Merry Christmas & I wish Santa would bring me some Astropulse work for Christmas.

please load some tapes before you guys leave for Christmas..

---

Lets all home so, because I won't make it till Thursday.

---

Merry Christmas & I wish Santa would bring me some Astropulse work for Christmas.

please load some tapes before you guys leave for Christmas..

We just had a rather nice AP run........
Methinks we have a lot of MB to run out before we get any more.

---

"Freedom is just Chaos, with better lighting." Alan Dean Foster

Merry Xmas to all.

As for AP's, I may just configure my 4 GPU's to crunch them as well so there will be even less to go around when they're produced.

Cheers.

Merry Xmas to all.

As for AP's, I may just configure my 4 GPU's to crunch them as well so there will be even less to go around when they're produced.

Cheers.

Well, that's the crux of the matter.......
I have all rigs configured to maximize AP production.
That means disabling CPU cores to feed the GPUs when AP can be had.
The only problem with that is that when AP is gone, I am running idle cores.

So, when MB is all that can be had, I am leaving blood on the table. A lot of it, across 9 rigs. But it would be a bugger to have to change over every rig from 2 to 3 cores on the CPU when I would estimate that AP would not be available for a few days.

I just don't have that much time.

---

"Freedom is just Chaos, with better lighting." Alan Dean Foster

Merry Xmas to all.

As for AP's, I may just configure my 4 GPU's to crunch them as well so there will be even less to go around when they're produced.

Cheers.

Well, that's the crux of the matter.......
I have all rigs configured to maximize AP production.
That means disabling CPU cores to feed the GPUs when AP can be had.
The only problem with that is that when AP is gone, I am running idle cores.

So, when MB is all that can be had, I am leaving blood on the table. A lot of it, across 9 rigs. But it would be a bugger to have to change over every rig from 2 to 3 cores on the CPU when I would estimate that AP would not be available for a few days.

I just don't have that much time.

I'm not greedy so I only do AP's on CPU's when I can get them and let my GPU's do MB's, but someone has to do them MB's or no more AP's. ;-)

[Edit] Maybe AP's should just be made CPU only and that will stop a lot of bitching about them. [/edit]

Cheers.

Merry Xmas to all.

As for AP's, I may just configure my 4 GPU's to crunch them as well so there will be even less to go around when they're produced.

Cheers.

Well, that's the crux of the matter.......
I have all rigs configured to maximize AP production.
That means disabling CPU cores to feed the GPUs when AP can be had.
The only problem with that is that when AP is gone, I am running idle cores.

So, when MB is all that can be had, I am leaving blood on the table. A lot of it, across 9 rigs. But it would be a bugger to have to change over every rig from 2 to 3 cores on the CPU when I would estimate that AP would not be available for a few days.

I just don't have that much time.

I'm not greedy so I only do AP's on CPU's when I can get them and let my GPU's do MB's, but someone has to do them MB's or no more AP's. ;-)

Cheers.

Who's greedy? I don't play the games that some do.
I take whatever the servers send me.
They send it, I crunch it.
I like AP, because it seems to work on my rigs the best, so I have them configured to crunch it the best.
I don't use anything to shift work around. I really don't have time or the ambition for it. I let the rigs run, find a stray GPU that has downclocked, and let 'em rip.

I'll take all the AP work that the servers will send me, but beyond that, the kitties are on their own.

---

"Freedom is just Chaos, with better lighting." Alan Dean Foster

But Mark, it is those that play silly games with AP's that really make it necessary to look into some sort of limiting factor with them.

Now if I set my GPU's to do them then there would be even less to go around and more would be forced to crunch those MB's to clear those files quicker.

If everyone played fair I wouldn't write what I did. ;-)

Cheers.

Some of us have not had stock AP GPU work (opencl_ati_100) since late November.

---

Some of us have not had stock AP GPU work (opencl_ati_100) since late November.

I`m wondering why you dont use the Lunatics installer.

---

With each crime and every kindness we birth our future.

Mark - Use app_info.xml to TEMPORARILY use CPU cores for AP on GPU and they won't be idle when you don't have APs for GPU. It will do that if you assign 1 cpu core to AP on GPU.

---

Yes, and no...
Yes every 50Gb "tape" contains the data that is split in two different ways.
Multi-Beam (MB) are smaller sections of the data, with a fairly large overlap.
Asto-Pulse (AP) are larger sections of the data, with a fairly small overlap.
The AP splitters have less work to do, and produce larger files so they get through a tape much quicker than the MB splitters.
There is a difference in the way that any known earth-sourced interference is handled, this is mostly due to the way the AP and MB processes we run work.

