Hi,
It is posted in news "Texas Instruments and UT Austin collaborate to deliver linear algebra library on TI's high performance multicore DSPs" that TI "have successfully ported UT Austin's libflame library, a dense linear algebra library for scientific computing, to TI's TMS320C6678 multicore digital signal processor (DSP)."
Where can I see (download or bay) this library and what a real performance is (any benchmarks, tests)?
Thanks in advance and sorry for my English.
Dmitry
Dmitry,
Let me check to see where we are on this. The basic library functions have been ported, but I believe there is still some on-going work to implement/optimize in our multicore environment.
Regards,
Travis
If you need more help, please reply back. If this answers the question, please click Verify Answer , below.
Did you have any luck locating it? If not, you can download the source at: http://z.cs.utexas.edu/wiki/flame.wiki/libflame/#head-07b59fe7e7ee10e1ff33b7e682009ab8b59f0822
I've been told that the TI build is on this website too, but I'm not exactly sure where. You can submit a question to flame@cs.utexas.edu and probably get a quick answer though.
Travis,
Thank you for your answer,
Dear Travis,
would you like to give me a brief introduction such as performace/benmark about libfalme on C6678?also tell me how to buy?
Best Regards,
Frank
frank.chen@eastdima.com
Frank,
I don't have any benchmarks at this point, but I think you can get the information you are looking for directly from the UT Austin folks using the link above. If you are asking for what the benchmarks encompass, LibFlame is a version of LINPACK written in C, used to solve Linear algebra matrix problems.
Forgot to mention that this library will be free. I made some inquires and found out the the LibFlame optimization effort for C66x should be complete in 3Q12.
Thanks for quickly respone.
If we want to buy the lib,who is we will contact for next step, UT Austin or Ti guys?
You mean library(libFlame) on C6676 is free,am I right?
It will be supported and delivered through UT Austin, not TI. My understanding is that it will be offered free of charge. I hope you meant to say C6678, not C6676.
Yes,C6678 not C6676.
So next I will contact UT austin.
Also would you please give me some recommendation about tools/sw for C6678 's multi-core code parallel optimization?
Thanks Travis！
Have you looked at the MCSDK for C66x?
http://software-dl.ti.com/sdoemb/sdoemb_public_sw/bios_mcsdk/latest/index_FDS.html