azure / AI-benchmarking-guide
File Size

The distribution of size of files (measured in lines of code).

Intro
Learn more...
File Size Overall
0% | 0% | 0% | 54% | 45%
Legend:
1001+
501-1000
201-500
101-200
1-100


explore: grouped by folders | grouped by size | sunburst | 3D view
File Size per Extension
1001+
501-1000
201-500
101-200
1-100
py0% | 0% | 0% | 54% | 45%
cfg0% | 0% | 0% | 0% | 100%
File Size per Logical Decomposition
primary
1001+
501-1000
201-500
101-200
1-100
Benchmarks0% | 0% | 0% | 54% | 45%
ROOT0% | 0% | 0% | 58% | 41%
Infra0% | 0% | 0% | 0% | 100%
Longest Files (Top 20)
File# lines# units
138 10
GEMMCublasLt.py
in Benchmarks/NVIDIA
 135 6
HBMBandwidth.py
in Benchmarks/NVIDIA
 123 8
CPUStream.py
in Benchmarks/NVIDIA
 114 8
RCCLBandwidth.py
in Benchmarks/AMD
 105 8
LLMBenchmark.py
in Benchmarks/NVIDIA
 103 6
GEMMHipblasLt.py
in Benchmarks/AMD
 101 7
96 8
HBMBandwidth.py
in Benchmarks/AMD
 94 8
LLMBenchmark.py
in Benchmarks/AMD
 93 5
NCCLBandwidth.py
in Benchmarks/NVIDIA
 84 7
NVBandwidth.py
in Benchmarks/NVIDIA
 54 6
TransferBench.py
in Benchmarks/AMD
 53 7
FlashAttention.py
in Benchmarks/AMD
 49 3
FIO.py
in Benchmarks/AMD
 40 2
FIO.py
in Benchmarks/NVIDIA
 38 2
Multichase.py
in Benchmarks/NVIDIA
 36 3
FlashAttention.py
in Benchmarks/NVIDIA
 26 2
tools.py
in Infra
 19 3
cfg
transferbench.cfg
in Benchmarks/AMD
 2 -
Files With Most Units (Top 19)
File# lines# units
138 10
96 8
CPUStream.py
in Benchmarks/NVIDIA
 114 8
HBMBandwidth.py
in Benchmarks/NVIDIA
 123 8
RCCLBandwidth.py
in Benchmarks/AMD
 105 8
HBMBandwidth.py
in Benchmarks/AMD
 94 8
NCCLBandwidth.py
in Benchmarks/NVIDIA
 84 7
GEMMHipblasLt.py
in Benchmarks/AMD
 101 7
TransferBench.py
in Benchmarks/AMD
 53 7
LLMBenchmark.py
in Benchmarks/NVIDIA
 103 6
GEMMCublasLt.py
in Benchmarks/NVIDIA
 135 6
NVBandwidth.py
in Benchmarks/NVIDIA
 54 6
LLMBenchmark.py
in Benchmarks/AMD
 93 5
Multichase.py
in Benchmarks/NVIDIA
 36 3
FlashAttention.py
in Benchmarks/AMD
 49 3
tools.py
in Infra
 19 3
FlashAttention.py
in Benchmarks/NVIDIA
 26 2
FIO.py
in Benchmarks/NVIDIA
 38 2
FIO.py
in Benchmarks/AMD
 40 2
Files With Long Lines (Top 17)

There are 17 files with lines longer than 120 characters. In total, there are 56 long lines.

