Commit graph

499 commits

Author SHA1 Message Date
0cc4m
7296c961d9
ggml : add CLBlast support (#1164)
* Allow use of OpenCL GPU-based BLAS using ClBlast instead of OpenBLAS for context processing

* Improve ClBlast implementation, avoid recreating buffers, remove redundant transfers

* Finish merge of ClBlast support

* Move CLBlast implementation to separate file

Add buffer reuse code (adapted from slaren's cuda implementation)

* Add q4_2 and q4_3 CLBlast support, improve code

* Double CLBlast speed by disabling OpenBLAS thread workaround

Co-authored-by: Concedo <39025047+LostRuins@users.noreply.github.com>
Co-authored-by: slaren <2141330+slaren@users.noreply.github.com>

* Fix device selection env variable names

* Fix cast in opencl kernels

* Add CLBlast to CMakeLists.txt

* Replace buffer pool with static buffers a, b, qb, c

Fix compile warnings

* Fix typos, use GGML_TYPE defines, improve code

* Improve btype dequant kernel selection code, add error if type is unsupported

* Improve code quality

* Move internal stuff out of header
* Use internal enums instead of CLBlast enums
* Remove leftover C++ includes and defines
* Make event use easier to read

Co-authored-by: Henri Vasserman <henv@hot.ee>

* Use c compiler for opencl files

* Simplify code, fix include

* First check error, then release event

* Make globals static, fix indentation

* Rename dequant kernels file to conform with other file names

* Fix import cl file name

---------

Co-authored-by: Concedo <39025047+LostRuins@users.noreply.github.com>
Co-authored-by: slaren <2141330+slaren@users.noreply.github.com>
Co-authored-by: Henri Vasserman <henv@hot.ee>
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2023-04-28 17:57:16 +03:00
Folko-Ven
78ec543733
Correcting link to w64devkit (#1214)
Correcting link to w64devkit (change seeto to skeeto).
2023-04-28 16:22:48 +02:00
Johannes Gäßler
92a6e13a31
Add Manjaro CUDA include and lib dirs to Makefile (#1212) 2023-04-28 15:40:32 +02:00
Yann Follet
04aaae1d79
add avx2 for dot_q8_0_q8_0, 2x faster than scalar (#1211) 2023-04-28 11:59:48 +00:00
Stephan Walter
0b2da20538
ggml : slightly faster AVX2 implementation for Q5 (#1197) 2023-04-26 23:26:42 +03:00
Georgi Gerganov
f9be42add0
readme : add quantization info 2023-04-26 23:24:42 +03:00
Georgi Gerganov
574406dc7e
ggml : add Q5_0 and Q5_1 quantization (#1187)
* ggml : add Q5_0 quantization (cuBLAS only)

* ggml : fix Q5_0 qh -> uint32_t

* ggml : fix q5_0 histogram stats

* ggml : q5_0 scalar dot product

* ggml : q5_0 ARM NEON dot

* ggml : q5_0 more efficient ARM NEON using uint64_t masks

* ggml : rename Q5_0 -> Q5_1

* ggml : adding Q5_0 mode

* quantize : add Q5_0 and Q5_1 to map

* ggml : AVX2 optimizations for Q5_0, Q5_1 (#1195)

---------

Co-authored-by: Stephan Walter <stephan@walter.name>
2023-04-26 23:14:13 +03:00
Ásgeir Bjarni Ingvarsson
87a6f846d3
Allow setting the rng seed after initialization. (#1184)
The llama_set_state_data function restores the rng state to what it
was at the time llama_copy_state_data was called. But users may want
to restore the state and proceed with a different seed.
2023-04-26 22:08:43 +02:00
DaniAndTheWeb
ea3ad7eb60
Updating build instructions to include BLAS support (#1183)
* Updated build information

First update to the build instructions to include BLAS.

* Update README.md

* Update information about BLAS

* Better BLAS explanation

Adding a clearer BLAS explanation and adding a link to download the CUDA toolkit.

