Hi.

I've not received any Asteroids@home tasks for the past six days. Every time my BOINC client asks for them, the event log goes as per the latest attempt:

Tue 18 Feb 15:55:13 2014 | Asteroids@home | Sending scheduler request: To fetch work.
Tue 18 Feb 15:55:13 2014 | Asteroids@home | Requesting new tasks for CPU
Tue 18 Feb 15:55:18 2014 | Asteroids@home | Scheduler request completed: got 0 new tasks
Tue 18 Feb 15:55:18 2014 | Asteroids@home | No tasks sent
Tue 18 Feb 15:55:18 2014 | Asteroids@home | Tasks for NVIDIA GPU are available, but your preferences are set to not accept them

Mac OS 10.9.1
BOINC 7.2.39
Only CPU tasks requested

The only thing that's changed here since returning the last batch of results is the upgrade to BOINC 7.2.39 three days ago. No problems with other projects.

Are there really only GPU tasks available at the moment?

NG

That's a new 'feature' of BOINC 7.2.39 + Mac OS

Read my PM to Kyong (I may post his answer only if he agrees)

BOINC 7.2.39 + Mac OS users - issue?


For Mac OS you created:
102.11 (sse3_osx)
102.10 (avx_osx)

Now I see this change in BOINC 7.2.34
"Fix: Normalize the CPU feature list on Mac and Haiku with other platforms supported by BOINC. (Mac)"
http://boinc.berkeley.edu/trac/wiki/VersionHistory

BOINC 7.2.39 is now 'Recommended'
I wonder if this will be problem

BOINC 7.2.39 will report to the server 'pni' and not 'SSE3' (or maybe both? - which will be good)

Maybe you need to check the BOINC sources.
I did and I'm not sure if both will be reported (and all is converted to lowercase):
http://boinc.berkeley.edu/trac/browser/boinc-v2/client/hostinfo_unix.cpp

// Convert Mac CPU features string to match that returned by Linux
731	    for(p=features; *p; p++) {
732	        *p = tolower(*p);
733	    }
734	
735	    host.p_features[0] = 0;
736	    for (p = strtok(features, sep); p; p = strtok(NULL, sep)) {
737	    if (p != features) safe_strcat(host.p_features, sep);
738	        if (!strcmp(p, "avx1.0")) {
739	            safe_strcat(host.p_features, "avx");
740	        } else if (!strcmp(p, "sse3")) {
741	            safe_strcat(host.p_features, "pni");
742	        } else if (!strcmp(p, "sse4.1")) {
743	            safe_strcat(host.p_features, "sse4_1");
744	        } else if (!strcmp(p, "sse4.2")) {
745	            safe_strcat(host.p_features, "sse4_2");
746	        } else {
747	            safe_strcat(host.p_features, p);
748	        }
749	    }

***

At the time I reported the issue to David Anderson but I think it was supposed to be fixed server-side and not on the client.

My PM to David Anderson on SETI@home site:
http://setiathome.berkeley.edu/view_profile.php?userid=432107

432107 (David Anderson)

BOINC server: Predefined <plan_class> for SSE3 do not work for Mac OSX


Problem:

BOINC client: 'Processor features' list is generated by very different means depending on OS:
http://asteroidsathome.net/boinc/forum_thread.php?id=93&postid=2083#2083

On Linux and Win strings are: pni and avx
Mac OSX sysctlbyname reports: SSE3 and AVX1.0
Example output:
http://stackoverflow.com/questions/14586258/gcc-doesnt-want-to-use-avx-on-mac

Problem noted here:
http://asteroidsathome.net/boinc/forum_thread.php?id=219&postid=2138#2138

Your note here:
http://boinc.berkeley.edu/trac/browser/boinc-v2/sched/plan_class_spec.xml.sample

scheduler: sample code for the SSE3 plan class must check for "pni" rather than "sse3"; clients report "pni"

---

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

My answer from PM, just corrected some typos:

Well, we'll see, what reports will be at forum. Without the hardware it is difficult to check it even just looking through source codes. I was even ready to disable all mac os apps because there was a new problem. We have many bad results which were computed by os x apps, but there is a problem which validator couldn't verify. Now it is fixed, so there are more invalid results from os x than before. But not all. For compilation was used the same source code which is for linux and win, so I don't know where is the problem and I can't find it out. I will try to check what will be reported at forum but solving os x problems have the minimum priority for now. We also wrote some e-mails to Apple for some hardware for scientific purposes but got no answers so they probably don't care about it if we got no responses from them. And buying appropriate hardware isn't cheap, so the os x apps are now with minimum priority.

If you're running OSX then you have effectively excluded yourself from many BOINC projects. One easy and very effective way for OSX users to get around that problem is to install VirtualBox and create a virtual Windows or Linux machine on OSX, then install BOINC in that virtual machine and crunch. Yes, the virtual machine has overhead and you probably can't crunch GPU tasks in a virtual machine unless you run Zen and your CPU has VT-d.

