aboutsummaryrefslogtreecommitdiff
path: root/ggml-cuda.cu
AgeCommit message (Collapse)Author
2023-07-25Fix Q4_K and Q5_K for QK_K = 64 on CUDA (#2359)Kawrakow
* Fix Q4_K and Q5_K for QK_K = 64 * Very slightly better Q5_K bit fiddling --------- Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-07-24make rms_norm_eps a parameter (#2374)slaren
* make rms_norm_eps a parameter * add rms_norm_eps to command line * fix baby llama, test-grad0 * use scientific notation for eps param in the help ggml-ci
2023-07-24ggml : sync (unary ops refactor, static-correctness) (#2370)Georgi Gerganov
* ggml : sync (unary ops, tests) ggml-ci * tests : remove unnecessary funcs
2023-07-24Some more Q4_K and Q5_K speedup on CUDA (#2346)Kawrakow
* Faster Q5_K on CUDA * Small Q5_K improvement on older GPUs * Spped up Q4_K on CUDA GTX1660: 29.5 ms/t -> 25.6 ms/t RTX4080: 8.40 ms/t -> 8.25 ms/t * Spped up Q4_K on CUDA GTX1660: 36.7 ms/t -> 35.6 ms/t RTX4080: 9.8 ms/t -> 9.5 ms/t * Address PR comments * Add some comments to satisfy PR reviewer --------- Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-07-23ggml: move op parameters from tensors to ggml_tensor::op_params (#2333)slaren
* ggml: move op parameters from tensors to ggml_tensor::op_params * alibi: use memcpy for float params * remove `src[1] = NULL` in ops
2023-07-23llama : grouped-query attention + LLaMAv2 70B support (#2276)Georgi Gerganov
* CUDA: GQA implementation * llama : support for GQA and LLaMAv2 70B ggml-ci * py : fix hparams parsing (if-else blocks) ggml-ci * py : oh boy .. ggml-ci * help : fix gqa value for 70B ggml-ci --------- Co-authored-by: JohannesGaessler <johannesg@5d6.de>
2023-07-23Speed up Q4_K (#2322)Kawrakow
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-07-22CUDA: Fixed 7b q3_K_S with mul_mat_vec_q (#2313)Johannes Gäßler
2023-07-21Custom RoPE + bettter memory management for CUDA (#2295)Kawrakow
* Custom RoPE + bettter memory management for CUDA * Adjusted look ahead in ggml_cuda_pool_malloc to 5% This is sufficient it seems. We end up using about 200 MB less VRAM that way when running the 13B model with context 8192. --------- Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-07-21llama : make tensor_split ptr instead of array (#2272)Georgi Gerganov
2023-07-17Support dup & cont ops on CUDA (#2242)Jiahao Li
2023-07-14cuda : allocate all temporary ggml_tensor_extra_gpu from a fixed-size buffer ↵Bach Le
(#2220)
2023-07-14cuda : support broadcast add & mul (#2192)Jiahao Li
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2023-07-14CUDA: mul_mat_vec_q kernels for k-quants (#2203)Johannes Gäßler
2023-07-14ggml : sync (ggml_conv_2d, fix mul_mat bug, CUDA GLM rope)Georgi Gerganov
2023-07-13Fix compile error on Windows CUDA (#2207)Howard Su
2023-07-12cuda : add gelu supportGeorgi Gerganov
2023-07-12Fixed __dp4a compute capability: 6.0 -> 6.1 (#2189)Johannes Gäßler
2023-07-12ggml : revert CUDA broadcast changes from #2183 (#2191)Georgi Gerganov
2023-07-11ggml : sync (abort callback, mul / add broadcast, fix alibi) (#2183)Georgi Gerganov
2023-07-11ggml : remove src0 and src1 from ggml_tensor and rename opt to src (#2178)Spencer Sutton
* Add ggml changes * Update train-text-from-scratch for change * mpi : adapt to new ggml_tensor->src --------- Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2023-07-08Fixed OpenLLaMA 3b CUDA mul_mat_vec_q (#2144)Johannes Gäßler
2023-07-08CUDA: add __restrict__ to mul mat vec kernels (#2140)Johannes Gäßler
2023-07-05Quantized dot products for CUDA mul mat vec (#2067)Johannes Gäßler
2023-07-03Fix crash of test-tokenizer-0 under Debug build (#2064)Howard Su
* Fix crash of test-tokenizer-0 under Debug build * Change per comment
2023-07-01Better CUDA synchronization logic (#2057)Johannes Gäßler
2023-06-28cuda : remove nchannels_x argument from mul_mat_vec_nc_f16_f32 (#2028)Salvador E. Tropea
- Not used
2023-06-28cuda : fix missing const qualifier in casts (#2027)Salvador E. Tropea
2023-06-28CUDA GPU acceleration for LoRAs + f16 models (#1970)Johannes Gäßler
2023-06-26k-quants : support for super-block size of 64 (#2001)Kawrakow
