mirror-github-bspeice/fractorium
1
0
Fork 0
mirror of https://bitbucket.org/mfeemster/fractorium.git synced 2025-07-14 20:24:54 -04:00
Code Issues Actions Packages Projects Releases Wiki Activity

--User changes

Browse Source 
-Remove some warnings about interpolation type on first and last flames.

--Code changes
 -Make DE block size always be 16x16, this should help stability on some Nvidia cards. No changes for AMD cards since they were that size already.
-Since the block size is now so small, do not reduce it further when supersampling.
 -Clean up some variable names and documentation around OpenCL DE to be more clear.
This commit is contained in:
mfeemster
2016-03-12 19:25:19 -08:00
parent 65be0143ff
commit 19cb27b83a
6 changed files with 530 additions and 553 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
Source/EmberCL/OpenCLWrapper.cpp
View File

@ -940,12 +940,20 @@ size_t OpenCLWrapper::GlobalMemSize() const { return m_GlobalMemSize; }
size_t OpenCLWrapper::MaxAllocSize() const { return m_MaxAllocSize; }
/// <summary>
/// Makes the even grid dims.
/// Make even grid dimensions.
/// The size of the blocks in terms of threads must divide evenly into the total number of threads in the grid.
/// In the case of a remainder, expand the width and height of the grid to the next highest evenly divisible value.
/// Ex:
/// blockW = 5, blockH = 5
/// gridW = 18, gridH = 27
///
/// To make these even:
/// gridW = 20, gridH = 30
/// </summary>
/// <param name="blockW">The block w.</param>
/// <param name="blockH">The block h.</param>
/// <param name="gridW">The grid w.</param>
/// <param name="gridH">The grid h.</param>
/// <param name="blockW">The width of each block in terms of threads.</param>
/// <param name="blockH">The height of each block in terms of threads.</param>
/// <param name="gridW">The width of the entire grid in terms of threads.</param>
/// <param name="gridH">The width of the entire grid in terms of threads.</param>
void OpenCLWrapper::MakeEvenGridDims(size_t blockW, size_t blockH, size_t& gridW, size_t& gridH)
{
if (gridW % blockW != 0)