Fix system sol units, remove units if None value. (#773)
* Fix system sol units.
[ROCm/rocprofiler-compute commit: 74e9853f8e]
Этот коммит содержится в:
+9
-9
@@ -22,56 +22,56 @@ Panel Config:
|
||||
metric:
|
||||
VALU FLOPs:
|
||||
value: None # No perf counter
|
||||
unit: GFLOPs
|
||||
unit: GFLOP/s
|
||||
peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
|
||||
pop: None # No perf counter
|
||||
tips:
|
||||
VALU IOPs:
|
||||
value: None # No perf counter
|
||||
unit: GIOPs
|
||||
unit: GIOP/s
|
||||
peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
|
||||
pop: None # No perf counter
|
||||
tips:
|
||||
MFMA FLOPs (F8):
|
||||
value: None # No HW module
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: None # No HW module
|
||||
pop: None # No HW module
|
||||
tips:
|
||||
MFMA FLOPs (BF16):
|
||||
value: None # No perf counter
|
||||
unit: GFLOPs
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 512) / 1000)
|
||||
pop: None # No perf counter
|
||||
tips:
|
||||
MFMA FLOPs (F16):
|
||||
value: None # No perf counter
|
||||
unit: GFLOPs
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 1024) / 1000)
|
||||
pop: None # No perf counter
|
||||
tips:
|
||||
MFMA FLOPs (F32):
|
||||
value: None # No perf counter
|
||||
unit: GFLOPs
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
|
||||
pop: None # No perf counter
|
||||
tips:
|
||||
MFMA FLOPs (F64):
|
||||
value: None # No perf counter
|
||||
unit: GFLOPs
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
|
||||
pop: None # No perf counter
|
||||
tips:
|
||||
# TODO: Fix baseline comparision logic to handle non existent metrics, then remove this
|
||||
MFMA FLOPs (F6F4):
|
||||
value: None
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: None
|
||||
pop: None
|
||||
tips:
|
||||
MFMA IOPs (Int8):
|
||||
value: None # No perf counter
|
||||
unit: GOPs
|
||||
unit: GIOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 1024) / 1000)
|
||||
pop: None # No perf counter
|
||||
tips:
|
||||
|
||||
+9
-9
@@ -22,56 +22,56 @@ Panel Config:
|
||||
metric:
|
||||
VALU FLOPs:
|
||||
value: None # No perf counter
|
||||
unit: GFLOPs
|
||||
unit: GFLOP/s
|
||||
peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
|
||||
pop: None # No perf counter
|
||||
tips:
|
||||
VALU IOPs:
|
||||
value: None # No perf counter
|
||||
unit: GIOPs
|
||||
unit: GIOP/s
|
||||
peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
|
||||
pop: None # No perf counter
|
||||
tips:
|
||||
MFMA FLOPs (F8):
|
||||
value: None # No HW module
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: None # No HW module
|
||||
pop: None # No HW module
|
||||
tips:
|
||||
MFMA FLOPs (BF16):
|
||||
value: None # No perf counter
|
||||
unit: GFLOPs
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 512) / 1000)
|
||||
pop: None # No perf counter
|
||||
tips:
|
||||
MFMA FLOPs (F16):
|
||||
value: None # No perf counter
|
||||
unit: GFLOPs
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 1024) / 1000)
|
||||
pop: None # No perf counter
|
||||
tips:
|
||||
MFMA FLOPs (F32):
|
||||
value: None # No perf counter
|
||||
unit: GFLOPs
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
|
||||
pop: None # No perf counter
|
||||
tips:
|
||||
MFMA FLOPs (F64):
|
||||
value: None # No perf counter
|
||||
unit: GFLOPs
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
|
||||
pop: None # No perf counter
|
||||
tips:
|
||||
# TODO: Fix baseline comparision logic to handle non existent metrics, then remove this
|
||||
MFMA FLOPs (F6F4):
|
||||
value: None
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: None
|
||||
pop: None
|
||||
tips:
|
||||
MFMA IOPs (Int8):
|
||||
value: None # No perf counter
|
||||
unit: GOPs
|
||||
unit: GIOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 1024) / 1000)
|
||||
pop: None # No perf counter
|
||||
tips:
|
||||
|
||||
+9
-9
@@ -26,7 +26,7 @@ Panel Config:
|
||||
+ SQ_INSTS_VALU_TRANS_F32) + (2 * SQ_INSTS_VALU_FMA_F32)))) + (64 * (((SQ_INSTS_VALU_ADD_F64
|
||||
+ SQ_INSTS_VALU_MUL_F64) + SQ_INSTS_VALU_TRANS_F64) + (2 * SQ_INSTS_VALU_FMA_F64))))
|
||||
/ (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
|
||||
pop: ((100 * AVG(((((64 * (((SQ_INSTS_VALU_ADD_F16 + SQ_INSTS_VALU_MUL_F16)
|
||||
+ SQ_INSTS_VALU_TRANS_F16) + (2 * SQ_INSTS_VALU_FMA_F16))) + (64 * (((SQ_INSTS_VALU_ADD_F32
|
||||
@@ -37,41 +37,41 @@ Panel Config:
|
||||
tips:
|
||||
VALU IOPs:
|
||||
value: AVG(((64 * (SQ_INSTS_VALU_INT32 + SQ_INSTS_VALU_INT64)) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GIOP
|
||||
unit: GIOP/s
|
||||
peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
|
||||
pop: ((100 * AVG(((64 * (SQ_INSTS_VALU_INT32 + SQ_INSTS_VALU_INT64)) / (End_Timestamp
|
||||
- Start_Timestamp)))) / (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000))
|
||||
tips:
|
||||
MFMA FLOPs (F8):
|
||||
value: None
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: None
|
||||
pop: None
|
||||
tips:
|
||||
MFMA FLOPs (BF16):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_BF16 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 1024) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_BF16 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 1024) / 1000))
|
||||
tips:
|
||||
MFMA FLOPs (F16):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F16 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 1024) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F16 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 1024) / 1000))
|
||||
tips:
|
||||
MFMA FLOPs (F32):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F32 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F32 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 256) / 1000))
|
||||
tips:
|
||||
MFMA FLOPs (F64):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F64 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F64 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 256) / 1000))
|
||||
@@ -79,13 +79,13 @@ Panel Config:
|
||||
# TODO: Fix baseline comparision logic to handle non existent metrics, then remove this
|
||||
MFMA FLOPs (F6F4):
|
||||
value: None
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: None
|
||||
pop: None
|
||||
tips:
|
||||
MFMA IOPs (Int8):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_I8 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GIOP
|
||||
unit: GIOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 1024) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_I8 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 1024) / 1000))
|
||||
|
||||
+9
-9
@@ -26,7 +26,7 @@ Panel Config:
|
||||
+ SQ_INSTS_VALU_TRANS_F32) + (2 * SQ_INSTS_VALU_FMA_F32)))) + (64 * (((SQ_INSTS_VALU_ADD_F64
|
||||
+ SQ_INSTS_VALU_MUL_F64) + SQ_INSTS_VALU_TRANS_F64) + (2 * SQ_INSTS_VALU_FMA_F64))))
|
||||
/ (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
|
||||
pop: ((100 * AVG(((((64 * (((SQ_INSTS_VALU_ADD_F16 + SQ_INSTS_VALU_MUL_F16)
|
||||
+ SQ_INSTS_VALU_TRANS_F16) + (2 * SQ_INSTS_VALU_FMA_F16))) + (64 * (((SQ_INSTS_VALU_ADD_F32
|
||||
@@ -37,42 +37,42 @@ Panel Config:
|
||||
tips:
|
||||
VALU IOPs:
|
||||
value: AVG(((64 * (SQ_INSTS_VALU_INT32 + SQ_INSTS_VALU_INT64)) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GIOP
|
||||
unit: GIOP/s
|
||||
peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
|
||||
pop: ((100 * AVG(((64 * (SQ_INSTS_VALU_INT32 + SQ_INSTS_VALU_INT64)) / (End_Timestamp
|
||||
- Start_Timestamp)))) / (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000))
|
||||
tips:
|
||||
MFMA FLOPs (F8):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F8 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 4096) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F8 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 4096) / 1000))
|
||||
tips:
|
||||
MFMA FLOPs (BF16):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_BF16 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 2048) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_BF16 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 2048) / 1000))
|
||||
tips:
|
||||
MFMA FLOPs (F16):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F16 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 2048) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F16 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 2048) / 1000))
|
||||
tips:
|
||||
MFMA FLOPs (F32):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F32 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F32 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 256) / 1000))
|
||||
tips:
|
||||
MFMA FLOPs (F64):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F64 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F64 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 256) / 1000))
|
||||
@@ -80,13 +80,13 @@ Panel Config:
|
||||
# TODO: Fix baseline comparision logic to handle non existent metrics, then remove this
|
||||
MFMA FLOPs (F6F4):
|
||||
value: None
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: None
|
||||
pop: None
