http://developer.amd.com/gpu/PerfStudioWorks very well with Horde. As in PIX, you can inspect drawcalls, the current API state, the active textures and you can see what is output to render targets. This makes debugging the internal graphics code a lot easier. All this works on my NVidia card. The profiling and display of internal hardware counters is limited to ATI hardware though.
To make the usage more convenient, I have added a camera freeze to the samples (hit space twice). That way you can enable the "Application Supports Self-Pause" option in PerfStudio.