doc: doc fixups

Author: vpirogov

doc/mainpage.md

@@ -19,6 +19,7 @@ on Intel CPUs and GPUs.
 
  * @ref dev_guide_basic_concepts
  * @ref cpu_getting_started_cpp
+ * @ref gpu_getting_started_cpp
  * @ref cpu_memory_format_propagation_cpp
  * @ref dev_guide_inference_and_training_aspects
    * @ref dev_guide_inference
@@ -30,8 +31,6 @@ on Intel CPUs and GPUs.
    * @ref dev_guide_attributes_post_ops
  * @ref dev_guide_data_types
  * @ref dev_guide_c_and_cpp_apis
- * @ref dev_guide_opencl_interoperability
-
 
 # Primitives
 
@@ -53,7 +52,6 @@ Memory bandwidth limited operations:
 Data manipulation:
  * [Reorder](@ref dev_guide_reorder)
 
-
 # Performance Benchmarking and Inspection
 
  * @ref dev_guide_verbose
@@ -67,6 +65,7 @@ Data manipulation:
  * @ref dev_guide_transition_to_v1
  * @ref dev_guide_understanding_memory_formats
  * @ref dev_guide_int8_computations
+ * @ref dev_guide_opencl_interoperability
 
 # Examples
 
@@ -76,7 +75,7 @@ Data manipulation:
 |                |          | @ref cpu_memory_format_propagation_cpp |                              |
 |                |          | @ref cpu_performance_profiling_cpp     |                              |
 |                | GPU      | @ref gpu_getting_started_cpp           | @ref gpu_getting_started_c   |
-|                | GPU      | @ref gpu_opencl_interop_cpp            | 			          |
+|                | GPU      | @ref gpu_opencl_interop_cpp            |                              |
 | f32 inference  | CPU      | @ref cpu_cnn_inference_f32_cpp         | @ref cpu_cnn_inference_f32_c |
 |                | CPU      | @ref cpu_rnn_inference_f32_cpp         |                              |
 | int8 inference | CPU      | @ref cpu_cnn_inference_int8_cpp        |                              |

examples/cpu_cnn_inference_f32.cpp

@@ -19,7 +19,6 @@
 /// > Annotated version: @ref cpu_cnn_inference_f32_cpp
 
 /// @page cpu_cnn_inference_f32_cpp CNN f32 inference example
-
 /// This C++ API example demonstrates how to build an AlexNet neural
 /// network topology for forward-pass inference.
 ///

tests/benchdnn/doc/driver_conv.md

@@ -66,15 +66,12 @@ The table below shows supported name configurations for this driver:
 | s8   | s8   | s32  | s32  | s8s8s32s32      | same as above
 | s8   | s8   | s8   | s32  | s8s8s8s32       | same as above
 | s8   | s8   | u8   | s32  | s8s8u8s32       | same as above
-
 | f32  | f32  | f32  | f32  | f32_wino        | Winograd-based convolution.
 | u8   | s8   | f32  | s32  | u8s8f32s32_wino | sabe as above
 | u8   | s8   | s32  | s32  | u8s8s32s32_wino | same as above
 | u8   | s8   | s8   | s32  | u8s8s8s32_wino  | same as above
 | u8   | s8   | u8   | s32  | u8s8u8s32_wino  | same as above
-
 | f16  | f16  | f16  | f16  | f16             | Only for GPU
-
 | bf16 | bf16 | bf16 | f32  | bf16bf16bf16    | optimized for processors with support of avx512vl + VNNI
 | bf16 | bf16 | f32  | f32  | bf16bf16f32     | same as above
 | bf16 | f32  | bf16 | f32  | bf16f32bf16     | same as above

tests/benchdnn/doc/driver_rnn.md

@@ -51,7 +51,6 @@ The table below shows supported name configurations for this driver:
 | u8     | u8    | u8        | f32            | u8u8u8f32   | TBA
 | f32    | u8    | f32       | u8             | f32u8f32u8  | TBA
 | f32    | u8    | f32       | f32            | f32u8f32f32 | TBA
-
 | f16    | f16   | f16       | f16            | f16         | Only for GPU
 
 

tests/benchdnn/doc/knobs_perf_report.md

@@ -104,4 +104,3 @@ not a special symbol here; any other delimiter can be used:
 Output template: %desc%,%-time%,%-Gflops%
 mb112oc1000ic2048n"resnet:ip1",0.521973,878.881
 ```
-