aboutsummaryrefslogtreecommitdiff
AgeCommit message (Collapse)Author
2023-06-06llama : fix vram_scratch varGeorgi Gerganov
2023-06-06llama : fix compile warningsGeorgi Gerganov
2023-06-06Multi GPU support, CUDA refactor, CUDA scratch buffer (#1703)Johannes Gäßler
* CUDA multi GPU + scratch ggml_cuda_compute_forward Tensor parallelism ggml_cuda_add ggml_cuda_rms_norm ggml_cuda_silu CUDA scratch buffer --main-gpu CLI option
2023-06-06metal : add f16 supportGeorgi Gerganov
2023-06-06Clblast fixes + enhancements to save VRAM and offload more layers (#1675)LostRuins
* Use events instead of clFinish, where possible * OpenCL: Don't load gpu layers into RAM, add mul_f32 kernel * Reduce queueing overhead for contiguous tensors by using single mul kernel call * Adapt to #1612 cl_mem malloc changes * Reduce code duplication between cuda and opencl branches * Improve implementation * Clblast fixes + enhancements to save VRAM: 1. Change all Clblast buffers to CL_MEM_READ_WRITE, as the pool malloc currently doesn't properly handle them. 2. When recycling buffers in pool malloc, always assign the SMALLEST available buffer that fits, instead of the FIRST available buffer 3. When failing to recycle a buffer in pool malloc (all too small), instead recycle the largest available free buffer by resizing it. * change max value size_t to use limits * removed flags from the CL pool malloc, apply code tidying suggestions.
2023-06-06ggml : fix builds, add ggml-quants-k.o (close #1712, close #1710)Georgi Gerganov
2023-06-06gitignore : add .clang-tidyGeorgi Gerganov
2023-06-06llama : temporary disable Q6_K output quantization (#1711)Georgi Gerganov
2023-06-06metal : add checks for buffer size (#1706)Spencer Sutton
Co-authored-by: Spencer Sutton <Spencer.Sutton@precisely.com>
2023-06-05docs : add performance troubleshoot + example benchmark documentation (#1674)Yuval Peled
* test anchor link * test table * add benchmarks * Add performance troubleshoot & benchmark * add benchmarks * remove unneeded line --------- Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2023-06-05readme : fix typo (#1700)Foul-Tarnished
Fix a typo in a command in README.md
2023-06-05llama : consistently catch and throw only exceptions deriving from ↵mgroeber9110
std::exception (#1599) Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2023-06-05metal : use shared buffers between CPU and GPU (#1696)kiltyj
* Use MTLDevice.newBufferWithBytesNoCopy to share buffers between CPU and GPU * Page-align buffers used by Metal * Remove trailing whitespace * Only import unistd.h for Metal builds * metal : remove unnecessary copies --------- Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2023-06-05ggml : fix internal overflow in ggml_time_us on Windows (#1702)grahameth
Co-authored-by: grahameth <->
2023-06-05ci : disable auto tidy (#1705)Georgi Gerganov
2023-06-05ggml : add SOTA 2,3,4,5,6 bit k-quantizations (#1684)Kawrakow
* Starting to add k-quantization to ggml I think it is better to have quantization separate from ggml. For now just adding the k-quants there, but it would be better to also factor out the existing ggml quantizations. * Adding Q3_K and Q8_K (de)-quantization * Q3_K now working on CUDA and AVX2/scalar CUDA is not ideal - ~50% slower than Q4_0 for single token prediction, about the same in batch mode (perplexity). CPU single token is ~55 ms (on Ryzen 7950X). * Some improvement for Q3_K on CUDA It is now ~22.5 ms/token on my GPU, so ~30% slower than Q4_0. * Some more CUDA optimizations for Q3_K Single token is now 20.5 ms/token (~20% slower than Q4_0). Perplexity is on par with Q4_0. * Adding Q4_K - scalar, AVX2, CUDA Performance is the same or perhaps very slightly better than Q4_0 on the CPU. On the GPU, single token prediction is ~10% better than Q4_0, batch mode (perplexity is about the same). * Adding Q6_K - scalar, AVX2, CUDA Performance is ~40% lower compared to Q4_K on the CPU. This is to be expected, considering that we are memory bound on the CPU and the 6-bit model is ~44% larger than the 4-bit. On the GPU, single token prediction is ~6% lower than Q4_0, batch mode (perplexity) is even closer (but still slower). * Adding Q5_K - scalar, AVX2, CUDA Performance is ~20% lower compared to Q4_K on the CPU. This is to be expected, considering that we are memory bound on the CPU and the 5-bit model is ~22% larger than the 4-bit. On the GPU, single token prediction is about the same as Q4_0 for both, single token and batch prediction. * Per convention, all QX_K quantizations use Q5_K for output.weight * Adding quantization mixes * Quantization mixes: didn't quite get what I wanted in the last commit * Q4_K dot product for ARM_NEON * Q6_K dot product for ARM_NEON * Q5_K dot product for ARM_NEON * Adding Q3_K dot for ARM_NEON It is 22% slower than Q4_K, despite the smaller model size. On x86_64, where we are memory bound, the Q3_K model is quite a bit faster than Q4_K. * A very slightly faster ARM_NEON Q3_K dot * Adding Q2_K - just CUDA for now Token prediction is pretty good - about 15.5 ms on a RTX 4080. Perplexity is about the same as Q4_K. * Adding scalar and AVX2 Q2_K dot * Adding ARM_NEON Q2_K dot About the same performance as Q4_K. * A slightly faster ARM_NEON Q2_K dot Single token prediction is now ~36 ms on M2 Max. The code is much simpler too. * Fixed bug in Q2_K CUDA dot product kernel Stranegly enough, for the few prompts I tried with the 7B model the responses looked perfectly reasonable. Only realized something is not quite right when I tried the larger models and started getting nonse back. In any case, Q2_K single token evaluation time on an RTX 4080 in a Ryzen7950X box iusing CUDA and model fully loaded on the GPU are ~15.5 ms for 7B, ~25.4 ms for 13B, and ~55.8 ms for 30B. The max number of layers that fit in VRAM for The 65B is 32. With that, we get ~330 ms per token, which is not that much faster than just running on the CPU (~470 ms per token). * Don't print zeros/NaNs when no count histogram has been collected * A 10% faster CUDA vector dot kernel for Q3_K Q3_K is now running at ~18.5 ms / token on CUDA, so the gap to Q4_0 is only 10%. It seems memory acccess pattern is more important for performance than the amount of computation the kernel does. * A slightly daster Q4_K AVX2 dot product For perplexity, where we are less memory bound, time per pass drops by ~5%. Barely measurable difference for single token prediction. * A slightly faster ARM_NEON A4_K dot product * Minor * Fix quantization error test We cannot possibly be expecting rmse < 0.002 for 2- and 3-bit quantization variants. * Fix docker build I have been sloppy with vector reinterpret casts on ARM_NEON. It seems clang is very forgiving in that regard. * Added forgotten ggml.o dependence on k_quants.h to the Makefile * Had unintentionally committed the Makefile with -Ofast enabled * ggml : rename k_quants -> ggml-quants-k, use lowercase in code --------- Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2023-06-05Increase 3B scratch buffers. (#1698)Henri Vasserman
The 128 MB was too optimistic. Too bad it is not dynamically computed.
