From ed8ac6e005d95e051bd03a182bbe0aa09a3c2266 Mon Sep 17 00:00:00 2001 From: David Nieto Date: Mon, 16 Oct 2017 12:24:59 -0700 Subject: gpu: nvgpu: fix smid generation of perf tables SMID tables were generated according with the local tpc and the pagepool and cb buffers from a different chip and did not take performance in consideration, which made compute kernels hang with CTAs on the fly. This change ensures we are using the right sizes and adds proper enumeration of smids. JIRA: NVGPUGV100-36 bug 2004378 Change-Id: Ic8f50c325d6d6720cca41d9740ae4f5f51e1100a Signed-off-by: David Nieto Reviewed-on: https://git-master.nvidia.com/r/1581664 Reviewed-by: mobile promotions Tested-by: mobile promotions --- drivers/gpu/nvgpu/Makefile | 1 + drivers/gpu/nvgpu/gv100/gr_gv100.c | 289 ++++++++++++++++++++++++++++++++++++ drivers/gpu/nvgpu/gv100/gr_gv100.h | 36 +++++ drivers/gpu/nvgpu/gv100/hal_gv100.c | 10 +- 4 files changed, 331 insertions(+), 5 deletions(-) create mode 100644 drivers/gpu/nvgpu/gv100/gr_gv100.c create mode 100644 drivers/gpu/nvgpu/gv100/gr_gv100.h diff --git a/drivers/gpu/nvgpu/Makefile b/drivers/gpu/nvgpu/Makefile index 8f1d42da..d6d60147 100644 --- a/drivers/gpu/nvgpu/Makefile +++ b/drivers/gpu/nvgpu/Makefile @@ -28,6 +28,7 @@ nvgpu-y += \ $(nvgpu-t19x)/gv100/fb_gv100.o \ $(nvgpu-t19x)/gv100/bios_gv100.o \ $(nvgpu-t19x)/gv100/fifo_gv100.o \ + $(nvgpu-t19x)/gv100/gr_gv100.o \ $(nvgpu-t19x)/gv100/hal_gv100.o nvgpu-$(CONFIG_TEGRA_GK20A) += $(nvgpu-t19x)/gv11b/platform_gv11b_tegra.o diff --git a/drivers/gpu/nvgpu/gv100/gr_gv100.c b/drivers/gpu/nvgpu/gv100/gr_gv100.c new file mode 100644 index 00000000..4b2038ba --- /dev/null +++ b/drivers/gpu/nvgpu/gv100/gr_gv100.c @@ -0,0 +1,289 @@ +/* + * GV100 GPU GR + * + * Copyright (c) 2017, NVIDIA CORPORATION. All rights reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + */ + +#include +#include +#include + +#include "gk20a/gk20a.h" +#include "gk20a/gr_gk20a.h" + +#include "gv100/gr_gv100.h" +#include "gv11b/subctx_gv11b.h" + +#include +#include + +/* + * Estimate performance if the given logical TPC in the given logical GPC were + * removed. + */ +static int gr_gv100_scg_estimate_perf(struct gk20a *g, + unsigned long *gpc_tpc_mask, + u32 disable_gpc_id, u32 disable_tpc_id, + int *perf) +{ + struct gr_gk20a *gr = &g->gr; + int err = 0; + u32 scale_factor = 512UL; /* Use fx23.9 */ + u32 pix_scale = 1024*1024UL; /* Pix perf in [29:20] */ + u32 world_scale = 1024UL; /* World performance in [19:10] */ + u32 tpc_scale = 1; /* TPC balancing in [9:0] */ + u32 scg_num_pes = 0; + u32 min_scg_gpc_pix_perf = scale_factor; /* Init perf as maximum */ + u32 average_tpcs = 0; /* Average of # of TPCs per GPC */ + u32 deviation; /* absolute diff between TPC# and + * average_tpcs, averaged across GPCs + */ + u32 norm_tpc_deviation; /* deviation/max_tpc_per_gpc */ + u32 tpc_balance; + u32 scg_gpc_pix_perf; + u32 