|
|
|
@2883
|
[2883]
|
3 years |
max |
allow usage of the 4. entry in Vector3::mData. This can be used to move …
|
|
|
|
@2878
|
[2878]
|
3 years |
max |
fixed cmake syntax
|
|
|
|
@2866
|
[2866]
|
3 years |
max |
optimization: according to Intel div_ss ist faster then div_ps
|
|
|
|
@2865
|
[2865]
|
3 years |
max |
optimization
|
|
|
|
@2864
|
[2864]
|
3 years |
max |
get rid of the compiler-warning "DCOLLIDE_USE_SSE not found" if not …
|
|
|
|
@2858
|
[2858]
|
3 years |
horst |
Introduced a CMake Parameter USE_SSE which switches between the different …
|
|
|
|
@2857
|
[2857]
|
3 years |
horst |
According to the MSVC help we have to compiler/linker flags which …
|
|
|
|
@2856
|
[2856]
|
3 years |
max |
BUGFIX: _mm_hadd4_ps delivered wrong results.
added a sse2-version of …
|
|
|
|
@2854
|
[2854]
|
3 years |
max |
added also a sse2-enabled version of length (from 149 cycles/call …
|
|
|
|
@2853
|
[2853]
|
3 years |
horst |
Added corrected determination of the SSE2 define for ICC and MSVC, but …
|
|
|
|
@2852
|
[2852]
|
3 years |
horst |
- Added better optimization flags for MSVC and ICC under Windows (incl. …
|
|
|
|
@2850
|
[2850]
|
3 years |
max |
BUGFIX: also add 4. column of the matrix to the returned vector
|
|
|
|
@2849
|
[2849]
|
3 years |
max |
added a sse2-Version of Matrix::transform. Improves performance from 563 …
|
|
|
|
@2848
|
[2848]
|
3 years |
max |
fixed some const-stuff
|
|
|
|
@2847
|
[2847]
|
3 years |
max |
dotProduct: We can not be sure that mData is 16Bit aligned --> we …
|
|
|
|
@2846
|
[2846]
|
3 years |
max |
vector: dotProduct (SSE2-Version) much faster now using correctly aligned …
|
|
|
|
@2845
|
[2845]
|
3 years |
horst |
Going back to direct acces to member variables instead of using getters…
|
|
|
|
@2844
|
[2844]
|
3 years |
max |
added sse2-implementation of dotProduct. If SSE2 is not defined, we …
|
|
|
|
@2843
|
[2843]
|
3 years |
max |
don't use the getters in the dotProduct --> 41 cycles per call
CMakeLists: …
|
|
|
|
@2842
|
[2842]
|
3 years |
horst |
Try to avoid dereferencing as bottle neck.
|
|
|
|
@2838
|
[2838]
|
3 years |
max |
make it compile again.
|
|
|
|
@2837
|
[2837]
|
3 years |
horst |
Reworked the Vector class further more. Now we should easily be able to …
|
|
|
|
@2823
|
[2823]
|
3 years |
horst |
corrected line ending
CR+LF -> LF
|
|
|
|
@2822
|
[2822]
|
3 years |
horst |
missed one file
|
|
|
|
@2821
|
[2821]
|
3 years |
horst |
Finished the rework of the internal data representation of the Vector …
|
|
|
|
@2816
|
[2816]
|
3 years |
horst |
replaced the last occurencies of the [] operator. everything should …
|
|
|
|
@2815
|
[2815]
|
3 years |
gregor |
replace Vector[012] with Vector.getXYZ() in sphere-triangle, box-sphere …
|
|
|
|
@2814
|
[2814]
|
3 years |
horst |
Adjusted the MAtrix class
|
|
|
|
@2812
|
[2812]
|
3 years |
horst |
Changed the internal data representation of the vector
|
|
|
|
@2811
|
[2811]
|
3 years |
horst |
Created a branch for some optimization testing
|
|
copied from d-collide/trunk/d-collide:
|
|
|
|
@2810
|
[2810]
|
3 years |
gregor |
reverted my changes from the last commit
@max ok, my fault. I did a …
|
