Pushed 3 through my HD4000 in my Mac Mini last night. They have been "inconclusive" thus far but that's better than some alternatives lol...

---

Hopefully this will be the last post prior to the release of the Stock Mac GPU Apps. Maybe this week? You can see the list of Apps here, http://setiweb.ssl.berkeley.edu/beta/apps.php

Mac OS X 	        7.00 	25 Mar 2013, 17:37:42 UTC       	2 GigaFLOPS
Mac OS X/Intel  	7.00 	25 Mar 2013, 17:37:42 UTC              79 GigaFLOPS
Mac OS X        	6.02 	3 Apr 2012, 16:53:28 UTC        	0 GigaFLOPS
Mac OS X/Intel  	6.02 	1 Apr 2012, 0:20:18 UTC         	0 GigaFLOPS
Mac OS X/64-bit Intel 	6.08    (opencl_ati_mac) 7 Apr 2014     	0 GigaFLOPS
Mac OS X        	7.00 	11 Jul 2014, 21:42:36 UTC       	0 GigaFLOPS
Mac OS X/64-bit Intel 	7.01    (sse3) 	29 Jul 2014     	       62 GigaFLOPS
Mac OS X/64-bit Intel 	7.04    (opencl_ati_mac) 	19 Sep 2014   235 GigaFLOPS
Mac OS X/64-bit Intel 	7.04    (opencl_intel_gpu_mac) 	19 Sep 2014    27 GigaFLOPS
Mac OS X/64-bit Intel 	7.04    (opencl_nvidia_mac) 	19 Sep 2014     0 GigaFLOPS

My Mac is still going, still running the GPU App that was released at Lunatics over a year ago;

ID: 6796479 #14 TBar  73,771.90  19,801,003  7.2.42  Intel(R) Xeon(R) CPU E5462 @ 2.80GHz  [3] AMD ATI Radeon Barts XT Prototype (1024MB) OpenCL: 1.2  Darwin 13.4.0

Did they fix the glitch with v7.04 NV? I turned off my Mac NV crunching after it started throwing out errors. I know Urs said they were going to release a work around but I haven't restarted crunching with the NV to see if it will work.

You know where to look;

Message 52512 - Posted: 20 Sep 2014, 21:16:34 UTC - in response to Message 52510.

Thanks for trying Zalster.
A new version of the AstroPulse OpenCL app for Mac OS X with NV GPUs that has a workaround for the "-w"-problem should be up in a few days.

Thought I would bump this back to the front page. There are now apps for Mac out here on Main. If you haven't tried them, please do. If you are currently running a Mac with Nvidia, please be aware there is a problem with the Astropulse 7.05 estimated time to complete. Many Mac users downloaded large cache of AstroPulse work units and they are almost all Erroring out with Exceeded Time to Complete messages. I've linked 2 threads, one from Raistmer's thread here in Number Crunching that TBar discusses it and how to fix it and one from over in Question and Answers but both are deal with the same issue and how to correct this. After 11 validated Astropulses, the times should go to a more reasonable time estimation.

http://setiathome.berkeley.edu/forum_thread.php?id=75831&postid=1585145

http://setiathome.berkeley.edu/forum_thread.php?id=75865


Happy Crunching...

Zalster

Looks as though we may be close to having a Mac AMD/ATI MB App. Seems as though the Volunteer Developer working on the project has a working App. Hopefully it will find it's way to Beta soon.
http://lunatics.kwsn.net/10-alternate-hardware-platforms/osx-multibeam-opencl-question.msg57788.html#msg57788

Have you seen the front page recently? http://setiathome.berkeley.edu/top_hosts.php
That's what happens when you can only run APs and the AP database blows up ;-)

It's a Hit!
Well, it's a hit in Yosemite...so far, Re: OSX Multibeam OpenCL
Seems to be working well on My Mac at Beta, Computer 63959.
Apparently it doesn't work very well with VLARs though, so VLARs are not sent to Macs at Beta...which is Good. Unfortunately Beta seems to have mostly VLARs at present. Oh well.