* k_quants: WIP super-blocks with 64 weights * k_quants: WIP super-blocks with 64 weights Q6_K scalar and AVX2 works * k_quants: WIP super-blocks with 64 weights Q4_K scalar and AVX2 works * k_quants: WIP super-blocks with 64 weights Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower than the scalar implementation) * k_quants: WIP super-blocks with 64 weights Q3_K scalar and AVX2 works. * k_quants: WIP super-blocks with 64 weights Q5_K scalar and AVX2 works, and with that all k_quants are done on AVX2 and scalar * k_quants: WIP super-blocks with 64 weights Q6_K working on CUDA. Cannot make it run quite as gast as with super-blocks with 256 weigths: 8% slower on 4080, 20% slower on the 1660 (but there we fit 1 less layer on the GPU because pf the larger model size), so some fraction of these 20% is due to that, * k_quants: WIP super-blocks with 64 weights Q4_K working on CUDA. ~10% slower on GTX-1660, 16% slower on 4080. * k_quants: WIP super-blocks with 64 weights Q2_K working on CUDA. ~3% slower on GTX-1660, 10% slower on 4080. * k_quants: WIP super-blocks with 64 weights Q3_K working on CUDA. * k_quants: WIP super-blocks with 64 weights Q5_K working on CUDA, and with this CUDA is done. * k_quants: WIP super-blocks with 64 weights Q6_K working on ARM_NEON * k_quants: WIP super-blocks with 64 weights Q4_K working on ARM_NEON, but quite a bit slower than 256 weights * k_quants: WIP super-blocks with 64 weights Q2_K working on ARM_NEON, but quite a bit slower than 256 weights * k_quants: WIP super-blocks with 64 weights Q3_K working on ARM_NEON, but quite a bit slower than 256 weights. * k_quants: WIP super-blocks with 64 weights Q5_K working on ARM_NEON, but quite a bit slower than 256 weights. With that, we have full support for ARM_NEON, although performance is not quite there. * k_quants: WIP super-blocks with 64 weights Slightly more efficient Q3_K and Q5_K * k_quants: WIP super-blocks with 64 weights Another small improvement for Q3_K and Q5_K on ARM_NEON * k_quants: WIP super-blocks with 64 weights Yet another speedup for Q5_K on ARM_NEON. We are now within 10% of the QK_K = 256 version. * k_quants: WIP super-blocks with 64 weights * We are able to pass preprocessor macros to the Metal compiler * Q6_K works and is actually slightly more efficient than the QK_K = 256 version (25.2 ms vs 25.8 ms) * k_quants: WIP super-blocks with 64 weights Q4_K works on Metal and is actually slightly faster than QK_K = 256 (21.95 ms vs 24.0 ms). * k_quants: WIP super-blocks with 64 weights Q2_K works on Metal and is very slightly faster than QK_K = 256 (23.8 ms vs 24.2 ms). * k_quants: WIP super-blocks with 64 weights Q3_K works on Metal and is slightly faster than QK_K = 256 (26.6 ms vs 28.3 ms). * k_quants: WIP super-blocks with 64 weights Q5_K works on Metal and is slightly faster than QK_K = 256 (23.7 ms vs 26.3 ms). * k_quants: call them _K, not _k, also on Metal * k_quants: correctly define QK_K in llama.cpp * Fixed bug in q4_K quantization added with the 64-block addition * Simplify via lambda * k_quants: swicth Q3_K to 4-bit scales when QK_K = 64 Otherwise there isn't much benefit from this quantization type. There is some very slight loss in accuracy, but we reduce size by ~7%. E.g., for OpenLLaMA-3B, Q3_K_S perplexity is 8.6131 with 8-bit scales and 8.6352 with 4-bit, while file size decreases from 1.53G to 1.44G. * k_quants: switch Q4_K to 4-bit scales when QK_K = 64 Here the loss in accuracy is greater than for Q3_K, but the Q4_K points still move further to the left on the perplexity vs size curve. * k_quants: forgot to add the Metal changes in last commit * k_quants: change Q5_K to be type 0 when QK_K = 64 Still needs AVX2 implementation * k_quants: AVX2 implementation for new 64-weight Q5_K * k_quants: 10% faster ARM_NEON Q5_K dot product * k_quants: fixed issue caused by merging with master --------- Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26Fix assert when free invalid cuda pointer (#2005)Howard Su
Fix assert via initializing extra structure always. CUDA error 1 at C:\GPT\llama.cpp\ggml-cuda.cu:2536: invalid argument
2023-06-24#1869 Fix null reference errors when training from scratch with CUDA (#1907)Robyn
