[FFmpeg-devel] [PATCH] WMA Voice decoder

[Måns Rullgård]

"Ronald S. Bultje" <rsbultje at gmail.com> writes:

> Hi,
>
> On Fri, Jan 22, 2010 at 12:30 PM, Uoti Urpala <uoti.urpala at pp1.inet.fi> wrote:
>> On Fri, 2010-01-22 at 11:19 -0500, Ronald S. Bultje wrote:
>>> On Fri, Jan 22, 2010 at 11:15 AM, Uoti Urpala <uoti.urpala at pp1.inet.fi> wrote:
>>> > x*49995 / 41 = x*(1219*41 + 16) / 41 = 1219*x + x * 16 / 41
>>> > In the last form x*16 is at most 1048544.
>>>
>>> In fastdiv form, this'd be 3 muls, an add and a shift. Is that still
>>> faster than 1 mul + 1 div?
>>
>> You can premultiply the *16*INV(41) into one table value, so two
>> multiplications. Also if the processor can directly calculate "upper 32
>> bits of multiply" then there's no need for an actual shift operation.
>
> Good point. Does MULL take care of that automatically if supported?

MULH returns the high 32 bits of a 32x32->64 multiplication.

-- 
M?ns Rullg?rd
mans at mansr.com