Testing new Ranges for Sierpinski/Riesel

[pepi37]

Originally Posted by rogue

I cannot compile for linux as I do not have a linux machine. I have already explained earlier in this thread how to compile for linux. All you should need is the GNU Compiler Chain (gcc).

gcc *.cpp -lstdc++ -O2 -m64 -o srbsieve
main.cpp: In function ‘void getInputs()’:
main.cpp:160:40: error: expected ‘)’ before ‘SCNu64’
main.cpp: In function ‘void removeTrivial()’:
main.cpp:273:34: error: expected ‘)’ before ‘PRId64’
main.cpp: In function ‘void removeGFN()’:
main.cpp:330:27: error: expected ‘)’ before ‘PRId64’
main.cpp: In function ‘void removeMOB()’:
main.cpp:423:30: error: expected ‘)’ before ‘PRId64’
main.cpp: In function ‘void processNewpgenFile(uint32_t)’:
main.cpp:481:27: error: expected ‘)’ before ‘PRId64’
main.cpp: In function ‘void doPhase(int)’:
main.cpp:519:31: error: expected ‘)’ before ‘PRId64’
main.cpp: In function ‘void call_srsieve(uint32_t, uint32_t, uint64_t)’:
main.cpp:555:46: error: expected ‘)’ before ‘PRIu64’
main.cpp: In function ‘void prp_test_ABC_file()’:
main.cpp:601:32: warning: deprecated conversion from string constant to ‘char*’ [-Wwrite-strings]
main.cpp: In function ‘void output_remain(char*)’:
main.cpp:710:29: error: expected ‘)’ before ‘PRId64’

I have compiled srsieve ,srsieve1 and srsieve2 for Linux without any problem on same machine: so it looks like I have whatever I need

This is your last code before adding recovery...

[KEP]

Originally Posted by VBCurtis

That won't be ideal for bases like R3 that lose a bunch of k's to primes, as those sieve files have most of their tests run well below the midpoint of the exponent range. But, it still prevents cases like my first couple tries at using srbsieve, where I ended up having it sieve for longer than it took the resulting file to PFGW completely. If one wanted to allow tinkering, one could make the 10% a setting in the .ini, so users can adjust the relative time spent in sieve; this allows stuff like R3 to run at 7%, or even lower, to compensate for the high frequency of primed k's.

No that won't be ideal. I think that you should wait a few days and see how much faster my srbsieve.ini file is. I have 18 phases and sieve to approximately 2% of the candidates remaining (This was found to be optimal, by using srsieve and sieve the given n-range for each phase, so the value wich I use is accurate to the removal rate on R3). I'm not sure, how much faster it is to run with 18 phases, but if it is as fast as it looks like, then I will be able to do about 100G (maybe more)for base 3 per year on my Hasswell (QuadCore). Also I should add, that my sievedepth, wich for phase 1 (33000 k's) is only p=70000 (n=33 to n=100).

@Rogue: Since I'm sieving to n=32 using NewPGen and might sieve to n=33 for the next tests for R3, could you please have the limit at nothing lower than 33 NewPGen files, for base 3?

[lalera]

Originally Posted by KEP

No that won't be ideal. I think that you should wait a few days and see how much faster my srbsieve.ini file is. I have 18 phases and sieve to approximately 2% of the candidates remaining (This was found to be optimal, by using srsieve and sieve the given n-range for each phase, so the value wich I use is accurate to the removal rate on R3). I'm not sure, how much faster it is to run with 18 phases, but if it is as fast as it looks like, then I will be able to do about 100G (maybe more)for base 3 per year on my Hasswell (QuadCore). Also I should add, that my sievedepth, wich for phase 1 (33000 k's) is only p=70000 (n=33 to n=100).

@Rogue: Since I'm sieving to n=32 using NewPGen and might sieve to n=33 for the next tests for R3, could you please have the limit at nothing lower than 33 NewPGen files, for base 3?

haswell - https://en.wikipedia.org/wiki/Haswel...rchitecture%29

[lalera]

Originally Posted by VBCurtis

If the target time were found on the same machine with the same load, it would not matter how many tasks were running, as the result would still be "sieve until PFGW is faster on an avg candidate."

... looks like ... "speech theory" ...

[rogue]

Originally Posted by pepi37

main.cpp:710:29: error: expected ‘)’ before ‘PRId64’

I have compiled srsieve ,srsieve1 and srsieve2 for Linux without any problem on same machine: so it looks like I have whatever I need

You will need to find the header that has a #define for that then modify the code to include it in main.cpp. I suggest searching /usr/include for that string.

[rogue]

Originally Posted by KEP

@Rogue: Since I'm sieving to n=32 using NewPGen and might sieve to n=33 for the next tests for R3, could you please have the limit at nothing lower than 33 NewPGen files, for base 3?

I see no reason for force that requirement.

[KEP]

Originally Posted by rogue

I see no reason for force that requirement.

Great, thank you

[pepi37]

Originally Posted by rogue

You will need to find the header that has a #define for that then modify the code to include it in main.cpp. I suggest searching /usr/include for that string.

Got it!
Thanks!
Only got one error, but with build binaries:
Now I run test what same data from Windows for reference!

[rogue]

Originally Posted by pepi37

Got it!
Thanks!
Only got one error, but with build binaries:
Now I run test what same data from Windows for reference!

Which header did you need to include?

[pepi37]

Originally Posted by rogue

Which header did you need to include?

None.
It looks like my gcc was "old" it was 4.7
I now install 4.9 and it compile whit that one error.
I found that yours decalration are same as in MiniGW inttypes.hr , and those I get in Linux are little different
( file is herehttp://www.dropbox.com/s/s4yj44kho6g...nttypes.h?dl=0 )

So if I follow correctly

#  define __PRI64_PREFIX    "l"
#  define __PRIPTR_PREFIX   "l"
....
# define PRId64     __PRI64_PREFIX "d"

will be equivalent of

# define PRId64 "ld" ?

I am correct?

[pepi37]

It attachment is cross- reference between Windows, Linux on I5-3570K and on Linux on Haswell.
All three has 7328 primes, remains are same, as GFN and MOB and trivial.
But I seen just one "error" on print screen on all three "windows" ( look at copy-scrren.txt)

It says

 Completed phase 5: 159 remaining ( on Linux and Windows),

but both Linux have 158 remain.
I dont know what was status of Windows at that time, since it is now on 8.8K .
In "win" directory are Pgen files, and settings for first set of S810 in srbsieve.ini ( I divide work on four cores)

UPDATE
I look at task on four cores on Windows
On first core it shows one prime more ( show 119, have 118) , on other show correct number
So it looks like "missing prime is 1*810^n+1" since I start from 2 in srbsieve.ini
810^1+1 is prime! (3 decimal digits, Trial divisions) Time : 0.002 ms.

And since cross reference was from first set, I think all is OK.