You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: README.md
+5-5Lines changed: 5 additions & 5 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -79,11 +79,11 @@ will produce what is referred to as the "Public" version of GPUPerfAPI. This ver
79
79
using a set of hardware counters. Until now, only the Public the version of GPUPerfAPI was available on the AMD Developer website. As part of the open-source effort,
80
80
we are also providing the ability to build the "Internal" versions of GPUPerfAPI. In addition to exposing the same counters as the Public version, the Internal version
81
81
also exposes some of the hardware Counters available in the GPU/APU. It's important to note that not all hardware counters receive the same validation as other parts of
82
-
the hardware on all GPUs, so in some cases accuracy of counter data cannot be guaranteed. The usage of the Internal version is identical to the Public version. The only difference will be in the
83
-
name of the library an application loads at runtime and the list of counters exposed by the library. See the [Build Instructions](BUILD.md) for more information on how
84
-
to build and use the Internal version. In the future, we see there being only a single version of GPUPerfAPI, with perhaps a change in the API to allow users of GPA to
85
-
indicate whether the library exposes just the Derived counters or both the Derived and the Hardware counters. We realize using the term "Internal" for something which is
86
-
no longer actually Internal-to-AMD can be a bit confusing, and we will aim to change this in the future.
82
+
the hardware on all GPUs, so in some cases accuracy of counter data cannot be guaranteed. The usage of the Internal version is identical to the Public version. The only
83
+
difference will be in the name of the library an application loads at runtime and the list of counters exposed by the library. See the [Build Instructions](BUILD.md) for
84
+
more information on how to build and use the Internal version. In the future, we see there being only a single version of GPUPerfAPI, with perhaps a change in the API to
85
+
allow users of GPA to indicate whether the library exposes just the Derived counters or both the Derived and the Hardware counters. We realize using the term "Internal"
86
+
for something which is no longer actually Internal-to-AMD can be a bit confusing, and we will aim to change this in the future.
0 commit comments