Commit Graph

46 Commits

Author SHA1 Message Date
Johannes Gäßler
d50f8897a7
CUDA: stream-k decomposition for MMQ (#8018)
* CUDA: stream-k decomposition for MMQ

* fix undefined memory reads for small matrices
2024-06-20 14:39:21 +02:00
Calvin Laurenson
43b35e38ba
Add support for sqrt on CUDA (#7953)
* cuda sqrt support

* enable cuda in pca

* fix comments in pca

* add test

* add sqrt to ggml_backend_cuda_supports_op

* fix test

* new line

* Use F32 sqrtf instead of F64 sqrt

Co-authored-by: Johannes Gäßler <johannesg@5d6.de>

---------

Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
2024-06-17 00:23:04 +02:00
Georgi Gerganov
19b7a836f6
cuda : fix bounds check for src0 rows in MMVQ kernel (whisper/2231)
* cuda : fix bounds check for src0 rows in MMVQ kernel

* Update ggml-cuda/mmvq.cu

Co-authored-by: Johannes Gäßler <johannesg@5d6.de>

---------

Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
2024-06-16 20:32:49 +03:00
Johannes Gäßler
76d66ee0be
CUDA: faster q2_K, q3_K MMQ + int8 tensor cores (#7921)
* CUDA: faster q2_K, q3_K MMQ + int8 tensor cores

* try CI fix

* try CI fix

* try CI fix

* fix data race

* rever q2_K precision related changes
2024-06-14 18:41:49 +02:00
Johannes Gäßler
963552903f
CUDA: fix broken oob check for FA vec f32 kernel (#7904) 2024-06-12 17:41:51 +02:00
Georgi Gerganov
a9cae48003
tests : add non-cont unary tests (#7857)
* tests : add non-cont unary tests

* ggml : update unary asserts and "supports_op"

ggml-ci
2024-06-12 16:00:22 +03:00
Johannes Gäßler
bdcb8f4222
CUDA: int8 tensor cores for MMQ (q4_K, q5_K, q6_K) (#7860) 2024-06-11 08:26:07 +02:00
Johannes Gäßler
1f0dabda8d
CUDA: use tensor cores for MMQ (#7676)
* CUDA: int8 tensor cores for MMQ (legacy quants)

* fix out-of-bounds writes

* __builtin_assume -> GGML_CUDA_ASSUME

* fix writeback returning too early
2024-06-10 11:45:13 +02:00
Johannes Gäßler
42b53d192f
CUDA: revise q8_1 data layout for mul_mat_q (#7824) 2024-06-09 09:42:25 +02:00
Johannes Gäßler
7d1a378b8f
CUDA: refactor mmq, dmmv, mmvq (#7716)
* CUDA: refactor mmq, dmmv, mmvq

* fix out-of-bounds write

* struct for qk, qr, qi

* fix cmake build

* mmq_type_traits
2024-06-05 16:53:00 +02:00
Georgi Gerganov
2b3389677a
ggml : refactor rope norm/neox (#7634)
* ggml : unify rope norm/neox (CPU)

* ggml : fix compile warning

* ggml : remove GLM rope mode

ggml-ci

* metal : better rope implementation

ggml-ci

* cuda : better rope implementation

ggml-ci

* naming : n_orig_ctx -> n_ctx_orig

ggml-ci

* dev : add reminders to update backends

ggml-ci

* vulkan : fix ggml_rope_ext() usage

* cuda : fix array size + indents

ggml-ci
2024-06-05 11:29:20 +03:00
Johannes Gäßler
e141ce624a
Fix FlashAttention debug test, FP32 assert (#7684) 2024-06-01 23:26:10 +02:00
Johannes Gäßler
750f60c03e
CUDA: fix Pascal FA, deq. KV to FP16 for batch > 8 (#7681) 2024-06-01 15:47:04 +02:00
Johannes Gäßler
9b596417af
CUDA: quantized KV support for FA vec (#7527)
* CUDA: quantized KV support for FA vec