If someone wants to test the App at Beta, my working app_info.xml is;

<app_info>
    <app>
        <name>setiathome_v7</name>
    </app>
    <file_info>
        <name>seti_boinc-hd5xxx</name>
        <executable/>
    </file_info>
    <file_info>
        <name>MultiBeam_Kernels_r2760.cl</name>
    </file_info>
    <file_info>
        <name>MultiBeam_Kernels_additional_r2760.cl</name>
    </file_info>
    <file_info>
 	<name>mb_cmdline_mac_OpenCL_ATi_HD5.txt</name>
    </file_info>
    <app_version>
        <app_name>setiathome_v7</app_name>
        <platform>x86_64-apple-darwin</platform>
        <version_num>703</version_num>
        <plan_class>opencl_ati5_mac</plan_class>
        <avg_ncpus>0.04</avg_ncpus>
        <max_ncpus>0.1</max_ncpus>
        <coproc>
            <type>ATI</type>
            <count>1</count>
        </coproc>
        <file_ref>
            <file_name>seti_boinc-hd5xxx</file_name>
            <main_program/>
        </file_ref>
        <file_ref>
            <file_name>MultiBeam_Kernels_r2760.cl</file_name>
        </file_ref>
        <file_ref>
            <file_name>MultiBeam_Kernels_additional_r2760.cl</file_name>
        </file_ref>
        <file_ref>
            <file_name>mb_cmdline_mac_OpenCL_ATi_HD5.txt</file_name>
            <open_name>mb_cmdline.txt</open_name>
        </file_ref>
     </app_version>
</app_info>

You will have to make a file and name it mb_cmdline_mac_OpenCL_ATi_HD5.txt.
My mb_cmdline file reads;
-sbs 256 -oclfft_tune_gr 256 -period_iterations_num 24

Good Luck.

Gave up on Yosemite. Went back to ML.

It kept crashing on my MacBook Pro (the new one not the old one). Let me know if he gets one for ML.

At least we are making progress, lol

Hmmm, Yosemite seems to work for most people. Maybe it's a nVidia thing? When I first tested the Mac Apps over a year ago I decided AMD was the way to go ;-)

As far as I know, No One has tested the nVidia MB App, other than the Developer. You might give it a go...it might work.
Stranger Things Have Happened...

I haven't had any issues with Yosemite. It definitely runs much better on 16GB of RAM versus just 4GB of RAM, but that would probably apply to any other version of Mac OS as well.

Well I have it installed on my ML laptop, but still waiting for the Beta servers to send me any wu's. Just keeps telling me it is getting no new tasks. 3000 or so haven't been set yet so I dunno if something else is squirrely or what.

