For all those who have not yet installed...
Keep crunching ! :)
BOINC Manager 7.2.39

---

where is changelog?
why is reason to upgrade boinc vercion ( not boincmgr alone, but all boinc is a new version, FYI. i often use ./boinc , not ./boincmgr )

I just put 7.2.33 on a test rig and I can tell you that there is not a hope in hell that I'm going to upgrade my main rigs from 6.10.60.

Cheers.

where is changelog?

Short version
Long version

why is reason to upgrade boinc vercion ( not boincmgr alone, but all boinc is a new version, FYI. i often use ./boinc , not ./boincmgr )

Mainly to support new hardware and technologies - AMD R9, OpenCL on CPUs and Intel GPUs, VBox. But they concern other projects more than SETI.

Thanks for info. but there not better to give release notes for aech verison on download page, for can get to see fast, whats changed and decide - it was a must to upgrade, or not?
or there is philosophy "get them dumb persons file, any information for them is not necessary..." ? :D

---

Thanks for info. but there not better to give release notes for aech verison on download page, for can get to see fast, whats changed and decide - it was a must to upgrade, or not?
or there is philosophy "get them dumb persons file, any information for them is not necessary..." ? :D

Well, the download page does have two separate links:

Release notes
Version history

(in the navigation bar across the bottom - at least in the Windows version of the page I see)

The second one is the 'short version' I linked earlier. I think having the information available one click away is a reasonable compromise, without overwhelming first-time users with information that will be meaningless to them.

yes, sorry, you right about the webpage.
but about "not overhelming" - imho in those days all that strong care to be not overwhelming anything, in all directions, that there all environment become to more and more worse.
there is less and less possibilities and depth explanations, and more and more "press that key, and all must be work"...:(

---

The Biggest interesting feature is now the ability to specify <app_versions> in the app_config.xml enabling you to specify a different number of tasks per GPU for a different <app_version>,
ie run two or three tasks per GPU on a Nvidia or AMD GPU, and only one on an Intel GPU, etc,
or if you're running Stock you can run more tasks on the Cuda32 or Cuda42 app_version to drive the APR down, so in future the scheduler will just send you Cuda50 work because that APR is higher.

Claggy

SuSE downloaded to my Linux box first 7.2.38 then 7.2.41, all necessary to run the latest version of Test4Theory@home with Virtual Box.
Tullio

---

Downloaded and installed.
I noticed that it does not log the "restarting" WUs so looking at the log I at first thought nothing had started running. I then looked in Tasks and they were all running.
Log seems to work when starting a new task.

---

SETI@home classic workunits 4,019
SETI@home classic CPU time 34,348 hours

The Biggest interesting feature is now the ability to specify <app_versions> in the app_config.xml enabling you to specify a different number of tasks per GPU for a different <app_version>,
ie run two or three tasks per GPU on a Nvidia or AMD GPU, and only one on an Intel GPU, etc,
or if you're running Stock you can run more tasks on the Cuda32 or Cuda42 app_version to drive the APR down, so in future the scheduler will just send you Cuda50 work because that APR is higher.

Claggy

Just curiosity could it be used to run diferent number of WU in diferent GPU´s (all Nvidia for example) in a 2xGPU host? If no that could be an interesting feature to add.

Example a host with 780+670, enable 3 at a time on the 770 and only 2 on the 670?

Or you could do only if the GPUS comes from diferent arquitectures of GPU (Intel/AMD/Nvidia)

---

what is a point to run boinc in virtual PC? :-O

---

IIRC there are some projects who need that, for now, SETI not use it, so we must install the version without virtual box.

---

No project runs BOINC in a virtual machine. There may be users that do so, but they then just run more than one OS simultaneously on their PC.

The projects using virtual box do this so they only have to make an application for ONE Linux version, and afterwards only update that one application. You running BOINC with Vbox will then run the project applications in a hyper visor in Vbox, which will communicate with BOINC about progress etc. BOINC will still do the scheduling of work, amount of CPUs to use etc. But the actual work runs in an environment that the project set up. Makes it simpler for them.

---

The Biggest interesting feature is now the ability to specify <app_versions> in the app_config.xml enabling you to specify a different number of tasks per GPU for a different <app_version>,
ie run two or three tasks per GPU on a Nvidia or AMD GPU, and only one on an Intel GPU, etc,
or if you're running Stock you can run more tasks on the Cuda32 or Cuda42 app_version to drive the APR down, so in future the scheduler will just send you Cuda50 work because that APR is higher.

Wonder if you can use a max concurrent line along with <app_versions> or <plan_class> to set different limits for cpu & gpu versions of the same short name? That would help me when I'm running multiple projects.

---

Another Fred
Support SETI@home when you search the Web with GoodSearch or shop online with GoodShop.

There is a newer version available.
7.3.2 Pre Release

http://www.bc-team.org/downloads.php?cat=4&sid=067ca1872a450e16a35e82c10d8d46c9

Just for your info

There is a newer version available.
7.3.2 Pre Release

Yes, as long as you keep in mind that it's a pre-release, it's an ALPHA. It's been built with the whole new wxWidgets 3.0 version and therefore will be buggy. Don't use it as if it's a normal version, it may harm your BOINC (Data) directory.

There will always be newer versions available, but not all are for the general public. You can see that it's an alpha version, because of its numbering. Major.odd.revision means Alpha/test phase, meaning buggy, meaning it can break on your computer.
Most of the major.even.revision mean Safe for general use, although if not recommended, still with possible major breaks and bugs.

So 7.2.39 is safe, all of 7.3 is definitely not safe.

---

The Biggest interesting feature is now the ability to specify <app_versions> in the app_config.xml enabling you to specify a different number of tasks per GPU for a different <app_version>,
ie run two or three tasks per GPU on a Nvidia or AMD GPU, and only one on an Intel GPU, etc,
or if you're running Stock you can run more tasks on the Cuda32 or Cuda42 app_version to drive the APR down, so in future the scheduler will just send you Cuda50 work because that APR is higher.

Claggy

Just curiosity could it be used to run diferent number of WU in diferent GPU´s (all Nvidia for example) in a 2xGPU host? If no that could be an interesting feature to add.

Example a host with 780+670, enable 3 at a time on the 770 and only 2 on the 670?

Or you could do only if the GPUS comes from diferent arquitectures of GPU (Intel/AMD/Nvidia)

No, i don't think you'll be able to run a different number of tasks on different GPUs from the same vendor,
But what you can do is specify a different set of x41zc parameters per GPU so you can drive the app harder on certain GPUs via Jason's mbcuda.cfg

Claggy

For some reason 7.2.39 is now auto reporting on each workunit finishing. It never was before. Probably an unintended by product.

It's a deliberate change. Tasks are now reported one hour after they complete (or earlier, if new work is needed), rather than hanging around for up to 24 hours.

7.2.39 has the new one hour report function. Work is no longer reported first after 24 hours, but after 1 hour. So if your computer finishes work every hour, or just under the hour, it'll report it around/at that time as well.

---