membench
Memory bandwidth, Memory latency
Memory copy bandwidth (sc-membench)
Measures aggregate memory copy bandwidth across all threads using OpenMPOpenMPAn API for shared-memory parallel programming in C, C++, and Fortran, made of compiler directives, library routines, and environment variables. A program marks loops or regions to run in parallel, and the OpenMP runtime splits them across a team of threads, which can be bound to specific CPU cores.OpenMPGlossary parallelization.
Benchmark details
- Benchmark ID
membench:bandwidth_copy- Unit of measurement
- MB/s
- Server types evaluated
- 2,393
- Total measurements
- 25,660
- Range of measurements
- 940.8 – 17.1M MB/s
- Better
- ↑ Higher is better
Configuration options
| Field | Description | Example values |
|---|---|---|
scope | RAMRAM: Random-Access MemoryThe main memory of a computer, which holds the running programs and their data. It is much faster than disk storage, and its content is lost when the power is off.WikipediaGlossary: working set exceeds total CPUCPU: Central Processing UnitThe general-purpose processor of a computer, which runs the operating system and programs. A CPU has one or more physical cores, and with SMT each core runs two or more hardware threads.WikipediaGlossary cache (L3, else L2, else L1d). | RAM |
size_kb | Per-thread buffer size in KiB. | 16 32 64 128 256 512 1024 2048 4096 8192 16384 32768 65536 131072 262144 524288 1048576 2097152 |
Score distribution (25,660 measurements)
Memory read bandwidth (sc-membench)
Measures aggregate sequential read bandwidth across all threads using OpenMPOpenMPAn API for shared-memory parallel programming in C, C++, and Fortran, made of compiler directives, library routines, and environment variables. A program marks loops or regions to run in parallel, and the OpenMP runtime splits them across a team of threads, which can be bound to specific CPU cores.OpenMPGlossary parallelization.
Benchmark details
- Benchmark ID
membench:bandwidth_read- Unit of measurement
- MB/s
- Server types evaluated
- 2,393
- Total measurements
- 25,660
- Range of measurements
- 1.9K – 37M MB/s
- Better
- ↑ Higher is better
Configuration options
| Field | Description | Example values |
|---|---|---|
scope | RAM: working set exceeds total CPU cache (L3, else L2, else L1d). | RAM |
size_kb | Per-thread buffer size in KiB. | 16 32 64 128 256 512 1024 2048 4096 8192 16384 32768 65536 131072 262144 524288 1048576 2097152 |
Score distribution (25,660 measurements)
Memory write bandwidth (sc-membench)
Measures aggregate sequential write bandwidth across all threads using OpenMPOpenMPAn API for shared-memory parallel programming in C, C++, and Fortran, made of compiler directives, library routines, and environment variables. A program marks loops or regions to run in parallel, and the OpenMP runtime splits them across a team of threads, which can be bound to specific CPU cores.OpenMPGlossary parallelization.
Benchmark details
- Benchmark ID
membench:bandwidth_write- Unit of measurement
- MB/s
- Server types evaluated
- 2,393
- Total measurements
- 25,660
- Range of measurements
- 1.7K – 40.6M MB/s
- Better
- ↑ Higher is better
Configuration options
| Field | Description | Example values |
|---|---|---|
scope | RAM: working set exceeds total CPU cache (L3, else L2, else L1d). | RAM |
size_kb | Per-thread buffer size in KiB. | 16 32 64 128 256 512 1024 2048 4096 8192 16384 32768 65536 131072 262144 524288 1048576 2097152 |
Score distribution (25,660 measurements)
Memory latency (sc-membench)
Measures median memory latency using pointer chasing with randomized access to defeat hardware prefetching.
Benchmark details
- Benchmark ID
membench:latency- Unit of measurement
- ns
- Server types evaluated
- 2,393
- Total measurements
- 25,660
- Range of measurements
- 0.73 – 860.6 ns
- Better
- ↓ Lower is better
Configuration options
| Field | Description | Example values |
|---|---|---|
scope | RAM: working set exceeds total CPU cache (L3, else L2, else L1d). | RAM |
size_kb | Per-thread buffer size in KiB. | 16 32 64 128 256 512 1024 2048 4096 8192 16384 32768 65536 131072 262144 524288 1048576 2097152 |
Score distribution (25,660 measurements)