|
||||
tips:
|
||||
MFMA IOPs (Int8):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_I8 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GIOP
|
||||
unit: GIOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 4096) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_I8 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 4096) / 1000))
|
||||
|
||||
+9
-9
@@ -26,7 +26,7 @@ Panel Config:
|
||||
+ SQ_INSTS_VALU_TRANS_F32) + (2 * SQ_INSTS_VALU_FMA_F32)))) + (64 * (((SQ_INSTS_VALU_ADD_F64
|
||||
+ SQ_INSTS_VALU_MUL_F64) + SQ_INSTS_VALU_TRANS_F64) + (2 * SQ_INSTS_VALU_FMA_F64))))
|
||||
/ (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
|
||||
pop: ((100 * AVG(((((64 * (((SQ_INSTS_VALU_ADD_F16 + SQ_INSTS_VALU_MUL_F16)
|
||||
+ SQ_INSTS_VALU_TRANS_F16) + (2 * SQ_INSTS_VALU_FMA_F16))) + (64 * (((SQ_INSTS_VALU_ADD_F32
|
||||
@@ -37,42 +37,42 @@ Panel Config:
|
||||
tips:
|
||||
VALU IOPs:
|
||||
value: AVG(((64 * (SQ_INSTS_VALU_INT32 + SQ_INSTS_VALU_INT64)) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GIOP
|
||||
unit: GIOP/s
|
||||
peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
|
||||
pop: ((100 * AVG(((64 * (SQ_INSTS_VALU_INT32 + SQ_INSTS_VALU_INT64)) / (End_Timestamp
|
||||
- Start_Timestamp)))) / (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000))
|
||||
tips:
|
||||
MFMA FLOPs (F8):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F8 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 4096) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F8 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 4096) / 1000))
|
||||
tips:
|
||||
MFMA FLOPs (BF16):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_BF16 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 2048) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_BF16 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 2048) / 1000))
|
||||
tips:
|
||||
MFMA FLOPs (F16):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F16 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 2048) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F16 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 2048) / 1000))
|
||||
tips:
|
||||
MFMA FLOPs (F32):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F32 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F32 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 256) / 1000))
|
||||
tips:
|
||||
MFMA FLOPs (F64):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F64 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F64 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 256) / 1000))
|
||||
@@ -80,13 +80,13 @@ Panel Config:
|
||||
# TODO: Fix baseline comparision logic to handle non existent metrics, then remove this
|
||||
MFMA FLOPs (F6F4):
|
||||
value: None
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: None
|
||||
pop: None
|
||||
tips:
|
||||
MFMA IOPs (Int8):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_I8 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GIOP
|
||||
unit: GIOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 4096) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_I8 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 4096) / 1000))
|
||||
|
||||
+9
-9
@@ -26,7 +26,7 @@ Panel Config:
|
||||
+ SQ_INSTS_VALU_TRANS_F32) + (2 * SQ_INSTS_VALU_FMA_F32)))) + (64 * (((SQ_INSTS_VALU_ADD_F64
|
||||
+ SQ_INSTS_VALU_MUL_F64) + SQ_INSTS_VALU_TRANS_F64) + (2 * SQ_INSTS_VALU_FMA_F64))))
|
||||
/ (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
|
||||
pop: ((100 * AVG(((((64 * (((SQ_INSTS_VALU_ADD_F16 + SQ_INSTS_VALU_MUL_F16)
|
||||
+ SQ_INSTS_VALU_TRANS_F16) + (2 * SQ_INSTS_VALU_FMA_F16))) + (64 * (((SQ_INSTS_VALU_ADD_F32
|
||||
@@ -37,42 +37,42 @@ Panel Config:
|
||||
tips:
|
||||
VALU IOPs:
|
||||
value: AVG(((64 * (SQ_INSTS_VALU_INT32 + SQ_INSTS_VALU_INT64)) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GIOP
|
||||
unit: GIOP/s
|
||||
peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
|
||||
pop: ((100 * AVG(((64 * (SQ_INSTS_VALU_INT32 + SQ_INSTS_VALU_INT64)) / (End_Timestamp
|
||||
- Start_Timestamp)))) / (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000))
|
||||
tips:
|
||||
MFMA FLOPs (F8):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F8 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 4096) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F8 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 4096) / 1000))
|
||||
tips:
|
||||
