Wall-clock time.

---

Wallclock, hmmm, yet to find the reasoning of why this change, but a strong downside is that it makes any slow running highly obscured, since few will actually drill into the job properties to find out the CPU time. At any case, last night on a quad/Vista with 6.6.15 client, I remotely looked in BOINCview and saw that 3 of 4 processes were running at ~32% CPU efficiency. Walking over and checking what else was active found that somehow 3 of the 4 sciences were running of 1 core, with 2 cores idle ?!? Whether BOINC allowed this or the OS did this is entirely outside my ability to diagnose. The story is, that showing elapsed time aka wallclock is very misleading.

The change arose out of CUDA development, where the opposite problem applies: showing the CPU time of an application running mainly on a GPU is likewise misleading.

The trouble is, the quick change (and it was an extremely quick, spur of the moment, decision, if you read the boinc_alpha mailing list) means that no consideration was given to the side-effects such as the one you've described.

Another consequence of the BOINC 'new feature' development model.

Wallclock, hmmm, yet to find the reasoning of why this change, but a strong downside is that it makes any slow running highly obscured, since few will actually drill into the job properties to find out the CPU time. At any case, last night on a quad/Vista with 6.6.15 client, I remotely looked in BOINCview and saw that 3 of 4 processes were running at ~32% CPU efficiency. Walking over and checking what else was active found that somehow 3 of the 4 sciences were running of 1 core, with 2 cores idle ?!? Whether BOINC allowed this or the OS did this is entirely outside my ability to diagnose. The story is, that showing elapsed time aka wallclock is very misleading.

Totally disagree - showing CPU time for CUDA tasks confuses many users. And although I have come to understand what is going on, I find it much more useful to have the "wallclock" time shown since I know most CUDA tasks take about 10 minutes on my rig so if there is significant variation from that I go investigate why. Note that the "wallclock" is also suspended when the tasks are suspended when one with an earlier deadline is downloaded so it is showing "wallclock" equivalent of the time the task is actually being crunched.

F.

The properties of the task will still show:
- CPU time at last checkpoint
- CPU time

Although that's a snapshot at the time you open that properties window. It won't update the timers in the window.

---

The thing is, even with CPU run tasks, the wall clock time and actual CPU time differ a great deal. For instance, one of my Seti tasks shows it ran for 3h 29m 00s wall-clock time, but actual time the CPU has taken on it is only 2h 59m 59s.

In my opinion it's best to always show the actual time as perceived by me that a task has run, not the time achieved by the CPU, as it'll never run 100% all of the time. Your computer is doing other things with the CPU(s) as well.

But I am sure that in the future this will change again. Perhaps two columns side-by-side, making doubly sure to confuse people. :-)

---

BOINCview is dead in development come next month 2 years. That interface has allot that could be adopted. Add a column for efficiency. That will signal people who are not throttling that something is not right and those that are throttling an indication that 60% is ~60%. 2 birds with 1 stone.

And also there's an awful lot of really useful design in BoincView which BOINC has failed to imitate in the two years since development stopped:

User-selectable columns (have both CPU and GPU available, let the user decide which one to hide/show - or keep both if their screen is large enough).
Totals/counts/averages for columns.
Filtering by project (recently introduced - in a less sophisticated way - for the messages tab only).

I have requested that a BOINC Manager be made with two columns, one showing CPU time Elapsed and one showing Wall-Clock Elapsed. Just as a test, to see if it clutters more. Won't know the outcome until they actually mail me back on that, if the mail didn't go to the "Oh it's only Jord, let's do this in 6-12 months time" pile. ;-)

---

Please revert and get the CUDA people to look at properties. They are the fastest minority!

In your opinion...

But OK, I just got word that there is a good possibility that the Elapsed column is going away. With nothing in return.

---

The Developers didn't want to build a BOINC Manager that shows both CPU time and Wall time, so I did it myself. Bow down to me. :-)

Go fetch it from here. It's a RAR file, so unRAR it. Then just add it to your BOINC directory and run it. It's called Boincmgr26.exe and is for 32bit Windows only (although it may happily run on x64 for all I know).

