verzeichnis)
Total time: 0.21 hours.
Factorization parameters were as follows:
n: 1522605027922533360535618378132637429718068114961380688657908494580122963258952897654000350692006139
# norm 1.402930e-13 alpha -6.325236 e 1.281e-08 rroots 2
skew: 3483965.44
c0: -36738379143255767765564663397
c1: 75967609570279467412754
c2: -784169456850679
c3: -3486221646
c4: 1368
Y0: -1027140279095180418809608
Y1: 16592488498117
type: gnfs
Factor base limits: 1800000/1800000
Large primes per side: 3
Large prime bits: 26/26
Sieved algebraic special-q in [0, 0)
Total raw relations: 5073376
Relations: 254576 relations
Pruned matrix : 155741 x 155971
Polynomial selection time: 0.01 hours.
Total sieving time: 0.17 hours.
Total relation processing time: 0.01 hours.
Matrix solve time: 0.02 hours.
time per square root: 0.00 hours.
Prototype def-par.txt line would be: gnfs,99,4,58,1500,0.003,0.4,220,15,10000,2000,1800000,1800000,26,26,48,48,2.5,2.5,100000
total time: 0.21 hours.
x86_64
processors: 48, speed: 1.20GHz
Linux-4.4.0-197-generic-x86_64-with-Ubuntu-16.04-xenial
Running Python 2.7

and whit 2 cpu:

Number: example
N = 1522605027922533360535618378132637429718068114961380688657908494580122963258952897654000350692006139 (100 digits)
Divisors found:
r1=37975227936943673922808872755445627854565536638199 (pp50)
r2=40094690950920881030683735292761468389214899724061 (pp50)
Version: Msieve v. 1.52 (SVN Nicht versioniertes Verzeichnis)
Total time: 0.20 hours.
Factorization parameters were as follows:
n: 1522605027922533360535618378132637429718068114961380688657908494580122963258952897654000350692006139
# norm 9.462249e-14 alpha -4.664878 e 1.057e-08 rroots 4
skew: 3782058.60
c0: 35793799656782527909425094900
c1: -19487356755197005852048
c2: -28534739231977525
c3: 3449886256
c4: 804
Y0: -1173089344615252621251927
Y1: 4312706706875
type: gnfs
Factor base limits: 1800000/1800000
Large primes per side: 3
Large prime bits: 26/26
Sieved algebraic special-q in [0, 0)
Total raw relations: 5021071
Relations: 284824 relations
Pruned matrix : 173370 x 173598
Polynomial selection time: 0.00 hours.
Total sieving time: 0.16 hours.
Total relation processing time: 0.01 hours.
Matrix solve time: 0.03 hours.
time per square root: 0.01 hours.
Prototype def-par.txt line would be: gnfs,99,4,58,1500,0.003,0.4,220,15,10000,2000,1800000,1800000,26,26,48,48,2.5,2.5,100000
total time: 0.20 hours.
x86_64
processors: 48, speed: 1.20GHz
Linux-4.4.0-197-generic-x86_64-with-Ubuntu-16.04-xenial
Running Python 2.7

by the clock wall only 30sec of diference for factorization does it normal? sould it be more diference whit 2 cpu?
also this line "processors: 48, speed: 1.20GHz" is not right because my cpu clock is 2.3GHZ and not 1.2 anyone have an ideia?

[retina]

Both results say the exact same CPU count:

processors: 48, speed: 1.20GHz

So I would expect them to finish in a similar time.

[EdH]

The speed is a "right now" speed rather than a listed speed. The 30 second difference is probably due to the polynomial rather than the hardware.

What was the command line that invoked msieve? Did it include "-t <number of threads to use>"? If msieve was called from another program/script, does that p/s provide the hardware values to use?

[cardmaker]

@retina

processors: 48, speed: 1.20GHz

this line i think is a reading of system not the total cores i set in script, because when executing script only the number of thread i set in script are executed.
example when set to 12 log appear t0 -> t23
when set 24 log appear t0 ->47

@EdH
yes the number of thread are invoked correctly in the msieve comand line.
i use the script version 84, and i see its correct.

[charybdis]

I presume your system has 24 physical cores with hyperthreading.

Setting

Originally Posted by cardmaker

NUM_CORES = 12
THREADS_PER_CORE = 2

will run 24 processes for most of the factorization, and unless you force factmsieve.py to only use 12 physical cores, the scheduler will assign one process to each core. The speedup you get from NUM_CORES = 24 relative to NUM_CORES = 12 is only due to the use of hyperthreading, so we shouldn't expect speed to double.

Your factorization with NUM_CORES = 24 had a substantially worse polynomial and still finished quicker than the one with NUM_CORES = 12, so 24 is definitely faster - just nowhere near double the speed.