Chris

Thu Dec 4 09:38:32 2014 | | cc_config.xml not found - using defaults
Thu Dec 4 09:38:32 2014 | | Starting BOINC client version 7.2.42 for x86_64-apple-darwin
Thu Dec 4 09:38:32 2014 | | log flags: file_xfer, sched_ops, task
Thu Dec 4 09:38:32 2014 | | Libraries: libcurl/7.26.0 OpenSSL/1.0.1e zlib/1.2.5 c-ares/1.9.1
Thu Dec 4 09:38:32 2014 | | Data directory: /Library/Application Support/BOINC Data
Thu Dec 4 09:38:32 2014 | | OpenCL: AMD/ATI GPU 0: ATI Radeon HD 6750M (driver version 1.2 (Jun 10 2014 22:41:22), device version OpenCL 1.2, 512MB, 512MB available, 288 GFLOPS peak)
Thu Dec 4 09:38:32 2014 | | OpenCL CPU: Intel(R) Core(TM) i7-2675QM CPU @ 2.20GHz (OpenCL driver vendor: Apple, driver version 1.1, device version OpenCL 1.2)
Thu Dec 4 09:38:32 2014 | SETI@home | Found app_info.xml; using anonymous platform
Thu Dec 4 09:38:32 2014 | SETI@home Beta Test | Found app_info.xml; using anonymous platform
Thu Dec 4 09:38:32 2014 | | Host name: adamek-chris.ndc.nasa.gov
Thu Dec 4 09:38:32 2014 | | Processor: 8 GenuineIntel Intel(R) Core(TM) i7-2675QM CPU @ 2.20GHz [x86 Family 6 Model 42 Stepping 7]
Thu Dec 4 09:38:32 2014 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clfsh ds acpi mmx fxsr sse sse2 ss htt tm pbe pni pclmulqdq dtes64 mon dscpl vmx est tm2 ssse3 cx16 tpr pdcm sse4_1 sse4_2 x2apic popcnt aes pcid xsave osxsave tsctmr avx
Thu Dec 4 09:38:32 2014 | | OS: Mac OS X 10.9.4 (Darwin 13.3.0)
Thu Dec 4 09:38:32 2014 | | Memory: 8.00 GB physical, 568.40 GB virtual
Thu Dec 4 09:38:32 2014 | | Disk: 697.52 GB total, 568.15 GB free
Thu Dec 4 09:38:32 2014 | | Local time is UTC -6 hours
Thu Dec 4 09:38:32 2014 | | VirtualBox version: 4.2.12
Thu Dec 4 09:38:32 2014 | | A new version of BOINC is available. (7.4.26) <a href=http://boinc.berkeley.edu/download.php>Download</a>
Thu Dec 4 09:38:32 2014 | rosetta@home | URL http://boinc.bakerlab.org/rosetta/; Computer ID 1139640; resource share 100
Thu Dec 4 09:38:32 2014 | Collatz Conjecture | URL http://boinc.thesonntags.com/collatz/; Computer ID 65631; resource share 100
Thu Dec 4 09:38:32 2014 | Einstein@Home | URL http://einstein.phys.uwm.edu/; Computer ID 3008314; resource share 0
Thu Dec 4 09:38:32 2014 | SETI@home | URL http://setiathome.berkeley.edu/; Computer ID 7159696; resource share 5000
Thu Dec 4 09:38:32 2014 | SETI@home Beta Test | URL http://setiweb.ssl.berkeley.edu/beta/; Computer ID 59989; resource share 100
Thu Dec 4 09:38:32 2014 | SETI@home | General prefs: from SETI@home (last modified 25-Sep-2014 11:09:45)
Thu Dec 4 09:38:32 2014 | SETI@home | Computer location: work
Thu Dec 4 09:38:32 2014 | | General prefs: using separate prefs for work
Thu Dec 4 09:38:32 2014 | | Reading preferences override file
Thu Dec 4 09:38:32 2014 | | Preferences:
Thu Dec 4 09:38:32 2014 | | max memory usage when active: 4096.00MB
Thu Dec 4 09:38:32 2014 | | max memory usage when idle: 7372.80MB
Thu Dec 4 09:38:32 2014 | | max disk usage: 100.00GB
Thu Dec 4 09:38:32 2014 | | max CPUs used: 7
Thu Dec 4 09:38:32 2014 | | (to change preferences, visit a project web site or select Preferences in the Manager)
Thu Dec 4 09:38:32 2014 | | Not using a proxy
Thu Dec 4 09:38:34 2014 | SETI@home | Sending scheduler request: To fetch work.
Thu Dec 4 09:38:34 2014 | SETI@home | Requesting new tasks for CPU and ATI
Thu Dec 4 09:38:35 2014 | SETI@home | Scheduler request completed: got 0 new tasks
Thu Dec 4 09:38:35 2014 | SETI@home | Project has no tasks available
Thu Dec 4 09:38:37 2014 | SETI@home Beta Test | project resumed by user
Thu Dec 4 09:38:40 2014 | SETI@home Beta Test | update requested by user
Thu Dec 4 09:38:41 2014 | SETI@home Beta Test | Master file download succeeded
Thu Dec 4 09:38:46 2014 | SETI@home Beta Test | Sending scheduler request: Requested by user.
Thu Dec 4 09:38:46 2014 | SETI@home Beta Test | Requesting new tasks for ATI
Thu Dec 4 09:38:47 2014 | SETI@home Beta Test | Scheduler request completed: got 0 new tasks