* Better BLAS explanation

* BLAS for Mac

Specifying that BLAS is already supported on Macs using the Accelerate Framework.

* Clarify the effect of BLAS

* Windows Make instructions

Added the instructions to build with Make on Windows

* Fixing typo

* Fix trailing whitespace
2023-04-26 22:03:03 +02:00
Pavol Rusnak
859fee6dfb
quantize : use map to assign quantization type from string (#1191)
instead of `int` (while `int` option still being supported)

This allows the following usage:

`./quantize ggml-model-f16.bin ggml-model-q4_0.bin q4_0`

instead of:

`./quantize ggml-model-f16.bin ggml-model-q4_0.bin 2`
2023-04-26 18:43:27 +02:00
Stephan Walter
4afcc37869
Update SHA256SUMS after quantization change (#1181)
Co-authored-by: Pavol Rusnak <pavol@rusnak.io>
2023-04-25 23:41:56 +02:00
ostix360
667c501334
py : cast lora_alpha to int in convert-lora-to-ggml (#1170)
Co-authored-by: Pavol Rusnak <pavol@rusnak.io>
2023-04-25 23:33:08 +02:00
Pavol Rusnak
bb98e77be7
nix: use convert.py instead of legacy wrapper convert-pth-to-ggml.py (#981) 2023-04-25 23:19:57 +02:00
Georgi Gerganov
7a32fcb3b2
ggml : add Q8_0 quantization format (rename the old one to Q8_1) (ARM NEON) (#1179)
* ggml : add Q8_0 quantization format (rename the old one to Q8_1)

* tests : fix test-quantize-fns

* ggml : finalize Q8_0 implementation

* ggml : use q4_0_q8_0 and q4_2_q8_0

* ggml : fix Q8_0 dot product bug (ARM)

* ggml : Q8_0 unroll x2

* ggml : fix bug - using wrong block type

* ggml : extend quantize_fns_t with "vec_dot_type"

* ggml : fix Q8_0 to use 255 values out of 256

* ggml : fix assert using wrong QK4_2 instead of QK4_3
2023-04-25 23:40:51 +03:00
unbounded
dd0eabc049
ggml : use full range for Q4_0 and Q4_2 quantization (#729)
* Use full range for q4_0 quantization

By keeping the sign of the highest magnitude, we can make sure the
highest value maps to -8, which is currently unused.
This is a bit of a freebie since it is fully backwards compatible with
the current format.

* Update quantize_row_q4_0 for AVX/AVX2

* Update quantize_row_q4_0 for WASM

Untested

* Update quantize_row_q4_0 for Arm NEON

* Update quantize_row_q4_0 for PowerPC

Untested

* Use full range for q4_2 quantization
2023-04-25 20:20:46 +03:00
xaedes
54bb60e268
ggml : fix bug in ggml_compute_forward_sum_f32 (#1162)
The sum over all rows is now computed instead of just the last row
2023-04-24 23:02:02 +02:00
Georgi Gerganov
8a0f8673ba
ggml : export symbols (#1155) 2023-04-24 22:18:25 +03:00
xaedes
0c5692345d
examples : add save_load_state example (#1150)
* add save_load_state example

* use <cstdio> instead of <iostream> and fprintf / printf instead of cout

* renamed save-load-state example files replacing underscores by dashes
2023-04-24 19:23:31 +03:00
Georgi Gerganov
957c8ae21d
llama : increase scratch buffer size for 65B (ref #1152)
Temporary solution
2023-04-24 18:47:30 +03:00
mgroeber9110
9b0a4d4214
examples/main README improvements and some light refactoring (#1131) 2023-04-24 15:45:32 +00:00
Stephan Walter
2ec83428de
Fix build for gcc 8 and test in CI (#1154) 2023-04-24 15:38:26 +00:00
slaren
e4cf982e0d
Fix cuda compilation (#1128)
* Fix: Issue with CUBLAS compilation error due to missing -fPIC flag

