Memory Latency Test Changelog

What's new in Memory Latency Test 0.3.2

Oct 31, 2021
  • Added:
  • Program now prints out its version, platform, and architecture when running
  • Fixed:
  • Print out actual path used to call executable when running help command
  • Inconsistent results on i386 platforms when testing L1

New in Memory Latency Test 0.3.0 (Oct 29, 2021)

  • Added:
  • Help option (-h)
  • Option to set max region size in megabytes (-m)
  • Option to set base iterations to run for each region size (-i)
  • Option to force the C version of the test (-c)
  • Assembly version of the critical section of the test
  • 384 KB now included in default region sizes
  • Changed:
  • x86 and x86-64 platforms now run the assembly version of the test by default
  • Use 32-bit data types when running on x86
  • Use time_t for calculation step
  • Fixed:
  • Test now ensures that memory was properly allocated before proceeding
  • Errors are now printed to sterr rather than stdout

New in Memory Latency Test 0.2.0 (Oct 23, 2021)

  • Added:
  • Parameter handling - options will be added in a future release
  • "Press ENTER key to continue" on Windows
  • Changed:
  • Significantly improve precision of latency calculations
  • Don't print trailing zeroes of nanosecond counts and print 5 significant figures maximum
  • New fine-grained heuristic for scaling iteration counts based on region size
  • Removed:
  • Unused mobile wait mode

New in Memory Latency Test 0.1.1 (Oct 20, 2021)

  • Removed unused tz structs
  • Use musl-gcc to build static Linux x86_64 executables, which should now run on the vast majority of Linux-based operating systems