On Mon, Mar 30, 2015 at 11:54 AM, Matt Turner wrote:
> On Sat, Mar 28, 2015 at 11:03 AM, Jason Ekstrand wrote:
>> On Fri, Mar 27, 2015 at 9:22 PM, Matt Turner wrote:
>>> The typical case of mat4*mat4*vec4 is 80 scalar multiplications, but
>>> mat4*(mat4*vec4) is only 32.
>>>
>>> On HSW (with vec
On Sat, Mar 28, 2015 at 11:03 AM, Jason Ekstrand wrote:
> On Fri, Mar 27, 2015 at 9:22 PM, Matt Turner wrote:
>> The typical case of mat4*mat4*vec4 is 80 scalar multiplications, but
>> mat4*(mat4*vec4) is only 32.
>>
>> On HSW (with vec4 vertex shaders):
>> instructions in affected programs:
On Fri, Mar 27, 2015 at 9:22 PM, Matt Turner wrote:
> The typical case of mat4*mat4*vec4 is 80 scalar multiplications, but
> mat4*(mat4*vec4) is only 32.
>
> On HSW (with vec4 vertex shaders):
> instructions in affected programs: 4420 -> 3194 (-27.74%)
>
> On BDW (with scalar vertex shaders):
For the series:
Reviewed-by: Chris Forbes
On Sat, Mar 28, 2015 at 5:22 PM, Matt Turner wrote:
> The typical case of mat4*mat4*vec4 is 80 scalar multiplications, but
> mat4*(mat4*vec4) is only 32.
>
> On HSW (with vec4 vertex shaders):
> instructions in affected programs: 4420 -> 3194 (-27.7
The typical case of mat4*mat4*vec4 is 80 scalar multiplications, but
mat4*(mat4*vec4) is only 32.
On HSW (with vec4 vertex shaders):
instructions in affected programs: 4420 -> 3194 (-27.74%)
On BDW (with scalar vertex shaders):
instructions in affected programs: 12756 -> 6726 (-47.27%)
I
