package tensorflow.profiler.hlo_stats

Get desktop application:
View/edit binary Protocol Buffers messages

A database of HloStats records.

• repeated HloStatsRecord hlo_stats_record = 1

  All HloStats records, one for each HLO operation.

There is one HloStatsRecord for each HLO operation profiled. Next ID: 40

Used in: HloStatsDatabase

• optional uint64 rank = 1

  The rank by self time

• optional uint64 program_id = 30

  program_id for this op

• optional string hlo_category = 17

  The HLO category name.

• optional string hlo_expression = 2

  The HLO expression.

• optional string tf_op_name = 21

  The framework op name (TF Op, JAX Op)

• optional int64 occurrences = 3

  Number of occurrences of the operation.

• optional double total_time_in_us = 4

  Total "accumulated" time in micro-seconds that the operation took. If this operation has any children operations, the "accumulated" time includes the time spent inside children.

• optional double avg_time_in_us = 5

  Average "accumulated" time in micro-seconds that each occurrence of the operation took.

• optional double total_self_time_in_us = 6

  Total "self" time in micro-seconds that the operation took. If this operation has any children operations, the "self" time doesn't include the time spent inside children.

• optional double avg_self_time_in_us = 7

  Average "self" time in micro-seconds that the operation took.

• optional double total_self_time_as_fraction = 8

• optional double cumulative_total_self_time_as_fraction = 9

• optional double dma_stall_fraction = 10

  Percentage of the total "accumulated" time that was caused by DMA stall.

• optional double measured_flop_rate = 13

  Total floating-point operations (FLOPs) performed per second normalized to the bf16 peak capacity.

• optional double model_flop_rate = 34

  Total Floating-point operations for the op per second.

• optional double measured_memory_bw = 14

  Number of total bytes (including both read and write) accessed per second.

• optional double hbm_bw = 22

  Number of bytes accessed from HBM (including both read and write) per second.

• optional double cmem_read_bw = 23

  Number of bytes read from CMEM per second.

• optional double cmem_write_bw = 24

  Number of bytes written to CMEM per second.

• optional double vmem_read_bw = 35

  Number of bytes read from VMEM per second.

• optional double vmem_write_bw = 36

  Number of bytes written to VMEM per second.

• optional double operational_intensity = 15

  Overall operational intensity in FLOP/Byte.

• optional double hbm_operational_intensity = 26

  Operational intensity based on HBM in FLOP/Byte.

• optional double cmem_read_operational_intensity = 27

  Operational intensity based on CMEM read in FLOP/Byte.

• optional double cmem_write_operational_intensity = 28

  Operational intensity based on CMEM write in FLOP/Byte.

• optional double vmem_read_operational_intensity = 37

  Operational intensity based on VMEM read in FLOP/Byte.

• optional double vmem_write_operational_intensity = 38

  Operational intensity based on VMEM write in FLOP/Byte.

• optional double bottleneck_operational_intensity = 29

  Operational intensity based on the bottleneck resource in FLOP/Byte.

• optional string bound_by = 16

  Whether this operation is "Compute", "HBM", "CMEM Read", "CMEM Write" bound, according to the Roofline Model.

• optional bool rematerialization = 20

  Whether this operation is for HLO or Framework rematerialization.

• optional bool outside_compilation = 25

  Whether this op is for outside compilation.

• optional bool autotuned = 31

  Whether this op is autotuned.

• optional uint64 flops = 32

  Flops for the record

• optional uint64 bytes_accessed = 33

  Bytes accessed for the record

• optional SourceInfo source_info = 39

  Infrmation about the corresponding source code.