Punchy
I haven't compared points as I only have CPUs, but from discussions I have read in the past, won't GPUs outperform CPUs by a huge margin (like 10x or more)?  If that is true, it seems that we should keep our CPUs on CSG and POGS whilst using all the GPU power we have to attack PG.

This post at PG from a couple years ago when the credit for PPS Sieve was adjusted sums it up
 

Message 78306 - Posted: 25 Jul 2014 | 11:17:55 UTC - in response to Message 78304.

I'm confused as to why the Nvidia cuda PPS Sieve is fairly short but gives so much credit: it seems to be 3,371 per task! May I ask exactly why this is?

The GPU program is very efficient and utilizes the GPU very well. That's why it runs so quickly. The task itself is actually quite a large task.

If you were to run the very same task on a cpu core you wouldn't think it's a very short task. You might even be asking why its credit is so low and is only 3,371.

Were you to read through the forums, you would see that we lowered the credit on this sieve a couple of years ago, which was widely hailed throughout the user community as a fantastic move. (Okay, maybe not so much. "Angry mob with pitchforks" might be a better description.)

Bottom line is we're happy with where the credit is right now and think it's the correct level. Understand that there's no "perfect" credit formula -- but there are certainly ways of assigning credit that are far worse that what we use.
____________

Sorry gang not up much on this anymore, this is what we are going to run now?
Proth Prime Search Sieve (PPS-Sieve) on the GPU and not the CPU.

Will do, I am setting them up now!
I will run 4 980 and 2 290x cards
 
My 980 Settings:
app_config.xml
 
<app_config>
<app>
  <name>pps_sr2sieve</name>
  <max_concurrent>8</max_concurrent>
    <gpu_versions>
    <gpu_usage>0.5</gpu_usage>
    <cpu_usage>1.0</cpu_usage>
    </gpu_versions>
</app>
</app_config>
 
or
 
<app_config>
<app>
  <name>pps_sr2sieve</name>
    <gpu_versions>
    <gpu_usage>1.0</gpu_usage>
    <cpu_usage>1.0</cpu_usage>
    </gpu_versions>
</app>
</app_config>
 
 Some completed 980 Tasks Run Time @ 1500 MHz 13:40 Minutes running 2 Tasks per GPU.
Run time: 829.64 CPU time: 66.85 Credit 3,371.00
Run time: 831.70 CPU time: 67.25 Credit: 3,371.00
 
My 290X Settings:
Run time: 836.61  CPU time: 468.39  Credit: 3,371.00
 
app_info.xml Note the Name here for the AMD 290 or CAL Hawaii and above cards
 AMD Tasks Run Time @ 1080 MHz 13:45 Minutes running 2 Tasks per GPU. * May changes all to One Per GPU
 
<app_info>
 <app>
 <name>pps_sr2sieve</name>
 <user_friendly_name>Proth Prime Search (Sieve)</user_friendly_name>
 </app>
 <file_info>
 <name>primegrid_tpsieve_1.40_windows_intelx86__atiPPSsieve.exe</name>
 <executable/>
 </file_info>
 <app_version>
 <app_name>pps_sr2sieve</app_name>
 <version_num>140</version_num>
 <platform>windows_x86_64</platform>
 <avg_ncpus>1</avg_ncpus>
 <max_ncpus>1</max_ncpus>
 <flops>1.0e11</flops>
 <coproc>
 <type>ATI</type>
 <count>.5</count>
 </coproc>
 <cmdline></cmdline>
 <file_ref>
 <file_name>primegrid_tpsieve_1.40_windows_intelx86__atiPPSsieve.exe</file_name>
 <main_program/>
 </file_ref>
 </app_version>
 </app_info>
 
OR 
 
<app_info>
 <app>
 <name>pps_sr2sieve</name>
 <user_friendly_name>Proth Prime Search (Sieve)</user_friendly_name>
 </app>
 <file_info>
 <name>primegrid_tpsieve_1.40_windows_intelx86__atiPPSsieve.exe</name>
 <executable/>
 </file_info>
 <app_version>
 <app_name>pps_sr2sieve</app_name>
 <version_num>140</version_num>
 <platform>windows_x86_64</platform>
 <avg_ncpus>1</avg_ncpus>
 <max_ncpus>1</max_ncpus>
 <flops>1.0e11</flops>
 <coproc>
 <type>ATI</type>
 <count>1</count>
 </coproc>
 <cmdline></cmdline>
 <file_ref>
 <file_name>primegrid_tpsieve_1.40_windows_intelx86__atiPPSsieve.exe</file_name>
 <main_program/>
 </file_ref>
 </app_version>
 </app_info>
 

I really do not like bunkering, but if the team is doing it I will start after my current testing.

We moved to 15th place from 22nd place
https://www.seti-germany.de/boinc_pentathlon/statistiken/pentathlon.php

OK, I did get 400 tasks for the 980 and then PrimeGrid ran out of tasks.
OK, I did get 200 tasks for the AMD and then PrimeGrid ran out of tasks.

Opolis
Ok I'm in for PPS Sieve on GPU for our PrimeGrid effort.

Did you get any Tasks?

robbysites
So I have to fly out of here Friday and return Sunday night. Let me know the settings and when I should start bunkering given my schedule. Until further instructions I am running CSG and POGS.
Thanks my EVGA broheims!
rs

robby, given your sched, just start running pps sieve about 8PM tomorrow nite your time or sooner if you want to bunker.

I am crunching some now, 400 or so. Can start unloading tomorrow night (Wednesday) 8PM eastern time.
5 min on my GTX980Ti  10 min on my GTX970

Will release a batch of POGS and GFN17 tomorrow.

Is it time to dump for PrimeGrid?
Also https://www.seti-germany.de website is down.

We are in 13TH overall, not bad at all.

Bunkers released, also I have switched from GFN to Sieve.
Two 780Ti's has been added to the sieving. (They cannot run GFN's)

Orange_1050
Bunkers released, also I have switched from GFN to Sieve.
Two 780Ti's has been added to the sieving. (They cannot run GFN's)

What was the time to complete and points for the GFN(?) that you ran?

Having so many pogs waiting for validation (700+), and not getting CSG for some reason.
I am running SGS primegrid just to keep busy.
One SGS took more than twice the time the others did, maybe I found a prime.
 
I did 16 universe yesterday and turned them in to get a time/point idea, they are still waiting validation.
Maybe after this round of PG SGS I'll bunker some universe.