|
1 | | -Version 5.1.3 -- 2011-01-13 |
| 1 | +Version 6.0.0 -- 2025-05-21 |
2 | 2 |
|
3 | 3 | ================================================================ |
4 | 4 | To use Unixbench: |
@@ -27,6 +27,41 @@ For information on adding tests into the benchmark, see "WRITING_TESTS". |
27 | 27 |
|
28 | 28 | ===================== RELEASE NOTES ===================================== |
29 | 29 |
|
| 30 | +======================== May 21 ========================== |
| 31 | +v6.0.0 |
| 32 | + |
| 33 | +Build System Modernization |
| 34 | +Introduced support for custom compiler flags via the UB_GCC_OPTIONS environment variable. |
| 35 | +Architecture-aware compiler optimization flags (e.g., -march=native, -mcpu=native) for x86, ARM, RISC-V, PowerPC, etc. |
| 36 | +Cleaner and more portable Makefile, compatible with GNU Make. |
| 37 | +Improved separation of CFLAGS and LDFLAGS with autodetection of platform-specific needs. |
| 38 | + |
| 39 | +Platform & Architecture Enhancements |
| 40 | +Added detailed ARM64 CPU detection with support for multiple vendors (ARM, Ampere, Cavium, Apple, etc.). |
| 41 | +Improved support for macOS using sysctl and Darwin-specific logic. |
| 42 | +Portable CPU core detection using nproc, getconf, and fallback logic. |
| 43 | +Improved fallback behavior when /proc/cpuinfo is unavailable. |
| 44 | + |
| 45 | +System Detection and Configuration |
| 46 | +Run script now uses FindBin for consistent directory resolution. |
| 47 | +Default directories (pgms, results, tmp) are created automatically if missing. |
| 48 | +Log file names can now be customized via UB_OUTPUT_FILE_NAME environment variable. |
| 49 | +Improved cross-platform compatibility and shell behavior. |
| 50 | + |
| 51 | +Documentation and Licensing |
| 52 | +Added official GPL v2 license file. |
| 53 | +Rewritten and modernized README.md in Markdown format. |
| 54 | +Clear explanation of each benchmark test and history of the project. |
| 55 | +Updated contributor credits and project guidelines. |
| 56 | + |
| 57 | + |
| 58 | +Performance Comparison Guidelines |
| 59 | +To ensure meaningful and repeatable benchmarking results, it is strongly recommended to maintain consistency in the following areas when comparing performance |
| 60 | +Operating System: Use the same OS distribution and version. |
| 61 | +Compiler Options: Use consistent compiler flags across test cases (e.g., -O3, -march=native). |
| 62 | +System Configuration: Ensure similar runtime environments (CPU governor, I/O scheduler, background load). |
| 63 | +Benchmark Version: Always compare results using the same version of UnixBench |
| 64 | + |
30 | 65 | ======================== Jan 13 ========================== |
31 | 66 |
|
32 | 67 | v5.1.3 |
|
0 commit comments