If you don't want to pay that overhead and if you run apps that run only on OSX and you cannot live without those apps then install Linux and VirtualBox then create a virtual OSX machine on Linux and run your must have OSX apps in that virtual OSX machine. Yes, that shifts overhead to the OSX apps but in the grander scheme it might be the most productive option. Another plus is that you could crunch GPU tasks this way.

---

BOINC FAQ Service
Official BOINC wiki
Installing BOINC on Linux

My answer from PM, just corrected some typos:

Well, we'll see, what reports will be at forum. Without the hardware it is difficult to check it even just looking through source codes. I was even ready to disable all mac os apps because there was a new problem. We have many bad results which were computed by os x apps, but there is a problem which validator couldn't verify. Now it is fixed, so there are more invalid results from os x than before. But not all. For compilation was used the same source code which is for linux and win, so I don't know where is the problem and I can't find it out. I will try to check what will be reported at forum but solving os x problems have the minimum priority for now. We also wrote some e-mails to Apple for some hardware for scientific purposes but got no answers so they probably don't care about it if we got no responses from them. And buying appropriate hardware isn't cheap, so the os x apps are now with minimum priority.

The problem is the OSX avx app (Intel 64-bit Mac OS 10.5 or later 102.10(avx_osx) ) ,which was compiled using llvm. I've send you a PM couple of weeks ago,that it needs to be removed! Somehow that didn't happen. I do have a new OSX avx app here that validated on my xeon 1240, however it's way slower than the sseX OSX app, so no need for a avx app on osx unless you wanna waste time!

---

Join BOINC United now!

Thank you, I completely forgot about your message. I have depreceated the avx application.

Thanks for the replies. It was indeed the sse3 tasks I was getting before, none of which ever gave bad results. The only problem I've ever had with Asteroids@home tasks was a small batch of bad sse2s sometime last year.

Dagorath wrote:

If you're running OSX then you have effectively excluded yourself from many BOINC projects.

An entertaining inversion. :) Running BOINC tasks is just an interesting secondary activity for me. It would never influence my choice of computer. While I do consider it a privilege to be able to contribute to scientific research in some small way, if a particular project doesn't support Mac OS, or if it regards Macs as more of an irritation than mobile phones, then it's a loss of crunchers to the project rather than any loss to me. My machines — even my old G5 — still get plenty of tasks from other projects.

That said, the paucity of BOINC projects with Mac OS applications is interesting. In scientific documentaries broadcast on UK television, most of the scientists are shown using MacBook Pros. And yet this prediliction for Macs isn't reflected in the BOINC world. :\

NG

And now if you add the same apps in duplicate sections with 'standard' <plan_class> 'sse3' == 'pni'

<plan_classes>
   <plan_class>
      <name>sse3</name>
      <cpu_feature>pni</cpu_feature>
   </plan_class>
</plan_classes>

... this will serve both users using older and newer BOINC versions on Mac OS.

So you'll have something like:

Mac OS 10.4 or later running on Intel	102.11 (sse3_osx)
Intel 64-bit Mac OS 10.5 or later	102.11 (sse3_osx)
Mac OS 10.4 or later running on Intel	102.11 (sse3)
Intel 64-bit Mac OS 10.5 or later	102.11 (sse3)

---

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

Nigel,

I share your sentiments regarding mobile phones vs. macs. I wouldn't dream of supporting phones but the thing is they're ubiquitous and easy to develop for due to free tools. If I understand correctly, one can buy OSX and run it on any x86 based computer, it doesn't have to have an ostentatious machined aluminum case to work and I hear it's reasonably priced compared to Windows. Maybe it's the compiler that is the tripping stone? I dunno but I do wish Mac support was better and I see it's not as bad here as we thought it was. I should think Kyong's deprecating the AVX version will get you crunching :-)

---

BOINC FAQ Service
Official BOINC wiki
Installing BOINC on Linux

Dagorath wrote:

If I understand correctly, one can buy OSX and run it on any x86 based computer

Hi Dagorath.

Last I heard, Apple itself wasn't keen on the idea! Looking around the Web quickly this morning, I see plenty of sources describing how to do it, but the machine used has to have compatible components and drivers. The current version of OS X is free, but you can only get it by buying a Mac with it already installed or by downloading it from Apple's App Store using a machine already running a recent version. Otherwise, you have to buy (or otherwise acquire) a DVD copy of a recent version (which will cost more than it did when new), perform the installation process (which looks pretty long-winded), then, if it works, register with the App Store and upgrade. I get the impression Kyong doesn't have the time to research, carry out, and trouble-shoot all this.

Maybe it's the compiler that is the tripping stone?