MFMA FLOPs (BF16):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_BF16 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 2048) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_BF16 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 2048) / 1000))
|
||||
tips:
|
||||
MFMA FLOPs (F16):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F16 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 2048) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F16 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 2048) / 1000))
|
||||
tips:
|
||||
MFMA FLOPs (F32):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F32 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F32 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 256) / 1000))
|
||||
tips:
|
||||
MFMA FLOPs (F64):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F64 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F64 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 256) / 1000))
|
||||
@@ -80,13 +80,13 @@ Panel Config:
|
||||
# TODO: Fix baseline comparision logic to handle non existent metrics, then remove this
|
||||
MFMA FLOPs (F6F4):
|
||||
value: None
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: None
|
||||
pop: None
|
||||
tips:
|
||||
MFMA IOPs (Int8):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_I8 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GIOP
|
||||
unit: GIOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 4096) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_I8 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 4096) / 1000))
|
||||
|
||||
+9
-9
@@ -26,7 +26,7 @@ Panel Config:
|
||||
+ SQ_INSTS_VALU_TRANS_F32) + (2 * SQ_INSTS_VALU_FMA_F32)))) + (64 * (((SQ_INSTS_VALU_ADD_F64
|
||||
+ SQ_INSTS_VALU_MUL_F64) + SQ_INSTS_VALU_TRANS_F64) + (2 * SQ_INSTS_VALU_FMA_F64))))
|
||||
/ (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
|
||||
pop: ((100 * AVG(((((64 * (((SQ_INSTS_VALU_ADD_F16 + SQ_INSTS_VALU_MUL_F16)
|
||||
+ SQ_INSTS_VALU_TRANS_F16) + (2 * SQ_INSTS_VALU_FMA_F16))) + (64 * (((SQ_INSTS_VALU_ADD_F32
|
||||
@@ -37,56 +37,56 @@ Panel Config:
|
||||
tips:
|
||||
VALU IOPs:
|
||||
value: AVG(((64 * (SQ_INSTS_VALU_INT32 + SQ_INSTS_VALU_INT64)) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GIOP
|
||||
unit: GIOP/s
|
||||
peak: (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000)
|
||||
pop: ((100 * AVG(((64 * (SQ_INSTS_VALU_INT32 + SQ_INSTS_VALU_INT64)) / (End_Timestamp
|
||||
- Start_Timestamp)))) / (((($max_sclk * $cu_per_gpu) * 64) * 2) / 1000))
|
||||
tips:
|
||||
MFMA FLOPs (F8):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F8 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 4096) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F8 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 8192) / 1000))
|
||||
tips:
|
||||
MFMA FLOPs (BF16):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_BF16 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 4096) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_BF16 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 4096) / 1000))
|
||||
tips:
|
||||
MFMA FLOPs (F16):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F16 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 4096) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F16 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 4096) / 1000))
|
||||
tips:
|
||||
MFMA FLOPs (F32):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F32 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F32 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 256) / 1000))
|
||||
tips:
|
||||
MFMA FLOPs (F64):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F64 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F64 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 256) / 1000))
|
||||
tips:
|
||||
MFMA FLOPs (F6F4):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_F6F4 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GFLOP
|
||||
unit: GFLOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 256) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_F6F4 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 256) / 1000))
|
||||
tips:
|
||||
MFMA IOPs (Int8):
|
||||
value: AVG(((SQ_INSTS_VALU_MFMA_MOPS_I8 * 512) / (End_Timestamp - Start_Timestamp)))
|
||||
unit: GIOP
|
||||
unit: GIOP/s
|
||||
peak: ((($max_sclk * $cu_per_gpu) * 8192) / 1000)
|
||||
pop: ((100 * AVG(((SQ_INSTS_VALU_MFMA_MOPS_I8 * 512) / (End_Timestamp - Start_Timestamp))))
|
||||
/ ((($max_sclk * $cu_per_gpu) * 8192) / 1000))
|
||||
|
||||
Ссылка в новой задаче
Block a user