---

Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

One CPU-thread reserved for one GPU AP OpenCL app:

<avg_ncpus>1.0</avg_ncpus>
<max_ncpus>1.0</max_ncpus>

Example:

    <app>
        <name>astropulse_v6</name>
    </app>
    <file_info>
        <name>AP6_win_x86_SSE2_OpenCL_NV_r1843.exe</name>
        <executable/>
    </file_info>
    <file_info>
        <name>libfftw3f-3.dll</name>
        <executable/>
    </file_info>
    <file_info>
        <name>ap_cmdline_win_x86_SSE2_OpenCL_NV.txt</name>
    </file_info>
    <file_info>
        <name>AstroPulse_Kernels_r1843.cl</name>
    </file_info>
    <app_version>
        <app_name>astropulse_v6</app_name>
        <version_num>604</version_num>
        <flops> bla </flops>
        <avg_ncpus>1.0</avg_ncpus>
        <max_ncpus>1.0</max_ncpus>
        <plan_class>opencl_nvidia_100</plan_class>
 	<cmdline> bla bla </cmdline>
        <coproc>
            <type>CUDA</type>
            <count> bla </count>
        </coproc>
        <file_ref>
            <file_name>AP6_win_x86_SSE2_OpenCL_NV_r1843.exe</file_name>
            <main_program/>
        </file_ref>
        <file_ref>
            <file_name>libfftw3f-3.dll</file_name>
        </file_ref>
        <file_ref>
            <file_name>ap_cmdline_win_x86_SSE2_OpenCL_NV.txt</file_name>
            <open_name>ap_cmdline.txt</open_name>
        </file_ref>
        <file_ref>
            <file_name>AstroPulse_Kernels_r1843.cl</file_name>
        </file_ref>
    </app_version>

For 'bla' take your own numbers/settings.

If no GPU AP OpenCL app/task, CPU-thread crunch other tasks.

Or use 0.5 or 0.34 (it must be slightly larger than/or= 1.0), depend of how much GPU/app/s or CPU-threads your PC have.

* Best regards! :-) * Philip J. Fry, team seti.international founder. * Optimize your PC for higher RAC. * SETI@home needs your help. *

---

.......
I have all rigs configured to maximize AP production.
That means disabling CPU cores to feed the GPUs when AP can be had.
The only problem with that is that when AP is gone, I am running idle cores.
....

create an app_config.xml file in your C:\ProgramData\BOINC\projects\setiathome.berkeley.edu directory like this:

<app_config>
<app>
<name>setiathome_v7</name>
<gpu_versions>
<gpu_usage>0.5</gpu_usage>
<cpu_usage>0.5</cpu_usage>
</gpu_versions>
</app>

<app>
<name>setiathome_enhanced</name>
<max_concurrent>6</max_concurrent>
<gpu_versions>
<gpu_usage>0.5</gpu_usage>
<cpu_usage>0.5</cpu_usage>
</gpu_versions>
</app>

<app>
<name>astropulse_v6</name>
<max_concurrent>6</max_concurrent>
<gpu_versions>
<gpu_usage>.50</gpu_usage>
<cpu_usage>1.0</cpu_usage>
</gpu_versions>
</app>

</app_config>

Then you will not have use computing preferences. the app_config.xml file overrides the settings in app_info.xml so when you upgrade lunatic's apps you don't need to edit this stuff all over again, and it is smaller.

---

Oh, it's much simpler than some of you have made out. I would only have to go to each rig and modify the local setting of how much CPU to use. They are all set to spare CPU power to assist the GPUs for AP. You can find the proof in the pudding in my RAC zooming up in the last 48 hours after the recent AP binge.
It's just the marching though 9 rigs to do it. And then setting it back when AP comes back online.

Just call me an AP kitty.

MeowAP.

---

"Freedom is just Chaos, with better lighting." Alan Dean Foster

you can set <cpu_usage> for astropulse on the gpu's to what ever you want (perhaps 2.0 would suit you), it is a leave it forget it setting. no manual adjusting like you are doing. unless i don't understand you.

---

.......
I have all rigs configured to maximize AP production.
That means disabling CPU cores to feed the GPUs when AP can be had.
The only problem with that is that when AP is gone, I am running idle cores.
....

create an app_config.xml file in your C:\ProgramData\BOINC\projects\setiathome.berkeley.edu directory like this:

<app_config>
<app>
<name>setiathome_v7</name>
<gpu_versions>
<gpu_usage>0.5</gpu_usage>
<cpu_usage>0.5</cpu_usage>
</gpu_versions>
</app>

<app>
<name>setiathome_enhanced</name>
<max_concurrent>6</max_concurrent>
<gpu_versions>
<gpu_usage>0.5</gpu_usage>
<cpu_usage>0.5</cpu_usage>
</gpu_versions>
</app>

<app>
<name>astropulse_v6</name>
<max_concurrent>6</max_concurrent>
<gpu_versions>
<gpu_usage>.50</gpu_usage>
<cpu_usage>1.0</cpu_usage>
</gpu_versions>
</app>

</app_config>

Then you will not have use computing preferences. the app_config.xml file overrides the settings in app_info.xml so when you upgrade lunatic's apps you don't need to edit this stuff all over again, and it is smaller.

If Im reading this right, This will run 2 AP's on a GPU? Im only asking because Im not sure My GTX 550 Ti's are up to running two AP at once. My setting at the moment are for one MB and one AP to run on my Cards.

Please bear with me on some of my questions as Im am just learning the app_congig.xml tweaking.

---

[/quote]

Old James

I run 2 APs per gpu, and 2 idle cores to support the gpus.

When the APs run out, the rigs go down to 50-60% cpu usage because MB on the gpus does not require cpu support.

---

"Freedom is just Chaos, with better lighting." Alan Dean Foster