ARGOMENTO:

CUDA vs nVidia OpenCL User Poll
Testing is finishing up on the Collatz v6.xx application for CUDA. Before I release it, I'd like your opinion on how I should configure the scheduler on the server. Should it:

a) Send the CUDA if available and only send an nVidia OpenCL version if there is no CUDA version for the platform

b) Send the nVidia OpenCL version if available and only send the CUDA version if there is no nVidia OpenCL version for the platform

c) Let the BOINC client learn which is faster over time

Note: The CUDA application uses less than 1% CPU whereas the OpenCL versions use considerably more - sometimes up to 80%.
14 Feb 2014, 16:50:57 UTC

Dopo l'aggiornamento del server news interessante per utenti NVIDIA:

CUDA vs nVidia OpenCL User Poll
Testing is finishing up on the Collatz v6.xx application for CUDA. Before I release it, I'd like your opinion on how I should configure the scheduler on the server. Should it:

a) Send the CUDA if available and only send an nVidia OpenCL version if there is no CUDA version for the platform

b) Send the nVidia OpenCL version if available and only send the CUDA version if there is no nVidia OpenCL version for the platform

c) Let the BOINC client learn which is faster over time

Note: The CUDA application uses less than 1% CPU whereas the OpenCL versions use considerably more - sometimes up to 80%.
14 Feb 2014, 16:50:57 UTC

While most people wanted only CUDA applications, some had some good reasons for using OpenCL applications which means it really can't be one or the other. As a result, you can use whichever you want or even both by selecting your preferred plan class(es) for each venue.

The Windows and Linux CUDA applications, both 32 and 64 bit, have been released so your preferences should take effect the next time your computer requests work. The default is that all plan classes are selected.
18 Feb 2014, 0:34:59 UTC

The Brook+/CAL application, otherwise known as plan class "ATI14", are now available for Windows and Windows 64. They are based on the 2.09 applications, but with the additional solo_collatz verification added. They utilize the same command line settings as the Collatz v2.09 ATI applications. Anyone with an AMD HD 2400 or later GPU can once again participate. In addition, those with AMD OpenCL compatible GPUs can choose via their preferences whether to run the ATI14 apps, the OpenCL AMD apps, or both.

Recent AMD video drivers, especially those installed for Windows 8 and even some for Windows 7, may not install the necessary CAL DLLs and therefore cannot run the Brook+/CAL application. Try as I might to manually install them, I had to downgrade a laptop with an ATI Mobility HD 3200 GPU from Windows 8.1 to Windows 7 and also install an older version of the AMD video drivers in order to even test the solo_collatz Brook+/CAL application. Given that Brook+ has not been supported by AMD for a number of years and new AMD GPU hardware cannot run CAL applications, it is not all that surprising. So, if you have an older AMD Radeon HD series GPU that cannot run OpenCL applications and you want to continue to use it to run ATI applications with Collatz or other BOINC projects, you may want to avoid future AMD video driver upgrades.
19 Feb 2014, 4:51:19 UTC

Per gli user NVIDIA sembra che il poll si sia chiuso a favore di CUDA ?

Variable Sized Wokunits & New CPU Applications
I have implemented BOINC's multi-size workunits which adjusts the size of the workunit to the speed of the processor. The workunits are all named solo_collatz but there will be 4 different sizes. The smallest will be the size of that the mini_collatz workunits used to be. The largest will remain the size of the original collatz and solo_collatz workunits or 8 times the size of the smallest. There are also intermediate sizes which are 25% and 50% of the size of the largest. You can think of it as mini, small, medium, and large or 100 billion, 200 bilion, 400 billion and 800 billion numbers per workunit. That also means that even though the workinit name is solo_collatz, the credit awarded is directly proportional to the work done. The mini workunit does 1/8th the work and is awarded 1/8th the credit.

The smaller size workunits allows me to release non-OpenCL versions of the CPU applications which use a single core/processor per workunit just like the old mini_collatz CPU applications.

Lastly, there still seems to be an issue with multiple venues and plan class choices. A fix will be put in place as soon as I figure out why it isn't working properly.
20 Feb 2014, 20:32:03 UTC

Ora invece è la volta delle news per gli utenti ATI:

Solo Collatz Brook+/CAL (a.k.a. ATI) Applications Released for Windows

astroale ha scritto:

Ora invece è la volta delle news per gli utenti ATI:

Solo Collatz Brook+/CAL (a.k.a. ATI) Applications Released for Windows

Ciao a tutti!
Premettendo che conosco benissimo i limiti della mia HD3850, vorrei capire perché non riesco più a farla lavorare.
Questo è un estratto dal log del BM:
CAL: ATI GPU 0: ATI Radeon HD 3800 (RV670) (CAL version 1.4.1332, 512MB, 491MB available, 922 GFLOPS peak)
e questo e cosa risponde il progetto:
Messaggio dal server: Il tuo file app_info.xml non ha una versione utilizzabile di solo_collatz.

Dove sbaglio? Ma soprattutto come posso risolvere? Grazie.

Solo Collatz applications are now available for Android. The Android application supports ARM, x86, and MIPS (Little Endian) processors. Assuming Murphy's Law will prevail as usual even though the apps work on my hardware, please let me know of any issues.

Aggiunta anche l'applicazione per android

Solo Collatz applications are now available for Android. The Android application supports ARM, x86, and MIPS (Little Endian) processors. Assuming Murphy's Law will prevail as usual even though the apps work on my hardware, please let me know of any issues.

boinc.thesonntags.com/collatz/forum_thread.php?id=1134

@Daniele

si l'avevo vista ieri e ti dirò che speravo fosse dovuta al 1° di aprile, evidentemente non lo è.
Peccato perche stato cercando di raggiungere rapidamente i 25 M

tutto sommato direi che sia giusto, pagava un po troppo (ce ne sono ancora tanti che pagano un casino !!!!!

Finally, because new volunteers who join the project won't ever be able to catch up to the credit leaders due to the switch to using CreditNew, the total credits will also be adjusted for all users, hosts, and teams. You will see an 80% reduction. But, on the bright side, you will be able to achieve those mega milestones tracked at FreeDC all over again!

se mai succederà rimpiangerò tutti 'sti cicli Gpu tolti a gpugrid !