---------

Co-authored-by: B1gM8c <89020353+B1gM8c@users.noreply.github.com>
2023-04-24 17:29:58 +02:00
Georgi Gerganov
c4fe84fb0d
llama : refactor get / set state + remove redundant kv cache API (#1143) 2023-04-24 07:40:02 +03:00
slaren
1d78fecdab
Fix LoRA acronym (#1145) 2023-04-23 23:03:44 +02:00
Georgi Gerganov
284685f169
scripts : add helper scripts to synch ggml repo 2023-04-23 19:57:09 +03:00
DannyDaemonic
edce63baa9
Added README.md for main with examples and explanations (#1139) 2023-04-23 15:37:02 +00:00
Georgi Gerganov
ec9cdb6752
ggml : do not print perf ops that have not been used at all 2023-04-23 18:32:52 +03:00
Georgi Gerganov
e4422e299c
ggml : better PERF prints + support "LLAMA_PERF=1 make" 2023-04-23 18:15:39 +03:00
Stephan Walter
53c8434398
Improve AVX2 for vec_dot_q4_3_q8_0 (#1138) 2023-04-23 11:01:03 +00:00
Pavol Rusnak
c6524f46eb
readme : update gpt4all instructions (#980) 2023-04-23 10:21:26 +02:00
Yishuo Wang
c9e2c26f41
A better packNibbles and mul_sum_i8_pairs_float implementation using AVX512 (#1119) 2023-04-23 07:57:05 +00:00
Georgi Gerganov
0e018fe008
ggml : fix Q4_3 cuBLAS 2023-04-22 16:32:07 +03:00
Stephan Walter
857308d1e8
ci : trigger CI for drafts, but not most PR actions (#1125) 2023-04-22 16:12:29 +03:00
Stephan Walter
c50b628810
Fix CI: ARM NEON, quantization unit tests, editorconfig (#1122) 2023-04-22 10:54:13 +00:00
unbounded
5f939498d5
ggml : unit test for quantization functions (#953)
* Unit test for quantization functions

Use the ggml_internal_get_quantize_fn function to loop through all
quantization formats and run a sanity check on the result.

Also add a microbenchmark that times these functions directly without
running the rest of the GGML graph.

* test-quantize-fns: CI fixes

Fix issues uncovered in CI
 - need to use sizes divisible by 32*8 for loop unrolling
 - use intrinsic header that should work on Mac

* test-quantize: remove

Per PR comment, subsumed by test-quantize-fns

* test-quantize: fix for q8_0 intermediates
2023-04-22 12:10:39 +03:00
wbpxre150
36b4f7e064
llama : print timings on ctrl+c exit (#1021)
* print timings on ctrl+c exit

* remove redundant free memory call.

* add global pointer to ctx.
2023-04-22 11:56:35 +03:00
eiery
10f19c1121
llama : have n_batch default to 512 (#1091)
* set default n_batch to 512 when using BLAS

* spacing

* alternate implementation of setting different n_batch for BLAS

* set n_batch to 512 for all cases
2023-04-22 11:27:05 +03:00
Howard Su
7e312f165c
cmake : fix build under Windows when enable BUILD_SHARED_LIBS (#1100)
* Fix build under Windows when enable BUILD_SHARED_LIBS

* Make AVX512 test on Windows to build the shared libs
2023-04-22 11:18:20 +03:00
Georgi Gerganov
872c365a91 ggml : fix AVX build + update to new Q8_0 format 2023-04-22 11:08:12 +03:00
Georgi Gerganov
955ef9a5d5
ggml : alternative Q4_3 implementation using modified Q8_0 (#1109)
* ggml : prefer vzip to vuzp

This way we always use the same type of instruction across all quantizations

* ggml : alternative Q4_3 implementation using modified Q8_0

* ggml : fix Q4_3 scalar imlpementation

* ggml : slight improvement of Q4_3 - no need for loop unrolling

* ggml : fix AVX paths for Q8_0 quantization
2023-04-22 10:55:35 +03:00
Stephan Walter
c5aa5e5777
ggml : AVX2 optimization for vec_dot_q4_3_q8_0 and refactoring (#1099)
* AVX2 optimization for vec_dot_q4_3_q8_0 and refactoring

* finish AVX vectorization of quantize_row_q8_0

* Rename hsum_int_8 to hsum_i32_8
2023-04-22 10:37:05 +03:00
Clint Herron
e9a9cb0c54
examples : Improve Alpaca Default Repeat Penalty: Better Match Alpaca.cpp Experience (#1107)
* Moving parameters to separate lines for readability.