Thu Dec 4 09:38:47 2014 | SETI@home Beta Test | No tasks sent
Thu Dec 4 09:38:47 2014 | SETI@home Beta Test | No tasks are available for AstroPulse v7
Thu Dec 4 09:38:47 2014 | SETI@home Beta Test | Tasks for CPU are available, but your preferences are set to not accept them
Thu Dec 4 09:38:47 2014 | SETI@home Beta Test | Tasks for NVIDIA GPU are available, but your preferences are set to not accept them
Thu Dec 4 09:38:57 2014 | SETI@home Beta Test | Sending scheduler request: To fetch work.
Thu Dec 4 09:38:57 2014 | SETI@home Beta Test | Requesting new tasks for ATI
Thu Dec 4 09:38:58 2014 | SETI@home Beta Test | Scheduler request completed: got 0 new tasks
Thu Dec 4 09:38:58 2014 | SETI@home Beta Test | No tasks sent
Thu Dec 4 09:38:58 2014 | SETI@home Beta Test | No tasks are available for AstroPulse v7
Thu Dec 4 09:38:58 2014 | SETI@home Beta Test | Tasks for CPU are available, but your preferences are set to not accept them
Thu Dec 4 09:38:58 2014 | SETI@home Beta Test | Tasks for NVIDIA GPU are available, but your preferences are set to not accept them
Thu Dec 4 09:39:26 2014 | SETI@home Beta Test | update requested by user
Thu Dec 4 09:39:29 2014 | SETI@home Beta Test | Sending scheduler request: Requested by user.
Thu Dec 4 09:39:29 2014 | SETI@home Beta Test | Requesting new tasks for ATI
Thu Dec 4 09:39:30 2014 | SETI@home Beta Test | Scheduler request completed: got 0 new tasks
Thu Dec 4 09:39:30 2014 | SETI@home Beta Test | No tasks sent
Thu Dec 4 09:39:30 2014 | SETI@home Beta Test | No tasks are available for AstroPulse v7
Thu Dec 4 09:39:30 2014 | SETI@home Beta Test | Tasks for CPU are available, but your preferences are set to not accept them
Thu Dec 4 09:39:30 2014 | SETI@home Beta Test | Tasks for NVIDIA GPU are available, but your preferences are set to not accept them
Thu Dec 4 09:39:40 2014 | SETI@home Beta Test | Sending scheduler request: To fetch work.
Thu Dec 4 09:39:40 2014 | SETI@home Beta Test | Requesting new tasks for ATI
Thu Dec 4 09:39:41 2014 | SETI@home Beta Test | Scheduler request completed: got 0 new tasks
Thu Dec 4 09:39:41 2014 | SETI@home Beta Test | No tasks sent
Thu Dec 4 09:39:41 2014 | SETI@home Beta Test | No tasks are available for AstroPulse v7
Thu Dec 4 09:39:41 2014 | SETI@home Beta Test | Tasks for CPU are available, but your preferences are set to not accept them
Thu Dec 4 09:39:41 2014 | SETI@home Beta Test | Tasks for NVIDIA GPU are available, but your preferences are set to not accept them
Thu Dec 4 09:40:22 2014 | SETI@home Beta Test | update requested by user
Thu Dec 4 09:40:26 2014 | SETI@home Beta Test | Sending scheduler request: Requested by user.
Thu Dec 4 09:40:26 2014 | SETI@home Beta Test | Requesting new tasks for ATI
Thu Dec 4 09:40:27 2014 | SETI@home Beta Test | Scheduler request completed: got 0 new tasks
Thu Dec 4 09:40:27 2014 | SETI@home Beta Test | No tasks sent
Thu Dec 4 09:40:27 2014 | SETI@home Beta Test | Tasks for CPU are available, but your preferences are set to not accept them
Thu Dec 4 09:40:27 2014 | SETI@home Beta Test | Tasks for NVIDIA GPU are available, but your preferences are set to not accept them
Thu Dec 4 09:40:27 2014 | SETI@home Beta Test | Tasks for Intel GPU are available, but your preferences are set to not accept them
Thu Dec 4 09:40:37 2014 | SETI@home Beta Test | Sending scheduler request: To fetch work.
Thu Dec 4 09:40:37 2014 | SETI@home Beta Test | Requesting new tasks for ATI
Thu Dec 4 09:40:38 2014 | SETI@home Beta Test | Scheduler request completed: got 0 new tasks
Thu Dec 4 09:40:38 2014 | SETI@home Beta Test | No tasks sent
Thu Dec 4 09:40:38 2014 | SETI@home Beta Test | Tasks for CPU are available, but your preferences are set to not accept them
Thu Dec 4 09:40:38 2014 | SETI@home Beta Test | Tasks for NVIDIA GPU are available, but your preferences are set to not accept them
Thu Dec 4 09:40:38 2014 | SETI@home Beta Test | Tasks for Intel GPU are available, but your preferences are set to not accept them

