| GPULib 1.8 released [message #90150] |
Mon, 02 February 2015 11:53  |
Michael Galloy
Messages: 1114
Registered: April 2006
|
Senior Member |
|
|
GPULib 1.8 has been released with updates to the underlying libraries as
well as many other features in many areas of the library. For
information about purchasing, see:
http://www.txcorp.com/home/gpulib
It has been updated to use the most recent versions of IDL and CUDA, IDL
8.4 and CUDA 6.5. The new features are:
* Support for integer data types. I have been wanting to support
integer types in GPULib for awhile and now GPULib supports all the
numeric types provided by IDL! We can finally do:
dx = gpuIndgen(10)
* Added `GPUREPMAT` routine. This is a handy routine to create a new
array by repeating a 2-dimensional array in a grid.
* Added `GPUCREATEKERNEL` routine to create the source code of a
simple kernel. This is a code generation routine that can be loaded with
`GPULOADMODULE`/`GPULOADFUNCTION` and executed with `GPUEXECUTEFUNCTION`.
* Added `GPUFINITE` routine similar to IDL's library routine.
* Added linear algebra routines `GPULUDC`, `GPULUSOL`, and
`GPULEAST_SQUARES`. This fills out more of the GPU equivalent of the
convenience routines provided by IDL so that the LAPACK interface of
MAGMA is not required to perform linear algebra computations.
* Added support for `RHO` and `THETA` keywords in `GPURADON`.
* Added `GPUMEAN` routine. This routine has `DIMENSION` and `NAN`
keywords with the same functionality as IDL's library routine.
Mike
--
Michael Galloy
www.michaelgalloy.com
Modern IDL: A Guide to IDL Programming (http://modernidl.idldev.com)
Research Mathematician
Tech-X Corporation
|
|
|
|
|
|
|
|
|
|
|
|
| Re: GPULib 1.8 released [message #90175 is a reply to message #90170] |
Thu, 05 February 2015 12:35  |
Michael Galloy
Messages: 1114
Registered: April 2006
|
Senior Member |
|
|
On 2/5/15, 3:41 AM, superchromix wrote:
>
> hi Mike,
>
> I notice that there is a new function called GPULEASTSQUARES. Is
> this an implementation of the GPU-based least squares fitting which
> you have mentioned in the past? It would be great if you could
> include a usage example in the documentation.
>
> best Mark
>
The GPULib distribution includes a unit test for GPULEAST_SQUARES shown
below (ignore the ASSERT statements, that is just checking to make sure
things are as the test expects along the way):
function gpuleast_squares_ut::test_float
compile_opt strictarr
assert, !gpu.mode ne 0, 'GPULEAST_SQUARES not available in emulation
mode', /skip
assert, !gpu.mode eq 0L || gpulmgr(/full), 'GPULib not licensed for
LAPACK calculations', /skip
assert, !gpu.mode eq 0L || gpuMagmaPresent(), 'MAGMA not present', /skip
a = [[1., 2., 1.], [4., 10., 15.], [3., 7., 1.]]
dims = size(a, /dimensions)
m = dims[0]
n = dims[1]
b = [2., 3., 7.]
standard = la_least_squares(a, b)
da = gpuputarr(a, ERROR=err)
assert, err eq 0, 'CUDA error: %s', gpu_errormessage(err)
db = gpuputarr(b, ERROR=err)
assert, err eq 0, 'CUDA error: %s', gpu_errormessage(err)
dresult = gpuleast_squares(da, db, ERROR=err)
assert, err eq 0, 'CUDA error: %s', gpu_errormessage(err)
result = gpugetarr(dresult, ERROR=err)
assert, err eq 0, 'CUDA error: %s', gpu_errormessage(err)
error = total(abs(standard - result), /preserve_type)
assert, error lt self.tolerance * n * 10.0, 'incorrect result: error
= %g', error
gpuFree, [da, db, dresult]
return, 1
end
function gpuleast_squares_ut::test_double
compile_opt strictarr
assert, !gpu.mode ne 0, 'GPULEAST_SQUARES not available in emulation
mode', /skip
assert, !gpu.mode eq 0L || gpulmgr(/full), 'GPULib not licensed for
LAPACK calculations', /skip
assert, !gpu.mode eq 0L || gpuMagmaPresent(), 'MAGMA not present', /skip
assert, !gpu.mode eq 0L || gpuDoubleCapable(), 'CUDA device not
double capable', /skip
a = [[1.0D, 2.0D, 1.0D], [4.0D, 10.0D, 15.0D], [3.0D, 7.0D, 1.0D]]
dims = size(a, /dimensions)
m = dims[0]
n = dims[1]
b = [2.0D, 3.0D, 7.0D]
standard = la_least_squares(a, b, /DOUBLE)
da = gpuputarr(a, ERROR=err)
assert, err eq 0, 'CUDA error: %s', gpu_errormessage(err)
db = gpuputarr(b, ERROR=err)
assert, err eq 0, 'CUDA error: %s', gpu_errormessage(err)
dresult = gpuleast_squares(da, db, ERROR=err)
assert, err eq 0, 'CUDA error: %s', gpu_errormessage(err)
result = gpugetarr(dresult, ERROR=err)
assert, err eq 0, 'CUDA error: %s', gpu_errormessage(err)
error = total(abs(standard - result), /preserve_type)
assert, error lt self.tolerance * n * 10.0, 'incorrect result: error
= %g', error
gpuFree, [da, db, dresult]
return, 1
end
Mike
--
Michael Galloy
www.michaelgalloy.com
Modern IDL: A Guide to IDL Programming (http://modernidl.idldev.com)
Research Mathematician
Tech-X Corporation
|
|
|
|
| Re: GPULib 1.8 released [message #90178 is a reply to message #90175] |
Fri, 06 February 2015 01:51 |
markb77
Messages: 217
Registered: July 2006
|
Senior Member |
|
|
On Thursday, February 5, 2015 at 9:35:52 PM UTC+1, Mike Galloy wrote:
> On 2/5/15, 3:41 AM, superchromix wrote:
>>
>> hi Mike,
>>
>> I notice that there is a new function called GPULEASTSQUARES. Is
>> this an implementation of the GPU-based least squares fitting which
>> you have mentioned in the past? It would be great if you could
>> include a usage example in the documentation.