* #1869 Fix null reference errors when training from scratch with CUDA build Calling ggml_compute_forward when node->src0 was null was causing train-text-from-scratch.exe to terminate unexpectedly. * ggml : do not dereference src0 if NULL --------- Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2023-06-19cuda : faster k-quants on older GPUs (#1930)Kawrakow
* k_quants: hopefully much faster Q4_K on older GPUs On the GTX-1660 that I have available to represent "old GPUs", token prediction drops from 65.5 ms/tok to 41.5 ms/tok! * k_quants: hopefully much faster Q3_K on older GPUs On the GTX-1660 that I have available to represent "old GPUs", token prediction drops from 60.3 ms/tok to 41.0 ms/tok! * k_quants: faster Q2_K on older GPUs It looks like I didn't need to change anything compared to what we already had, so this is just adding clarifying comments. But I now measure 36.3 ms/tok on the GTX-1660, instead fo the 47.2 ms/tok that I have written in the faster k-quants PR. * k_quants: faster Q5_K on older GPUs 68.5 ms/tok -> 62.0 ms/tok on GTX-1660. For some reason the same access pattern that leads to such resounding success for Q2_K to Q4_K did not work at all for Q5_K. It is also more difficult to measure because for Q5_K_S we only have 32 layers on the GTX-1660, so output, tok embeddings and kv cache are done on the CPU. --------- Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-19Convert vector to f16 for dequantize mul mat vec (#1913)Johannes Gäßler
* Convert vector to f16 for dmmv * compile option * Added compilation option description to README * Changed cmake CUDA_ARCHITECTURES from "OFF" to "native"
2023-06-17Only one CUDA stream per device for async compute (#1898)Johannes Gäßler
2023-06-17ggml : fix warnings under MSVC (#1908)Howard Su
2023-06-16CUDA : faster k-quant dot kernels (#1862)Kawrakow
* cuda : faster k-quant dot kernels * Imrove Q2_K dot kernel on older GPUs We now have a K_QUANTS_PER_ITERATION macro, which should be set to 1 on older and to 2 on newer GPUs. With this, we preserve the performance of the original PR on RTX-4080, and are faster compared to master on GTX-1660. * Imrove Q6_K dot kernel on older GPUs Using the same K_QUANTS_PER_ITERATION macro as last commit, we preserve performance on RTX-4080 and speed up Q6_K on a GTX-1660. * Add LLAMA_CUDA_KQUANTS_ITER to CMakeLists.txt and Makefile Allowed values are 1 or 2. 2 gives the best performance on modern GPUs and is set as default. On older GPUs 1 may work better. * PR comments --------- Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-15Fixed CUDA runtime version check (#1879)Johannes Gäßler
2023-06-15Fix the validation of main device (#1872)Howard Su
2023-06-14CUDA full GPU acceleration, KV cache in VRAM (#1827)Johannes Gäßler
* Fixed CUDA RoPE * ggml_cuda_mul_mat_vec_p021 * ggml_cuda_scale * ggml_cuda_diag_mask_inf * ggml_is_permuted * ggml_cuda_cpy * flatten rows for ggml_cuda_op * Added a --low-vram option * Fixed Windows performance * Fixed LLAMA_CUDA_DMMV_Y > 1 for WizardLM
2023-06-12Leverage mmap for offloading tensors to GPU (#1597)Howard Su
* Rebase to latest * Show progress * Add assert to make sure we only allocate temp buffer for non-CPU backend tensor Co-authored-by: Johannes Gäßler <johannesg@5d6.de> --------- Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
2023-06-11Fixed WSL cuda's OOM error (#1594)Kyle Liang
* In the function , add the cuda error bypass. * remove excessive codes and prints --------- Co-authored-by: liang <liangmanlai@126.com>
2023-06-09Windows nvcc workaround (#1753)Johannes Gäßler
Fix gibberish output on Windows when using CUDA
2023-06-07k-quants : allow to optionally disable at compile time (#1734)Georgi Gerganov
* k-quants : put behind optional compile flag LLAMA_K_QUANTS * build : enable k-quants by default
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
2023-06-05ggml : 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>
2023-05-26cuda : 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
2023-05-20cuda : loading models directly into VRAM, norm calculation on GPU, ↵Johannes Gäßler