File# lines# units# long lines
LLMBenchmark.py
in Benchmarks/NVIDIA
 103 6 11
NCCLBandwidth.py
in Benchmarks/NVIDIA
 84 7 7
96 8 5
NVBandwidth.py
in Benchmarks/NVIDIA
 54 6 5
RCCLBandwidth.py
in Benchmarks/AMD
 105 8 5
138 10 3
GEMMHipblasLt.py
in Benchmarks/AMD
 101 7 3
TransferBench.py
in Benchmarks/AMD
 53 7 3
LLMBenchmark.py
in Benchmarks/AMD
 93 5 3
FlashAttention.py
in Benchmarks/NVIDIA
 26 2 2
HBMBandwidth.py
in Benchmarks/AMD
 94 8 2
FlashAttention.py
in Benchmarks/AMD
 49 3 2
CPUStream.py
in Benchmarks/NVIDIA
 114 8 1
Multichase.py
in Benchmarks/NVIDIA
 36 3 1
GEMMCublasLt.py
in Benchmarks/NVIDIA
 135 6 1
FIO.py
in Benchmarks/NVIDIA
 38 2 1
FIO.py
in Benchmarks/AMD
 40 2 1
Correlations

File Size vs. Commits (all time): 20 points

Benchmarks/NVIDIA/NCCLBandwidth.py x: 9 commits (all time) y: 84 lines of code Benchmarks/NVIDIA/GEMMCublasLt.py x: 6 commits (all time) y: 135 lines of code Benchmarks/NVIDIA/Multichase.py x: 3 commits (all time) y: 36 lines of code NVIDIA_runner.py x: 9 commits (all time) y: 138 lines of code Benchmarks/NVIDIA/LLMBenchmark.py x: 6 commits (all time) y: 103 lines of code Benchmarks/NVIDIA/CPUStream.py x: 2 commits (all time) y: 114 lines of code Benchmarks/NVIDIA/NVBandwidth.py x: 7 commits (all time) y: 54 lines of code AMD_runner.py x: 12 commits (all time) y: 96 lines of code Benchmarks/AMD/RCCLBandwidth.py x: 7 commits (all time) y: 105 lines of code Benchmarks/AMD/HBMBandwidth.py x: 7 commits (all time) y: 94 lines of code Benchmarks/AMD/LLMBenchmark.py x: 9 commits (all time) y: 93 lines of code Benchmarks/AMD/GEMMHipblasLt.py x: 8 commits (all time) y: 101 lines of code Benchmarks/AMD/FlashAttention.py x: 10 commits (all time) y: 49 lines of code Benchmarks/AMD/FIO.py x: 4 commits (all time) y: 40 lines of code Infra/tools.py x: 4 commits (all time) y: 19 lines of code Benchmarks/AMD/TransferBench.py x: 8 commits (all time) y: 53 lines of code Benchmarks/NVIDIA/FIO.py x: 4 commits (all time) y: 38 lines of code Benchmarks/NVIDIA/FlashAttention.py x: 5 commits (all time) y: 26 lines of code Benchmarks/NVIDIA/HBMBandwidth.py x: 4 commits (all time) y: 123 lines of code Benchmarks/AMD/transferbench.cfg x: 1 commits (all time) y: 2 lines of code
138.0
lines of code
  min: 2.0
  average: 75.15
  25th percentile: 38.5
  median: 88.5
  75th percentile: 104.5
  max: 138.0
0 12.0
commits (all time)
min: 1.0 | average: 6.25 | 25th percentile: 4.0 | median: 6.5 | 75th percentile: 8.75 | max: 12.0

