marktwtn

> Till now, there are 152 hyperlinks which point to Microsoft website. > > ``` > $ grep microsoft sse2neon.h | wc -l > 152 > ``` I was trying...

The CI tests failed. It seems like the failed tests is related to the cross compilation since it has the `bazel` option `--crosstool_top` and `--host_crosstool_top`?

The command `$ bazel build --build_tests_only --config x86_64 //...` ran well on my desktop. However, if I tried the command like the CI runs, it failed. `$ ./tools/ci_test_setup && bazel...

> @marktwtn I was not fully wake up. The `--config x86_64` uses the toolchain provided by bazel and another is the toolchain provide by [bootlin](https://toolchains.bootlin.com/releases_x86-64-core-i7.html) for x86-64-core-i7. I wonder what...

I fixed a bug and refactored some code in the updated branch. The bug was not detected since the input data size of `trits_to_trytes` and `trytes_to_trits` testing are not large...

I found another bug and fixed it in the updated branch. Sorry for not noticing that there are other error logs showing the memory mismatching problem. I also increased the...

If I change the intrinsic function from `__mm_store_si128()` to `__mm_storeu_si128()`, the errors will be fixed. Even with the bootlin toolchain, every test is passed. The main difference of the intrinsic...

**Conclusion:** The load and store intrinsic function are unified to the unalignment type since we can not control the input pointer value. The value is based on 8-byte alignment, but...

> @marktwtn Could you write benchmarks for `trits_to_trytes`, `trytes_to_trits` and put it in `//common/trinary/benchmark`? thanks. It seems that there are no other benchmark in the other directories for me to...

> Hey, @marktwtn how are you? do you need any help on Bazel or else? @oopsmonk I found out that performance might not behave as I expected. The input size...