LO, CHIN-HAO
LO, CHIN-HAO
When building `llama.cpp` with Metal support on my Mac M1, the `ggml-metal.metal` file is placed in the `bin` directory correctly. However, when building it as a shared library: the [`pathForResource`](https://github.com/ggerganov/llama.cpp/blob/213d1439fadefe182f69c5f7e8dd3b4b6572ebcb/ggml-metal.m#L285C38-L285C53)...
Fixes #2615 The problem is that `_lpython_str_compare_eq` requires pointer to pointer to char as parameters. However, the previous implementation passed a null pointer, leading to a segmentation fault.
``` def main0(): s_var1: str = "Hello" s_var2: str = "LPython" print(s_var1 or s_var2) (lp) hank@MacBook-Pro lpython % ./src/bin/lpython test.py --backend=c expr2__tmp__generated__.c: In function ‘main0’: expr2__tmp__generated__.c:21:14: warning: format ‘%s’ expects...
In Linux, we can read `/proc/meminfo` to get the huge page size. Please node that this will get only **one** possible size, but Linux can supports more than one huge...