* try CI fix

* fix commented-out kernel variants

* add q8_0 q4_0 tests

* fix nwarps > batch size

* split fattn compile via extern templates

* fix flake8

* fix metal tests

* fix cmake

* make generate_cu_files.py executable

* add autogenerated .cu files

* fix AMD

* error if type_v != FP16 and not flash_attn

* remove obsolete code
2024-06-01 08:44:14 +02:00
Georgi Gerganov
fb76ec31a9
ggml : fix YARN + add tests + add asserts (#7617)
* tests : add rope tests

ggml-ci

* ggml : fixes (hopefully)

ggml-ci

* tests : add non-cont tests

ggml-ci

* cuda : add asserts for rope/norm + fix DS2

ggml-ci

* ggml : assert contiguousness

* tests : reduce RoPE tests

ggml-ci
2024-05-29 20:17:31 +03:00
Georgi Gerganov
cce3dcffc5
cuda : non-cont concat support (#7610)
* tests : add non-cont concat tests

* cuda : non-cont concat support

ggml-ci
2024-05-29 15:38:26 +03:00
Georgi Gerganov
0548a4187f
ggml : generalize GGML_OP_CONCAT (#7563)
* ggml : generalize GGML_OP_CONCAT (WIP)

ggml-ci

* tests : add dim != 2 tests

* metal : generalize concat kernel

* tests : naming

* cuda : generalize concat kernel

ggml-ci

* sycl : add warning and assert

* ggml : fix op params handling

* metal : bugfix kernel

ggml-ci

* ggml : reimplement CPU and Metal

* cuda : add asserts

ggml-ci

* ggml : fix ptrs

ggml-ci
2024-05-28 11:04:19 +03:00
Djip007
852aafb163
update HIP_UMA #7399 (#7414)
* update HIP_UMA #7399

add use of hipMemAdviseSetCoarseGrain when LLAMA_HIP_UMA is enable.
- get x2 on prompte eval and x1.5 on token gen with rocm6.0 on ryzen 7940HX iGPU (780M/gfx1103)

* simplify code, more consistent style

---------

Co-authored-by: slaren <slarengh@gmail.com>
2024-05-28 01:40:47 +02:00
Georgi Gerganov
e84b71c2c6
ggml : drop support for QK_K=64 (#7473)
* ggml : drop support for QK_K=64

ggml-ci

* opencl : restore QK_K=256 define
2024-05-23 10:00:21 +03:00
Johannes Gäßler
cd93a28cb1
CUDA: fix FA out-of-bounds reads (#7479) 2024-05-23 00:31:20 +02:00
Johannes Gäßler
38c03478a3
CUDA: fix FA out-of-bounds writes (#7465) 2024-05-22 17:58:25 +02:00
Georgi Gerganov
9b3d833189
cuda : fix compile warning (#7454) 2024-05-22 12:36:37 +03:00
Johannes Gäßler
95fb0aefab
CUDA: remove incorrect precision check (#7454) 2024-05-22 10:24:29 +02:00
Georgi Gerganov
3e5faa8503
cuda : fix rope + add tests (#7452)
* cuda : fix rope pos data

ggml-ci

* ggml : drop mode & 1 == 1 support for ggml_rope

ggml-ci

* ggml : support freq_factors for f16 rope (CPU)

ggml-ci

* tests : add rope tests using frequency factors

ggml-ci
2024-05-22 11:01:35 +03:00
liuwei-git
201cc11afa
llama : add phi3 128K model support (#7225)
* add phi3 128k support in convert-hf-to-gguf

* add phi3 128k support in cuda

* address build warnings on llama.cpp

* adjust index value in cuda long rope freq factors

* add long rope support in ggml cpu backend

* make freq factors only depend on ctx size

* remove unused rope scaling type 'su' frin gguf converter

* fix flint warnings on convert-hf-to-gguf.py

* set to the short freq factor when context size is small than trained context size

* add one line of comments

* metal : support rope freq_factors

* ggml : update ggml_rope_ext API to support freq. factors

* backends : add dev messages to support rope freq. factors

* minor : style

* tests : update to use new rope API

* backends : fix pragma semicolons

* minor : cleanup

* llama : move rope factors from KV header to tensors

* llama : remove tmp assert

* cuda : fix compile warning

* convert : read/write n_head_kv

* llama : fix uninitialized tensors

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2024-05-21 23:28:32 +03:00
Johannes Gäßler
fcf6538ba6
CUDA: fix unused warning in mmq.cu (#7442) 2024-05-21 20:27:12 +03:00
Johannes Gäßler
d8ee902227
CUDA: deduplicate mmq code (#7397) 2024-05-21 16:02:12 +02:00
Johannes Gäßler
133d99c599
CUDA: deduplicate FlashAttention code (#7352) 2024-05-18 12:36:25 +02:00
Engininja2
d233b507cd
cuda : add half2 __shfl_xor() for ROCm 5.5 (#7263) 2024-05-18 10:05:17 +02:00
Johannes Gäßler
0fc1e820a9
CUDA: faster large batch FA without tensor cores (#7314) 2024-05-17 18:54:52 +02:00
John Balis
48aa8fd1f2
ggml : add ggml_upscale_ext (ggml/814)
* initial commit with CPU implementation of upscale to shape and test, cuda implementation next

