* Very minor speedup via simd-group synchronization in f16 x f32
* Another very minor speedup on metal
* Quite significant PP speedup on metal
* Another attempt
* Minor
* Massive improvement for TG for fp16
* ~4-5% improvement for Q8_0 TG on metal
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* make : remove unused -DGGML_BIG_ENDIAN
* make : put preprocessor stuff in CPPFLAGS
* make : pass Raspberry Pi arch flags to g++ as well
* make : support overriding CFLAGS/CXXFLAGS/CPPFLAGS/LDFLAGS
* make : fix inverted conditional
"bug fix, probably solves the 'ggml_allocr_alloc: not enough space in the buffer' issue"
"alloc was freeing an externally allocated tensor, because it calculated the end of allocator memory as alloc->data + alloc->max_size instead of alloc->data + alloc->size."
This is intentional to reduce the risk of freeing external tensors when measuring. Unless max_size is not properly calculated, I don't see why this is an issue.
* ggml_metal_init: Show all Metal device instances in the system
Also show the default Metal device that was picked.
* Update ggml-metal.m
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* k-quants : fix build on armv7
* ggml : cleanup unused arm32 specific impl
* k-quants : avoid some unused vzero / mzero define
* ggml-alloc : use 4g for MEASURE_MAX_SIZE in 32-bit arm
* Allow quantize tool to only copy tensors to allow repackaging models.
* Slightly better logic when requantizing.
* Change help message to go to `stdout`.
remove '--n_examples N' parameter, as it no longer makes sense to call optimization process multiple times in a loop.
add '--only_write_lora' command line option: will skip tokenization and training, to only write a llama.cpp comptabile LORA adapter.
remove memory buffer related command line options.
improve iteration console output.
* llama2c : fix segfault if vocab is not found
* llama2c : fix mismatch between new[] and delete
* llama2c : fix basename on Windows
* llama2c : use a destructor to prevent memory leaks
* Somewhat faster f16 x f32 matrix multiply kernel
* Better use 32 thread groups for f16 x f32
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
* convert : fix python 3.8 support
* convert : sort imports
* convert : fix required parameters in convert-llama-ggmlv3-to-gguf
* convert : fix mypy errors in convert-llama-ggmlv3-to-gguf
* convert : use PEP 585 generics and PEP 604 unions
Now that we have `from __future__ import annotations`, we can use this
modern syntax in Python 3.7 instead of restricting support to Python 3.9
or 3.10 respectively.
* gguf.py : a tuple is already a tuple
* add mypy.ini
* convert : add necessary `type: ignore` comments
* gguf-py: bump version
