![for (i=0; i<n_pixels; i++)
{
out_pixel[0] = (in_pixel[0] - 0.5f) * contrast + brightness + 0.5;
out_pixel[1] = (in_pixel[1] - 0.5f) * contrast + brightness + 0.5;
out_pixel[2] = (in_pixel[2] - 0.5f) * contrast + brightness + 0.5;
out_pixel[3] = in_pixel[3]; /* copy the alpha */
in_pixel += 4;
out_pixel += 4;
}
__kernel void gegl_brightness_contrast(__global const float4 *in,
__global float4 *out,
float contrast,
float brightness)
{
int gid = get_global_id(0);
float4 in_v = in[gid];
float4 out_v;
out_v.xyz = (in_v.xyz - 0.5f) * contrast + brightness + 0.5f;
out_v.w = in_v.w;
out[gid] = out_v;
}](https://image.slidesharecdn.com/victor-lgm-gegl-opencl-120508004256-phpapp01/85/Implementing-OpenCL-support-in-GEGL-and-GIMP-11-320.jpg)
Implementing OpenCL support in GEGL and GIMP
- 1. OpenCL in GEGL and GIMP Victor Oliveira Libre Graphics Meeting 2012
- 2. The goats are ?nally running!
- 3. GIMP was designed to process images with 8-bits per channel. ? Artifacts ? Noise GEGL is optimized to do it in high-depth formats, like ?oating-point.
- 4. Processing images in ?oating-point is intensive. How can we improve GEGL performance?
- 5. The solution: making ?oating-poing processing faster using GPUs, multi-threading and vectorization! (All to do with OpenCL)
- 6. But what is OpenCL?
- 7. ? A language for writing functions that execute in OpenCL devices: ? GPUs ? Multi-core CPUs
- 8. ? An API to control these OpenCL devices
- 10. ? A shared library (you can use it in your C/C++ code)
- 11. for (i=0; i<n_pixels; i++) { out_pixel[0] = (in_pixel[0] - 0.5f) * contrast + brightness + 0.5; out_pixel[1] = (in_pixel[1] - 0.5f) * contrast + brightness + 0.5; out_pixel[2] = (in_pixel[2] - 0.5f) * contrast + brightness + 0.5; out_pixel[3] = in_pixel[3]; /* copy the alpha */ in_pixel += 4; out_pixel += 4; } __kernel void gegl_brightness_contrast(__global const float4 *in, __global float4 *out, float contrast, float brightness) { int gid = get_global_id(0); float4 in_v = in[gid]; float4 out_v; out_v.xyz = (in_v.xyz - 0.5f) * contrast + brightness + 0.5f; out_v.w = in_v.w; out[gid] = out_v; }
- 12. Common ?ow in a OpenCL program: ? Write code in the OpenCL language ? Detect the device ? Build your code ? Transfer data to the device ? Run your OpenCL program ? Transfer data back from device
- 13. AMD, Intel and Nvidia proprietary implementations. Mesa open-source implementation is progressing fast.
- 14. What I have done:
- 15. ? Automatic data transference to the OpenCL Device when necessary
- 16. ? Conversion between color spaces
- 17. ? A simple API for writing OpenCL ?lters in GEGL
- 18. ? Benchmarking
- 19. ? Integrating this work in the next release of GIMP (2.8)
- 20. Sharing data in the OpenCL Device among many operations
- 25. Color conversion in the OpenCL Device
- 29. Benchmarks (GEGL single-threaded x GEGL w/ OpenCL)
- 35. ? OpenCL is a portable and e?cient way to write ?lters in GEGL/GIMP.
- 36. ? There is a huge performance boost using GPUs.
- 37. ? On CPUs, we gained multi-core support for free.
- 38. ? 91 commits - 14,161 lines changed, code in GEGL 0.20, which is required by GIMP 2.8.
- 39. What is left to be done:
- 40. ? Testing by users on di?erent cards/drivers/cpus and feedback
- 41. ? Porting more ops to OpenCL
- 42. ? Multiple GPUs support
- 43. ? OpenGL output node for GEGL
- 44. ? The more GIMP uses GEGL, more performance gains
- 45. Thanks! Victor Oliveira (victorm) victormatheus@gmail.com