* experimental commit to see if dst shape is correct

* test version

* test

* removed unnecessary params

* refactor

* fixed tests

* ggml : metal impl + cleanup + sycl dev warnings

* patched ggml_upscale cuda op to handle non-contiguous tensors, added test for non-contiguous behavior

* metal : fix upsacle op to support nb00 + style

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2024-05-15 13:23:33 +03:00
Johannes Gäßler
dc685be466
CUDA: add FP32 FlashAttention vector kernel (#7188)
* CUDA: add FP32 FlashAttention vector kernel

* fixup! CUDA: add FP32 FlashAttention vector kernel

* fixup! fixup! CUDA: add FP32 FlashAttention vector kernel

* fixup! fixup! fixup! CUDA: add FP32 FlashAttention vector kernel
2024-05-12 19:40:45 +02:00
Justina Cho
f5ef34e428 feat: implemented sigmoid function (ggml/806)
* added sigmoid function

* implemented metal kernel for sigmoid

* implemented cuda kernel for sigmoid

* added sigmoid unary op and incremented count
2024-05-11 15:38:34 +03:00
Georgi Gerganov
9cb317f77e
ggml : full ALiBi support (#7192)
* ggml : full ALiBi support

* ggml : update ggml_soft_max_ext() CUDA, SYCL

* ggml : ggml_flash_attn_ext() support ALiBi (CPU)

* ggml : ggml_flash_attn_ext() support ALiBi (Metal)

* ggml : fix warning

* ggml : ggml_flash_attn_ext() support ALiBi (CUDA)

ggml-ci

* ggml : fix assert message

* vulkan : add dev notes

* ggml : require mask when using ALiBi

ggml-ci

* convert : fix convert for refact models
2024-05-11 10:32:41 +03:00
Johannes Gäßler
a743d76a01
CUDA: generalize FP16 fattn vec kernel (#7061)
* CUDA: generalize FP16 fattn vec kernel

* disable unsupported head sizes for AMD in test

* try AMD fix

* fix batch size 2-8

* partially revert changes
2024-05-09 14:32:02 +02:00
agray3
bc4bba364f
Introduction of CUDA Graphs to LLama.cpp (#6766)
* DRAFT: Introduction of CUDA Graphs to LLama.cpp

* FIx issues raised in comments

* Tidied to now only use CUDA runtime (not mixed with driver calls)

* disable for multi-gpu and batch size > 1

* Disable CUDA graphs for old GPU arch and with env var

* added missing CUDA_CHECKs

* Addressed comments

* further addressed comments

* limit to GGML_ALLOW_CUDA_GRAPHS defined in llama.cpp cmake

* Added more comprehensive graph node checking

* With mechanism to fall back if graph capture fails

* Revert "With mechanism to fall back if graph capture fails"

This reverts commit eb9f15fb6f.

* Fall back if graph capture fails and address other comments

* - renamed GGML_ALLOW_CUDA_GRAPHS to GGML_CUDA_USE_GRAPHS

- rename env variable to disable CUDA graphs to GGML_CUDA_DISABLE_GRAPHS

- updated Makefile build to enable CUDA graphs

- removed graph capture failure checking in ggml_cuda_error
  using a global variable to track this is not thread safe, but I am also not safistied with checking an error by string
  if this is necessary to workaround some issues with graph capture with eg. cuBLAS, we can pass the ggml_backend_cuda_context to the error checking macro and store the result in the context

- fixed several resource leaks

- fixed issue with zero node graphs

- changed fixed size arrays to vectors

- removed the count of number of evaluations before start capturing, and instead changed the capture mode to relaxed

- removed the check for multiple devices so that it is still possible to use a single device, instead checks for split buffers to disable cuda graphs with -sm row

