Jaroslav Hajek wrote:
> On Wed, Mar 10, 2010 at 9:52 PM, David Bateman wrote:
>
>> Yeah, things like UMFPACK internally block the sparse matrices and treat the
>> sub-blocks with blas (ie dense kernels), so its a well known trick.
>>
>
> Internally? Or can UMFPACK be forced to solve a bloc
On Wed, Mar 10, 2010 at 9:52 PM, David Bateman wrote:
> Jaroslav Hajek wrote:
>>
>> On Tue, Mar 9, 2010 at 11:14 PM, David Bateman
>> wrote:
>>
>>>
>>> Jaroslav Hajek wrote:
>>>
I think I looked into those sources briefly (because I also noticed
that it implements basic ops), but i
On Wed, Mar 10, 2010 at 03:37, Jaroslav Hajek wrote:
> so it seems that block multiplication is really a big advantage (with
> larger blocks), so much that it even outweighs the extra work & memory
> needed by the m-code. It's not that surprising, considering that
> optimized BLAS also get their s
Jaroslav Hajek wrote:
> On Tue, Mar 9, 2010 at 11:14 PM, David Bateman wrote:
>
>> Jaroslav Hajek wrote:
>>
>>> I think I looked into those sources briefly (because I also noticed
>>> that it implements basic ops), but indeed there seems to be nothing
>>> magical (read: no assembler). Some
On Tue, Mar 9, 2010 at 11:14 PM, David Bateman wrote:
> Jaroslav Hajek wrote:
>>
>> I think I looked into those sources briefly (because I also noticed
>> that it implements basic ops), but indeed there seems to be nothing
>> magical (read: no assembler). Some operations we need (such as
>> transp
Jaroslav Hajek wrote:
I think I looked into those sources briefly (because I also noticed
that it implements basic ops), but indeed there seems to be nothing
magical (read: no assembler). Some operations we need (such as
transpose-multiply or diagonal scaling) were not even there, I think.
That d
On Tue, Mar 9, 2010 at 6:47 PM, David Bateman wrote:
> Jaroslav Hajek wrote:
>>
>> hi all,
>>
>> I contributed a toy implementation of block sparse matrices to the
>> linear-algebra package (also attached).
>> This is probably not meant for serious business, as all the code is
>> m-files only, but
Jaroslav Hajek wrote:
> hi all,
>
> I contributed a toy implementation of block sparse matrices to the
> linear-algebra package (also attached).
> This is probably not meant for serious business, as all the code is
> m-files only, but is rather a demonstration of what can now be done in
> Octave.
>
hi all,
I contributed a toy implementation of block sparse matrices to the
linear-algebra package (also attached).
This is probably not meant for serious business, as all the code is
m-files only, but is rather a demonstration of what can now be done in
Octave.
Basic operations (+, -, *, ', .') a
