* vulkan: Reuse conversion results in prealloc_y
Cache the pipeline and tensor that were most recently used to fill prealloc_y,
and skip the conversion if the current pipeline/tensor match.
* don't use shared pointer for prealloc_y_last_pipeline_used
* musa: fix build warnings
Signed-off-by: Xiaodong Ye <xiaodong.ye@mthreads.com>
* fix warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
Signed-off-by: Xiaodong Ye <xiaodong.ye@mthreads.com>
---------
Signed-off-by: Xiaodong Ye <xiaodong.ye@mthreads.com>
While working on the [whisper-cpp](https://conan.io/center/recipes/whisper-cpp) Conan package for ConanCenter, I noticed that enabling the `with_blas` option fails to build due to an issue in the _MKL_ detection logic.
The problem is that the CMake condition currently expands `BLAS_INCLUDE_DIRS` without quotes:
```cmake
if (${BLAS_INCLUDE_DIRS} MATCHES "mkl" AND (${GGML_BLAS_VENDOR} MATCHES "Generic" OR ${GGML_BLAS_VENDOR} MATCHES "Intel"))
```
When `BLAS_INCLUDE_DIRS` is a list (as Conan provides it), the `if()` command receives multiple arguments and produces a CMake error:
```bash
...
-- BLAS found, Includes: /root/.conan2/p/b/openb034c5a6ca927b/p/include;/root/.conan2/p/b/openb034c5a6ca927b/p/include/openblas
CMake Error at ggml/src/ggml-blas/CMakeLists.txt:77 (if):
if given arguments:
"/root/.conan2/p/b/openb034c5a6ca927b/p/include" "/root/.conan2/p/b/openb034c5a6ca927b/p/include/openblas" "MATCHES" "mkl" "AND" "(" "OpenBLAS" "MATCHES" "Generic" "OR" "OpenBLAS" "MATCHES" "Intel" ")"
Unknown arguments specified
...
```
This PR fixes the issue by quoting the variable:
```cmake
if ("${BLAS_INCLUDE_DIRS}" MATCHES "mkl" AND (${GGML_BLAS_VENDOR} MATCHES "Generic" OR ${GGML_BLAS_VENDOR} MATCHES "Intel"))
```
With this change, the whole list is treated as a single string and the regex still works correctly.
On busybox-based systems like Alpine Linux, wget does not have
certain CLI flags such as '--no-config'. Thus, search for the
existence of 'curl' first in the PATH before wget. wget2 is
still the preferred download tool.
This commit remove the brew install of cmake for macos-latest
as this now seems to be pre-installed on the runner.
The motivation for this is that this job is failing with the following
error:
```console
Error: cmake was installed from the local/pinned tap
but you are trying to install it from the homebrew/core tap.
Formulae with the same name from different taps cannot be installed at the same time.
```
This commit modifies the test-vad and test-vad-full tests to use CMake
definitions for the model and sample paths.
The motivation for this is that currently the tests use relative paths
which might not always be correct depending on the working directory.
With the changes in this commit the tests can be run usins ctest:
```console
$ ctest -R ^test-vad$ --test-dir build
```
Or directly (which is not currently possible without this fix):
```
./build/bin/test-vad
```
Resolves: https://github.com/ggml-org/whisper.cpp/issues/3404
this might happen depending on the way the $stderr.winsize is defined. If the expression "$stderr.winsize[1] - line.size" in Line 114 gets negative, we will get a "negative argument" exception in the padding calculation
* vulkan: Use larger workgroups for mul_mat_vec when M is small
Also use subgroup instructions for (part of) the reduction when supported.
Without this, the more expensive reductions would eat into the benefits of
the larger workgroups.
* update heuristic for amd/intel
Co-authored-by: 0cc4m <picard12@live.de>
---------
Co-authored-by: 0cc4m <picard12@live.de>
- Launch an appropriate number of invocations (next larger power of two).
32 invocations is common and the barrier is much cheaper there.
- Specialize for "needs bounds checking" vs not.
- Make the code less branchy and [[unroll]] the loops. In the final code,
I see no branches inside the main loop (only predicated stores) when
needs_bounds_check is false.
- Always sort ascending, then apply the ascending vs descending option when
doing the final stores to memory.
- Copy the values into shared memory, makes them slightly cheaper to access.
* vulkan: fuse adds
Fuse adds that have the same shape, which are common in MoE models.
It will currently fuse up to 6 adds, because we assume no more than
8 descriptors per dispatch. But this could be changed.
* check runtimeDescriptorArray feature
* disable multi_add for Intel due to likely driver bug
* vulkan: Add missing bounds checking to scalar/coopmat1 mul_mat_id
* vulkan: Support mul_mat_id with f32 accumulators, but they are not hooked up
- There's no explicit way to request f32 precision for mul_mat_id, but there
probably should be, and this gets the code in place for that.
- A couple fixes to check_results.
- Remove casts to fp16 in coopmat1 FA shader (found by inspection).
* add F16/F16 fa support
* fix kernel init
* use mad instead of fma
* use inline function
* mark FA with sinks as unsupported for now
* add pragma unroll to loops
add expicit conversion operator to support older versions of rocm
Switch over to hip_bf16 from legacy hip_bfloat16
Simplify RDNA3 define
Reduce swap over of new hipblas api to rocm 6.5 as this version is used for rocm 7.0 previews
---------
Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
* vulkan: perf_logger improvements
- Account for batch dimension in flops calculation.
- Fix how "_VEC" is detected for mat_mul_id.
- Fix "n" dimension for mat_mul_id (in case of broadcasting).
- Include a->type in name.
