Hercules vs z/PDT - some observations
I received some time ago results from a handful of
s370_perf
runs done on a z/PDT V1.7 system. Took some time to analyze,
the data and full analysis are now under
Performance testing of z/PDT is explicitly disallowed in the z/PDT license conditions. s370_perf would be not the proper tool anyway. The page cited above merely gives some observations on general features of z/PDT, the key ones are repeated here
- z/PDT is based on on-the-fly binary translation, see section background.
- z/PDT does an optimizing compilation, see section code optimization.
- the same code is sometimes compiled, sometimes not, see section to compile or not to compile.
- if code is compiled, compilation can happen with substantial delay, see section compilation delay.
- performance in plain interpretive mode seems similar to Hercules, see section interpreter mode performance.
- RR instructions are the easy part, see section RR instructions.
- RX instructions are the hard part, see section RX instructions.
- z/PDT vs Hercules comparisons are difficult to interpret, see section z/PDT vs Hercules.
- little gain for floating-point arithmetic, see section floating-point performance.
- a bit faster for decimal packed arithmetic, see section decimal packed performance.
EXis apparently always interpreted, see section EX instruction.- z/PDT performs well for some instructions which are slow on Hercules, see sections CLCL, MVCIN, and TRT.
- overall JIT gain depends heavily on workload, see bottom line.
For any further background and detail follow the links.
Any remarks and comments are very welcome.
2018-12-27: MVCIN, CLCL and TRT performance significantly improved
in Hercules Release V4.1, see
this short note.
For original posting to
Yahoo! Group - Hercules-390
see
topic 83576.
Dead link since 2020-12-15: Yahoo! Groups
was discontinued by Verizon.