Since it comes without an upgraded boinc.dll it may not work on older versions of BOINC. So far it's been tested on 6.4.5 where it works without a problem. I am hesitant on also releasing boinc.dll as it may break older versions of BOINC.

Let me know what you think. ;-)

---

Go fetch it from here. It's a RAR file, so unRAR it. Then just add it to your BOINC directory and run it. It's called Boincmgr26.exe and is for 32bit Windows only (although it may happily run on x64 for all I know).

Let me know what you think. ;-)

I've built the 32 bit Linux version of this. It's simply called boincmgr. I have no idea if it'll run on anything but my machine. I've never built binaries for distribution. It was compiled with the --enable-unicode flag for wxWidgets. My suggestion is if you want to try it, download it, check the md5sum (4fa7a16a56dadb2e26a61a81b97591ca) and then run ldd on it.

The output of ldd is as follows:

[kathryn@Galaxy clientgui]$ ldd boincmgr                                   
        linux-gate.so.1 =>  (0x00793000)                                   
        libgtk-x11-2.0.so.0 => /usr/lib/libgtk-x11-2.0.so.0 (0x02ef6000)   
        libgdk-x11-2.0.so.0 => /usr/lib/libgdk-x11-2.0.so.0 (0x034a6000)   
        libatk-1.0.so.0 => /usr/lib/libatk-1.0.so.0 (0x004a6000)           
        libgdk_pixbuf-2.0.so.0 => /usr/lib/libgdk_pixbuf-2.0.so.0 (0x004c5000)
        libpango-1.0.so.0 => /usr/lib/libpango-1.0.so.0 (0x002c1000)          
        libgobject-2.0.so.0 => /lib/libgobject-2.0.so.0 (0x00311000)          
        libgmodule-2.0.so.0 => /lib/libgmodule-2.0.so.0 (0x0048a000)          
        libgthread-2.0.so.0 => /lib/libgthread-2.0.so.0 (0x00221000)          
        librt.so.1 => /lib/librt.so.1 (0x0080d000)                            
        libglib-2.0.so.0 => /lib/libglib-2.0.so.0 (0x00110000)                
        libXinerama.so.1 => /usr/lib/libXinerama.so.1 (0x00469000)            
        libSM.so.6 => /usr/lib/libSM.so.6 (0x00480000)                        
        libpng12.so.0 => /usr/lib/libpng12.so.0 (0x0040d000)                  
        libjpeg.so.62 => /usr/lib/libjpeg.so.62 (0x038d3000)                  
        libcurl.so.4 => /usr/lib/libcurl.so.4 (0x0526f000)                    
        libssl.so.7 => /lib/libssl.so.7 (0x00226000)                          
        libcrypto.so.7 => /lib/libcrypto.so.7 (0x0784d000)                    
        libdl.so.2 => /lib/libdl.so.2 (0x006a8000)                            
        libnsl.so.1 => /lib/libnsl.so.1 (0x02e62000)                          
        libz.so.1 => /lib/libz.so.1 (0x006f5000)                              
        libpthread.so.0 => /lib/libpthread.so.0 (0x006da000)                  
        libstdc++.so.6 => /usr/lib/libstdc++.so.6 (0x0095b000)                
        libm.so.6 => /lib/libm.so.6 (0x006af000)                              
