From dd5ae06405c5565b99889bdb3f168f4351252cfb Mon Sep 17 00:00:00 2001 From: Kawrakow <48489457+ikawrakow@users.noreply.github.com> Date: Mon, 8 Jan 2024 16:02:32 +0100 Subject: SOTA 2-bit quants (#4773) * iq2_xxs: basics * iq2_xxs: scalar and AVX2 dot products Needed to change Q8_K to have quants in the -127...127 range, else the IQ2_XXS AVX implementation becomes very awkward. The alternative would have been to use Q8_0 instead. Perhaps I'll change later, for now this is what we have. * iq2_xxs: ARM_NEON dot product Somehow strangely slow (112 ms/token). * iq2_xxs: WIP Metal Dequantize works, something is still wrong with the dot product. * iq2_xxs: Metal dot product now works We have PP-512 = 475 t/s TG-128 = 47.3 t/s Not the greatest performance, but not complete garbage either. * iq2_xxs: slighty faster dot product TG-128 is now 48.4 t/s * iq2_xxs: slighty faster dot product TG-128 is now 50.9 t/s * iq2_xxs: even faster Metal dot product TG-128 is now 54.1 t/s. Strangely enough, putting the signs lookup table into shared memory has a bigger impact than the grid values being in shared memory. * iq2_xxs: dequantize CUDA kernel - fix conflict with master * iq2_xxs: quantized CUDA dot product (MMVQ) We get TG-128 = 153.1 t/s * iq2_xxs: slightly faster CUDA dot product TG-128 is now at 155.1 t/s. * iq2_xxs: add to llama ftype enum * iq2_xxs: fix MoE on Metal * Fix missing MMQ ops when on hipBLAS I had put the ggml_supports_mmq call at the wrong place. * Fix bug in qequantize_row_iq2_xxs The 0.25f factor was missing. Great detective work by @ggerganov! * Fixing tests * PR suggestion --------- Co-authored-by: Iwan Kawrakow --- llama.cpp | 3 +++ 1 file changed, 3 insertions(+) (limited to 'llama.cpp') diff --git a/llama.cpp b/llama.cpp index 63853d1c..8e0717db 100644 --- a/llama.cpp +++ b/llama.cpp @@ -2222,6 +2222,7 @@ struct llama_model_loader { case GGML_TYPE_Q4_K: ftype = LLAMA_FTYPE_MOSTLY_Q4_K_M; break; case GGML_TYPE_Q5_K: ftype = LLAMA_FTYPE_MOSTLY_Q5_K_M; break; case GGML_TYPE_Q6_K: ftype = LLAMA_FTYPE_MOSTLY_Q6_K; break; + case GGML_TYPE_IQ2_XXS: ftype = LLAMA_FTYPE_MOSTLY_IQ2_XXS; break; default: { LLAMA_LOG_WARN("%s: unknown type %s\n", __func__, ggml_type_name(type_max)); @@ -2593,6 +2594,7 @@ static std::string llama_model_ftype_name(llama_ftype ftype) { case LLAMA_FTYPE_MOSTLY_Q5_K_S: return "Q5_K - Small"; case LLAMA_FTYPE_MOSTLY_Q5_K_M: return "Q5_K - Medium"; case LLAMA_FTYPE_MOSTLY_Q6_K: return "Q6_K"; + case LLAMA_FTYPE_MOSTLY_IQ2_XXS:return "IQ2_XSS - 2.0625 bpw"; default: return "unknown, may not work"; } @@ -9038,6 +9040,7 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s case LLAMA_FTYPE_MOSTLY_Q5_K_S: case LLAMA_FTYPE_MOSTLY_Q5_K_M: quantized_type = GGML_TYPE_Q5_K; break; case LLAMA_FTYPE_MOSTLY_Q6_K: quantized_type = GGML_TYPE_Q6_K; break; + case LLAMA_FTYPE_MOSTLY_IQ2_XXS:quantized_type = GGML_TYPE_IQ2_XXS; break; default: throw std::runtime_error(format("invalid output file type %d\n", ftype)); } -- cgit v1.2.3