The Mac OS developer tool package, Xcode, is also free — but again, only as an option from one of the older installer DVDs or by downloading the latest version from the App Store.

I should think Kyong's deprecating the AVX version will get you crunching :-)

It hasn't made the slightest difference. It's sse3 tasks I'm no longer getting.

NG

I should think Kyong's deprecating the AVX version will get you crunching :-)

It hasn't made the slightest difference. It's sse3 tasks I'm no longer getting.

Of course.
"deprecating the AVX version" is only to stop errors.

What I posted in Message 2555 need to be done
(If Kyong read it and have time to do the addition/edit. Which file on the server need to be edited I don't know but it seems an easy edit - copy the section and change 'sse3_osx' to 'sse3' in the duplicate)

---

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

Thanks for the info, Nigel. I didn't know Jobs was that stingy and sneaky. I guess the Asteroids team hasn't time for that but with Crunch3r assisting them it's only a matter of time and a few tweaks to the app distribution system (it's a tad complicated here) before you get what you need.

---

BOINC FAQ Service
Official BOINC wiki
Installing BOINC on Linux

I guess the Asteroids team hasn't time for that but with Crunch3r assisting them it's only a matter of time and a few tweaks to the app distribution system (it's a tad complicated here) before you get what you need.

Hopefully so. It's not an OS X problem anyway but, as BilBg's been saying, a BOINC one.

NG

I allowed apps with pni detection, so now you should be able to obtain tasks pro mac os with BOINC <=> 7.2.39.

Thanks, Kyong. Unfortunately, that hasn't made any difference either. :(

Sat 22 Feb 07:20:47 2014 | Asteroids@home | Sending scheduler request: To fetch work.
Sat 22 Feb 07:20:47 2014 | Asteroids@home | Requesting new tasks for CPU
Sat 22 Feb 07:20:59 2014 | Asteroids@home | Scheduler request completed: got 0 new tasks
Sat 22 Feb 07:20:59 2014 | Asteroids@home | No tasks sent
Sat 22 Feb 07:20:59 2014 | Asteroids@home | Tasks for NVIDIA GPU are available, but your preferences are set to not accept them

NG

Thanks, Kyong. Unfortunately, that hasn't made any difference either. :(

Just to let you know I successfully picked up an Asteroids@home task this morning. Its estimated completion time is over 111 hours, but I'm sure that'll settle down. :)

Thanks again.

NG

Edit: Ah. I see I got a new application too.

Edit: Ah. I see I got a new application too.

You mean you have new name for the application (Period Search Application v102.11 (sse3))
or/and you have new executable in Asteroids@home directory?:
<BOINC_Data>\projects\asteroidsathome.net_boinc\

If 'new' executable (with different name): I'm sure if you do a byte-compare of the 'old' and the 'new' executables they are the same.

http://asteroidsathome.net/boinc/download/

Probably you mean one of these two couples (look at the bottom of page/list):

period_search_10211_i686-apple-darwin__sse3		12-Nov-2013 16:32	385K	 
period_search_10211_i686-apple-darwin__sse3_osx		01-Dec-2013 11:19	385K	 

period_search_10211_x86_64-apple-darwin__sse3		12-Nov-2013 16:32	344K	 
period_search_10211_x86_64-apple-darwin__sse3_osx	01-Dec-2013 11:19	344K	 

---

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

period_search_10211_x86_64-apple-darwin__sse3		12-Nov-2013 16:32	344K	 
period_search_10211_x86_64-apple-darwin__sse3_osx	01-Dec-2013 11:19	344K	 

You're right. The two apps have exactly the same number of bytes. But it's rounded to 353K in the Finder display because Apple now defines a K as 1000 bytes, not 1024.

By the way, that first task finished in 4 hours 21 minutes (not 111 hours) and I've now received another.

NG

You're right. The two apps have exactly the same number of bytes.

By "do a byte-compare" I mean "compare the files by content"

I downloaded and compared the files and they are the same (identical) by content.
You can also compare the MD5
b3d4f38f549945937d6115310d26a61f *period_search_10211_x86_64-apple-darwin__sse3
b3d4f38f549945937d6115310d26a61f *period_search_10211_x86_64-apple-darwin__sse3_osx


I don't know how to do "compare by content" on Mac OS, maybe this?:
http://www.reedolsen.com/mac-os-x-file-comparison-software-no-longer-needed/

---

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

I don't know how to do "compare by content" on Mac OS

My preferred weapon's AppleScript.

set aahPath to (path to application support from local domain as text) & "BOINC Data:Projects:asteroidsathome.net_boinc:"
set data1 to (read file (aahPath & "period_search_10211_x86_64-apple-darwin__sse3") as data)
set data2 to (read file (aahPath & "period_search_10211_x86_64-apple-darwin__sse3_osx") as data)

return (data1 = data2) --> true

NG