Benchmark Command (b)

The b (benchmark) command tests CPU performance and compression efficiency by compressing and decompressing test data.

Syntax

7z b [number_of_iterations] [<switches>...]

Description

The benchmark command:

Measures compression and decompression speed
Tests CPU performance under load
Validates multi-threading efficiency
Generates reproducible performance data
Uses synthetic test data (does not require files)
Reports speed in MB/s and MIPS (million instructions per second)

Common Options

number

Number of benchmark passes to run. More iterations provide more accurate results.Default: Runs until Ctrl+C or sufficient data collectedExample: 7z b 10 (run 10 iterations)

number

Set number of CPU threads to use.

-mmt1 - Single thread
-mmt2 - Two threads
-mmt=on - All available threads (default)

Example: -mmt=4

string

Set dictionary size for testing.Example: -md=32m (32 MB dictionary)

string

Set compression method to test.Example: -m0=LZMA2 or -m0=PPMd

Examples

Basic benchmark

7z b

Output:

7-Zip 23.01 (x64) : Copyright (c) 1999-2023 Igor Pavlov : 2023-06-20

Intel(R) Core(TM) i7-9750H CPU @ 2.60GHz (6C12T)
RAM size:    16384 MB,  # CPU hardware threads:  12
RAM usage:    3418 MB,  # Benchmark threads:     12

                       Compressing  |                  Decompressing
Dict     Speed Usage    R/U Rating  |      Speed Usage    R/U Rating
         KB/s     %   MIPS   MIPS  |       KB/s     %   MIPS   MIPS

22:      30145  1200   2445  29336  |     389234  1200   2766  33195
23:      29234  1200   2481  29775  |     382145  1200   2751  33018
24:      27823  1199   2498  29952  |     374234  1199   2738  32826
25:      26234  1199   2491  29868  |     361234  1198   2712  32488
-----------------------------------|-----------------------------------
Avg:             1200   2479  29733  |              1199   2742  32882
Tot:             1199   2610  31307

Benchmark with specific thread count

7z b -mmt=4

Benchmark single-threaded performance

7z b -mmt=1

Run specific number of iterations

7z b 10

Benchmark with custom dictionary size

7z b -md=64m

Benchmark specific compression method

7z b -m0=LZMA2

Benchmark PPMd method

7z b -m0=PPMd

Quick benchmark (fewer passes)

7z b 3

Understanding Benchmark Output

Header Information

Intel(R) Core(TM) i7-9750H CPU @ 2.60GHz (6C12T)

CPU model and specifications
Core count (6C) and thread count (12T)

RAM size:    16384 MB,  # CPU hardware threads:  12
RAM usage:    3418 MB,  # Benchmark threads:     12

Total system RAM
Available hardware threads
RAM used for benchmark
Threads used in test

Performance Metrics

number

Dictionary size (2^N bytes). Larger = more memory, better compression.Example: 22 = 2^22 = 4 MB

number

Throughput in kilobytes per second.

number

CPU utilization percentage. 100% per thread.Example: 1200% = 12 threads at 100%

number

Performance rating in Million Instructions Per Second:

R (Rating) - Single-thread equivalent performance
U (Usage) - Multi-thread total performance

Compression vs Decompression

Compressing - Tests encoding speed
Decompressing - Tests decoding speed
Decompression is typically faster than compression

Average Scores

Avg:             1200   2479  29733  |              1199   2742  32882
Tot:             1199   2610  31307

Avg - Average for compression and decompression separately
Tot - Combined total rating (overall performance)

Advanced Benchmarking

Test Multi-Threading Scaling

#!/bin/bash
for threads in 1 2 4 8 12; do
    echo "Testing with $threads threads:"
    7z b -mmt=$threads 3 | grep "^Tot:"
done

Compare Compression Methods

echo "LZMA2:"
7z b -m0=LZMA2 5 | grep "^Tot:"

echo "PPMd:"
7z b -m0=PPMd 5 | grep "^Tot:"

echo "BZip2:"
7z b -m0=BZip2 5 | grep "^Tot:"

Test Different Dictionary Sizes

for size in 16m 32m 64m 128m; do
    echo "Dictionary: $size"
    7z b -md=$size 3 | grep "^Tot:"
done

Continuous Stress Test

# Run indefinitely until interrupted
7z b

Quick Performance Check

7z b 1 | tail -1

Use Cases

System Performance Testing

Test CPU and memory subsystem:

7z b -mmt=on 10

Thermal Testing

Run extended benchmark to test cooling:

# Run for extended period
7z b 100

Multi-Threading Efficiency

Compare single vs multi-threaded:

echo "Single thread:"
7z b -mmt=1 5 | grep "^Tot:"

echo "All threads:"
7z b -mmt=on 5 | grep "^Tot:"

Memory Bandwidth Test

Test with large dictionary:

