| Author |
Message |
streamVolunteer moderator Project administrator Volunteer developer Volunteer tester Send message
Joined: 1 Mar 14 Posts: 1182 ID: 301928 Credit: 777,527,559 RAC: 0
                              
|
|
To celebrate our move to new powerful servers, SGS project was converted to LLR2 fast DC. Enjoy!
|
|
|
Vato Volunteer tester
 Send message
Joined: 2 Feb 08 Posts: 989 ID: 18447 Credit: 2,389,235,210 RAC: 462,824
                                      
|
|
excellent - thank you
____________
|
|
|
robish Volunteer moderator Volunteer tester
 Send message
Joined: 7 Jan 12 Posts: 2264 ID: 126266 Credit: 10,773,116,553 RAC: 2,127,560
                                       
|
|
Haha! Cool 😎
Thanks stream!
____________
My lucky number 10590941048576+1 |
|
|
|
|
|
Cool change! Nice to see that even on a very fast computer the proof tasks aren't so short as to be a real burden like GFN15 running solo on a fast GPU.
Two bug notes:
1: #Threads is not being reported in the Task Returned chart
B: In the same chart it is counting the proof tasks in the duration stats and I think (really hard to check) also in the total number of tasks.
____________
Eating more cheese on Thursdays. |
|
|
Michael Goetz Volunteer moderator Project administrator
 Send message
Joined: 21 Jan 10 Posts: 14692 ID: 53948 Credit: 1,042,938,327 RAC: 12,726
                                           
|
Cool change! Nice to see that even on a very fast computer the proof tasks aren't so short as to be a real burden like GFN15 running solo on a fast GPU.
Two bug notes:
1: #Threads is not being reported in the Task Returned chart
B: In the same chart it is counting the proof tasks in the duration stats and I think (really hard to check) also in the total number of tasks.
A flag needed to be set.
Note that the chart now works for LLR2, but while they're still around, won't show the thread count for the old LLR tasks.
____________
My lucky number is 75898524288+1 |
|
|
streamVolunteer moderator Project administrator Volunteer developer Volunteer tester Send message
Joined: 1 Mar 14 Posts: 1182 ID: 301928 Credit: 777,527,559 RAC: 0
                              
|
|
Current SGS rate: 263000 tasks / day
Average number of tasks per 1000G: 1238000
Range left to test: 8700G-10000G.
Thus, 1238000/263000*13 = 61 day until end of sieve file (at current rate)
|
|
|
Davina   Send message
Joined: 13 Feb 12 Posts: 3485 ID: 130544 Credit: 2,886,089,950 RAC: 574,291
                                   
|
|
So are you saying it could end before 15 Dec? |
|
|
streamVolunteer moderator Project administrator Volunteer developer Volunteer tester Send message
Joined: 1 Mar 14 Posts: 1182 ID: 301928 Credit: 777,527,559 RAC: 0
                              
|
So are you saying it could end before 15 Dec?
Yes, at current rate we will run out of work before 15 Dec.
On the other hand, crunching rate is unpredictable. People may reach their goals and leave project. Or jump into leaving train with full power.
|
|
|
|
|
|
SGS no longer runs on X86 machines? |
|
|
streamVolunteer moderator Project administrator Volunteer developer Volunteer tester Send message
Joined: 1 Mar 14 Posts: 1182 ID: 301928 Credit: 777,527,559 RAC: 0
                              
|
SGS no longer runs on X86 machines?
I've installed 32-bit Windows version on SGS (and updated version on other application).
|
|
|
|
|
|
Thank you for allowing SGS x86 tasks to run.
Works fine!!!! |
|
|
|
|
|
If your hardware supports x64, but your OS (operating system), in your case Windows 10
Professional x86 Edition, does not, then will the hardware still be fully utilized? /JeppeSN |
|
|
Michael Goetz Volunteer moderator Project administrator
 Send message
Joined: 21 Jan 10 Posts: 14692 ID: 53948 Credit: 1,042,938,327 RAC: 12,726
                                           
|
If your hardware supports x64, but your OS (operating system), in your case Windows 10
Professional x86 Edition, does not, then will the hardware still be fully utilized? /JeppeSN
It shouldn’t make that much of a difference. In this context, 64 bits refers to memory addresses, not arithmetic ability. 64 bit math is supported in “32 bit” operating systems and existed in 32 bit x86 CPUS.
The ability to address more than 4 GB of memory by LLR isn’t needed.
____________
My lucky number is 75898524288+1 |
|
|
|
|
If your hardware supports x64, but your OS (operating system), in your case Windows 10
Professional x86 Edition, does not, then will the hardware still be fully utilized? /JeppeSN
x64 instructions cannot be executed on Windows 32-bit.
The number of AVX/SSE registers is 8 in x86 mode and 16 in x64 mode then 32-bit code is significantly slower.
The butterfly of a radix-4 FFT is 4 complex numbers and 2 or 3 twiddle factors, i.e. 12 or 14 floating-point numbers. |
|
|
Michael Goetz Volunteer moderator Project administrator
 Send message
Joined: 21 Jan 10 Posts: 14692 ID: 53948 Credit: 1,042,938,327 RAC: 12,726
                                           
