v0.2.4

This release contains bug fixes for Neoverse V1 and Neoverse V2 detection, and adds a function that returns the brand string of the host CPU.

What's Changed

• Add a function returning the brand string of the host by @alalazo in #156
• Update neoverse-v2 detection by @alalazo in #157
• Improve Microarchitecture.optimization_flags errors by @alalazo in #155
• Update neoverse-v1 detection by @alalazo in #158

Full Changelog: v0.2.3...v0.2.4

v0.2.3

This release:

• Adds the ability to detect microarchitectures on x86_64-Windows machines.
• Allows customizing the JSON files being used via environment variables

What's Changed

• Add unit tests to source distribution by @alalazo in #148
• Refactor and clean-up of detection by @alalazo in #149
• Add support for CPU detection on Windows by @alalazo and @isuruf in #150
• Add a Microarchitecture.from_dict constructor by @alalazo in #153
• Allow specifying a user defined JSON file by @alalazo in #146

Full Changelog: v0.2.2...v0.2.3

v0.2.2

This release introduces support for:

• Sapphire Rapids (#124)
• Neoverse V2 (#131)
• Power 10 (#130)

Other changes

• Ensure compiler versions are ranges by @alalazo in #117
• Cache the computation of Microarchitecture.ancestors by @alalazo in #122
• Add readthedocs configuration file by @alalazo in #126

Full Changelog: v0.2.1...v0.2.2

v0.2.1

This release introduces support for:

• The Zen 4 microarchitecture (#105,#115)

and adds compiler flags for the nvhpc compiler (#112). It also drops the dependency on click for the command line tool (#109)

Other changes

• Add unit tests for a64fx detection by @alalazo in #104
• Update TOX configuration to avoid use of deprecated options by @alalazo in #107
• Update JSON file with Apple M2 fixes by @alalazo in #111

New Contributors

• @althonos made their first contribution in #109

Full Changelog: v0.2.0...v0.2.1

v0.2.0

This release introduces support for:

• Neoverse v1 and generic Armv8.X architectures
• Apple M2

Breaking changes

• Compared to v0.1, the naming scheme for graviton chips changed to use generic microarchitecture names. This means that graviton is now named cortex_a72, graviton2 is now named neoverse_n1 and instead of graviton3 we introduced neoverse_v1 (due to being based on archspec-json v0.2)
• Support for Python 2.7 and Python 3.5 has been dropped (#103)

Features

• Added support for Python 3.11 (#103)
• Speedup microarchitecture comparison (#96)

Bug fixes

• Fixed wrong compiler flags for the Intel classic compiler when targeting x86_64 (due to being based on archspec-json v0.2)
• Removed the umip feature from icelake and cannonlake (due to being based on archspec-json v0.2)

v0.1.4

This release improves the detection and support for the Apple M1 CPU, both on macOS and on Asahi linux based systems. It also:

• Adds compiler flags for oneAPI Intel compilers
• Improves compiler flags for a64fx

Note: archspec was maintaining an "unofficial" compatibility with Python 2.6. This has been dropped in 478594c.

Full Changelog: v0.1.3...v0.1.4

v0.1.3

This release adds support for x86-64 generic architectures and introduces the following new microarchitectures:

• Apple M1
• Zen 3
• RiscV and SiFive u74mc

It also improves compilers support in particular for AMD and ARM compilers.

Full Changelog: v0.1.2...v0.1.3

v0.1.2

This release adds support for Graviton and Graviton 2 microarchitectures. Other changes:

• Added support for Apple Clang optimization flags on x86_64 microarchitectures
• Added a Dockerfile to package archspec into a container image
• A few misnamed features for icelake have been fixed

Basic command line interface

This release adds a basic command line interface based on click. Right now it just provides ways to query the current host architecture:

$ archspec cpu
broadwell

but will be expanded in the near future to give direct access to more features that are already in archspec's API and make it a valuable tool for scripting. Other modifications in this release are:

• Compilers are not distinguished by version suffix anymore, see #19
• JSON data file has been updated to latest version

Initial release of Archspec

The initial release of Archspec provides support for detection and manipulation of 43 CPU microarchitectures including:

• All Intel x86_64 microarchitectures up to icelake
• AMD Family 10h, 15h, 17h processors
• Power 7, 8 and 9
• ThunderX2 and a64fx processors