{"model": "reduced-order systolic-array GEMM dataflow -- roofline, PE utilization, energy (educational)", "inputs": {"rows": 128, "cols": 128, "freq_ghz": 1.0, "mac_pj": 0.2, "precision_bits": 8.0, "sram_mb": 24.0, "hbm_bw_gbs": 1600.0, "gemm_m": 2048, "gemm_k": 8192, "gemm_n": 8192, "dataflow": "ws"}, "outputs": {"peak_tops": 32.77, "achieved_tops": 29.15, "pe_utilization_percent": 89.0, "map_efficiency_percent": 100.0, "temporal_efficiency_percent": 89.0, "arithmetic_intensity": 2730.67, "ridge_point_flop_byte": 20.48, "bottleneck": "compute-bound", "latency_ms": 9.429, "compute_time_ms": 9.429, "memory_time_ms": 0.0629, "hbm_traffic_gb": 0.101, "tiles_k": 64, "tiles_n": 64, "tile_working_set_mb": 1.327, "sram_fits": true, "tops_per_watt": 9.86, "power_w": 3.0, "energy_per_inference_mj": 27.8904, "verdict": "compute-bound, well-mapped (near-peak MXU utilization)"}, "profile": {"roofline": [[0.25, 0.4], [0.3092, 0.495], [0.3825, 0.612], [0.4731, 0.757], [0.5851, 0.936], [0.7237, 1.158], [0.8952, 1.432], [1.1072, 1.772], [1.3695, 2.191], [1.6939, 2.71], [2.0951, 3.352], [2.5914, 4.146], [3.2052, 5.128], [3.9645, 6.343], [4.9036, 7.846], [6.0652, 9.704], [7.5019, 12.003], [9.2789, 14.846], [11.4768, 18.363], [14.1954, 22.713], [17.558, 28.093], [21.7171, 32.768], [26.8614, 32.768], [33.2243, 32.768], [41.0944, 32.768], [50.8288, 32.768], [62.869, 32.768], [77.7612, 32.768], [96.1811, 32.768], [118.9643, 32.768], [147.1443, 32.768], [181.9995, 32.768], [225.1111, 32.768], [278.435, 32.768], [344.39, 32.768], [425.9683, 32.768], [526.8708, 32.768], [651.6747, 32.768], [806.042, 32.768], [996.9754, 32.768], [1233.1367, 32.768], [1525.2394, 32.768], [1886.5346, 32.768], [2333.4127, 32.768], [2886.1463, 32.768], [3569.8102, 32.768], [4415.4188, 32.768], [5461.3333, 32.768]], "util_vs_dim": [[8, 0.9932], [30, 0.9658], [51, 0.949], [73, 0.9214], [94, 0.8991], [116, 0.8895], [138, 0.8634], [159, 0.8505], [181, 0.8233], [203, 0.8092], [224, 0.8023], [246, 0.7741], [267, 0.7776], [289, 0.7457], [311, 0.7306], [332, 0.7362], [354, 0.6914], [375, 0.7222], [397, 0.6963], [419, 0.6786], [440, 0.6721], [462, 0.6692], [484, 0.6736], [505, 0.6102], [527, 0.6236], [548, 0.6474], [570, 0.5901], [592, 0.6195], [613, 0.5703], [635, 0.6082], [657, 0.5607], [678, 0.52], [700, 0.5652], [721, 0.5264], [743, 0.4895], [765, 0.5428], [786, 0.5082], [808, 0.4751], [829, 0.5399], [851, 0.5064], [873, 0.4756], [894, 0.4485], [916, 0.5215], [938, 0.4917], [959, 0.4654], [981, 0.4399], [1002, 0.4173], [1024, 0.5002]], "op_ai": 2730.6667, "op_tops": 29.152, "peak_tops": 32.768, "ridge_ai": 20.48, "rows": 128, "cols": 128, "dataflow": "ws", "pe_util": 0.8897, "map_eff": 1.0, "temporal_eff": 0.8897, "tiles_k": 64, "tiles_n": 64, "bottleneck": "compute-bound"}, "steps": 96, "node": "ip-172-26-4-148", "compute_ms": 0.2}