2023-06-05llama : fix Metal KV cache sync (close #1695)Georgi Gerganov
2023-06-04readme : update hot topicsGeorgi Gerganov
2023-06-04llama : Metal inference (#1642)Georgi Gerganov
* mtl : export the LLaMA computation graph * ci : disable temporary * mtl : adapt the MNIST example as starter * mtl : no need for mtl-export tool, add cli arg for main instead * mtl : export just a small part of the graph for now to make it easier * mtl : move MSL code into separate file for easy editing * mtl : initial get_rows_q4_0 kernel * mtl : confirmed get_rows_q4_0 is working correctly * mtl : add rms_norm kernel + confirm working * mtl : add mul kernel + confirm working * mtl : initial mul_mat Q4 kernel (wrong results) * mtl : mul_mat fixes (still wrong) * mtl : another mul_mat Q4 (still does not work) * mtl : working mul_mat q4 * ggml : fix handling of "view" ops in ggml_graph_import() * mtl : add rope kernel * mtl : add reshape and transpose handling * ggml : store offset as opt arg for ggml_view_xd() operators * mtl : add cpy kernel + handle view ops * mtl : confirm f16 x f32 attention mul mat * mtl : add scale kernel * mtl : add diag_mask_inf kernel * mtl : fix soft_max kernel * ggml : update ggml_nbytes() to handle non-contiguous tensors * mtl : verify V tensor contents * mtl : add f32 -> f32 cpy kernel * mtl : add silu kernel * mtl : add non-broadcast mul kernel * mtl : full GPU inference of the computation graph * mtl : optimize rms_norm and soft_max kernels * mtl : add f16 mat x f32 vec multiplication kernel * mtl : fix bug in f16 x f32 mul mat + speed-up computation * mtl : faster mul_mat_q4_0_f32 kernel * mtl : fix kernel signature + roll inner loop * mtl : more threads for rms_norm + better timing * mtl : remove printfs from inner loop * mtl : simplify implementation * mtl : add save/load vocab to ggml file * mtl : plug Metal inference into llama.cpp (very quick-n-dirty) * mtl : make it work with main example Lots of hacks but at least now it generates text * mtl : preparing for merge * mtl : clean-up ggml mtl interface + suport scratch / inplace * mtl : remove temp / debug code * metal : final refactoring and simplification * Revert "ci : disable temporary" This reverts commit 98c267fc77fe811082f672538fc91bcfc9072d63. * metal : add comments * metal : clean-up stuff, fix typos * readme : add Metal instructions * readme : add example for main
2023-06-04OpenCL: Fix duplication of layers in VRAM and RAM, add GPU mul kernel (#1653)0cc4m
* Use events instead of clFinish, where possible * OpenCL: Don't load gpu layers into RAM, add mul_f32 kernel * Reduce queueing overhead for contiguous tensors by using single mul kernel call * Adapt to #1612 cl_mem malloc changes * Reduce code duplication between cuda and opencl branches * Improve implementation
2023-06-03Add info about CUDA_VISIBLE_DEVICES (#1682)Henri Vasserman
2023-06-03Docker: change to calling convert.py (#1641)Jiří Podivín
Deprecation disclaimer was added to convert-pth-to-ggml.py
2023-06-03Fix prompt cache saving and chat-persistent rollover (#1678)Evan Jones
* Fix prompt cache saving and chat-persistent rollover (fixes #1670) * clang-tidy Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> --------- Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
2023-05-30OpenLLaMA 3B support (#1588)Henri Vasserman
This adds support to llama.cpp to load the model. Currently missing are changes that are required from convert.py to convert the model correctly. It needs some changes to start reading the JSON configuration for HF models instead of deriving the values by guessing. Co-authored-by: FNsi <125447286+FNsi@users.noreply.github.com>
2023-05-29ggml : sync cgraph import / export APIGeorgi Gerganov
2023-05-29ggml : fix bug in ggml_alibiGeorgi Gerganov
2023-05-29Work around for recalculating logits in cached prompts (Fixes #1585) (#1609)DannyDaemonic
* Work around for recalculating logits in cached prompts
2023-05-28Adding git in container package dependencies (#1621)Jiří Podivín
Git added to build packages for version information in docker image Signed-off-by: Jiri Podivin <jpodivin@gmail.com>
2023-05-28LLAMA_DEBUG adds debug symbols (#1617)Johannes Gäßler
2023-05-28Only show -ngl option when relevant + other doc/arg handling updates (#1625)Kerfuffle
1. Add a `LLAMA_SUPPORTS_GPU_OFFLOAD` define to `llama.h` (defined when compiled with CLBlast or cuBLAS) 2. Update the argument handling in the common example code to only show the `-ngl`, `--n-gpu-layers` option when GPU offload is possible. 3. Add an entry for the `-ngl`, `--n-gpu-layers` option to the `main` and `server` examples documentation 4. Update `main` and `server` examples documentation to use the new style dash separator argument format 5. Update the `server` example to use dash separators for its arguments and adds `-ngl` to `--help` (only shown when compiled with appropriate support). It will still support `--memory_f32` and `--ctx_size` for compatibility. 6. Add a warning discouraging use of `--memory-f32` for the `main` and `server` examples `--help` text as well as documentation. Rationale: https://github.com/ggerganov/llama.cpp/discussions/1593#discussioncomment-6004356
2023-05-28examples : add --alias option to gpt_params to set use friendly model name ↵Vladimir Zorin
(#1614)
2023-05-28opencl : no need to allocate cl_mem on heap (#1612)Howard Su
2023-05-28opencl : use strstr to check if fp16 supported (#1611)Howard Su
* Use strstr to check if fp16 supported * Ensure ext_buffer is null terminated
2023-05-27ggml : add support for the RISCV architecture (#1616)apcameron
2023-05-27Include server in releases + other build system cleanups (#1610)Kerfuffle
Set `LLAMA_BUILD_SERVER` in workflow so the `server` example gets build. This currently only applies to Windows builds because it seems like only Windows binary artifacts are included in releases. Add `server` example target to `Makefile` (still uses `LLAMA_BUILD_SERVER` define and does not build by default) Fix issue where `vdot` binary wasn't removed when running `make clean`. Fix compile warnings in `server` example. Add `.hpp` files to trigger workflow (the server example has one).