7z b -md=256m -mmt=1

Interpreting Results

Good Performance Indicators

High CPU Usage - Near 100% per thread (e.g., 1200% for 12 threads)
Consistent Speeds - Little variation between iterations
Scaling - Multi-thread rating ≈ single-thread × thread count

Performance Issues

Low Usage - CPU not fully utilized (thermal throttling, power limits)
Varying Speeds - Inconsistent results (background processes)
Poor Scaling - Multi-thread not scaling linearly (memory bandwidth limit)

Benchmark Comparison

Entry-Level CPU

Tot:              300    800   4800

Older or low-power CPU
4 threads, moderate performance

Mid-Range CPU

Tot:              800   1800  14400

Modern mid-range processor
8 threads, good performance

High-End CPU

Tot:             1600   3500  28000

High-end desktop or workstation
16+ threads, excellent performance

Server CPU

Tot:             3200   7000  56000

Multi-socket server
32+ threads, extreme performance

Scripting Examples

Save Benchmark Results

#!/bin/bash
DATE=$(date +%Y%m%d-%H%M%S)
OUTPUT="benchmark-${DATE}.txt"

echo "7-Zip Benchmark - $(date)" > "$OUTPUT"
uname -a >> "$OUTPUT"
echo "" >> "$OUTPUT"
7z b 10 >> "$OUTPUT" 2>&1
echo "Results saved to $OUTPUT"

Compare Before/After Performance

#!/bin/bash
# Save baseline
echo "Baseline test:"
7z b 5 | grep "^Tot:" > baseline.txt

# Make system changes here
# ...

# Test again
echo "After changes:"
7z b 5 | grep "^Tot:" > after.txt

# Compare
echo "Baseline:"
cat baseline.txt
echo "After:"
cat after.txt

Automated Performance Log

#!/bin/bash
# Run weekly benchmark and log results

LOGFILE="/var/log/7zip-benchmark.log"

echo "=== Benchmark $(date) ===" >> "$LOGFILE"
7z b 5 >> "$LOGFILE" 2>&1
echo "" >> "$LOGFILE"

Performance Tuning

Optimize for Speed

7z b -mmt=on -md=16m

Smaller dictionary = faster, less memory

Optimize for Compression

7z b -mmt=on -md=128m

Larger dictionary = better compression, more memory

Test Specific Scenario

# Simulate typical 7z compression settings
7z b -m0=LZMA2 -md=32m -mmt=on

Benchmark results are relative to the specific CPU and system configuration. They provide a comparison metric but don’t directly predict real-world file compression times.

For consistent results:

Close other applications
Disable CPU frequency scaling (set to performance mode)
Run multiple iterations and average results
Ensure adequate cooling

Extended benchmarking can generate significant heat. Ensure your system has adequate cooling before running long benchmark sessions.

Add Command (a) - Real-world compression testing
Hash Command (h) - Test hashing performance

​Syntax

​Description

​Common Options

​Examples

​Basic benchmark

​Benchmark with specific thread count

​Benchmark single-threaded performance

​Run specific number of iterations

​Benchmark with custom dictionary size

​Benchmark specific compression method

​Benchmark PPMd method

​Quick benchmark (fewer passes)

​Understanding Benchmark Output

​Header Information

​Performance Metrics

​Compression vs Decompression

​Average Scores

​Advanced Benchmarking

​Test Multi-Threading Scaling

​Compare Compression Methods

​Test Different Dictionary Sizes

​Continuous Stress Test

​Quick Performance Check

​Use Cases

​System Performance Testing

​Thermal Testing

​Multi-Threading Efficiency

​Memory Bandwidth Test

​Interpreting Results

​Good Performance Indicators

​Performance Issues

​Benchmark Comparison

​Entry-Level CPU

​Mid-Range CPU

​High-End CPU

​Server CPU

​Scripting Examples

​Save Benchmark Results

​Compare Before/After Performance

​Automated Performance Log

​Performance Tuning

​Optimize for Speed

​Optimize for Compression

​Test Specific Scenario

​Related Commands

Syntax

Description

Common Options

Examples

Basic benchmark

Benchmark with specific thread count

Benchmark single-threaded performance

Run specific number of iterations

Benchmark with custom dictionary size

Benchmark specific compression method

Benchmark PPMd method

Quick benchmark (fewer passes)

Understanding Benchmark Output

Header Information

Performance Metrics

Compression vs Decompression

Average Scores

Advanced Benchmarking

Test Multi-Threading Scaling

Compare Compression Methods

Test Different Dictionary Sizes

Continuous Stress Test

Quick Performance Check

Use Cases

System Performance Testing

Thermal Testing

Multi-Threading Efficiency

Memory Bandwidth Test

Interpreting Results

Good Performance Indicators

Performance Issues

Benchmark Comparison

Entry-Level CPU

Mid-Range CPU

High-End CPU

Server CPU

Scripting Examples

Save Benchmark Results

Compare Before/After Performance

Automated Performance Log

Performance Tuning

Optimize for Speed

Optimize for Compression

Test Specific Scenario

Related Commands