Yes, it happened again. Even though I manually verified that the .cl file was in each application directory, they didn't end up being found. I guess I have to look to see if changes have been made to update_versions.

8.16 soon.

---

Yes, it happened again. Even though I manually verified that the .cl file was in each application directory, they didn't end up being found. I guess I have to look to see if changes have been made to update_versions.

8.16 soon.

Hmm, why does it download both setiathome_8.16_windows_intelx86__opencl_intel_gpu_sah.exe, and the more "Main like/Anonymous" MB8_win_x86_SSSE3_OpenCL_Intel_r3486.exe, for the iGPU?

Same size for both.

The one being run now for me here in Beta, is of course setiathome_8.16_windows_intelx86__opencl_intel_gpu_sah.exe, but both were downloaded into the Beta folder. I reset Beta, but same happens again. Not that any ill effects seems to come from it, but anyhow....

---

WARNING!! "THIS IS A SIGNATURE", of the "IT MAY CHANGE AT ANY MOMENT" type. It may, or may not be considered insulting, all depending upon HOW SENSITIVE THE VIEWER IS, to certain inputs to/from the nervous system.

I added my gpu as stock. I I don't understand why the app restarts as much sa it does with the guppi tasks as it does. I'm a hardware builder, software escapes me sometimes. help me please?

I added my gpu as stock. I I don't understand why the app restarts as much sa it does with the guppi tasks as it does. I'm a hardware builder, software escapes me sometimes. help me please?

The iGPU restarts because you still have some older versions onboard (SETI@home v8 v8.14 (opencl_intel_gpu_sah). That one is a crappy version for the iGPU.

Abort all SETI@home v8 v8.14 (opencl_intel_gpu_sah), and your iGPU will start running SETI@home v8 v8.15 (opencl_intel_gpu_sah), and SETI@home v8 v8.16 (opencl_intel_gpu_sah)

You might as well abort SETI@home v8 v8.15 (opencl_intel_gpu_sah)too, since SETI@home v8 v8.16 (opencl_intel_gpu_sah) have just been released.

That is for your computer with the INTEL Intel(R) HD Graphics 4400 (1603MB) OpenCL: 1.2. Your other two computers with the Nvidia GPU's, should not be affected by the restarts.

---

WARNING!! "THIS IS A SIGNATURE", of the "IT MAY CHANGE AT ANY MOMENT" type. It may, or may not be considered insulting, all depending upon HOW SENSITIVE THE VIEWER IS, to certain inputs to/from the nervous system.

You might as well abort SETI@home v8 v8.15 (opencl_intel_gpu_sah)too, since SETI@home v8 v8.16 (opencl_intel_gpu_sah) have just been released.

Is v8.16 a different version, or is it just v8.15 with the .cl file included?

---

Grant
Darwin NT.

You might as well abort SETI@home v8 v8.15 (opencl_intel_gpu_sah)too, since SETI@home v8 v8.16 (opencl_intel_gpu_sah) have just been released.

Is v8.16 a different version, or is it just v8.15 with the .cl file included?

It's still r3486.

---

WARNING!! "THIS IS A SIGNATURE", of the "IT MAY CHANGE AT ANY MOMENT" type. It may, or may not be considered insulting, all depending upon HOW SENSITIVE THE VIEWER IS, to certain inputs to/from the nervous system.

It's just 8.15 with the CL file, and in the case of the opencl_nvidia_SoG you (due to another script bug) get a copy of the MB8_win_x86_SSSE3_OpenCL_Intel_r3486.exe file but it will not be run.

---

You might as well abort SETI@home v8 v8.15 (opencl_intel_gpu_sah)too, since SETI@home v8 v8.16 (opencl_intel_gpu_sah) have just been released.

Is v8.16 a different version, or is it just v8.15 with the .cl file included?

It's still r3486.

Thanks.
Might as well let them run then.

---

Grant
Darwin NT.

It's just 8.15 with the CL file, and in the case of the opencl_nvidia_SoG you (due to another script bug) get a copy of the MB8_win_x86_SSSE3_OpenCL_Intel_r3486.exe file but it will not be run.

Thanks.
Will let the current ones run out then.

---

Grant
Darwin NT.

Await new version.

Hello Raistmer,

Thank for looking into high CPU usage with -use_sleep enabled.

Things have improved with 8.15/8.16: now at appr. 50% CPU usage (was close to 100% with 8.14).
However, with 8.12 is was closer to only 1% (same as CUDA tasks).

8.15/8.16 examples:

https://setiweb.ssl.berkeley.edu/beta/result.php?resultid=24297574
https://setiweb.ssl.berkeley.edu/beta/result.php?resultid=24302382
https://setiweb.ssl.berkeley.edu/beta/result.php?resultid=24302936
https://setiweb.ssl.berkeley.edu/beta/result.php?resultid=24298075

8.12 examples:

https://setiweb.ssl.berkeley.edu/beta/result.php?resultid=23935040
https://setiweb.ssl.berkeley.edu/beta/result.php?resultid=23934803
https://setiweb.ssl.berkeley.edu/beta/result.php?resultid=23935061
https://setiweb.ssl.berkeley.edu/beta/result.php?resultid=23934806

I have another question. From what I understand we are experimenting with 4-bit guppi tasks (as opposed to the regulard 2-bit tasks), but it is not possible to find out from the generated output (correct me if I am wrong). Would it be possible to add something to the output file that will show with what kind af task we are dealing ?

Thanks again,

Tom

5. @Zalster - tasks you posted are -v 6 ones.
Please do few runs with -sleep_quantum 3 for midAR VLAr, VHAR (w/o -v 6)

Will have to wait.

