SWDEV-518214: GPU Metrics 1.8 (#31)

* SWDEV-518214: GPU Metrics 1.8 (#31)

- Updates:
    - Adding the following metrics to allow new calculations for violation status:
        - Per XCP metrics gfx_below_host_limit_ppt_acc
        - Per XCP metrics gfx_below_host_limit_thm_acc
        - Per XCP metrics gfx_low_utilization_acc
        - Per XCP metrics gfx_below_host_limit_total_acc
    - Increasing available JPEG engines to 40. Current ASICs may not support all 40. These will be indicated as UINT16_MAX or N/A in CLI.

Signed-off-by: Juan Castillo <juan.castillo@amd.com>
Co-authored-by: Charis Poag <Charis.Poag@amd.com>

[ROCm/rocm_smi_lib commit: f69e65f7bd]

projects/rocm-smi-lib/python_smi_tools/rocm_smi.py

@@ -3795,9 +3795,41 @@ def showGPUMetrics(deviceList):
             },
             "xcp_stats.gfx_busy_acc": {
                 "value": gpu_metrics.xcp_stats,
-                "unit": percent_unit,
+                "unit": count,
             },
             "xcp_stats.gfx_below_host_limit_acc": {
+                "value": gpu_metrics.xcp_stats,
+                "unit": count,
+            },
+             "xcp_stats.gfx_below_host_limit_ppt_acc": {
+                "value": gpu_metrics.xcp_stats,
+                "unit": count,
+            },
+            "xcp_stats.gfx_below_host_limit_thm_acc": {
+                "value": gpu_metrics.xcp_stats,
+                "unit": count,
+            },
+            "xcp_stats.gfx_low_utilization_acc": {
+                "value": gpu_metrics.xcp_stats,
+                "unit": count,
+            },
+            "xcp_stats.gfx_below_host_limit_total_acc": {
+                "value": gpu_metrics.xcp_stats,
+                "unit": count,
+            },
+            "xcp_stats.gfx_below_host_limit_ppt_acc": {
+                "value": gpu_metrics.xcp_stats,
+                "unit": percent_unit,
+            },
+            "xcp_stats.gfx_below_host_limit_thm_acc": {
+                "value": gpu_metrics.xcp_stats,
+                "unit": percent_unit,
+            },
+            "xcp_stats.gfx_low_utilization_acc": {
+                "value": gpu_metrics.xcp_stats,
+                "unit": percent_unit,
+            },
+            "xcp_stats.gfx_below_host_limit_total_acc": {
                 "value": gpu_metrics.xcp_stats,
                 "unit": percent_unit,
             },
@@ -3841,6 +3873,30 @@ def showGPUMetrics(deviceList):
                         for _, val in enumerate(item.gfx_below_host_limit_acc):
                             print_xcp_detail.append(validateIfMaxUint(val, UIntegerTypes.UINT64_T))
                         printLog(device, k + " (" + str(v["unit"]) + ")", str(print_xcp_detail), xcp=str(curr_xcp))
+                if 'xcp_stats.gfx_below_host_limit_ppt_acc' in k:
+                    for curr_xcp, item in enumerate(v['value']):
+                        print_xcp_detail = []
+                        for _, val in enumerate(item.gfx_below_host_limit_ppt_acc):
+                            print_xcp_detail.append(validateIfMaxUint(val, UIntegerTypes.UINT64_T))
+                        printLog(device, k + " (" + str(v["unit"]) + ")", str(print_xcp_detail), xcp=str(curr_xcp))
+                if 'xcp_stats.gfx_below_host_limit_thm_acc' in k:
+                    for curr_xcp, item in enumerate(v['value']):
+                        print_xcp_detail = []
+                        for _, val in enumerate(item.gfx_below_host_limit_thm_acc):
+                            print_xcp_detail.append(validateIfMaxUint(val, UIntegerTypes.UINT64_T))
+                        printLog(device, k + " (" + str(v["unit"]) + ")", str(print_xcp_detail), xcp=str(curr_xcp))
+                if 'xcp_stats.gfx_low_utilization_acc' in k:
+                    for curr_xcp, item in enumerate(v['value']):
+                        print_xcp_detail = []
+                        for _, val in enumerate(item.gfx_low_utilization_acc):
+                            print_xcp_detail.append(validateIfMaxUint(val, UIntegerTypes.UINT64_T))
+                        printLog(device, k + " (" + str(v["unit"]) + ")", str(print_xcp_detail), xcp=str(curr_xcp))
+                if 'xcp_stats.gfx_below_host_limit_total_acc' in k:
+                    for curr_xcp, item in enumerate(v['value']):
+                        print_xcp_detail = []
+                        for _, val in enumerate(item.gfx_below_host_limit_total_acc):
+                            print_xcp_detail.append(validateIfMaxUint(val, UIntegerTypes.UINT64_T))
+                        printLog(device, k + " (" + str(v["unit"]) + ")", str(print_xcp_detail), xcp=str(curr_xcp))
 
             if int(device) < (len(deviceList) - 1):
                 printLogSpacer()

projects/rocm-smi-lib/python_smi_tools/rsmiBindings.py

@@ -662,10 +662,14 @@ class amdgpu_xcp_metrics_t(Structure):
 # amdgpu_xcp_metrics_t._pack_ = 1 # source:False
 amdgpu_xcp_metrics_t._fields_ = [
     ('gfx_busy_inst', c_uint32 * 8),
-    ('jpeg_busy', c_uint16 * 32),
+    ('jpeg_busy', c_uint16 * 40),
     ('vcn_busy', c_uint16 * 4),
     ('gfx_busy_acc', c_uint64 * 8),
     ('gfx_below_host_limit_acc', c_uint64 * 8),
+    ('gfx_below_host_limit_ppt_acc', c_uint64 * 8),
+    ('gfx_below_host_limit_thm_acc', c_uint64 * 8),
+    ('gfx_low_utilization_acc', c_uint64 * 8),
+    ('gfx_below_host_limit_total_acc', c_uint64 * 8),
 ]
 xcp_stats_t = amdgpu_xcp_metrics_t