- changed the op for checking batch size to GGML_OP_ADD, should be more reliable than GGML_OP_SOFT_MAX

- code style fixes

- things to look into
  - VRAM usage of the cudaGraphExec_t, if it is significant we may need to make it optional
  - possibility of using cudaStreamBeginCaptureToGraph to keep track of which ggml graph nodes correspond to which cuda graph nodes

* fix build without cuda graphs

* remove outdated comment

* replace minimum cc value with a constant

---------

Co-authored-by: slaren <slarengh@gmail.com>
2024-05-08 22:55:49 +02:00
Johannes Gäßler
1613ef8d8e
CUDA: CUDART < 11.7 workaround for __hmax, __hmax2 (#7019) 2024-05-01 14:46:37 +02:00
Georgi Gerganov
9c67c2773d
ggml : add Flash Attention (#5021)
* ggml : add ggml_flash_attn_ext API

* ggml : fix GQA support in ggml_flash_attn_ext

* ggml : online attention (CPU)

* metal : initial implementation

* metal : f16 precision

* metal : reduce branches

* metal : specialize for head size

* wip : 8 rows per simd group

* wip : 4 rows per simd group

* wip : template for rows per warp

* metal : parallelize across KV size

* metal : parallel reduce across heads

* metal : efficient flash_attn_f16 implementation

* metal : avoid redundant loads of the attention

* metal : scale and mask in matrix form

* metal : fix comment

* llama : avoid ggml_cast, use F32 query

* metal : add parallel reduce version (disabled)

* metal : move output into local memory + optimize

- the result from each simdgroup now stays in the registers
- significantly reduced SRAM usage
- more efficient skipping of -INF blocks
- avoid simdgroup barrier in hot loop
- add comments

* metal : add tests, fix scaling, support C > 32

* metal : improve precision

* ggml : fix f16 mad

* metal : minor

* metal : support Q > 8

* tests : add ATTN tests

* metal : disable buffer allocation logs

* tests : more

* metal : faster inner loop for C == 32

* metal : fix array initialization

* tests : ifdef

* ggml : switch to padded F16 mask for ggml_soft_max, ggml_flash_attn_ext

* ggml : fix ggml_soft_max mask requirement

* cuda : fix soft_max to use correct mask size

* cuda : add flash_attn kernel (wip)

* metal : optimize softmax for C > 32

* metal : optimize softmax

* tests : minor fix

* cuda : avoid zeroing fragments

* tests : update dims

* cuda : fix __hisinf() result check

* cuda : avoid warp_reduce for smax

* cuda : use int instead of int64_t

Noticeably improves performance (thanks to Johannes)

* cuda : make loops use the same loop values

Thanks Johannes again for the tip

* cuda : unroll some of the loops

* cuda : avoid __hisinf branches

* cuda : use half2 in softmax

* cuda : switch to 1 warp for bs > 16

* cuda : speed-up reduce part of the kernel

* cuda : unroll Q*K^T loop

* cuda : fix -INF block check

* cuda : simplify softmax

* cuda : fix matrix names

* cuda : minor

* llama : adapt to F16 KQ_pos

* llama : adapt new models to F16 KQ_mask

* ggml : fix F16 store (ARM NEON)

* llama : fix type of KQ_mask and KQ_pos

* ggml : fix CPU soft_max

* tests : add hs=256

* cuda : fix build

* metal : improve perf via smaller int registers

* cuda : adapt soft_max to F16 mask and pos

* CUDA: faster FlashAttention, kernel for bs == 1

* 16 cols for Phi-2

* no vec for hs, no hs==256 ncols==32 for Volta

* adjust kernel selection logic

* 4 warps, 256 stride for all D

* no ncols == 64

* Multiple parallel blocks for batch size 1

* fix compile warnings

* fix excessive KQ_b loads

* fix cmake build

* fix KV cache padding, NaN from INFINITY (#6438)

* llama : flash_attn cparam + fix defrag

* server: support flash_attn param

* server: bench: enable flash_attn param

* CUDA: refactor host code, dyn. par. blocks

* fix flash_attn_vec_f16 race condition

* flush softmax exp below threshold to 0

* store temp KQ in registers

* Calculate KQ as FP32 if KQV has GGML_PREC_F32

* Add __hgt2_mask implementation for CUDA 11

* fix KQ FP32 precision fpr parallel_blocks > 1

* llama-bench : add -fa,--flash-attn arg

* metal : add BS=1 kernel for flash attention (#6508)