I'm away from those computers for the last few days, busy with a rebuild at another location and won't get back to that machine until late tomorrow.

I need to look at what was just released. Is it the same as your last revision?

If so then I will run the test with -sleep_quantum 3 if not, then I will try with the new build

Oh joy, update 149 of 224....Going to be a long day...

w with what kind af task we are dealing ?

Thanks again,

Tom

Besides few rare resends we are delaing with 4-bit only now. Absolutely no sense to send 2-bit currently.

---

News about SETI opt app releases: https://twitter.com/Raistmer

However, with 8.12 is was closer to only 1% (same as CUDA tasks).

and GPU performance with such low CPU usage for OpenCL app was pretty bad, not?
Now sleep more agile and changes with kernel time change.
And all this tunable. Look ReadMe for new options description. I think it's possible to get that ~1% after some tuning in that direction.

http://setiathome.berkeley.edu/forum_thread.php?id=79765&postid=1802005

---

News about SETI opt app releases: https://twitter.com/Raistmer

However, with 8.12 is was closer to only 1% (same as CUDA tasks).

and GPU performance with such low CPU usage for OpenCL app was pretty bad, not?

Hm, looking into sample results you provided actually not.
I'll propose some tuning line that would resemble 8.12 behavior for your GPU bit later.

---

News about SETI opt app releases: https://twitter.com/Raistmer

However, with 8.12 is was closer to only 1% (same as CUDA tasks).

and GPU performance with such low CPU usage for OpenCL app was pretty bad, not?

Hm, looking into sample results you provided actually not.
I'll propose some tuning line that would resemble 8.12 behavior for your GPU bit later.

For beginning try to add to your current tuning line these options (all others remain the same as currently):

-sleep_quantum 1 -tt 60

---

News about SETI opt app releases: https://twitter.com/Raistmer

"For all but iGPU only cosmetic change - no need to provide -use_sleep directly if some of options from sleep set defined."

I haven't got a clue, what he means....

it means that any of -use_sleep, -use_sleep_ex, -sleep_quantum, -sleep_margin will enable sleep so no need to provide 2 options instead of just one.

Does it mean that sleep is always on, or what???

sleep still governed by conditional use_sleep. ON by default only for low-end path.

And I think also that -high_perf, is automatic with the newer versions. I think Raistmer wrote something about that somewhere.

last revs implement some of high-perf path automatically where seems fit (basing on on-fly profiling). Few places where manual -high-perf option will change behavior still remain (also, this will override app's decisions so will switch path where app itself still thinks GPU not fast enough to make switch).

---

News about SETI opt app releases: https://twitter.com/Raistmer

However, with 8.12 is was closer to only 1% (same as CUDA tasks).

and GPU performance with such low CPU usage for OpenCL app was pretty bad, not?

Hm, looking into sample results you provided actually not.
I'll propose some tuning line that would resemble 8.12 behavior for your GPU bit later.

For beginning try to add to your current tuning line these options (all others remain the same as currently):

-sleep_quantum 1 -tt 60

Thanks, I will try. I'll post results later (probably tomorrow).

Tom

About to start testing v8.16


Notice 2 changes to standard app

new name
MB8_win_x86_SSSE3_OpenCL_Intel_r3486.exe

There is an extra S now

second was the name of the commandline text

mb_cmdline-8.16_windows_intel__opencl_nvidia_SoG.txt

Had to reset the project on my computer, once I get it all squared away I'll post result w/o -v6

About to start testing v8.16


Notice 2 changes to standard app

new name
MB8_win_x86_SSSE3_OpenCL_Intel_r3486.exe

There is an extra S now

second was the name of the commandline text

mb_cmdline-8.16_windows_intel__opencl_nvidia_SoG.txt

Had to reset the project on my computer, once I get it all squared away I'll post result w/o -v6

Hey, you're running the iGPU version on your Nvidia GPU. It is the iGPU version that is the SSSE3 version. SoG is SSE3´.

http://setiweb.ssl.berkeley.edu/beta/result.php?resultid=24316277

See:
Build features: SETI8 Non-graphics OpenCL USE_OPENCL_INTEL OCL_SYNCHED OCL_ZERO_COPY OCL_CHIRP3 FFTW USE_SSSE3 x86
CPUID: Intel(R) Core(TM) i7-5960X CPU @ 3.00GHz

The right one reads: Build features: SETI8 Non-graphics OpenCL USE_OPENCL_NV OCL_ZERO_COPY SIGNALS_ON_GPU OCL_CHIRP3 FFTW USE_SSE3 x86


MB8_win_x86_SSSE3_OpenCL_Intel_r3486.exe is not the SoG version, it's the iGPU version.


Eric wrote the wrong app name when he wrote: " in the case of the opencl_nvidia_SoG you (due to another script bug) get a copy of the MB8_win_x86_SSSE3_OpenCL_Intel_r3486.exe. It should have read setiathome_8.16_windows_intelx86__opencl_intel_gpu_sah.exe (due to another script bug) get a copy of the MB8_win_x86_SSSE3_OpenCL_Intel_r3486.exe.

---

WARNING!! "THIS IS A SIGNATURE", of the "IT MAY CHANGE AT ANY MOMENT" type. It may, or may not be considered insulting, all depending upon HOW SENSITIVE THE VIEWER IS, to certain inputs to/from the nervous system.

Ok, I'm not running any iGPU.

So, that is very interesting.

Thinking I should just download directly from Raistmer's mail site and amend from there.

Ok, this is going to take a while, lol