| Age | Commit message (Collapse) | Author |
|---|
|
This changes the native emitter to access qstr values using the qstr
indirection table qstr_table, but only when generating native code that
will be saved to a .mpy file. This makes the resulting native code fully
static, ie it does not require any fix-ups or rewriting when it is
imported.
The performance of native code is more or less unchanged. Benchmark
results on PYBv1.0 (using --via-mpy and --emit native) are:
N=100 M=100 baseline -> this-commit diff diff% (error%)
bm_chaos.py 407.16 -> 411.85 : +4.69 = +1.152% (+/-0.01%)
bm_fannkuch.py 100.89 -> 101.20 : +0.31 = +0.307% (+/-0.01%)
bm_fft.py 3521.17 -> 3441.72 : -79.45 = -2.256% (+/-0.00%)
bm_float.py 6707.29 -> 6644.83 : -62.46 = -0.931% (+/-0.00%)
bm_hexiom.py 55.91 -> 55.41 : -0.50 = -0.894% (+/-0.00%)
bm_nqueens.py 5343.54 -> 5326.17 : -17.37 = -0.325% (+/-0.00%)
bm_pidigits.py 603.89 -> 632.79 : +28.90 = +4.786% (+/-0.33%)
core_qstr.py 64.18 -> 64.09 : -0.09 = -0.140% (+/-0.01%)
core_yield_from.py 313.61 -> 311.11 : -2.50 = -0.797% (+/-0.03%)
misc_aes.py 654.29 -> 659.75 : +5.46 = +0.834% (+/-0.02%)
misc_mandel.py 4205.10 -> 4272.08 : +66.98 = +1.593% (+/-0.01%)
misc_pystone.py 3077.79 -> 3128.39 : +50.60 = +1.644% (+/-0.01%)
misc_raytrace.py 388.45 -> 393.71 : +5.26 = +1.354% (+/-0.01%)
viper_call0.py 576.83 -> 566.76 : -10.07 = -1.746% (+/-0.05%)
viper_call1a.py 550.39 -> 540.12 : -10.27 = -1.866% (+/-0.11%)
viper_call1b.py 438.32 -> 432.09 : -6.23 = -1.421% (+/-0.11%)
viper_call1c.py 442.96 -> 436.11 : -6.85 = -1.546% (+/-0.08%)
viper_call2a.py 536.31 -> 527.37 : -8.94 = -1.667% (+/-0.04%)
viper_call2b.py 378.99 -> 377.50 : -1.49 = -0.393% (+/-0.08%)
Signed-off-by: Damien George <damien@micropython.org>
|
|
This patch adds full support for unwinding jumps to the native emitter.
This means that return/break/continue can be used in try-except,
try-finally and with statements. For code that doesn't use unwinding jumps
there is almost no overhead added to the generated code.
|
|
Prior to this patch, native code would use a full nlr_buf_t for each
exception handler (try-except, try-finally, with). For nested exception
handlers this would use a lot of C stack and be rather inefficient.
This patch changes how exceptions are handled in native code by setting up
only a single nlr_buf_t context for the entire function, and then manages a
state machine (using the PC) to work out which exception handler to run
when an exception is raised by an nlr_jump. This keeps the C stack usage
at a constant level regardless of the depth of Python exception blocks.
The patch also fixes an existing bug when local variables are written to
within an exception handler, then their value was incorrectly restored if
an exception was raised (since the nlr_jump would restore register values,
back to the point of the nlr_push).
And it also gets nested try-finally+with working with the viper emitter.
Broadly speaking, efficiency of executing native code that doesn't use
any exception blocks is unchanged, and emitted code size is only slightly
increased for such function. C stack usage of all native functions is
either equal or less than before. Emitted code size for native functions
that use exception blocks is increased by roughly 10% (due in part to
fixing of above-mentioned bugs).
But, most importantly, this patch allows to implement more Python features
in native code, like unwind jumps and yielding from within nested exception
blocks.
|
|
Instead of emitnative.c having configuration code for each supported
architecture, and then compiling this file multiple times with different
macros defined, this patch adds a file per architecture with the necessary
code to configure the native emitter. These files then #include the
emitnative.c file.
This simplifies emitnative.c (which is already very large), and simplifies
the build system because emitnative.c no longer needs special handling for
compilation and qstr extraction.
|
