I'm setting it in Shader Analyzer, in the compilation settings, in "Additional Custom Arguments". And I'm getting this error:
*************** Build Began for 1 Devices***************
X.hlsl(57,10-20): error X1507: failed to open source file: 'Y.hlsl'
And Y.hlsl is inside of the path specified in the commandline ("-I <path>).
Thanks!
Solved! Go to Solution.
I've been off hobnobbing with this engineering team. Here's what we have.
We're treating this like a bug for now. At the very least, the -I command does not work as expected, or as the documentation describes.
The work around for this is to explicitly include the file in your code. Modify the Shader code’s “#include” line to point to the local include file.
// add or modify this
#include “C:\temp\MyIncludes.h”
Using your file(s) obviously. Or use a relative path.
I've been off hobnobbing with this engineering team. Here's what we have.
We're treating this like a bug for now. At the very least, the -I command does not work as expected, or as the documentation describes.
The work around for this is to explicitly include the file in your code. Modify the Shader code’s “#include” line to point to the local include file.
// add or modify this
#include “C:\temp\MyIncludes.h”
Using your file(s) obviously. Or use a relative path.
Thanks for the reply.
Solving this would be a huge productivity boost.
Specially when you work with version control software that make files read-only, using the suggested workaround means to check out multiple files (in my case I've many nested files that needs to have the includes replaced), and the chances of messing/breaking something. Having the ability to just open your pixel shader file and have it compiled would be a huge time saver. I could for example work in visual studio compiling for the visuals, then check the performance implications both with timers (in the application) and assembly (in GPU PerfStudio); this all without having to constantly copy/paste code into the GPU PerfStudio.
Thanks again for such a great tool.
Update: The team tells me the bug is fixed, and fix is checked in. They're synching up a couple of others and expect we may see an updated version soon. I never predict dates.
The current version is 3.2.16, released April 1. In a week or two, perhaps, go visit http://developer.amd.com/tools-and-sdks/graphics-development/gpu-perfstudio/
See if there's an updated release. If so, it should resolve this problem.
Thanks for the report. I wish things were always this easy.
This is awesome, thanks