* metal : add BS=1 kernel for flash attention (wip)

* metal : support more than 1 warps

* metal : opts

* metal : opt

* metal : switch to parallel reduce

* metal : reduce registers

* metal : simplify

* metal : initial FA vec kernel

* metal : use F32 attention accumulators

* batched-bench : add fattn arg

* llama : simplify llama_build_kv_store

ggml-ci

* llama : adapt build_olmo to changes

* ggml : fix arm fp16 store on windows

* metal : clean-up

* metal : clean-up kernel code

* metal : minor

* tests : remove benchmarks

ggml-ci

* ggml : fix avx512 const correctness

ggml-ci

* ggml : fix soft_max with bias on CPU

ggml-ci

* common : print --flash-attn in help

* ggml : fix num dimensions in ggml_flash_attn_ext

* llama : force disable flash attention for incompatible models

* ggml : ggml_soft_max support F16/F32 mask/pos

ggml-ci

* cuda : uint -> uint32_t

* cuda : "constexpr dim3" -> "const dim3"

ggml-ci

* cuda : try to fix __hgt2_mask

ggml-ci

* ggml : add TODO's for F16/F32 mask/pos support in other backends

* llama : replace bool need_kq_pos with use_alibi

* llama : prep ALiBi support for BERT models

ggml-ci

* llama : fix n_batch requirements

ggml-ci

* cont

* server : add help for --flash-attn arg

* llama : disable FA for AMD

* tests : remove TMP_ATTN_BENCH

ggml-ci

* llama : support save/load state with FA enabled

ggml-ci

* ci : add CUDA save-load-state tests

ggml-ci

* llama : llama_kv_cache_clear zeroes data + fix save-load seq

ggml-ci

* llama : fix copy-paste errors, add TODO

* llama : disallow incompatible states

* llama : update llama_state_get_size after v_trans field

* metal : remove tmp log

* llama : add static reminder for llama_state_get_size

* metal : fix max nsg

ggml-ci

* ci : fix arg order

ggml-ci

---------

Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
Co-authored-by: Pierrick HYMBERT <pierrick.hymbert@gmail.com>
2024-04-30 12:16:08 +03:00
DAN™
e00b4a8f81
Fix more int overflow during quant (PPL/CUDA). (#6563)
* Fix more int overflow during quant.

* Fix some more int overflow in softmax.

* Revert back to int64_t.
2024-04-29 00:38:44 +02:00
slaren
0d56246f4b
ggml : group all experts in a single ggml_mul_mat_id (#6505)
* ggml : group all experts in a single ggml_mul_mat_id
cuda : improve mmid row copy

* cuda : fix bin bcast with non-cont src0

* test-backend-ops : only run all mul mat tests for base types

* llama : disable moe offloading with SYCL

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2024-04-18 15:18:48 +02:00
Carolinabanana
5dc9dd7152
llama : add Command R Plus support (#6491)
* Add Command R Plus GGUF

* Add Command R Plus GGUF

* Loading works up to LayerNorm2D

* Export new tensors in 1D so they are not quantized.

* Fix embedding layer based on Noeda's example

* Whitespace

* Add line

* Fix unexpected tokens on MPS. Re-add F16 fix. ((Noeda)

* dranger003: Fix block index overflow in CUDA dequantizing.

* Reverted blocked multiplication code as it still has issues and could affect other Llama arches

* export norms as f32

* fix overflow issues during quant and other cleanup

* Type convention

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>

* dranger003: Fix more int overflow during quant.

---------

Co-authored-by: S <seast@Ss-Mac-Studio.local>
Co-authored-by: S <s@example.com>
Co-authored-by: slaren <slarengh@gmail.com>
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2024-04-09 11:16:13 +03:00
slaren
08a0c02060
ggml : mul_mat_id use the same tensor for all the experts (#6387)
* ggml : update mul_mat_id to use the same tensor for all the experts

* update cuda

* minor

* update metal

* update test-backend-ops

* fix cuda

* Update ggml-metal.m

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>

* update convert.py

* update convert-hf-to-gguf.py

* update convert.py for mixtral hf models

* Update convert-hf-to-gguf.py

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>

* cuda : support non-pow-2 number of experts

* allow quantize to work for split and merged experts models in the same way

* cleanup + disable mmap automatically with split tensors models

* update imatrix

* test-backend-ops : test qwen argsort

* update grok model loading

* llama : add merged experts tensors to the grok tensor map

* minor

* gguf : bump version

* fix quantizing of merged experts

* convert-hf-to-gguf.py : update grok (untested)