|
If your hardware supports x64, but your OS (operating system), in your case Windows 10
Professional x86 Edition, does not, then will the hardware still be fully utilized? /JeppeSN
x64 instructions cannot be executed on Windows 32-bit.
The number of AVX/SSE registers is 8 in x86 mode and 16 in x64 mode then 32-bit code is significantly slower.
The butterfly of a radix-4 FFT is 4 complex numbers and 2 or 3 twiddle factors, i.e. 12 or 14 floating-point numbers.
Oops. Right, I forgot that you lose the register extensions too. The regular 64 bit math will work, but not the vector instructions specific to the x64 extensions.
____________
My lucky number is 75898524288+1 |
|
|
|
|
|
Allen Paschke, it seems you should consider installing an x64 OS on your old machine. /JeppeSN |
|
|
|
|
|
Question: Just found my first LLR2 prime (was surprised at the drop in find rate), but the compute time is just as long as all my non-primes instead of doubled. Are we no longer doing the extra prime test once a discovery has been made?
____________
Eating more cheese on Thursdays. |
|
|
Vato Volunteer tester
 Send message
Joined: 2 Feb 08 Posts: 989 ID: 18447 Credit: 2,389,235,210 RAC: 462,824
                                      
|
|
correct, all the additional prime checks are now done on the server
____________
|
|
|
|
|
|
Ah, okay, cool beans!
____________
Eating more cheese on Thursdays. |
|
|
|
|
|
Also note that since this project works with c = -1 candidates, and since we want to have Gerbicz hardware error checks with LLR2, we do a Fermat PRP test. That is the reason you see:
Gerbicz check is requested, switching to PRP.
Starting probable prime test of [...]
So when you found a prime, in fact initially it was known only that it was a PRP. So the server would now also do a deterministic N+1 test of the same number you found. And also check for twin and Sophie Germain as always.
/JeppeSN |
|
|
Conan   Send message
Joined: 24 Mar 09 Posts: 1179 ID: 37336 Credit: 296,684,190 RAC: 356,654
                         
|
SGS no longer runs on X86 machines?
Yes this is now correct as my 32 bit XP machine which has been running SGS fine can no longer run it.
It appears that I can no longer run any Primegrid, all GFN and PPSE work fails as well.
There are still few other projects I can run so will set Primegrid to no new work.
Conan
____________
|
|
|
|
|
SGS no longer runs on X86 machines?
Yes this is now correct as my 32 bit XP machine which has been running SGS fine can no longer run it.
It appears that I can no longer run any Primegrid, all GFN and PPSE work fails as well.
There are still few other projects I can run so will set Primegrid to no new work.
Conan
Not quite accurate. As in an above post, a 32 bit binary does exist and works. The issue is XP, which is not capable of running any of the current apps.
PG stopped officially supporting XP several years ago and the OS has been out of MS support for a decade.
____________
Eating more cheese on Thursdays. |
|
|
Conan   Send message
Joined: 24 Mar 09 Posts: 1179 ID: 37336 Credit: 296,684,190 RAC: 356,654
                         
|
SGS no longer runs on X86 machines?
Yes this is now correct as my 32 bit XP machine which has been running SGS fine can no longer run it.
It appears that I can no longer run any Primegrid, all GFN and PPSE work fails as well.
There are still few other projects I can run so will set Primegrid to no new work.
Conan
Not quite accurate. As in an above post, a 32 bit binary does exist and works. The issue is XP, which is not capable of running any of the current apps.
PG stopped officially supporting XP several years ago and the OS has been out of MS support for a decade.
Thanks Grebuloner,
Yes I know about that, but things still worked until the recent update of the SGS app, so it was working only a few days ago.
All good I still have three Linux machines that are going strong.
The computer in question is an AMD Phenom 2 X4 machine that is at least 10 years old and is running XP because that is what it was installed with.
It still has things on it I can't do on my Linux machines so that is why it is still here.
I will see if I can do those things in Linux if I can I might update to 64 bit Linux and problem goes away.
Conan
____________
|
|
|
|
|
|
Even updating to 32-bit (x86) Windows Vista should be sufficient to get something to run. But a 64-bit operating system should be better! /JeppeSN |
|
|
|
|
The computer in question is an AMD Phenom 2 X4 machine that is at least 10 years old and is running XP because that is what it was installed with.
Virtual machines support 64-bit guest OS on 32-bit host OS.
|
|
|