File Size vs. Contributors (all time): 20 points

Benchmarks/NVIDIA/NCCLBandwidth.py x: 4 contributors (all time) y: 84 lines of code Benchmarks/NVIDIA/GEMMCublasLt.py x: 3 contributors (all time) y: 135 lines of code Benchmarks/NVIDIA/Multichase.py x: 2 contributors (all time) y: 36 lines of code NVIDIA_runner.py x: 5 contributors (all time) y: 138 lines of code Benchmarks/NVIDIA/LLMBenchmark.py x: 3 contributors (all time) y: 103 lines of code Benchmarks/NVIDIA/CPUStream.py x: 2 contributors (all time) y: 114 lines of code Benchmarks/NVIDIA/NVBandwidth.py x: 4 contributors (all time) y: 54 lines of code AMD_runner.py x: 4 contributors (all time) y: 96 lines of code Benchmarks/AMD/RCCLBandwidth.py x: 3 contributors (all time) y: 105 lines of code Benchmarks/AMD/HBMBandwidth.py x: 3 contributors (all time) y: 94 lines of code Benchmarks/AMD/LLMBenchmark.py x: 3 contributors (all time) y: 93 lines of code Benchmarks/AMD/GEMMHipblasLt.py x: 2 contributors (all time) y: 101 lines of code Benchmarks/AMD/FlashAttention.py x: 3 contributors (all time) y: 49 lines of code Benchmarks/AMD/FIO.py x: 3 contributors (all time) y: 40 lines of code Infra/tools.py x: 1 contributors (all time) y: 19 lines of code Benchmarks/AMD/TransferBench.py x: 3 contributors (all time) y: 53 lines of code Benchmarks/NVIDIA/FIO.py x: 3 contributors (all time) y: 38 lines of code Benchmarks/NVIDIA/FlashAttention.py x: 3 contributors (all time) y: 26 lines of code Benchmarks/NVIDIA/HBMBandwidth.py x: 3 contributors (all time) y: 123 lines of code Benchmarks/AMD/transferbench.cfg x: 1 contributors (all time) y: 2 lines of code
138.0
lines of code
  min: 2.0
  average: 75.15
  25th percentile: 38.5
  median: 88.5
  75th percentile: 104.5
  max: 138.0
0 5.0
contributors (all time)
min: 1.0 | average: 2.9 | 25th percentile: 2.25 | median: 3.0 | 75th percentile: 3.0 | max: 5.0

File Size vs. Commits (30 days): 4 points

Benchmarks/NVIDIA/NCCLBandwidth.py x: 1 commits (30d) y: 84 lines of code Benchmarks/NVIDIA/GEMMCublasLt.py x: 1 commits (30d) y: 135 lines of code Benchmarks/NVIDIA/Multichase.py x: 1 commits (30d) y: 36 lines of code NVIDIA_runner.py x: 1 commits (30d) y: 138 lines of code
138.0
lines of code
  min: 36.0
  average: 98.25
  25th percentile: 48.0
  median: 109.5
  75th percentile: 137.25
  max: 138.0
0 1.0
commits (30d)
min: 1.0 | average: 1.0 | 25th percentile: 1.0 | median: 1.0 | 75th percentile: 1.0 | max: 1.0

File Size vs. Contributors (30 days): 4 points

Benchmarks/NVIDIA/NCCLBandwidth.py x: 1 contributors (30d) y: 84 lines of code Benchmarks/NVIDIA/GEMMCublasLt.py x: 1 contributors (30d) y: 135 lines of code Benchmarks/NVIDIA/Multichase.py x: 1 contributors (30d) y: 36 lines of code NVIDIA_runner.py x: 1 contributors (30d) y: 138 lines of code
138.0
lines of code
  min: 36.0
  average: 98.25
  25th percentile: 48.0
  median: 109.5
  75th percentile: 137.25
  max: 138.0
0 1.0
contributors (30d)
min: 1.0 | average: 1.0 | 25th percentile: 1.0 | median: 1.0 | 75th percentile: 1.0 | max: 1.0