        libgcc_s.so.1 => /lib/libgcc_s.so.1 (0x00840000)                      
        libc.so.6 => /lib/libc.so.6 (0x0053d000)                              
        libX11.so.6 => /usr/lib/libX11.so.6 (0x00850000)                      
        libpangocairo-1.0.so.0 => /usr/lib/libpangocairo-1.0.so.0 (0x00490000)
        libXcomposite.so.1 => /usr/lib/libXcomposite.so.1 (0x00353000)        
        libXfixes.so.3 => /usr/lib/libXfixes.so.3 (0x00396000)                
        libcairo.so.2 => /usr/lib/libcairo.so.2 (0x02573000)                  
        libfontconfig.so.1 => /usr/lib/libfontconfig.so.1 (0x00b81000)        
        libXext.so.6 => /usr/lib/libXext.so.6 (0x00384000)                    
        libXrender.so.1 => /usr/lib/libXrender.so.1 (0x00437000)              
        libXi.so.6 => /usr/lib/libXi.so.6 (0x00442000)                        
        libXrandr.so.2 => /usr/lib/libXrandr.so.2 (0x00308000)
        libXcursor.so.1 => /usr/lib/libXcursor.so.1 (0x0046e000)
        /lib/ld-linux.so.2 (0x0051d000)
        libICE.so.6 => /usr/lib/libICE.so.6 (0x00be6000)
        libuuid.so.1 => /lib/libuuid.so.1 (0x0047a000)
        libidn.so.11 => /lib/libidn.so.11 (0x02e7e000)
        libssh2.so.1 => /usr/lib/libssh2.so.1 (0x0039d000)
        libldap-2.4.so.2 => /usr/lib/libldap-2.4.so.2 (0x07ae5000)
        libgssapi_krb5.so.2 => /usr/lib/libgssapi_krb5.so.2 (0x001f2000)
        libkrb5.so.3 => /usr/lib/libkrb5.so.3 (0x079c5000)
        libk5crypto.so.3 => /usr/lib/libk5crypto.so.3 (0x00d3e000)
        libcom_err.so.2 => /lib/libcom_err.so.2 (0x00271000)
        libssl3.so => /lib/libssl3.so (0x00cce000)
        libsmime3.so => /lib/libsmime3.so (0x00274000)
        libnss3.so => /lib/libnss3.so (0x029cf000)
        libplds4.so => /lib/libplds4.so (0x0029b000)
        libplc4.so => /lib/libplc4.so (0x004ef000)
        libnspr4.so => /lib/libnspr4.so (0x02b97000)
        libresolv.so.2 => /lib/libresolv.so.2 (0x00b6a000)
        libxcb-xlib.so.0 => /usr/lib/libxcb-xlib.so.0 (0x00380000)
        libxcb.so.1 => /usr/lib/libxcb.so.1 (0x00362000)
        libpangoft2-1.0.so.0 => /usr/lib/libpangoft2-1.0.so.0 (0x003bf000)
        libfreetype.so.6 => /usr/lib/libfreetype.so.6 (0x00ad8000)
        libpixman-1.so.0 => /usr/lib/libpixman-1.so.0 (0x00709000)
        libexpat.so.1 => /lib/libexpat.so.1 (0x00a67000)
        libXau.so.6 => /usr/lib/libXau.so.6 (0x00819000)
        liblber-2.4.so.2 => /usr/lib/liblber-2.4.so.2 (0x00bc9000)
        libsasl2.so.2 => /usr/lib/libsasl2.so.2 (0x03c40000)
        libkrb5support.so.0 => /usr/lib/libkrb5support.so.0 (0x00d33000)
        libkeyutils.so.1 => /lib/libkeyutils.so.1 (0x00c8c000)
        libnssutil3.so => /lib/libnssutil3.so (0x00cb3000)
        libXdmcp.so.6 => /usr/lib/libXdmcp.so.6 (0x00953000)
        libcrypt.so.1 => /lib/libcrypt.so.1 (0x03b65000)
        libselinux.so.1 => /lib/libselinux.so.1 (0x0029f000)

