Hello MW@Home,

Please let me know if these runs are causing you errors. They are twice as thick as other wedges.

Thanks,
John Vickers

ps_sgr_210F_2s_1_4369_1244753324_0 - 45s on 4830 - 27,7 Credits
ps_sgr_226F_2s_1_5_1244752853_0 - 45s on 4830 - 27,7 Credits

ok so far, but dont call them big ....

Does thicker mean bigger and hence supposedly more credits? Not in this case. All up, so far, all good.

I presume he means thicker stripes, i.e. chunks of the sky; 5° rather than 2.5°?

Yes,

Thicker means that I've doubled the nu range (5 degrees instead of 2.5 degrees). I dont think this affects credit gain as that is based on the number of steps in each integral if I remember correctly.

That is a good point though-- I should have doubled the integral step size for that parameter and that would logically give you double credits.

I'll try to do that next time :P

Cheers,
John Vickers

Thicker means that I've doubled the nu range (5 degrees instead of 2.5 degrees). I dont think this affects credit gain as that is based on the number of steps in each integral if I remember correctly.

That's right. Credit is based on the amount of work needed to compute the integrals. So it is linear with every number of steps you have. It is proportional to
mu_steps * nu_steps * r_steps * convolve * (number_of_streams + 1)
with the last factor being only a rough guide (as the background integral is a bit different than a stream integral).

That is a good point though-- I should have doubled the integral step size for that parameter and that would logically give you double credits.

Actually the step size should stay the same, but for that you have to double the number of steps, if the range is twice as large ;)

I put a load of these WUs through on both CPUs and a GPU. All seem to run OK, report OK and be given credit. So, I assume no problems are showing ATM.

---

Go away, I was asleep

That is a good point though-- I should have doubled the integral step size for that parameter and that would logically give you double credits.

Actually the step size should stay the same, but for that you have to double the number of steps, if the range is twice as large ;)

@John Vickers:
Just seen you have done just that, 5 degrees nu range (instead of the 2.5 degrees of the common WUs), and now 320 nu steps instead of 160. Takes twice the time, gives twice the credit (55.5 credits) and halves the server load. Nice!

Its not even letting me get any WUs, on my main PC.

(I just woke up to it having finished all WUs and it didn't send them till I just hit update and now I can't get new ones to crunch..)

Edit: It finally downloaded some..looks to be mostly 210's and a few 208's. With about the same crunch time estimated.

ok so far

........... Works well for me, and the site is much more responsive.

Hi
I have just finished ps-sgr-230f-2s-2 worked ok but took 3 hours 8 minutes and 8 seconds.

most previously averaged about 58 minutes or 1 hour 40 minutes at most.
is the credits the same and is the normal time for projects now?
thanks

paul

Hi
I have just finished ps-sgr-230f-2s-2 worked ok but took 3 hours 8 minutes and 8 seconds.

most previously averaged about 58 minutes or 1 hour 40 minutes at most.
is the credits the same and is the normal time for projects now?
thanks

paul

These WUs are twice as large (and give twice the credit) as the longest running so far. But you could try one of the optimized apps to speed it up quite a bit by the use of the SSSE3 version for your Core2. It should finish such a WU within an hour.

But you could try one of the optimized apps to speed it up quite a bit by the use of the SSSE3 version for your Core2. It should finish such a WU within an hour.

But nothing there for Mac. From what I remember of the last time we had this discussion, the problem is that no one - volunteers or project admin - is willing to pay $600-odd for the Intel compiler. Still the same?

Pretty much nobody has the money for it.

Jedirock took a shot at it, but only managed about a 2% improvement since he used the same tools as Travis.

http://www.arkayn.us/milkyway/milkyway_0.18_mac.zip

---

Yeah, it's Jedirock's app I'm using.

Pretty much nobody has the money for it.

Jedirock took a shot at it, but only managed about a 2% improvement since he used the same tools as Travis.

http://www.arkayn.us/milkyway/milkyway_0.18_mac.zip

Unfortunately, even on OS X, GCC is not compiled for Intel specific processors, so it runs slower than the Intel compiler, which generates slower code for other processors and faster for Intel processors.

---

Hi further to my previous message about it taking over three hours. the latest 2 are taking over 4 hours.
re the message But you could try one of the optimized apps to speed it up quite a bit by the use of the SSSE3 version for your Core2.
what do I do when i download the file when i just ran it I got this message Can't open init data file - running in standalone mode
Running Milkyway@home version 0.19 by Gipsel
Couldn't find input file [astronomy_parameters.txt] to read astronomy parameters.
APP: error reading astronomy parameters: 1
so i must have done something wrong any advice

thanks Paul

All smiles here! Actually even getting more than one unit at a time occasionally!

I vote to make them 16x.
That should make for a more average length runtime. Maybe shoot for ~12 hours on CPU. I know I am going to get told NOOOOOOO! But I really want some longer units. Anyway, thanks for all the hard work. Never had a host achieve >20k rac before. Keep up the great work! Aloha

---

Put the files app_info.xml and astronomy_0.19_SSSE3.exe that you got in the zip file in this directory.....C:\Documents and Settings\All Users\Application Data\BOINC\projects\milkyway.cs.rpi.edu_milkyway see how that goes.

Edit: I forgot to say shut down BOINC copy the files start BOINC again

---