CUDA 3.0 released has CUBLAS functions you need!

[oscarbg]

Perhaps you don't know but seems Nvidia shipped a better CUBLAS to CUlatools as they have near all this extensions from 30 september!
You can grab the basic and get cublas.dll,cublas.so with all the functions without a .lib in Windows for linking to it.
Anyway with 3.0 you have this
functions that seem to be so useful:
* cublasCtrsm()
* cublasCtrmm()
* cublasCsyrk()
* cublasZtrsm()
* cublasZtrmm()
* cublasZsyrk()

It would be good if you can use them to release (at least in later build than 14) with complex data types and single and double precision
- Added the BLAS1 functions:
* cublasZaxpy()
* cublasZcopy()
* cublasZswap()
- Added the BLAS2 functions:
* cublasDtrmv()
* cublasCtrmv()
* cublasCgemv()
* cublasCgeru()
* cublasCgerc()
* cublasZtrmv()
* cublasZgemv()
* cublasZgeru()
* cublasZgerc()
- Added the BLAS3 functions:
* cublasCtrsm()
* cublasCtrmm()
* cublasCsyrk()
* cublasCsymm()
* cublasCherk()
* cublasZtrsm()
* cublasZtrmm()
* cublasZsyrk()
* cublasZsymm()
* cublasZherk()

[Stan Tomov]

I knew about the new CUDA 3.0 and the additional complex functionality. This is quite impressive! Thanks for pointing it out.
I haven't tried it out yet though. Actually, we just made our second MAGMA release (version 0.2) and we added the complex versions of the one sided factorizations. We don't use yet the new complex routines from CUBLAS as CUDA 3.0 is still available for registered developers only.
Regards,
Stan