You can download it where ever you like (your home directory would be a good place), you'll probably have to make it executable (chmod +x) and then run it (./boincmgr). You can connect it to an already running client by choosing "Select Computer" in the Advanced Menu, filling in localhost for the computer and the gui_rpc password.

---

Kathryn :o)

So if you make the application column showing as in the old days, ideally allow opting, I'll bow..

Opting what columns to see and what not is the next thing the developers want to add. So by then the CPU time column will probably return.

Thanks though. :-)

---

Yes, but many will like the application column {snip}.

How many? Do you have any threads that show the diversity of people who want this? You do know it's added to be user friendly, hence the name "user friendly name"?

Besides, I like the user friendly names as well as it is way easier to help people on various project forums (or here) with the full name of the application than to go figure out what the next abbreviation could possibly be. But that's me. :-)

And are there any projects still sending out the short name with the applications and tasks? As far as I can find, they are these days only used internally.

---

Yes, but many will like the application column {snip}.

How many? Do you have any threads that show the diversity of people who want this? You do know it's added to be user friendly, hence the name "user friendly name"?

Besides, I like the user friendly names as well as it is way easier to help people on various project forums (or here) with the full name of the application than to go figure out what the next abbreviation could possibly be. But that's me. :-)

And are there any projects still sending out the short name with the applications and tasks? As far as I can find, they are these days only used internally.

Yet again, why have this argument about what (unspecified numbers of) (unspecified) people may or may not prefer, depending whether there is an 'R' in the month or what direction the wind is blowing from?

Yet again, BoincView solves it: allow the user a choice.

This all reminds me why BoincView is by so much a better program than BOINC Manager: it was designed from the beginning with the user in mind, to show the user whatever that user might want to know. It doesn't try to pre-judge what the user 'ought' to want to know.

Yet again, BoincView solves it: allow the user a choice.

Great for BoincView. Too bad it is for Windows only. Also too bad it isn't in development anymore and that the source code of it isn't available to anyone.

At least I tried something with the current BOINC Manager and maybe that I have more up my sleeve. Although I may also keep it to myself if the attitude is that to and fro arguments are useless as we all have to hail BoincView.

---

Yet again, BoincView solves it: allow the user a choice.

Great for BoincView. Too bad it is for Windows only. Also too bad it isn't in development anymore and that the source code of it isn't available to anyone.

At least I tried something with the current BOINC Manager and maybe that I have more up my sleeve. Although I may also keep it to myself if the attitude is that to and fro arguments are useless as we all have to hail BoincView.

Sorry, sorry. Misinterpreted again. I really must learn to write one of these days. ;-) <-- smiley

BoincView is no longer in existence. BoincView never was available except for Windows. And it never was perfect, even though it was good enough to let me stop looking for anything better. So the time has come to steal the best bits and build them in to something else

All I wanted to convey was that BV had a solution to the question you were discussing. Should we show technical names for tasks, or should we show friendly names? My answer (unashamedly stolen from BV) is that, if the answer isn't immediately obvious, then make both available but let the user choose which one not to clutter their screen with.

Now, if you really want to build a better mousetrap, the feature I'd really like someone to steal (this time from M$ Access) is "Remove Filter/Sort" - whichever column heading you've clicked on, forget it, and go back to showing the order the tasks arrived in.

Now, if you really want to build a better mousetrap, the feature I'd really like someone to steal (this time from M$ Access) is "Remove Filter/Sort" - whichever column heading you've clicked on, forget it, and go back to showing the order the tasks arrived in.

As far as I understand it, there will be some major changes to the Manager for 6.8 or 6.10, including the choice of columns to see, an RSS reader and more such useful things.

I still want to get rid of that big blob on the left side of the manager and have all the space then open for all the information. Question is, what to do with the project buttons? Put them on the bottom somewhere, in a scrollable manner?

---

There are reports from SETI that work requests for CUDA plan class are not being filled, and instead are being met with a 86400 second (1 day) backoff. Not clear from the log whether this is server-mandated or a client response:[/url]
Lunatics pre-release Problem downloading Seti Enhanced - Astropulse ok

FYI, [sched_op_debug] helps know what the server returned, in particular backoffs.

---

Wallclock, hmmm, yet to find the reasoning of why this change, but a strong downside is that it makes any slow running highly obscured, since few will actually drill into the job properties to find out the CPU time.

BOINC still doesn't know how to measure "GPU time", and CPU time is meaningless for CUDA apps.

---

All I wanted to convey was that BV had a solution to the question you were discussing. Should we show technical names for tasks, or should we show friendly names? My answer (unashamedly stolen from BV) is that, if the answer isn't immediately obvious, then make both available but let the user choose which one not to clutter their screen with.

Note "Preference" dialog boxes can get cluttered too...

Joel Spolsky wrote:

Software has a similar archaeological record, too: it's called the Options dialog. Pull up the Tools | Options dialog box and you will see a history of arguments that the software designers had about the design of the product. Should we automatically open the last file that the user was working on? Yes! No! There is a two week debate, nobody wants to hurt anyone's feelings, the programmer puts in an #ifdef in self defense while the designers fight it out. Eventually they just decide to make it an option.

Every time you provide an option, you're asking the user to make a decision.

See "User Interface Design For Programmers - Chapter 3 - Choices".

---