>>
>> best Mark
>>
>
> The GPULib distribution includes a unit test for GPULEAST_SQUARES shown
> below (ignore the ASSERT statements, that is just checking to make sure
> things are as the test expects along the way):
>
> function gpuleast_squares_ut::test_float
> compile_opt strictarr
>
> assert, !gpu.mode ne 0, 'GPULEAST_SQUARES not available in emulation
> mode', /skip
> assert, !gpu.mode eq 0L || gpulmgr(/full), 'GPULib not licensed for
> LAPACK calculations', /skip
> assert, !gpu.mode eq 0L || gpuMagmaPresent(), 'MAGMA not present', /skip
>
> a = [[1., 2., 1.], [4., 10., 15.], [3., 7., 1.]]
> dims = size(a, /dimensions)
> m = dims[0]
> n = dims[1]
>
> b = [2., 3., 7.]
>
> standard = la_least_squares(a, b)
>
> da = gpuputarr(a, ERROR=err)
> assert, err eq 0, 'CUDA error: %s', gpu_errormessage(err)
>
> db = gpuputarr(b, ERROR=err)
> assert, err eq 0, 'CUDA error: %s', gpu_errormessage(err)
>
> dresult = gpuleast_squares(da, db, ERROR=err)
> assert, err eq 0, 'CUDA error: %s', gpu_errormessage(err)
>
> result = gpugetarr(dresult, ERROR=err)
> assert, err eq 0, 'CUDA error: %s', gpu_errormessage(err)
>
> error = total(abs(standard - result), /preserve_type)
> assert, error lt self.tolerance * n * 10.0, 'incorrect result: error
> = %g', error
>
> gpuFree, [da, db, dresult]
>
> return, 1
> end
>
>
> function gpuleast_squares_ut::test_double
> compile_opt strictarr
>
> assert, !gpu.mode ne 0, 'GPULEAST_SQUARES not available in emulation
> mode', /skip
> assert, !gpu.mode eq 0L || gpulmgr(/full), 'GPULib not licensed for
> LAPACK calculations', /skip
> assert, !gpu.mode eq 0L || gpuMagmaPresent(), 'MAGMA not present', /skip
> assert, !gpu.mode eq 0L || gpuDoubleCapable(), 'CUDA device not
> double capable', /skip
>
> a = [[1.0D, 2.0D, 1.0D], [4.0D, 10.0D, 15.0D], [3.0D, 7.0D, 1.0D]]
> dims = size(a, /dimensions)
> m = dims[0]
> n = dims[1]
>
> b = [2.0D, 3.0D, 7.0D]
>
> standard = la_least_squares(a, b, /DOUBLE)
>
> da = gpuputarr(a, ERROR=err)
> assert, err eq 0, 'CUDA error: %s', gpu_errormessage(err)
>
> db = gpuputarr(b, ERROR=err)
> assert, err eq 0, 'CUDA error: %s', gpu_errormessage(err)
>
> dresult = gpuleast_squares(da, db, ERROR=err)
> assert, err eq 0, 'CUDA error: %s', gpu_errormessage(err)
>
> result = gpugetarr(dresult, ERROR=err)
> assert, err eq 0, 'CUDA error: %s', gpu_errormessage(err)
>
> error = total(abs(standard - result), /preserve_type)
> assert, error lt self.tolerance * n * 10.0, 'incorrect result: error
> = %g', error
>
> gpuFree, [da, db, dresult]
>
> return, 1
> end
>
> Mike
> --
> Michael Galloy
> www.michaelgalloy.com
> Modern IDL: A Guide to IDL Programming (http://modernidl.idldev.com)
> Research Mathematician
> Tech-X Corporation
I see - thanks. I guess I was looking for a non-linear least squares solver.
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