* Increasing repeate_penalty to 1.1 to make alpaca more usable by default.

* Adding trailing newline.
2023-04-22 09:54:33 +03:00
xaedes
b6e7f9b09e
llama : add api for getting/setting the complete state: rng, logits, embedding and kv_cache (#1105)
* reserve correct size for logits

* add functions to get and set the whole llama state:

including rng, logits, embedding and kv_cache

* remove unused variables

* remove trailing whitespace

* fix comment
2023-04-22 09:21:32 +03:00
slaren
50cb666b8a
Improve cuBLAS performance by using a memory pool (#1094)
* Improve cuBLAS performance by using a memory pool

* Move cuda specific definitions to ggml-cuda.h/cu

* Add CXX flags to nvcc

* Change memory pool synchronization mechanism to a spin lock
General code cleanup
2023-04-21 21:59:17 +02:00
apaz
25d7abbd1f
llama : fixed rlimit error message (#888) 2023-04-21 21:48:06 +03:00
源文雨
018f2279f5
cmake : link threads publicly to ggml (#1042)
* fix: ld link test-tokenizer-0 error

```
cmake3 --build . --config Release
[  5%] Built target ggml
[ 16%] Built target llama
[ 22%] Linking CXX executable ../bin/test-tokenizer-0
../libllama.a(ggml.c.o):在函数‘ggml_graph_compute’中:
ggml.c:(.text+0xf2db):对‘pthread_create’未定义的引用
ggml.c:(.text+0xf9d4):对‘pthread_join’未定义的引用
collect2: error: ld returned 1 exit status
gmake[2]: *** [bin/test-tokenizer-0] 错误 1
gmake[1]: *** [tests/CMakeFiles/test-tokenizer-0.dir/all] 错误 2
gmake: *** [all] 错误 2
```

* Update CMakeLists.txt

* Update CMakeLists.txt

* Update CMakeLists.txt
2023-04-21 21:27:06 +03:00
Alex Klinkhamer
9411288271
main : evaluate tokens in batches after swapping context (#1014)
* examples : evaluate tokens in batches after swapping context

* Update examples/main/main.cpp

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2023-04-21 21:18:09 +03:00
xaedes
8687c1f258
llama : remember and restore kv cache data pointers (#1104)
because their value is stored in buf and overwritten by memcpy
2023-04-21 18:25:21 +03:00
Kawrakow
1bfc153e2f
ggml : a faster version for Q4_1 x Q8_0 dot products (#1083)
* A faster version for Q4_1 x Q8_0 dot products

The idea nehind being that Q8_0 quantized
values get used many times in the matrix multiplications
where they are involved. In the current implementations,
when we are evaluating the dot products, we need to compute
the sum of the quants in the Q8_0 vector, so the same
operation is repeated many times. Here we pre-compute
the sum during Q8_0 quantization, store it in the
now modified block_q8_0 struct, and then reuse this
result in the subsequent dot products.

In a synthetic benchmark (just compute a bunch of dot
products), this change speeds up the Q4_1 * Q8_0 dot
product by 80%, making the performance identical to
Q4_0 * Q8_0.

In practical application, I see a ~15% gain in speed for
token prediction on M2, and ~5% gain on Ryzen 7950X.
The speed gain in the prompt evaluation is much bigger
(around 50%).

I have only done the change for the scalar version,
ARM_NEON, and AVX2, so we still need an AVX implementation.

* Cleaning up

---------

Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-04-21 18:18:26 +03:00
slaren
3d59769c3b
Show perplexity ETA in hours and minutes (#1096) 2023-04-21 14:57:57 +02:00