---

Well I got some, 2 have finished so far. Doesn't look like they errored so far, but waiting to see if they validated.

Chris

http://setiweb.ssl.berkeley.edu/beta/results.php?hostid=59989&offset=0&show_names=0&state=2&appid=8

---

Your first Valid was against my Mac...

Seems yours is taking much less CPU time. For some reason mine are taking a large amount of CPU time, especially with the longer tasks. The Beta server is sending out VLARs most of the time so there could be long periods where none are sent to the Macs. I'm running the new MB App r2760 on my Vista host and it is receiving GPU VLARs. When the Vista host is receiving VLARs my Mac goes dry. The new r2760 handles VLARs very well in Vista, too bad the Mac version doesn't. I tried a VLAR on my ATI 6870 in the Mac and it took over 2 hours, the 6970 in Vista does it in 25 minutes.

So, it seems the current Mac version works in Mavericks as well as Yosemite...nice.

Yeah I totally plagiarized all your settings in the mb config file and app_info so I'm not sure why mine would be using so much less CPU. I'm quite pleased to see the GPU finally being utilized for multi beam.=) I'll try it out on the iGPU in one of my other macs in the next day or so. I may end up needing some help getting the app_info formatted correctly for it as I've not seem many examples of that floating around. Something for the machines to do at least while we wait for wu's over here.

Thanks,

Chris

---

My attempt at a app_info for my iGPU mac doesn't seem to be working. I get an odd statement in my event log:

Thu Dec 4 17:47:52 2014 | SETI@home Beta Test | Found app_info.xml; using anonymous platform
Thu Dec 4 17:47:52 2014 | | app version refers to missing GPU type INTEL
Thu Dec 4 17:47:52 2014 | | app version refers to missing GPU type INTEL

So I suspect I have the GPU type wrong, but am unsure where to find the correct plan_class and type... Suggestions would be appreciated!

Thanks,

Chris

<app_info>
<app>
<name>setiathome_v7</name>
</app>
<file_info>
<name>seti_boinc-intel</name>
<executable/>
</file_info>
<file_info>
<name>MultiBeam_Kernels_r2760.cl</name>
</file_info>
<file_info>
<name>MultiBeam_Kernels_additional_r2760.cl</name>
</file_info>
<file_info>
<name>mb_cmdline_mac_OpenCL_Intel.txt</name>
</file_info>
<app_version>
<app_name>setiathome_v7</app_name>
<platform>x86_64-apple-darwin</platform>
<version_num>703</version_num>
<plan_class>opencl_intel_mac</plan_class>
<avg_ncpus>0.04</avg_ncpus>
<max_ncpus>0.1</max_ncpus>
<coproc>
<type>INTEL</type>
<count>1</count>
</coproc>
<file_ref>
<file_name>seti_boinc-intel</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>MultiBeam_Kernels_r2760.cl</file_name>
</file_ref>
<file_ref>
<file_name>MultiBeam_Kernels_additional_r2760.cl</file_name>
</file_ref>
<file_ref>
<file_name>mb_cmdline_mac_OpenCL_Intel.txt</file_name>
<open_name>mb_cmdline.txt</open_name>
</file_ref>
</app_version>
</app_info>

---

Plan_class is at your own choice, but you will find it much easier if you follow the conventions set out on the stock applications page: the plan_class string is shown in brackets in the 'Version' column. Choose, or make up, a plan class that matches the shape of the application you're trying to deploy.