File Size vs. Commits (90 days): 19 points

Benchmarks/NVIDIA/NCCLBandwidth.py x: 7 commits (90d) y: 84 lines of code Benchmarks/NVIDIA/GEMMCublasLt.py x: 4 commits (90d) y: 135 lines of code Benchmarks/NVIDIA/Multichase.py x: 3 commits (90d) y: 36 lines of code NVIDIA_runner.py x: 6 commits (90d) y: 138 lines of code Benchmarks/NVIDIA/LLMBenchmark.py x: 4 commits (90d) y: 103 lines of code Benchmarks/NVIDIA/CPUStream.py x: 2 commits (90d) y: 114 lines of code Benchmarks/NVIDIA/NVBandwidth.py x: 5 commits (90d) y: 54 lines of code AMD_runner.py x: 7 commits (90d) y: 96 lines of code Benchmarks/AMD/RCCLBandwidth.py x: 5 commits (90d) y: 105 lines of code Benchmarks/AMD/HBMBandwidth.py x: 5 commits (90d) y: 94 lines of code Benchmarks/AMD/LLMBenchmark.py x: 6 commits (90d) y: 93 lines of code Benchmarks/AMD/GEMMHipblasLt.py x: 6 commits (90d) y: 101 lines of code Benchmarks/AMD/FlashAttention.py x: 7 commits (90d) y: 49 lines of code Benchmarks/AMD/FIO.py x: 3 commits (90d) y: 40 lines of code Infra/tools.py x: 3 commits (90d) y: 19 lines of code Benchmarks/AMD/TransferBench.py x: 4 commits (90d) y: 53 lines of code Benchmarks/NVIDIA/FIO.py x: 2 commits (90d) y: 38 lines of code Benchmarks/NVIDIA/FlashAttention.py x: 3 commits (90d) y: 26 lines of code Benchmarks/NVIDIA/HBMBandwidth.py x: 3 commits (90d) y: 123 lines of code
138.0
lines of code
  min: 19.0
  average: 79.0
  25th percentile: 40.0
  median: 93.0
  75th percentile: 105.0
  max: 138.0
0 7.0
commits (90d)
min: 2.0 | average: 4.47 | 25th percentile: 3.0 | median: 4.0 | 75th percentile: 6.0 | max: 7.0

File Size vs. Contributors (90 days): 19 points

Benchmarks/NVIDIA/NCCLBandwidth.py x: 3 contributors (90d) y: 84 lines of code Benchmarks/NVIDIA/GEMMCublasLt.py x: 2 contributors (90d) y: 135 lines of code Benchmarks/NVIDIA/Multichase.py x: 2 contributors (90d) y: 36 lines of code NVIDIA_runner.py x: 3 contributors (90d) y: 138 lines of code Benchmarks/NVIDIA/LLMBenchmark.py x: 2 contributors (90d) y: 103 lines of code Benchmarks/NVIDIA/CPUStream.py x: 2 contributors (90d) y: 114 lines of code Benchmarks/NVIDIA/NVBandwidth.py x: 3 contributors (90d) y: 54 lines of code AMD_runner.py x: 2 contributors (90d) y: 96 lines of code Benchmarks/AMD/RCCLBandwidth.py x: 2 contributors (90d) y: 105 lines of code Benchmarks/AMD/HBMBandwidth.py x: 2 contributors (90d) y: 94 lines of code Benchmarks/AMD/LLMBenchmark.py x: 2 contributors (90d) y: 93 lines of code Benchmarks/AMD/GEMMHipblasLt.py x: 2 contributors (90d) y: 101 lines of code Benchmarks/AMD/FlashAttention.py x: 2 contributors (90d) y: 49 lines of code Benchmarks/AMD/FIO.py x: 2 contributors (90d) y: 40 lines of code Infra/tools.py x: 1 contributors (90d) y: 19 lines of code Benchmarks/AMD/TransferBench.py x: 2 contributors (90d) y: 53 lines of code Benchmarks/NVIDIA/FIO.py x: 2 contributors (90d) y: 38 lines of code Benchmarks/NVIDIA/FlashAttention.py x: 2 contributors (90d) y: 26 lines of code Benchmarks/NVIDIA/HBMBandwidth.py x: 2 contributors (90d) y: 123 lines of code
138.0
lines of code
  min: 19.0
  average: 79.0
  25th percentile: 40.0
  median: 93.0
  75th percentile: 105.0
  max: 138.0
0 3.0
contributors (90d)
min: 1.0 | average: 2.11 | 25th percentile: 2.0 | median: 2.0 | 75th percentile: 2.0 | max: 3.0