scg_world_perf; + u32 gpc_id; + u32 pes_id; + int diff; + bool is_tpc_removed_gpc = false; + bool is_tpc_removed_pes = false; + u32 max_tpc_gpc = 0; + u32 num_tpc_mask; + u32 *num_tpc_gpc = nvgpu_kzalloc(g, sizeof(u32) * + nvgpu_get_litter_value(g, GPU_LIT_NUM_GPCS)); + + if (!num_tpc_gpc) + return -ENOMEM; + + /* Calculate pix-perf-reduction-rate per GPC and find bottleneck TPC */ + for (gpc_id = 0; gpc_id < gr->gpc_count; gpc_id++) { + num_tpc_mask = gpc_tpc_mask[gpc_id]; + + if ((gpc_id == disable_gpc_id) && num_tpc_mask & + (0x1 << disable_tpc_id)) { + /* Safety check if a TPC is removed twice */ + if (is_tpc_removed_gpc) { + err = -EINVAL; + goto free_resources; + } + /* Remove logical TPC from set */ + num_tpc_mask &= ~(0x1 << disable_tpc_id); + is_tpc_removed_gpc = true; + } + + /* track balancing of tpcs across gpcs */ + num_tpc_gpc[gpc_id] = hweight32(num_tpc_mask); + average_tpcs += num_tpc_gpc[gpc_id]; + + /* save the maximum numer of gpcs */ + max_tpc_gpc = num_tpc_gpc[gpc_id] > max_tpc_gpc ? + num_tpc_gpc[gpc_id] : max_tpc_gpc; + + /* + * Calculate ratio between TPC count and post-FS and post-SCG + * + * ratio represents relative throughput of the GPC + */ + scg_gpc_pix_perf = scale_factor * num_tpc_gpc[gpc_id] / + gr->gpc_tpc_count[gpc_id]; + + if (min_scg_gpc_pix_perf > scg_gpc_pix_perf) + min_scg_gpc_pix_perf = scg_gpc_pix_perf; + + /* Calculate # of surviving PES */ + for (pes_id = 0; pes_id < gr->gpc_ppc_count[gpc_id]; pes_id++) { + /* Count the number of TPC on the set */ + num_tpc_mask = gr->pes_tpc_mask[pes_id][gpc_id] & + gpc_tpc_mask[gpc_id]; + + if ((gpc_id == disable_gpc_id) && (num_tpc_mask & + (0x1 << disable_tpc_id))) { + + if (is_tpc_removed_pes) { + err = -EINVAL; + goto free_resources; + } + num_tpc_mask &= ~(0x1 << disable_tpc_id); + is_tpc_removed_pes = true; + } + if (hweight32(num_tpc_mask)) + scg_num_pes++; + } + } + + if (!is_tpc_removed_gpc || !is_tpc_removed_pes) { + err = -EINVAL; + goto free_resources; + } + + if (max_tpc_gpc == 0) { + *perf = 0; + goto free_resources; + } + + /* Now calculate perf */ + scg_world_perf = (scale_factor * scg_num_pes) / gr->ppc_count; + deviation = 0; + average_tpcs = scale_factor * average_tpcs / gr->gpc_count; + for (gpc_id =0; gpc_id < gr->gpc_count; gpc_id++) { + diff = average_tpcs - scale_factor * num_tpc_gpc[gpc_id]; + if (diff < 0) + diff = -diff; + deviation += diff; + } + + deviation /= gr->gpc_count; + + norm_tpc_deviation = deviation / max_tpc_gpc; + + tpc_balance = scale_factor - norm_tpc_deviation; + + if ((tpc_balance > scale_factor) || + (scg_world_perf > scale_factor) || + (min_scg_gpc_pix_perf > scale_factor) || + (norm_tpc_deviation > scale_factor)) { + err = -EINVAL; + goto free_resources; + } + + *perf = (pix_scale * min_scg_gpc_pix_perf) + + (world_scale * scg_world_perf) + + (tpc_scale * tpc_balance); +free_resources: + nvgpu_kfree(g, num_tpc_gpc); + return