PyPy JIT Now Running Well On 64-Bit ARM For Faster Performance
Thanks to funding from Arm Holdings and Crossbar, the PyPy folks working on their speedy Python JIT implementation have extended it to support 64-bit ARM (AArch64) with compelling performance results.
This performance-oriented Python implementation now supports x86, x86_64, PowerPC 64-bit, s390, ARM 32-bit, and now ARM 64-bit.
Like with other architectures, the early results from this AArch64 port of PyPy are very promising: "In the majority of benchmarks, the speedups achieved on AArch64 match those achieved on the x86_64 laptop. Over CPython, PyPy on AArch64 achieves speedups between 0.6x to 44.9x. These speedups are comparable to x86_64, where the numbers are between 0.6x and 58.9x."
Even still, the PyPy developers know of more areas to squeeze extra performance out of AArch64 and other modern CPUs moving forward. At this point PyPy is still engineered with x86 32-bit in mind but could improve code generation for newer CPUs especially those with more registers.
Those wanting to learn more about PyPy for AArch64 can do so via the project's blog.
This performance-oriented Python implementation now supports x86, x86_64, PowerPC 64-bit, s390, ARM 32-bit, and now ARM 64-bit.
Like with other architectures, the early results from this AArch64 port of PyPy are very promising: "In the majority of benchmarks, the speedups achieved on AArch64 match those achieved on the x86_64 laptop. Over CPython, PyPy on AArch64 achieves speedups between 0.6x to 44.9x. These speedups are comparable to x86_64, where the numbers are between 0.6x and 58.9x."
Even still, the PyPy developers know of more areas to squeeze extra performance out of AArch64 and other modern CPUs moving forward. At this point PyPy is still engineered with x86 32-bit in mind but could improve code generation for newer CPUs especially those with more registers.
Those wanting to learn more about PyPy for AArch64 can do so via the project's blog.
Add A Comment