Blabla

Author: cirosantilli

opencl/applications.md

@@ -7,20 +7,29 @@ For an application to experience speedup compared to the CPU, it must:
 
 Minimal example request: <http://stackoverflow.com/questions/7663343/simplest-possible-example-to-show-gpu-outperform-cpu-using-cuda>
 
-## Examples
+## Specific applications
 
 -   Monte Carlo
 
 -   PDEs
 
     - <https://en.wikipedia.org/wiki/Black%E2%80%93Scholes_model>
     - Reverse Time Migration: RTM <http://www.slb.com/services/seismic/geophysical_processing_characterization/dp/technologies/depth/prestackdepth/rtm.aspx>
+    - Interactive fluid simulation <https://www.youtube.com/watch?v=LGTxZRRwvsI>
 
 -   clMathLibraries organization, by AMD employees
 
     -   <https://github.com/clMathLibraries/clFFT> FFT
     -   <https://github.com/clMathLibraries/clRNG> random number generation
 
+-   Ray tracing
+
+    <https://github.com/nthend/cltracer>
+
+-   Computer vision
+
+    -   Hough transform: <https://www.youtube.com/watch?v=vn9y7t9iqC8>
+
 ### Matrix multiplication
 
 - <http://hpclab.blogspot.fr/2011/09/is-gpu-good-for-large-vector-addition.html>
@@ -42,8 +51,26 @@ Vector addition. Too little work per input byte (1 CPU cycle). <https://forums.k
 
 ## Projects using OpenCL
 
+Master list: <https://en.wikipedia.org/wiki/List_of_OpenCL_applications>
+
 Notable users:
 
-- OpenCV
-- Bullet physics
-- VP9 decoding 2013 by Ittiam: <http://malideveloper.arm.com/partner-showroom/ittiam-vp9-decoder-using-opencl/>
+-   OpenCV
+
+-   VP9 decoding 2013 by Ittiam: <http://malideveloper.arm.com/partner-showroom/ittiam-vp9-decoder-using-opencl/>
+
+-   <https://github.com/bulletphysics/bullet3>
+
+    - <https://www.youtube.com/watch?v=8jGZv1YYe2c>
+    - <https://www.youtube.com/watch?v=-6Sl5CCxp3Q>
+    - <https://www.youtube.com/watch?v=J9HaT23b-xc>
+
+-   Scientific computation: <http://scicomp.stackexchange.com/questions/376/mathematical-libraries-for-opencl>
+
+## Application areas
+
+-   <https://streamcomputing.eu/blog/2013-06-03/the-application-areas-opencl-can-be-used/>
+
+## General purpose wrappers
+
+-   <https://github.com/arrayfire/arrayfire>

opencl/bibliography.md

@@ -53,26 +53,6 @@ Resources:
 - The specifications of your hardware, e.g. <http://www.nvidia.com/object/nvs_techspecs.html>
 - <http://www.cmsoft.com.br/> several good examples, but not always with full source? By <https://www.linkedin.com/in/douglas-coimbra-de-andrade-1241aa34> who is a Brazilian from a famous Brazilian university.
 
-Big programs that use it:
-
--   <https://github.com/bulletphysics/bullet3>
-
-    - <https://www.youtube.com/watch?v=8jGZv1YYe2c>
-    - <https://www.youtube.com/watch?v=-6Sl5CCxp3Q>
-    - <https://www.youtube.com/watch?v=J9HaT23b-xc>
-
--   <https://github.com/nthend/cltracer>
-
--   Computer vision:
-
-    - Hough transform: <https://www.youtube.com/watch?v=vn9y7t9iqC8>
-
--   Interactive fluid simulations
-
-    - <https://www.youtube.com/watch?v=LGTxZRRwvsI>
-
--   Scientific computation: <http://scicomp.stackexchange.com/questions/376/mathematical-libraries-for-opencl>
-
 ## NVIDIA samples
 
 Used to be part of the GPU computing SDK, then renamed CUDA SDK.
@@ -85,10 +65,6 @@ But a header is missing and it does not compile: <https://github.com/sschaetz/nv
 
 I'm not the only one who noticed: <https://streamcomputing.eu/blog/2012-09-10/nvidias-industry-leading-support-for-opencl/>
 
-### CUDA examples
-
-
-
 ## AMD samples
 
 <http://developer.amd.com/tools-and-sdks/opencl-zone/amd-accelerated-parallel-processing-app-sdk>

opencl/implementations.md

@@ -8,8 +8,7 @@ There is a certain "installable client driver loader (ICD loader)" which forward
 
 <http://www.x.org/wiki/XorgEVoC/GalliumCompute/>
 
-Looks like the major open source implementation.
-MIT and LLVM based.
+Looks like the major open source implementation. MIT and LLVM based.
 
 ## NVIDIA
 
@@ -33,8 +32,6 @@ Threads:
 - <http://stackoverflow.com/questions/3271243/clcreatesubbuffer-not-found-oo>
 - <https://devtalk.nvidia.com/default/topic/486564/nvidia-39-s-opencl-1-1-and-clcreatesubbuffer/>
 
-## OpenCL
-
 OpenCL 2 announced in 2017:
 
 - <http://stackoverflow.com/questions/29219307/opencl-2-0-on-nvidia-graphics-cards>
@@ -65,6 +62,8 @@ Beignet looks like an open source implementation for Intel's embedded GPU: <http
 
 ## CPU implementations
 
+## Software implementations
+
 Slow but portable, and a sanity check that OpenCL is actually making anything faster:
 
 ### Clover
@@ -79,3 +78,6 @@ Portable OpenCL implementation.
 
 <http://portablecl.org/>
 
+### Oclgrind
+
+<https://github.com/jrprice/Oclgrind>