broadcasting for ggml_mul (#1483) * Broadcasting for ggml_mul * CUDA kernel for ggml_mul, norms in VRAM * GPU weights not in RAM, direct loading with cuFile * fixup! GPU weights not in RAM, direct loading with cuFile * fixup! GPU weights not in RAM, direct loading with cuFile * define default model path once, sync path with readme (#1366) * ~7% faster Q5_1 AVX2 code (#1477) * convert.py: Support models which are stored in a single pytorch_model.bin (#1469) * Support models in a single pytorch_model.bin * Remove spurious line with typo * benchmark-matmul: Print the average of the test results (#1490) * Remove unused n_parts parameter (#1509) * Fixes #1511 lambda issue for w64devkit (mingw) (#1513) * Fix for w64devkit and mingw * make kv_f16 the default for api users (#1517) * minor : fix compile warnings * readme : adds WizardLM to the list of supported models (#1485) * main : make reverse prompt option act as a stop token in non-interactive mode (#1032) * Make reverse prompt option act as a stop token in non-interactive scenarios * Making requested review changes * Update gpt_params_parse and fix a merge error * Revert "Update gpt_params_parse and fix a merge error" This reverts commit 2bb2ff1748513591ad45b175a75ed1d8089d84c8. * Update gpt_params_parse and fix a merge error take 2 * examples : add persistent chat (#1495) * examples : add persistent chat * examples : fix whitespace --------- Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * tests : add missing header * ggml : use F16 instead of F32 in Q4_0, Q4_1, Q8_0 (#1508) * ggml : use F16 instead of F32 in Q4_0, Q4_1 and Q8_0 * llama : bump LLAMA_FILE_VERSION to 3 * cuda : update Q4 and Q8 dequantize kernels * ggml : fix AVX dot products * readme : update performance table + hot topics * ggml : fix scalar implementation of Q4_1 dot * llama : fix compile warnings in llama_set_state_data() * llama : fix name shadowing and C4146 (#1526) * Fix name shadowing and C4146 * Fix if macros not using defined when required * Update llama-util.h Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> * Update llama-util.h Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> * Code style Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> --------- Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Fix for mingw (#1462) * llama : add llama_init_backend() API (close #1527) * feature : add blis and other BLAS implementation support (#1502) * feature: add blis support * feature: allow all BLA_VENDOR to be assigned in cmake arguments. align with whisper.cpp pr 927 * fix: version detection for BLA_SIZEOF_INTEGER, recover min version of cmake * Fix typo in INTEGER Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> --------- Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Revert "feature : add blis and other BLAS implementation support (#1502)" This reverts commit 07e9ace0f9da424d82e75df969642522880feb92. * GPU weights not in RAM, direct loading with cuFile * llama : code style fixes + progress print fix * ggml : ggml_mul better broadcast support * cmake : workarounds for cufile when CMake version < 3.25 * gg rebase fixup * Loop in llama.cpp, fixed progress callback * Attempt clang-tidy fix * llama : fix vram size computation * Add forgotten fclose() --------- Co-authored-by: András Salamon <ott2@users.noreply.github.com> Co-authored-by: Ilya Kurdyukov <59548320+ilyakurdyukov@users.noreply.github.com> Co-authored-by: Tom Jobbins <784313+TheBloke@users.noreply.github.com> Co-authored-by: rankaiyx <rankaiyx@rankaiyx.com> Co-authored-by: Stephan Walter <stephan@walter.name> Co-authored-by: DannyDaemonic <DannyDaemonic@gmail.com> Co-authored-by: Erik Scholz <Green-Sky@users.noreply.github.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> Co-authored-by: David Kennedy <dakennedyd@gmail.com> Co-authored-by: Jason McCartney <jmac@theroot.org> Co-authored-by: Evan Jones <evan.q.jones@gmail.com> Co-authored-by: Maxime <672982+maximegmd@users.noreply.github.com> Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Co-authored-by: Zenix <zenixls2@gmail.com>
2023-05-19ggml : use F16 instead of F32 in Q4_0, Q4_1, Q8_0 (#1508)Georgi Gerganov
* ggml : use F16 instead of F32 in Q4_0, Q4_1 and Q8_0 * llama : bump LLAMA_FILE_VERSION to 3 * cuda : update Q4 and Q8 dequantize kernels * ggml : fix AVX dot products * readme : update performance table + hot topics
2023-05-14cuda : deduplicated dequantization code (#1453)Johannes Gäßler