The coproc type is, strictly,

    <coproc>
        <type>intel_gpu</type>
        <count>1</count>
    </coproc>

Don't mess with that in any way - it's lower case, with _gpu.

My attempt at a app_info for my iGPU mac doesn't seem to be working. I get an odd statement in my event log:

Thu Dec 4 17:47:52 2014 | SETI@home Beta Test | Found app_info.xml; using anonymous platform
Thu Dec 4 17:47:52 2014 | | app version refers to missing GPU type INTEL
Thu Dec 4 17:47:52 2014 | | app version refers to missing GPU type INTEL

So I suspect I have the GPU type wrong, but am unsure where to find the correct plan_class and type... Suggestions would be appreciated!

Thanks,

Chris

<app_info>
<app>
<name>setiathome_v7</name>
</app>
<file_info>
<name>seti_boinc-intel</name>
<executable/>
</file_info>
<file_info>
<name>MultiBeam_Kernels_r2760.cl</name>
</file_info>
<file_info>
<name>MultiBeam_Kernels_additional_r2760.cl</name>
</file_info>
<file_info>
<name>mb_cmdline_mac_OpenCL_Intel.txt</name>
</file_info>
<app_version>
<app_name>setiathome_v7</app_name>
<platform>x86_64-apple-darwin</platform>
<version_num>703</version_num>
<plan_class>opencl_intel_mac</plan_class>
<avg_ncpus>0.04</avg_ncpus>
<max_ncpus>0.1</max_ncpus>
<coproc>
<type>INTEL</type>
<count>1</count>
</coproc>
<file_ref>
<file_name>seti_boinc-intel</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>MultiBeam_Kernels_r2760.cl</file_name>
</file_ref>
<file_ref>
<file_name>MultiBeam_Kernels_additional_r2760.cl</file_name>
</file_ref>
<file_ref>
<file_name>mb_cmdline_mac_OpenCL_Intel.txt</file_name>
<open_name>mb_cmdline.txt</open_name>
</file_ref>
</app_version>
</app_info>

From Lunatics_Win64_v0.43a_setup.exe

        <plan_class>opencl_intel_gpu_sah</plan_class>
        	<cmdline></cmdline>	
	<coproc>
            <type>intel_gpu</type>
            <count>1</count>
        </coproc>

---

 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 

My attempt at a app_info for my iGPU mac doesn't seem to be working...

Looks like you're now receiving some GPU Error. You might want to remove the cmdline settings and use the defaults. I'm not sure how those settings will work on the intel gpu;

<message>
process exited with code 255 (0xff, -1)
</message>
<stderr_txt>
Maximum single buffer size set to:256MB
oclFFT global radix override set to:256
...
ERROR: Possible wrong computation state on GPU, host needs reboot or maintenance...

Yup, as soon as I got those errors I removed those settings and it started to run, albeit not very fast. I've got two completed now, running about 4000-5000 seconds each. We'll see if they validate. I would have thought the wu's should be running a bit faster than that even on the igpu, but there may be some further tweaking to do if these validate properly.

Thanks,

Chris

---

I'm pretty sure the GPU load is Low. Normally when running near full load the Macs Fan Control shows the IO sensor at around 99F. It's been running around 91-93F running the ATI App. The longer tasks were taking around 48 minutes, much longer than in Linux on the same machine. A few hours ago I set it to run 2 at a time. Now the ones that were taking 48 minutes a piece are completing 2 in about an hour, and the Temp is still only 93F. I would go for 3 at a time but the CPU meter is bouncing between 40-100% CPU load just running the GPUs.

Well, it IS the first release. Not bad for a first release ;-)