2023-05-27Add documentation about CLBlast (#1604)Henri Vasserman
Installing, compiling and using.
2023-05-27[CI] Fix openblas (#1613)Henri Vasserman
* Fix OpenBLAS build * Fix `LLAMA_BLAS_VENDOR` CMake variable that should be a string and not a boolean.
2023-05-27ggml : add ggml_tensor_overhead()Georgi Gerganov
2023-05-27[CI] CLBlast: Fix directory name (#1606)Henri Vasserman
2023-05-27ggml : sync ggml core (minor additions, e.g. ggml_get_tensor_by_name())Georgi Gerganov
2023-05-25Some improvements to loading the session with --prompt-cache (#1550)Kerfuffle
Improvements to loading the session with `--prompt-cache` in the `main` example. 1. Fix an issue where the `--seed` parameter was ignored when loading a cached prompt. 2. When loading a cached prompt, you previously had to specify the saved prompt (or a prefix of it) again. This pull changes that behavior to default to the prompt that was cached if a prompt wasn't specified by the user.
2023-05-26cuda : performance optimizations (#1530)Johannes Gäßler
* xor hack * block y dim * loop unrolling * Fixed cmake LLAMA_CUDA_BY option * Removed hipblas compatibility code * Define GGML_CUDA_DMMV_BLOCK_Y if not defined * Fewer iters, more ops per iter * Renamed DMMV X/Y compilation options
2023-05-24Update CLBlast to 1.6.0 (#1580)Henri Vasserman
* Update CLBlast to 1.6.0
2023-05-24readme : add docs for chat-persistent.sh (#1568)Evan Jones
* readme : add docs for chat-persistent.sh * Update README.md
2023-05-24chat-persistent.sh : use bracket expressions in grep (#1564)Senemu
2023-05-23Fix handling of "invalid property" when creating OpenCL command queue (#1565)Maarten ter Huurne
The `clCreateCommandQueue()` function will return the code `CL_INVALID_QUEUE_PROPERTIES` when passed unsupported properties, not `CL_INVALID_PROPERTY` as the original code was checking for.
2023-05-23OpenCL Token Generation Acceleration (#1459)0cc4m
* Move back to C++ for OpenCL * Refactor OpenCL code to work more like the CUDA code, add missing functions * Deduplicate dequant kernels * Add OpenCL compile options * Use compile args for preprocessing constants * Restore default platform + device selection by id behavior --------- Co-authored-by: Johannes Gäßler <johannesg@5d6.de> Co-authored-by: Henri Vasserman <henv@hot.ee>
2023-05-21examples : add server example with REST API (#1443)Steward Garcia
* Added httplib support * Added readme for server example * fixed some bugs * Fix the build error on Macbook * changed json11 to nlohmann-json * removed some whitespaces * remove trailing whitespace * added support custom prompts and more functions * some corrections and added as cmake option
2023-05-21make : .PHONY clean (#1553)Stefan Sydow