* make linter happy

* cuda/argsort : use shared memory instead of pool memory

* convert : fix grok tensor names

* metal : add support for non-pow-2 argsort

* llama : more loader cleanup, better error checking

* cuda : fix warning

* llama : still use mmap for loading old models, but copy the data to a host buffer

* add review note

* llama : remove ffn tensor counting + add sanity check

ggml-ci

* convert : fix handling of n_experts == None

ggml-ci

* imatrix : fix ncall counters

* llama : produce error if imatrix size does not match

* quantize : terminate on errors + trace logs

ggml-ci

* metal : pad shared memory to 16 bytes

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2024-04-03 16:07:05 +03:00
Georgi Gerganov
d48ccf3ad4
sync : ggml (#6351)
* sync : ggml

ggml-ci

* cuda : move GGML_CUDA_DMMV constants to dmmv.cuh

---------

Co-authored-by: slaren <slarengh@gmail.com>
2024-03-29 17:45:46 +02:00
Kawrakow
55c1b2a3bb
IQ1_M: 1.75 bpw quantization (#6302)
* iq1_m: basics

* iq1_m: basics-2

* iq1_m: CUDA dequantize works

Very 1st shot I get PPL = 9.76 for LLaMA-v2-7B.

* iq1_m: separate shifts for each group of 8 in a block

We get
PPL(LLaMA-v2-7B ) = 9.2810
PPL(LLaMA-v2-13B) = 6.8105

Not bad, but slightly higher than
  sqrt(PPL(IQ1_S) * PPL(IQ2_XXS))
which is the expected outcome given that IQ1_M is
halfway between IQ1_S and IQ2_XXS in terms of bpw.
From this, we would expect
 PPL = 9.14 for LLaMA-v2-7B
 PPL = 6.63 for LLaMA-v2-13B

* iq1_m: go to 3-bit scales

There is slight increase in PPL, but the 0.0625 bpw reduction
in size is totally worth it.

We now have
PPL(LLaMA-v2-7B ) = 9.4469 at 1.96 bpw
PPL(LLaMA-v2-13B) = 6.8717 at 1.93 bpw
PPL(LLaMA-v2-70B) = 4.8568 at 1.85 bpw

* iq1_m: scalar dot product

* iq1_m: AVX2 dot product

* iq1_m: very slightly faster AVX2 dot product

* iq1_m: ARM_NEON dot product

Works, but very slow (10.5 t/s)

* iq1_m: Metal - dequantize works, dot product does not

* iq1_m: Metal now works

About the same performance as iq1_s.

* iq1_m: minor

* iq1_m: checking pure iq1_m quantization

It is pretty bad: PPL(LLaMA-v2-7B) = 34 if we quantize output.weight
with Q4_K.

* iiq1_m: slightly faster ARM_NEON dot product

10.5 t/s -> 11.65 t/s

* iq1_m: faster ARM_NEON dot product

11.65 t/s -> 14.9 t/s

* iq1_m: another minor ARM_NEON dot product improvement

14.9 -> 15.0 t/s

* iq1_m: small PPL improvement via super-block scale adjustment

After quantizing block scales redo the super-block scale fit.

PPL(LLaMA-v2-7B ) = 9.3346
PPL(LLaMA-v2-13B) = 6.8419
PPL(LLaMA-v2-70B) = 4.8294
PPL(Mistral-7B  ) = 8.1624

* iq1_m: adapt to CUDA refactoring

* iq1_m: remove unused variable

We have progressed to warnings being errors.

* iq1_m: add to backend-ops tests

* iq1_m: fix Windows ARM

* iq1_m: use common definition of iq1m_scale_t

* cuda: assert -> NO_DEVICE_CODE

* iq1_M: PR comments

---------

Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2024-03-26 15:21:27 +01:00
slaren
2f34b865b6
cuda : fix LLAMA_CUDA_F16 build (#6298) 2024-03-25 16:43:22 +02:00
slaren
ae1f211ce2
cuda : refactor into multiple files (#6269) 2024-03-25 13:50:23 +01:00