* use <=mul_mat_vec_max_cols rather than ==1
* examples/finetune -opt SGD (stochastic gradient descent) memory opt
add unit tested GGML_OPT_OPTIMIZER_SGD to ggml - avoids allocating
m, v tensors.
support finetune.cpp arg -opt SGD (or sgd). (default adamw as before)
llama 3.2-1b-F32 result: observed 11gb gpu ram (41 sec/epoch)
when using SGD instead of 19gb (55 sec/epoch) using adamw.
(wikipedia 100 lines finetune)
(
using the same GPU memory, adamw can only do before OOM 512
batch/context, reaching:
train: [███████▉] data=0000140/0000140 loss=0.02575±0.00099 acc=99.52±0.03% t=00:00:47 ETA=00:00:00
val: [███████▉] data=0000008/0000008 loss=4.76565±0.28810 acc=41.46±0.77% t=00:00:00 ETA=00:00:00
SGD is superior, though it converges slower, with max before OOM 1728
batch/context (esp see the better validation perf):
train: [███████▉] data=0000039/0000039 loss=0.00371±0.00010 acc=99.96±0.01% t=00:00:41 ETA=00:00:00
val: [███████▉] data=0000003/0000003 loss=5.11406±0.76034 acc=48.01±0.69% t=00:00:01 ETA=00:00:00
)
note: when finetuning long enough (or w/ enough -lr),
validation accuracy *eventually* drops ('catastrophic forgetting')
-lr-half (halflife) option useful for SGD to avoid oscillation or
super slow underdamped learning (makes setting -lr more forgiving).
terminal -lr for now is set by lr-halvings i.e. if you want at most
1/8 the inital -lr you set -lr-halvings 3.
note: objective loss not directly comparable between adamw, sgd? -
check perplexity or accuracy or consider relative improvements
for convergence
new finetune args -wd 1e-9 to enable weight decay in sgd or adamw,
and max -epochs N (default 2 as before)
cache (1 - wd*alpha) in 'adamw' opt struct -
no noticeable perf benefit, disabled (still done
for new SGD though)
since opt. memory is pre-allocated, the ggml_opt_get_optimizer_params
would probably be able to change between SGD and AdamW with each epoch
but would need to use adamw for the first (unconfirmed - no cmdline arg
to set such a policy yet)
test-opt checks adamw as before and now sgd (except for a few disabled
tests for sgd only; probably just needs logging values and adding
alternate reference values); tolerance on the 'regression'
test is broader for sgd (so we don't need many more epochs)
* Vulkan: Implement GGML_OP_OPT_STEP_SGD
* tests: Fix OPT_STEP_SGD test-backend-ops
* SGD op param store weight-decay and not 1-alpha*wd
* minor + cosmetic changes
* fix vulkan sgd
* try CI fix
---------
Co-authored-by: 0cc4m <picard12@live.de>
Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
* Factor out `reduce_rows_f32` from common.cuh
This increases iteration cycle speed by not having to recompile
every kernel all the time
* Hide memory-latency by loop unrolling in reduce_rows_f32
* Further optimizations to `reduce_rows_f32`
1. Increase threadblock size to better hide latency of memory requests.
As a consequence of bigger threadblocks, do 2-step summation, using
shared memory to communicate results between invocations
2. Use sum_temp array to reduce waits on sum
3. Adjust num_unroll to reflext bigger threadblock
4. Improve default block_dims, increase support for more block_dims
* Add perf tests for `reduce_rows_f32` kernel
* Add heuristic to toggle 128/512 threads based on sm count
Break even point was the minimum of the following multiples.
| GPU Model | Nrow SM Count Multiple |
| ----------- | ----------- |
| RTX 4000 SFF ADA | 2.0x |
| RTX 6000 ADA | 2.5x |
| RTX PRO 6000 Blackwell Max-Q | 3.04x |
| RTX PRO 4500 Blackwell | 3.15x |
* Ensure perf gains also for small ncols and large nrows
Alternative to this, one could have also made the number of unrollings
template-able, but that would require compiling the kernel multiple
times, increasing binary size unnecessarily
* Modify perf and unit-tests
* Apply auto-formatting by clang
* Fix CI build failure
See https://github.com/ggml-org/llama.cpp/actions/runs/16798370266/job/47573716079?pr=15132#step:7:486
Building with VS generator worked though.
* Remove sm_count property from `ggml_backend_cuda_context`
Requested by @JohannesGaessler, and should fix remaining CI issues as a
side-effect
* Add CUB-based implementation for GGML_OP_MEAN
Currently this branch is only executed for nrows==1
* Add heuristics to execute CUB branch only when it brings perf
Heuristics were determined on the following HW:
* RTX 4000 SFF ADA
* RTX 6000 ADA
* RTX PRO 6000 Blackwell Max-Q
* RTX PRO 4500 Blackwell
* Add unit-test for CUB-based mean
Tests should run with CUDA Graphs enabled per default on NVGPUs
* Rename `USE_CUB` to `GGML_CUDA_USE_CUB`
Suggested by @JohannesGaessler
* Unindent Preprocessor directives
See
https://github.com/ggml-org/llama.cpp/pull/15132#discussion_r2269213506
Diego Devesa
Acly
Dong Won Kim
Jeff Bolz
Xuan-Son Nguyen
R0CKSTAR
Johannes Gäßler
lhez
SHUAI YANG
Marvin Gießing
Georgi Gerganov
Carlos Zoido
Siva Mahadevan
Daniel Bevenius
Treboko
Thea Mukhi
Reese Levine
Aaron Teo
compilade
rmatif
uvos
Jason Ni
Sigbjørn Skjæret
Jonathan Graehl
Judd
Oliver Simons