err; +} + +void gr_gv100_bundle_cb_defaults(struct gk20a *g) +{ + struct gr_gk20a *gr = &g->gr; + + gr->bundle_cb_default_size = + gr_scc_bundle_cb_size_div_256b__prod_v(); + gr->min_gpm_fifo_depth = + gr_pd_ab_dist_cfg2_state_limit_min_gpm_fifo_depths_v(); + gr->bundle_cb_token_limit = + gr_pd_ab_dist_cfg2_token_limit_init_v(); +} + +void gr_gv100_cb_size_default(struct gk20a *g) +{ + struct gr_gk20a *gr = &g->gr; + + if (!gr->attrib_cb_default_size) + gr->attrib_cb_default_size = + gr_gpc0_ppc0_cbm_beta_cb_size_v_default_v(); + gr->alpha_cb_default_size = + gr_gpc0_ppc0_cbm_alpha_cb_size_v_default_v(); +} + +void gr_gv100_set_gpc_tpc_mask(struct gk20a *g, u32 gpc_index) +{ +} + +void gr_gv100_init_sm_id_table(struct gk20a *g) +{ + u32 gpc, tpc, sm, pes, gtpc; + u32 sm_id = 0; + u32 sm_per_tpc = nvgpu_get_litter_value(g, GPU_LIT_NUM_SM_PER_TPC); + u32 num_sm = sm_per_tpc * g->gr.tpc_count; + int perf, maxperf; + int err; + unsigned long *gpc_tpc_mask; + u32 *tpc_table, *gpc_table; + + gpc_table = nvgpu_kzalloc(g, g->gr.tpc_count * sizeof(u32)); + tpc_table = nvgpu_kzalloc(g, g->gr.tpc_count * sizeof(u32)); + gpc_tpc_mask = nvgpu_kzalloc(g, sizeof(unsigned long) * + nvgpu_get_litter_value(g, GPU_LIT_NUM_GPCS)); + + if (!gpc_table || !tpc_table || !gpc_tpc_mask) { + nvgpu_err(g, "Error allocating memory for sm tables"); + goto exit_build_table; + } + + for (gpc = 0; gpc < g->gr.gpc_count; gpc++) + for (pes = 0; pes < g->gr.gpc_ppc_count[gpc]; pes++) + gpc_tpc_mask[gpc] |= g->gr.pes_tpc_mask[pes][gpc]; + + for (gtpc = 0; gtpc < g->gr.tpc_count; gtpc++) { + maxperf = -1; + for (gpc = 0; gpc < g->gr.gpc_count; gpc++) { + for_each_set_bit(tpc, &gpc_tpc_mask[gpc], + g->gr.gpc_tpc_count[gpc]) { + perf = -1; + err = gr_gv100_scg_estimate_perf(g, + gpc_tpc_mask, gpc, tpc, &perf); + + if (err) { + nvgpu_err(g, + "Error while estimating perf"); + goto exit_build_table; + } + + if (perf >= maxperf) { + maxperf = perf; + gpc_table[gtpc] = gpc; + tpc_table[gtpc] = tpc; + } + } + } + gpc_tpc_mask[gpc_table[gtpc]] &= ~(0x1 << tpc_table[gtpc]); + } + + for (tpc = 0, sm_id = 0; sm_id < num_sm; tpc++, sm_id += sm_per_tpc) { + for (sm = 0; sm < sm_per_tpc; sm++) { + g->gr.sm_to_cluster[sm_id + sm].gpc_index = + gpc_table[tpc]; + g->gr.sm_to_cluster[sm_id + sm].tpc_index = + tpc_table[tpc]; + g->gr.sm_to_cluster[sm_id + sm].sm_index = sm; + g->gr.sm_to_cluster[sm_id + sm].global_tpc_index = tpc; + } + } + + g->gr.no_of_sm = num_sm; + nvgpu_log_info(g, " total number of sm = %d", g->gr.no_of_sm); +exit_build_table: + nvgpu_kfree(g, gpc_table); + nvgpu_kfree(g, tpc_table); + nvgpu_kfree(g, gpc_tpc_mask); +} + +void gr_gv100_load_tpc_mask(struct gk20a *g) +{ + u64 pes_tpc_mask = 0x0ULL; + u32 gpc, pes; + u32 num_tpc_per_gpc = nvgpu_get_litter_value(g, + GPU_LIT_NUM_TPC_PER_GPC); + + /* gv100 has 6 GPC and 7 TPC/GPC */ + for (gpc = 0; gpc < g->gr.gpc_count; gpc++) { + for (pes = 0; pes < g->gr.pe_count_per_gpc; pes++) { + pes_tpc_mask |= (u64) g->gr.pes_tpc_mask[pes][gpc] << + (num_tpc_per_gpc * gpc); + } + } + + nvgpu_log_info(g, "pes_tpc_mask: %016llx\n", pes_tpc_mask); + gk20a_writel(g, gr_fe_tpc_fs_r(0), u64_lo32(pes_tpc_mask)); + gk20a_writel(g, gr_fe_tpc_fs_r(1), u64_hi32(pes_tpc_mask)); +} diff --git a/drivers/gpu/nvgpu/gv100/gr_gv100.h b/drivers/gpu/nvgpu/gv100/gr_gv100.h new file mode 100644 index 00000000..460b05ae --- /dev/null +++ b/drivers/gpu/nvgpu/gv100/gr_gv100.h @@ -0,0 +1,36 @@ +/* + * GV100 GPU GR + * + * Copyright (c) 2017, NVIDIA CORPORATION. All rights reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + */ + +#ifndef _NVGPU_GR_GV100_H_ +#define _NVGPU_GR_GV100_H_ + +void gr_gv100_bundle_cb_defaults(struct gk20a *g); +void gr_gv100_cb_size_default(struct gk20a *g); +void gr_gv100_set_gpc_tpc_mask(struct gk20a *g, u32 gpc_index); +void gr_gv100_init_sm_id_table(struct gk20a *g); +void gr_gv100_program_sm_id_numbering(struct gk20a *g, + u32 gpc, u32 tpc, u32 smid); +int gr_gv100_load_smid_config(struct gk20a *g); + +#endif diff --git a/drivers/gpu/nvgpu/gv100/hal_gv100.c b/drivers/gpu/nvgpu/gv100/hal_gv100.c index f20d2dcf..cefaf1ae 100644 --- a/drivers/gpu/nvgpu/gv100/hal_gv100.c +++ b/drivers/gpu/nvgpu/gv100/hal_gv100.c @@ -78,7 +78,7 @@ #include "gv11b/dbg_gpu_gv11b.h" #include "gv11b/hal_gv11b.h" -#include "gv11b/gr_gv11b.h" +#include "gv100/gr_gv100.h" #include "gv11b/mc_gv11b.h" #include "gv11b/ltc_gv11b.h" #include "gv11b/gv11b.h" @@ -263,8 +263,8 @@ static const struct gpu_ops gv100_ops = { }, .gr = { .init_gpc_mmu = gr_gv11b_init_gpc_mmu, - .bundle_cb_defaults = gr_gv11b_bundle_cb_defaults, - .cb_size_default = gr_gv11b_cb_size_default, + .bundle_cb_defaults = gr_gv100_bundle_cb_defaults, + .cb_size_default = gr_gv100_cb_size_default, .calc_global_ctx_buffer_size = gr_gv11b_calc_global_ctx_buffer_size, .commit_global_attrib_cb = gr_gv11b_commit_global_attrib_cb, @@ -285,7 +285,7 @@ static const struct gpu_ops gv100_ops = { .set_hww_esr_report_mask = gv11b_gr_set_hww_esr_report_mask, .falcon_load_ucode = gr_gm20b_load_ctxsw_ucode_segments, .load_ctxsw_ucode = gr_gm20b_load_ctxsw_ucode, - .set_gpc_tpc_mask = gr_gv11b_set_gpc_tpc_mask, + .set_gpc_tpc_mask = gr_gv100_set_gpc_tpc_mask, .get_gpc_tpc_mask = gr_gm20b_get_gpc_tpc_mask, .free_channel_ctx = gk20a_free_channel_ctx, .alloc_obj_ctx = gk20a_alloc_obj_ctx, @@ -335,7 +335,7 @@ static const struct gpu_ops gv100_ops = { .resume_contexts = gr_gk20a_resume_contexts, .get_preemption_mode_flags = gr_gp10b_get_preemption_mode_flags, .fuse_override = gp10b_gr_fuse_override, - .init_sm_id_table = gr_gv11b_init_sm_id_table, + .init_sm_id_table = gr_gv100_init_sm_id_table, .load_smid_config = gr_gv11b_load_smid_config, .program_sm_id_numbering = gr_gv11b_program_sm_id_numbering, .is_ltcs_ltss_addr = gr_gm20b_is_ltcs_ltss_addr, -- cgit v1.2.2