You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I don't think this is a blocker for me. I just hit it again and decided to report here in case it can be fixed.
I have plans to write some optimized esp32 kernels using the https://github.com/espressif/esp-dsp library so this might just be something that is better fixed that way.
/home/runner/work/tensorflow-micropython-examples/tensorflow-micropython-examples/micropython-modules/microlite/tflm/tensorflow/lite/micro/kernels/l2_pool_2d.cpp: In function 'TfLiteStatus tflite::{anonymous}::L2Eval(TfLiteContext*, TfLiteNode*)':
/home/runner/work/tensorflow-micropython-examples/tensorflow-micropython-examples/micropython-modules/microlite/tflm/tensorflow/lite/micro/kernels/l2_pool_2d.cpp:128:1: error: insn does not satisfy its constraints:
}
^
(insn 753 18 593 28 (set (reg/v:SF 20 f1 [orig:77 sum_squares ] [77])
(mem/u/c:SF (symbol_ref/u:SI ("*.LC28") [flags 0x2]) [0 S4 A32])) "/home/runner/work/tensorflow-micropython-examples/tensorflow-micropython-examples/micropython-modules/microlite/tflm/tensorflow/lite/kernels/internal/reference/pooling.h":169 47 {movsf_internal}
(expr_list:REG_EQUAL (const_double:SF 0.0 [0x0.0p+0])
(nil)))
during RTL pass: postreload
/home/runner/work/tensorflow-micropython-examples/tensorflow-micropython-examples/micropython-modules/microlite/tflm/tensorflow/lite/micro/kernels/l2_pool_2d.cpp:128:1: internal compiler error: in extract_constrain_insn, at recog.c:2210
Please submit a full bug report,
with preprocessed source if appropriate.
The text was updated successfully, but these errors were encountered:
I'm going to close this issue. Currently I plan to try and add an optimized version or at least a different implementation for esp32 that will not trigger the compiler bug.
There is a compiler bug that doesn't allow compiling l2_pool_2d.cpp at the -O2 optimization level on esp32.
jcmvbkbc/gcc-xtensa#14
The work around is to compile this file at the -O3 optimization level.
mocleiri/tensorflow-micropython-examples@0967b90
I'm restructuring how I build tflm to use the generator and then compile as part of the micropython build.
Micropython just turned on -O2 for all of the esp32 port: micropython/micropython@549448e
I don't think this is a blocker for me. I just hit it again and decided to report here in case it can be fixed.
I have plans to write some optimized esp32 kernels using the https://github.com/espressif/esp-dsp library so this might just be something that is better fixed that way.
The text was updated successfully, but these errors were encountered: