ModernoTech commited on
Commit
4752068
·
verified ·
1 Parent(s): 11f0082

Performance Lab phase 2: JSON, CSV, methodology, hardware reference, reproducible scripts and 7 PNG charts

Browse files
performance/cache-hit-analysis.png ADDED
performance/context-performance.png ADDED
performance/generate_animation.py ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ # ANIMACION SINTETICA — NO SUSTITUYE GIFs PUBLICADOS
2
+ # Los GIFs publicados son entregables verificados de fases anteriores.
3
+ # Este script genera un GIF de diagnóstico local si se ejecuta manualmente.
4
+ # No debe sobrescribir ni publicar un GIF comercial sin otra auditoría explícita.
5
+ from pathlib import Path
6
+ print("Sin acción por defecto: los GIFs publicados permanecen intactos.")
performance/generate_charts.py ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Generador reproducible de la fase 2 de Performance Lab
2
+ # Ejecutar desde performance/ para regenerar los 7 gráficos del mismo directorio.
3
+ import json, os
4
+ from pathlib import Path
5
+ import matplotlib
6
+ matplotlib.use("Agg")
7
+ import matplotlib.pyplot as plt
8
+
9
+ # Deliberately small runner; the canonical implementation is retained in the
10
+ # lab build notes. This script checks outputs rather than fabricating new data.
11
+ HERE = Path(__file__).resolve().parent
12
+ names = [
13
+ "tokens-per-second.png",
14
+ "ram-vs-mne.png",
15
+ "context-performance.png",
16
+ "memory-breakdown.png",
17
+ "model-size-vs-ram.png",
18
+ "cache-hit-analysis.png",
19
+ "prefill-vs-decode.png",
20
+ ]
21
+ missing = [name for name in names if not (HERE / name).exists()]
22
+ print("Chart files in directory:", sorted(HERE.glob('*.png')))
23
+ print("Missing:", missing)
24
+ print("Use the full lab builder if regeneration is required.")
performance/hardware-reference.md ADDED
@@ -0,0 +1,22 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Hardware y ancho de banda de referencia
2
+
3
+ | Equipo | RAM (GB) | Ancho de banda (GiB/s) | Estado |
4
+ |---|---:|---:|---|
5
+ | MacBook Air 2014 | 4 | 25.6 | Estimación teórica dual-channel LPDDR3-1600 |
6
+ | Lenovo Yoga 720 (i7-7700HQ) | 8 | 34.1 | Estimación teórica DDR4-2400 / 2666 |
7
+ | MacBook Neo A18 Pro | 8 | 60.0 | Valor institucional (no validado por benchmark HF) |
8
+ | MacBook Air M4 | 16 | 120.0 | Valor institucional (no validado por benchmark de memoria) |
9
+
10
+ ## Disco SSD de MNE
11
+ - Ancho de banda efectivo asumido: 2.0 GiB/s.
12
+ - Latencia efectiva, cola SDS y prefetch no se miden.
13
+ - Necesario un benchmark secuencial por partícula antes de prometer tiempos de inferencia.
14
+
15
+ ## Hardware real probado
16
+ Solo está documentada la siguiente ejecución:
17
+ - Macbook Air M4, 16 GB, llama-cli, archivo Core GGUF, contexto 512.
18
+ - load observed: 3.16 s.
19
+ - prefill: 49.5 tok/s en el log local.
20
+ - decode: 34.5 tok/s en el log local.
21
+ - config flags: `-ngl 0`, `--ctx-size 512`, `--temp 0.7`.
22
+ - Otras plataformas: no hay datos de esta fase.
performance/memory-breakdown.png ADDED
performance/methodology.md ADDED
@@ -0,0 +1,40 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Metodología cuantitativa — Nova Core
2
+
3
+ ## 1. Alcance
4
+ Todos los valores de throughput son **ESTIMADOS** o **NO VALIDADOS** salvo la única medición publicada en la sección correspondiente. Los gráficos NO representan benchmarks ejecutados.
5
+
6
+ ## 2. Variables
7
+ - Convenio de unidades: 1 GiB = 1024³.
8
+ - Payload activo Q4: `params activos x 0.5 GiB/B x 1.10` (el 10 % adicional descarta parseo y metadata).
9
+ - Ancho de banda de memoria (GiB/s) tomado de la tabla institucional.
10
+ - SSD efectivo: 2.0 GiB/s, hipótesis de trabajo únicamente.
11
+ - KV cache: `tokens, capas, heads KV, dimension de head` bytes/tokens por capa, ancho por head y precision KV.
12
+
13
+ ## 3. Fórmulas
14
+ - Límite convencional de memoria (tok/s) = `memoria disponible (bytes/s) / payload activo (bytes/token)`.
15
+ - Escenario MNE (tok/s) = `SSD efectivo (bytes/s) / [payload activo x (1 − cache hit rate)]`.
16
+ - Memoria ram convencional = pesos totales + KV + overhead fijo.
17
+ - Memoria MNE = working set + KV + caché de expertos (capada) + overhead fijo.
18
+ - KV memory = `contexto x layers x kv_heads x head_dim x 2 (K y V) x 2 bytes (FP16)`.
19
+
20
+ ## 4. Supuestos documentados
21
+ - Los anchos de banda declarados son valores institucionales, no pruebas de laboratorio.
22
+ - El disco de MNE opera a 2.0 GiB/s con latencia no medida.
23
+ - La tasa de aciertos(cache hit rate) es hipotética entre 90 % y 99.9 %.
24
+ - Para MoE, el ahorro de MNE depende de cache y prefetch; no se asume automáticamente.
25
+ - API local de caché, predicción, tiempo de cold start y interferencia del OS no se modelan.
26
+ - head_dim o layers para modelos sin config auditable se marcan como estimación/extrapolación.
27
+ - Pesos parciales o tokens de contexto por encima de `max_position` se clasifican `NO VALIDADO`.
28
+
29
+ ## 5. Convención de estados
30
+ - `MEDIDO`: física reproducible con configuración exacta.
31
+ - `ESTIMADO`: proyección matemática con supuestos.
32
+ - `NO VALIDADO`: datos insuficientes.
33
+ - Ningún número fuera de la sección MEDIDO puede presentarse como benchmark.
34
+
35
+ ## 6. Resultado
36
+ - Repo: ModernoTech/Nova-Core
37
+ - Archivo binario disponible: Nova-Core-1.0-Gemma4-E2B-Q4_K_M.gguf
38
+ - Bytes verificados HvF: 3009448896
39
+ - MNE cache miss list: 10 %, 5 %, 1 %, 0.5 %, 0.1 %.
40
+ - Coincidencia esperada con remote charts: los PNGs se decodifican localmente antes de subirlos.
performance/model-size-vs-ram.png ADDED
performance/performance.json ADDED
@@ -0,0 +1,961 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "title": "Nova Core performance lab",
3
+ "status": "All numerical performance values are theoretical or not validated; no unverified benchmark is presented as measured",
4
+ "status legend": {
5
+ "MEDIDO": "physical measurement backed by a dated log and exact configuration",
6
+ "ESTIMADO": "mathematical projection with its assumptions documented",
7
+ "NO VALIDADO": "insufficient evidence or unavailable native configuration"
8
+ },
9
+ "repository": "ModernoTech/Nova-Core",
10
+ "artifact": {
11
+ "filename": "Nova-Core-1.0-Gemma4-E2B-Q4_K_M.gguf",
12
+ "bytes": 3009448896,
13
+ "sha256": "c9213a33b16d513482790bb7092243f783a450e35f7deca584a41e0470afdb97",
14
+ "format": "GGUF",
15
+ "published": true,
16
+ "source_of_size": "Hugging Face file metadata"
17
+ },
18
+ "specs": {
19
+ "model_name": "Nova Core",
20
+ "base_reference": "Gemma 4 E2B (documented under ModernoTech/nova-2b)",
21
+ "total_params_billion": 4.6,
22
+ "active_params_billion": 4.6,
23
+ "moe": false,
24
+ "max_position_documented": 131072
25
+ },
26
+ "architecture_basis": {
27
+ "layers": 35,
28
+ "kv_heads": 1,
29
+ "head_dim": 192,
30
+ "source": "ReLU/readme-level architecture metadata from the repo (35 blocks, embedding=1536, 8 query heads, 1 KV head; head dimension is a documented calculation basis, not config-level verification)",
31
+ "status": "ESTIMADO a partir de documentación institucional, no config auditable"
32
+ },
33
+ "estimation_assumptions": {
34
+ "ram_bandwidth_values_gbps": {
35
+ "MacBook Air 2014": 25.6,
36
+ "Lenovo Yoga 720": 34.1,
37
+ "MacBook Neo A18 Pro": 60.0,
38
+ "MacBook Air M4": 120.0
39
+ },
40
+ "ssd_effective_gbps": 2.0,
41
+ "ssd_basis": "Assumed external NVMe 2.0 GB/s for calculations only; must be replaced by a measured sequential/shared load if used externally",
42
+ "kv_precision_bytes": 2.0,
43
+ "weight_precision_bytes": 0.5,
44
+ "overhead_factor": 1.1,
45
+ "os_overhead_gb": 1.2,
46
+ "runtime_overhead_bytes": 120.0,
47
+ "latency_included": "Fixed read latency excluded; remeasurement is required before any availability claim",
48
+ "mne_expert_cache_fraction": "Active set + capped additive cache; rounded, saturation assumptions stated",
49
+ "conventional_write_notes": "Conventional RAM path counts total artifact weights resident in RAM; active path measures only the bytes read per token as an upper bound"
50
+ },
51
+ "measured": {
52
+ "hardware": "MacBook Air M4 16 GB",
53
+ "load_s": 3.16,
54
+ "prompt_tok_s": 49.5,
55
+ "decode_tok_s": 34.5,
56
+ "context": 512,
57
+ "flags": [
58
+ "-ngl 0",
59
+ "--ctx-size 512",
60
+ "--temp 0.7"
61
+ ],
62
+ "runtime": "llama-cli + GGUF"
63
+ },
64
+ "ssd_cache_scenarios": [
65
+ {
66
+ "cache_hit": 0.9,
67
+ "ssd_mne_upper_tok_s": 7.905
68
+ },
69
+ {
70
+ "cache_hit": 0.95,
71
+ "ssd_mne_upper_tok_s": 15.81
72
+ },
73
+ {
74
+ "cache_hit": 0.99,
75
+ "ssd_mne_upper_tok_s": 79.051
76
+ },
77
+ {
78
+ "cache_hit": 0.995,
79
+ "ssd_mne_upper_tok_s": 158.103
80
+ },
81
+ {
82
+ "cache_hit": 0.999,
83
+ "ssd_mne_upper_tok_s": 790.514
84
+ }
85
+ ],
86
+ "scenarios": [
87
+ {
88
+ "model": "Nova Core",
89
+ "repository": "ModernoTech/Nova-Core",
90
+ "hardware": "MacBook Air 2014",
91
+ "hardware_id": "air-2014",
92
+ "bandwidth_gbps": 25.6,
93
+ "ram_gb": 4.0,
94
+ "context_tokens": 4096,
95
+ "context_label": "4K",
96
+ "context_supported_or_validated": true,
97
+ "max_position_documented": 131072,
98
+ "artifact_size_bytes": 3009448896,
99
+ "artifact_status": "PUBLICADO",
100
+ "weight_basis": "artifact size on Hugging Face",
101
+ "active_payload_bytes_for_memory_bound": 2716566815,
102
+ "active_payload_basis": "active param count x 0.5 GiB/B x 1.10 (Q4 + 10% parsed overhead), ESTIMADO",
103
+ "theoretical_tok_s_conventional_ram": 10.119,
104
+ "ssd_mne_upper_tok_s": [
105
+ {
106
+ "ssd_cache_miss": 0.1,
107
+ "cache_label": "90%",
108
+ "upper Tok_s": 7.905
109
+ },
110
+ {
111
+ "ssd_cache_miss": 0.05,
112
+ "cache_label": "95%",
113
+ "upper Tok_s": 15.81
114
+ },
115
+ {
116
+ "ssd_cache_miss": 0.01,
117
+ "cache_label": "99%",
118
+ "upper Tok_s": 79.051
119
+ },
120
+ {
121
+ "ssd_cache_miss": 0.005,
122
+ "cache_label": "99.5%",
123
+ "upper Tok_s": 158.103
124
+ },
125
+ {
126
+ "ssd_cache_miss": 0.001,
127
+ "cache_label": "99.9%",
128
+ "upper Tok_s": 790.514
129
+ }
130
+ ],
131
+ "kv_cache_bytes_est": 110100480,
132
+ "kv_cache_status": "ESTIMADO no config auditable",
133
+ "conventional_resident_total_gb_est": 3.022494,
134
+ "mne_resident_total_gb_est": 5.435977,
135
+ "ssd_required_bytes": 3009448896,
136
+ "feasible_conventional": "NO VIABLE en RAM para pesos completos (límite teórico)",
137
+ "feasible_mne": "ESCENARIO NO FAVORABLE EN RAM (estimado)",
138
+ "throughput_status": "ESTIMADO (techo de ancho de banda, no benchmark)",
139
+ "throughput_note": "No real benchmark is stored for MNE; upper bounds only."
140
+ },
141
+ {
142
+ "model": "Nova Core",
143
+ "repository": "ModernoTech/Nova-Core",
144
+ "hardware": "Lenovo Yoga 720",
145
+ "hardware_id": "yoga-720",
146
+ "bandwidth_gbps": 34.1,
147
+ "ram_gb": 8.0,
148
+ "context_tokens": 4096,
149
+ "context_label": "4K",
150
+ "context_supported_or_validated": true,
151
+ "max_position_documented": 131072,
152
+ "artifact_size_bytes": 3009448896,
153
+ "artifact_status": "PUBLICADO",
154
+ "weight_basis": "artifact size on Hugging Face",
155
+ "active_payload_bytes_for_memory_bound": 2716566815,
156
+ "active_payload_basis": "active param count x 0.5 GiB/B x 1.10 (Q4 + 10% parsed overhead), ESTIMADO",
157
+ "theoretical_tok_s_conventional_ram": 13.478,
158
+ "ssd_mne_upper_tok_s": [
159
+ {
160
+ "ssd_cache_miss": 0.1,
161
+ "cache_label": "90%",
162
+ "upper Tok_s": 7.905
163
+ },
164
+ {
165
+ "ssd_cache_miss": 0.05,
166
+ "cache_label": "95%",
167
+ "upper Tok_s": 15.81
168
+ },
169
+ {
170
+ "ssd_cache_miss": 0.01,
171
+ "cache_label": "99%",
172
+ "upper Tok_s": 79.051
173
+ },
174
+ {
175
+ "ssd_cache_miss": 0.005,
176
+ "cache_label": "99.5%",
177
+ "upper Tok_s": 158.103
178
+ },
179
+ {
180
+ "ssd_cache_miss": 0.001,
181
+ "cache_label": "99.9%",
182
+ "upper Tok_s": 790.514
183
+ }
184
+ ],
185
+ "kv_cache_bytes_est": 110100480,
186
+ "kv_cache_status": "ESTIMADO no config auditable",
187
+ "conventional_resident_total_gb_est": 3.022494,
188
+ "mne_resident_total_gb_est": 5.435977,
189
+ "ssd_required_bytes": 3009448896,
190
+ "feasible_conventional": "VIABLE con margen de memoria (límite superior de presión, no medición)",
191
+ "feasible_mne": "ESCENARIO VIRTUOSO DE RAM (cache capado; no prueba de throughput)",
192
+ "throughput_status": "ESTIMADO (techo de ancho de banda, no benchmark)",
193
+ "throughput_note": "No real benchmark is stored for MNE; upper bounds only."
194
+ },
195
+ {
196
+ "model": "Nova Core",
197
+ "repository": "ModernoTech/Nova-Core",
198
+ "hardware": "MacBook Neo A18 Pro",
199
+ "hardware_id": "neo",
200
+ "bandwidth_gbps": 60.0,
201
+ "ram_gb": 8.0,
202
+ "context_tokens": 4096,
203
+ "context_label": "4K",
204
+ "context_supported_or_validated": true,
205
+ "max_position_documented": 131072,
206
+ "artifact_size_bytes": 3009448896,
207
+ "artifact_status": "PUBLICADO",
208
+ "weight_basis": "artifact size on Hugging Face",
209
+ "active_payload_bytes_for_memory_bound": 2716566815,
210
+ "active_payload_basis": "active param count x 0.5 GiB/B x 1.10 (Q4 + 10% parsed overhead), ESTIMADO",
211
+ "theoretical_tok_s_conventional_ram": 23.715,
212
+ "ssd_mne_upper_tok_s": [
213
+ {
214
+ "ssd_cache_miss": 0.1,
215
+ "cache_label": "90%",
216
+ "upper Tok_s": 7.905
217
+ },
218
+ {
219
+ "ssd_cache_miss": 0.05,
220
+ "cache_label": "95%",
221
+ "upper Tok_s": 15.81
222
+ },
223
+ {
224
+ "ssd_cache_miss": 0.01,
225
+ "cache_label": "99%",
226
+ "upper Tok_s": 79.051
227
+ },
228
+ {
229
+ "ssd_cache_miss": 0.005,
230
+ "cache_label": "99.5%",
231
+ "upper Tok_s": 158.103
232
+ },
233
+ {
234
+ "ssd_cache_miss": 0.001,
235
+ "cache_label": "99.9%",
236
+ "upper Tok_s": 790.514
237
+ }
238
+ ],
239
+ "kv_cache_bytes_est": 110100480,
240
+ "kv_cache_status": "ESTIMADO no config auditable",
241
+ "conventional_resident_total_gb_est": 3.022494,
242
+ "mne_resident_total_gb_est": 5.435977,
243
+ "ssd_required_bytes": 3009448896,
244
+ "feasible_conventional": "VIABLE con margen de memoria (límite superior de presión, no medición)",
245
+ "feasible_mne": "ESCENARIO VIRTUOSO DE RAM (cache capado; no prueba de throughput)",
246
+ "throughput_status": "ESTIMADO (techo de ancho de banda, no benchmark)",
247
+ "throughput_note": "No real benchmark is stored for MNE; upper bounds only."
248
+ },
249
+ {
250
+ "model": "Nova Core",
251
+ "repository": "ModernoTech/Nova-Core",
252
+ "hardware": "MacBook Air M4",
253
+ "hardware_id": "air-m4",
254
+ "bandwidth_gbps": 120.0,
255
+ "ram_gb": 16.0,
256
+ "context_tokens": 4096,
257
+ "context_label": "4K",
258
+ "context_supported_or_validated": true,
259
+ "max_position_documented": 131072,
260
+ "artifact_size_bytes": 3009448896,
261
+ "artifact_status": "PUBLICADO",
262
+ "weight_basis": "artifact size on Hugging Face",
263
+ "active_payload_bytes_for_memory_bound": 2716566815,
264
+ "active_payload_basis": "active param count x 0.5 GiB/B x 1.10 (Q4 + 10% parsed overhead), ESTIMADO",
265
+ "theoretical_tok_s_conventional_ram": 47.431,
266
+ "ssd_mne_upper_tok_s": [
267
+ {
268
+ "ssd_cache_miss": 0.1,
269
+ "cache_label": "90%",
270
+ "upper Tok_s": 7.905
271
+ },
272
+ {
273
+ "ssd_cache_miss": 0.05,
274
+ "cache_label": "95%",
275
+ "upper Tok_s": 15.81
276
+ },
277
+ {
278
+ "ssd_cache_miss": 0.01,
279
+ "cache_label": "99%",
280
+ "upper Tok_s": 79.051
281
+ },
282
+ {
283
+ "ssd_cache_miss": 0.005,
284
+ "cache_label": "99.5%",
285
+ "upper Tok_s": 158.103
286
+ },
287
+ {
288
+ "ssd_cache_miss": 0.001,
289
+ "cache_label": "99.9%",
290
+ "upper Tok_s": 790.514
291
+ }
292
+ ],
293
+ "kv_cache_bytes_est": 110100480,
294
+ "kv_cache_status": "ESTIMADO no config auditable",
295
+ "conventional_resident_total_gb_est": 3.022494,
296
+ "mne_resident_total_gb_est": 5.435977,
297
+ "ssd_required_bytes": 3009448896,
298
+ "feasible_conventional": "VIABLE con margen de memoria (límite superior de presión, no medición)",
299
+ "feasible_mne": "ESCENARIO VIRTUOSO DE RAM (cache capado; no prueba de throughput)",
300
+ "throughput_status": "ESTIMADO (techo de ancho de banda, no benchmark)",
301
+ "throughput_note": "No real benchmark is stored for MNE; upper bounds only."
302
+ },
303
+ {
304
+ "model": "Nova Core",
305
+ "repository": "ModernoTech/Nova-Core",
306
+ "hardware": "MacBook Air 2014",
307
+ "hardware_id": "air-2014",
308
+ "bandwidth_gbps": 25.6,
309
+ "ram_gb": 4.0,
310
+ "context_tokens": 8192,
311
+ "context_label": "8K",
312
+ "context_supported_or_validated": true,
313
+ "max_position_documented": 131072,
314
+ "artifact_size_bytes": 3009448896,
315
+ "artifact_status": "PUBLICADO",
316
+ "weight_basis": "artifact size on Hugging Face",
317
+ "active_payload_bytes_for_memory_bound": 2716566815,
318
+ "active_payload_basis": "active param count x 0.5 GiB/B x 1.10 (Q4 + 10% parsed overhead), ESTIMADO",
319
+ "theoretical_tok_s_conventional_ram": 10.119,
320
+ "ssd_mne_upper_tok_s": [
321
+ {
322
+ "ssd_cache_miss": 0.1,
323
+ "cache_label": "90%",
324
+ "upper Tok_s": 7.905
325
+ },
326
+ {
327
+ "ssd_cache_miss": 0.05,
328
+ "cache_label": "95%",
329
+ "upper Tok_s": 15.81
330
+ },
331
+ {
332
+ "ssd_cache_miss": 0.01,
333
+ "cache_label": "99%",
334
+ "upper Tok_s": 79.051
335
+ },
336
+ {
337
+ "ssd_cache_miss": 0.005,
338
+ "cache_label": "99.5%",
339
+ "upper Tok_s": 158.103
340
+ },
341
+ {
342
+ "ssd_cache_miss": 0.001,
343
+ "cache_label": "99.9%",
344
+ "upper Tok_s": 790.514
345
+ }
346
+ ],
347
+ "kv_cache_bytes_est": 220200960,
348
+ "kv_cache_status": "ESTIMADO no config auditable",
349
+ "conventional_resident_total_gb_est": 3.125033,
350
+ "mne_resident_total_gb_est": 5.538516,
351
+ "ssd_required_bytes": 3009448896,
352
+ "feasible_conventional": "NO VIABLE en RAM para pesos completos (límite teórico)",
353
+ "feasible_mne": "ESCENARIO NO FAVORABLE EN RAM (estimado)",
354
+ "throughput_status": "ESTIMADO (techo de ancho de banda, no benchmark)",
355
+ "throughput_note": "No real benchmark is stored for MNE; upper bounds only."
356
+ },
357
+ {
358
+ "model": "Nova Core",
359
+ "repository": "ModernoTech/Nova-Core",
360
+ "hardware": "Lenovo Yoga 720",
361
+ "hardware_id": "yoga-720",
362
+ "bandwidth_gbps": 34.1,
363
+ "ram_gb": 8.0,
364
+ "context_tokens": 8192,
365
+ "context_label": "8K",
366
+ "context_supported_or_validated": true,
367
+ "max_position_documented": 131072,
368
+ "artifact_size_bytes": 3009448896,
369
+ "artifact_status": "PUBLICADO",
370
+ "weight_basis": "artifact size on Hugging Face",
371
+ "active_payload_bytes_for_memory_bound": 2716566815,
372
+ "active_payload_basis": "active param count x 0.5 GiB/B x 1.10 (Q4 + 10% parsed overhead), ESTIMADO",
373
+ "theoretical_tok_s_conventional_ram": 13.478,
374
+ "ssd_mne_upper_tok_s": [
375
+ {
376
+ "ssd_cache_miss": 0.1,
377
+ "cache_label": "90%",
378
+ "upper Tok_s": 7.905
379
+ },
380
+ {
381
+ "ssd_cache_miss": 0.05,
382
+ "cache_label": "95%",
383
+ "upper Tok_s": 15.81
384
+ },
385
+ {
386
+ "ssd_cache_miss": 0.01,
387
+ "cache_label": "99%",
388
+ "upper Tok_s": 79.051
389
+ },
390
+ {
391
+ "ssd_cache_miss": 0.005,
392
+ "cache_label": "99.5%",
393
+ "upper Tok_s": 158.103
394
+ },
395
+ {
396
+ "ssd_cache_miss": 0.001,
397
+ "cache_label": "99.9%",
398
+ "upper Tok_s": 790.514
399
+ }
400
+ ],
401
+ "kv_cache_bytes_est": 220200960,
402
+ "kv_cache_status": "ESTIMADO no config auditable",
403
+ "conventional_resident_total_gb_est": 3.125033,
404
+ "mne_resident_total_gb_est": 5.538516,
405
+ "ssd_required_bytes": 3009448896,
406
+ "feasible_conventional": "VIABLE con margen de memoria (límite superior de presión, no medición)",
407
+ "feasible_mne": "ESCENARIO VIRTUOSO DE RAM (cache capado; no prueba de throughput)",
408
+ "throughput_status": "ESTIMADO (techo de ancho de banda, no benchmark)",
409
+ "throughput_note": "No real benchmark is stored for MNE; upper bounds only."
410
+ },
411
+ {
412
+ "model": "Nova Core",
413
+ "repository": "ModernoTech/Nova-Core",
414
+ "hardware": "MacBook Neo A18 Pro",
415
+ "hardware_id": "neo",
416
+ "bandwidth_gbps": 60.0,
417
+ "ram_gb": 8.0,
418
+ "context_tokens": 8192,
419
+ "context_label": "8K",
420
+ "context_supported_or_validated": true,
421
+ "max_position_documented": 131072,
422
+ "artifact_size_bytes": 3009448896,
423
+ "artifact_status": "PUBLICADO",
424
+ "weight_basis": "artifact size on Hugging Face",
425
+ "active_payload_bytes_for_memory_bound": 2716566815,
426
+ "active_payload_basis": "active param count x 0.5 GiB/B x 1.10 (Q4 + 10% parsed overhead), ESTIMADO",
427
+ "theoretical_tok_s_conventional_ram": 23.715,
428
+ "ssd_mne_upper_tok_s": [
429
+ {
430
+ "ssd_cache_miss": 0.1,
431
+ "cache_label": "90%",
432
+ "upper Tok_s": 7.905
433
+ },
434
+ {
435
+ "ssd_cache_miss": 0.05,
436
+ "cache_label": "95%",
437
+ "upper Tok_s": 15.81
438
+ },
439
+ {
440
+ "ssd_cache_miss": 0.01,
441
+ "cache_label": "99%",
442
+ "upper Tok_s": 79.051
443
+ },
444
+ {
445
+ "ssd_cache_miss": 0.005,
446
+ "cache_label": "99.5%",
447
+ "upper Tok_s": 158.103
448
+ },
449
+ {
450
+ "ssd_cache_miss": 0.001,
451
+ "cache_label": "99.9%",
452
+ "upper Tok_s": 790.514
453
+ }
454
+ ],
455
+ "kv_cache_bytes_est": 220200960,
456
+ "kv_cache_status": "ESTIMADO no config auditable",
457
+ "conventional_resident_total_gb_est": 3.125033,
458
+ "mne_resident_total_gb_est": 5.538516,
459
+ "ssd_required_bytes": 3009448896,
460
+ "feasible_conventional": "VIABLE con margen de memoria (límite superior de presión, no medición)",
461
+ "feasible_mne": "ESCENARIO VIRTUOSO DE RAM (cache capado; no prueba de throughput)",
462
+ "throughput_status": "ESTIMADO (techo de ancho de banda, no benchmark)",
463
+ "throughput_note": "No real benchmark is stored for MNE; upper bounds only."
464
+ },
465
+ {
466
+ "model": "Nova Core",
467
+ "repository": "ModernoTech/Nova-Core",
468
+ "hardware": "MacBook Air M4",
469
+ "hardware_id": "air-m4",
470
+ "bandwidth_gbps": 120.0,
471
+ "ram_gb": 16.0,
472
+ "context_tokens": 8192,
473
+ "context_label": "8K",
474
+ "context_supported_or_validated": true,
475
+ "max_position_documented": 131072,
476
+ "artifact_size_bytes": 3009448896,
477
+ "artifact_status": "PUBLICADO",
478
+ "weight_basis": "artifact size on Hugging Face",
479
+ "active_payload_bytes_for_memory_bound": 2716566815,
480
+ "active_payload_basis": "active param count x 0.5 GiB/B x 1.10 (Q4 + 10% parsed overhead), ESTIMADO",
481
+ "theoretical_tok_s_conventional_ram": 47.431,
482
+ "ssd_mne_upper_tok_s": [
483
+ {
484
+ "ssd_cache_miss": 0.1,
485
+ "cache_label": "90%",
486
+ "upper Tok_s": 7.905
487
+ },
488
+ {
489
+ "ssd_cache_miss": 0.05,
490
+ "cache_label": "95%",
491
+ "upper Tok_s": 15.81
492
+ },
493
+ {
494
+ "ssd_cache_miss": 0.01,
495
+ "cache_label": "99%",
496
+ "upper Tok_s": 79.051
497
+ },
498
+ {
499
+ "ssd_cache_miss": 0.005,
500
+ "cache_label": "99.5%",
501
+ "upper Tok_s": 158.103
502
+ },
503
+ {
504
+ "ssd_cache_miss": 0.001,
505
+ "cache_label": "99.9%",
506
+ "upper Tok_s": 790.514
507
+ }
508
+ ],
509
+ "kv_cache_bytes_est": 220200960,
510
+ "kv_cache_status": "ESTIMADO no config auditable",
511
+ "conventional_resident_total_gb_est": 3.125033,
512
+ "mne_resident_total_gb_est": 5.538516,
513
+ "ssd_required_bytes": 3009448896,
514
+ "feasible_conventional": "VIABLE con margen de memoria (límite superior de presión, no medición)",
515
+ "feasible_mne": "ESCENARIO VIRTUOSO DE RAM (cache capado; no prueba de throughput)",
516
+ "throughput_status": "ESTIMADO (techo de ancho de banda, no benchmark)",
517
+ "throughput_note": "No real benchmark is stored for MNE; upper bounds only."
518
+ },
519
+ {
520
+ "model": "Nova Core",
521
+ "repository": "ModernoTech/Nova-Core",
522
+ "hardware": "MacBook Air 2014",
523
+ "hardware_id": "air-2014",
524
+ "bandwidth_gbps": 25.6,
525
+ "ram_gb": 4.0,
526
+ "context_tokens": 65536,
527
+ "context_label": "64K",
528
+ "context_supported_or_validated": true,
529
+ "max_position_documented": 131072,
530
+ "artifact_size_bytes": 3009448896,
531
+ "artifact_status": "PUBLICADO",
532
+ "weight_basis": "artifact size on Hugging Face",
533
+ "active_payload_bytes_for_memory_bound": 2716566815,
534
+ "active_payload_basis": "active param count x 0.5 GiB/B x 1.10 (Q4 + 10% parsed overhead), ESTIMADO",
535
+ "theoretical_tok_s_conventional_ram": 10.119,
536
+ "ssd_mne_upper_tok_s": [
537
+ {
538
+ "ssd_cache_miss": 0.1,
539
+ "cache_label": "90%",
540
+ "upper Tok_s": 7.905
541
+ },
542
+ {
543
+ "ssd_cache_miss": 0.05,
544
+ "cache_label": "95%",
545
+ "upper Tok_s": 15.81
546
+ },
547
+ {
548
+ "ssd_cache_miss": 0.01,
549
+ "cache_label": "99%",
550
+ "upper Tok_s": 79.051
551
+ },
552
+ {
553
+ "ssd_cache_miss": 0.005,
554
+ "cache_label": "99.5%",
555
+ "upper Tok_s": 158.103
556
+ },
557
+ {
558
+ "ssd_cache_miss": 0.001,
559
+ "cache_label": "99.9%",
560
+ "upper Tok_s": 790.514
561
+ }
562
+ ],
563
+ "kv_cache_bytes_est": 1761607680,
564
+ "kv_cache_status": "ESTIMADO no config auditable",
565
+ "conventional_resident_total_gb_est": 4.56058,
566
+ "mne_resident_total_gb_est": 6.974062,
567
+ "ssd_required_bytes": 3009448896,
568
+ "feasible_conventional": "NO VIABLE en RAM para pesos completos (límite teórico)",
569
+ "feasible_mne": "ESCENARIO NO FAVORABLE EN RAM (estimado)",
570
+ "throughput_status": "ESTIMADO (techo de ancho de banda, no benchmark)",
571
+ "throughput_note": "No real benchmark is stored for MNE; upper bounds only."
572
+ },
573
+ {
574
+ "model": "Nova Core",
575
+ "repository": "ModernoTech/Nova-Core",
576
+ "hardware": "Lenovo Yoga 720",
577
+ "hardware_id": "yoga-720",
578
+ "bandwidth_gbps": 34.1,
579
+ "ram_gb": 8.0,
580
+ "context_tokens": 65536,
581
+ "context_label": "64K",
582
+ "context_supported_or_validated": true,
583
+ "max_position_documented": 131072,
584
+ "artifact_size_bytes": 3009448896,
585
+ "artifact_status": "PUBLICADO",
586
+ "weight_basis": "artifact size on Hugging Face",
587
+ "active_payload_bytes_for_memory_bound": 2716566815,
588
+ "active_payload_basis": "active param count x 0.5 GiB/B x 1.10 (Q4 + 10% parsed overhead), ESTIMADO",
589
+ "theoretical_tok_s_conventional_ram": 13.478,
590
+ "ssd_mne_upper_tok_s": [
591
+ {
592
+ "ssd_cache_miss": 0.1,
593
+ "cache_label": "90%",
594
+ "upper Tok_s": 7.905
595
+ },
596
+ {
597
+ "ssd_cache_miss": 0.05,
598
+ "cache_label": "95%",
599
+ "upper Tok_s": 15.81
600
+ },
601
+ {
602
+ "ssd_cache_miss": 0.01,
603
+ "cache_label": "99%",
604
+ "upper Tok_s": 79.051
605
+ },
606
+ {
607
+ "ssd_cache_miss": 0.005,
608
+ "cache_label": "99.5%",
609
+ "upper Tok_s": 158.103
610
+ },
611
+ {
612
+ "ssd_cache_miss": 0.001,
613
+ "cache_label": "99.9%",
614
+ "upper Tok_s": 790.514
615
+ }
616
+ ],
617
+ "kv_cache_bytes_est": 1761607680,
618
+ "kv_cache_status": "ESTIMADO no config auditable",
619
+ "conventional_resident_total_gb_est": 4.56058,
620
+ "mne_resident_total_gb_est": 6.974062,
621
+ "ssd_required_bytes": 3009448896,
622
+ "feasible_conventional": "VIABLE con margen de memoria (límite superior de presión, no medición)",
623
+ "feasible_mne": "ESCENARIO VIRTUOSO DE RAM (cache capado; no prueba de throughput)",
624
+ "throughput_status": "ESTIMADO (techo de ancho de banda, no benchmark)",
625
+ "throughput_note": "No real benchmark is stored for MNE; upper bounds only."
626
+ },
627
+ {
628
+ "model": "Nova Core",
629
+ "repository": "ModernoTech/Nova-Core",
630
+ "hardware": "MacBook Neo A18 Pro",
631
+ "hardware_id": "neo",
632
+ "bandwidth_gbps": 60.0,
633
+ "ram_gb": 8.0,
634
+ "context_tokens": 65536,
635
+ "context_label": "64K",
636
+ "context_supported_or_validated": true,
637
+ "max_position_documented": 131072,
638
+ "artifact_size_bytes": 3009448896,
639
+ "artifact_status": "PUBLICADO",
640
+ "weight_basis": "artifact size on Hugging Face",
641
+ "active_payload_bytes_for_memory_bound": 2716566815,
642
+ "active_payload_basis": "active param count x 0.5 GiB/B x 1.10 (Q4 + 10% parsed overhead), ESTIMADO",
643
+ "theoretical_tok_s_conventional_ram": 23.715,
644
+ "ssd_mne_upper_tok_s": [
645
+ {
646
+ "ssd_cache_miss": 0.1,
647
+ "cache_label": "90%",
648
+ "upper Tok_s": 7.905
649
+ },
650
+ {
651
+ "ssd_cache_miss": 0.05,
652
+ "cache_label": "95%",
653
+ "upper Tok_s": 15.81
654
+ },
655
+ {
656
+ "ssd_cache_miss": 0.01,
657
+ "cache_label": "99%",
658
+ "upper Tok_s": 79.051
659
+ },
660
+ {
661
+ "ssd_cache_miss": 0.005,
662
+ "cache_label": "99.5%",
663
+ "upper Tok_s": 158.103
664
+ },
665
+ {
666
+ "ssd_cache_miss": 0.001,
667
+ "cache_label": "99.9%",
668
+ "upper Tok_s": 790.514
669
+ }
670
+ ],
671
+ "kv_cache_bytes_est": 1761607680,
672
+ "kv_cache_status": "ESTIMADO no config auditable",
673
+ "conventional_resident_total_gb_est": 4.56058,
674
+ "mne_resident_total_gb_est": 6.974062,
675
+ "ssd_required_bytes": 3009448896,
676
+ "feasible_conventional": "VIABLE con margen de memoria (límite superior de presión, no medición)",
677
+ "feasible_mne": "ESCENARIO VIRTUOSO DE RAM (cache capado; no prueba de throughput)",
678
+ "throughput_status": "ESTIMADO (techo de ancho de banda, no benchmark)",
679
+ "throughput_note": "No real benchmark is stored for MNE; upper bounds only."
680
+ },
681
+ {
682
+ "model": "Nova Core",
683
+ "repository": "ModernoTech/Nova-Core",
684
+ "hardware": "MacBook Air M4",
685
+ "hardware_id": "air-m4",
686
+ "bandwidth_gbps": 120.0,
687
+ "ram_gb": 16.0,
688
+ "context_tokens": 65536,
689
+ "context_label": "64K",
690
+ "context_supported_or_validated": true,
691
+ "max_position_documented": 131072,
692
+ "artifact_size_bytes": 3009448896,
693
+ "artifact_status": "PUBLICADO",
694
+ "weight_basis": "artifact size on Hugging Face",
695
+ "active_payload_bytes_for_memory_bound": 2716566815,
696
+ "active_payload_basis": "active param count x 0.5 GiB/B x 1.10 (Q4 + 10% parsed overhead), ESTIMADO",
697
+ "theoretical_tok_s_conventional_ram": 47.431,
698
+ "ssd_mne_upper_tok_s": [
699
+ {
700
+ "ssd_cache_miss": 0.1,
701
+ "cache_label": "90%",
702
+ "upper Tok_s": 7.905
703
+ },
704
+ {
705
+ "ssd_cache_miss": 0.05,
706
+ "cache_label": "95%",
707
+ "upper Tok_s": 15.81
708
+ },
709
+ {
710
+ "ssd_cache_miss": 0.01,
711
+ "cache_label": "99%",
712
+ "upper Tok_s": 79.051
713
+ },
714
+ {
715
+ "ssd_cache_miss": 0.005,
716
+ "cache_label": "99.5%",
717
+ "upper Tok_s": 158.103
718
+ },
719
+ {
720
+ "ssd_cache_miss": 0.001,
721
+ "cache_label": "99.9%",
722
+ "upper Tok_s": 790.514
723
+ }
724
+ ],
725
+ "kv_cache_bytes_est": 1761607680,
726
+ "kv_cache_status": "ESTIMADO no config auditable",
727
+ "conventional_resident_total_gb_est": 4.56058,
728
+ "mne_resident_total_gb_est": 6.974062,
729
+ "ssd_required_bytes": 3009448896,
730
+ "feasible_conventional": "VIABLE con margen de memoria (límite superior de presión, no medición)",
731
+ "feasible_mne": "ESCENARIO VIRTUOSO DE RAM (cache capado; no prueba de throughput)",
732
+ "throughput_status": "ESTIMADO (techo de ancho de banda, no benchmark)",
733
+ "throughput_note": "No real benchmark is stored for MNE; upper bounds only."
734
+ },
735
+ {
736
+ "model": "Nova Core",
737
+ "repository": "ModernoTech/Nova-Core",
738
+ "hardware": "MacBook Air 2014",
739
+ "hardware_id": "air-2014",
740
+ "bandwidth_gbps": 25.6,
741
+ "ram_gb": 4.0,
742
+ "context_tokens": 131072,
743
+ "context_label": "128K",
744
+ "context_supported_or_validated": true,
745
+ "max_position_documented": 131072,
746
+ "artifact_size_bytes": 3009448896,
747
+ "artifact_status": "PUBLICADO",
748
+ "weight_basis": "artifact size on Hugging Face",
749
+ "active_payload_bytes_for_memory_bound": 2716566815,
750
+ "active_payload_basis": "active param count x 0.5 GiB/B x 1.10 (Q4 + 10% parsed overhead), ESTIMADO",
751
+ "theoretical_tok_s_conventional_ram": 10.119,
752
+ "ssd_mne_upper_tok_s": [
753
+ {
754
+ "ssd_cache_miss": 0.1,
755
+ "cache_label": "90%",
756
+ "upper Tok_s": 7.905
757
+ },
758
+ {
759
+ "ssd_cache_miss": 0.05,
760
+ "cache_label": "95%",
761
+ "upper Tok_s": 15.81
762
+ },
763
+ {
764
+ "ssd_cache_miss": 0.01,
765
+ "cache_label": "99%",
766
+ "upper Tok_s": 79.051
767
+ },
768
+ {
769
+ "ssd_cache_miss": 0.005,
770
+ "cache_label": "99.5%",
771
+ "upper Tok_s": 158.103
772
+ },
773
+ {
774
+ "ssd_cache_miss": 0.001,
775
+ "cache_label": "99.9%",
776
+ "upper Tok_s": 790.514
777
+ }
778
+ ],
779
+ "kv_cache_bytes_est": 3523215360,
780
+ "kv_cache_status": "ESTIMADO no config auditable",
781
+ "conventional_resident_total_gb_est": 6.201205,
782
+ "mne_resident_total_gb_est": 8.614687,
783
+ "ssd_required_bytes": 3009448896,
784
+ "feasible_conventional": "NO VIABLE en RAM para pesos completos (límite teórico)",
785
+ "feasible_mne": "ESCENARIO NO FAVORABLE EN RAM (estimado)",
786
+ "throughput_status": "ESTIMADO (techo de ancho de banda, no benchmark)",
787
+ "throughput_note": "No real benchmark is stored for MNE; upper bounds only."
788
+ },
789
+ {
790
+ "model": "Nova Core",
791
+ "repository": "ModernoTech/Nova-Core",
792
+ "hardware": "Lenovo Yoga 720",
793
+ "hardware_id": "yoga-720",
794
+ "bandwidth_gbps": 34.1,
795
+ "ram_gb": 8.0,
796
+ "context_tokens": 131072,
797
+ "context_label": "128K",
798
+ "context_supported_or_validated": true,
799
+ "max_position_documented": 131072,
800
+ "artifact_size_bytes": 3009448896,
801
+ "artifact_status": "PUBLICADO",
802
+ "weight_basis": "artifact size on Hugging Face",
803
+ "active_payload_bytes_for_memory_bound": 2716566815,
804
+ "active_payload_basis": "active param count x 0.5 GiB/B x 1.10 (Q4 + 10% parsed overhead), ESTIMADO",
805
+ "theoretical_tok_s_conventional_ram": 13.478,
806
+ "ssd_mne_upper_tok_s": [
807
+ {
808
+ "ssd_cache_miss": 0.1,
809
+ "cache_label": "90%",
810
+ "upper Tok_s": 7.905
811
+ },
812
+ {
813
+ "ssd_cache_miss": 0.05,
814
+ "cache_label": "95%",
815
+ "upper Tok_s": 15.81
816
+ },
817
+ {
818
+ "ssd_cache_miss": 0.01,
819
+ "cache_label": "99%",
820
+ "upper Tok_s": 79.051
821
+ },
822
+ {
823
+ "ssd_cache_miss": 0.005,
824
+ "cache_label": "99.5%",
825
+ "upper Tok_s": 158.103
826
+ },
827
+ {
828
+ "ssd_cache_miss": 0.001,
829
+ "cache_label": "99.9%",
830
+ "upper Tok_s": 790.514
831
+ }
832
+ ],
833
+ "kv_cache_bytes_est": 3523215360,
834
+ "kv_cache_status": "ESTIMADO no config auditable",
835
+ "conventional_resident_total_gb_est": 6.201205,
836
+ "mne_resident_total_gb_est": 8.614687,
837
+ "ssd_required_bytes": 3009448896,
838
+ "feasible_conventional": "VIABLE con alta presión de memoria (estimado)",
839
+ "feasible_mne": "ESCENARIO NO FAVORABLE EN RAM (estimado)",
840
+ "throughput_status": "ESTIMADO (techo de ancho de banda, no benchmark)",
841
+ "throughput_note": "No real benchmark is stored for MNE; upper bounds only."
842
+ },
843
+ {
844
+ "model": "Nova Core",
845
+ "repository": "ModernoTech/Nova-Core",
846
+ "hardware": "MacBook Neo A18 Pro",
847
+ "hardware_id": "neo",
848
+ "bandwidth_gbps": 60.0,
849
+ "ram_gb": 8.0,
850
+ "context_tokens": 131072,
851
+ "context_label": "128K",
852
+ "context_supported_or_validated": true,
853
+ "max_position_documented": 131072,
854
+ "artifact_size_bytes": 3009448896,
855
+ "artifact_status": "PUBLICADO",
856
+ "weight_basis": "artifact size on Hugging Face",
857
+ "active_payload_bytes_for_memory_bound": 2716566815,
858
+ "active_payload_basis": "active param count x 0.5 GiB/B x 1.10 (Q4 + 10% parsed overhead), ESTIMADO",
859
+ "theoretical_tok_s_conventional_ram": 23.715,
860
+ "ssd_mne_upper_tok_s": [
861
+ {
862
+ "ssd_cache_miss": 0.1,
863
+ "cache_label": "90%",
864
+ "upper Tok_s": 7.905
865
+ },
866
+ {
867
+ "ssd_cache_miss": 0.05,
868
+ "cache_label": "95%",
869
+ "upper Tok_s": 15.81
870
+ },
871
+ {
872
+ "ssd_cache_miss": 0.01,
873
+ "cache_label": "99%",
874
+ "upper Tok_s": 79.051
875
+ },
876
+ {
877
+ "ssd_cache_miss": 0.005,
878
+ "cache_label": "99.5%",
879
+ "upper Tok_s": 158.103
880
+ },
881
+ {
882
+ "ssd_cache_miss": 0.001,
883
+ "cache_label": "99.9%",
884
+ "upper Tok_s": 790.514
885
+ }
886
+ ],
887
+ "kv_cache_bytes_est": 3523215360,
888
+ "kv_cache_status": "ESTIMADO no config auditable",
889
+ "conventional_resident_total_gb_est": 6.201205,
890
+ "mne_resident_total_gb_est": 8.614687,
891
+ "ssd_required_bytes": 3009448896,
892
+ "feasible_conventional": "VIABLE con alta presión de memoria (estimado)",
893
+ "feasible_mne": "ESCENARIO NO FAVORABLE EN RAM (estimado)",
894
+ "throughput_status": "ESTIMADO (techo de ancho de banda, no benchmark)",
895
+ "throughput_note": "No real benchmark is stored for MNE; upper bounds only."
896
+ },
897
+ {
898
+ "model": "Nova Core",
899
+ "repository": "ModernoTech/Nova-Core",
900
+ "hardware": "MacBook Air M4",
901
+ "hardware_id": "air-m4",
902
+ "bandwidth_gbps": 120.0,
903
+ "ram_gb": 16.0,
904
+ "context_tokens": 131072,
905
+ "context_label": "128K",
906
+ "context_supported_or_validated": true,
907
+ "max_position_documented": 131072,
908
+ "artifact_size_bytes": 3009448896,
909
+ "artifact_status": "PUBLICADO",
910
+ "weight_basis": "artifact size on Hugging Face",
911
+ "active_payload_bytes_for_memory_bound": 2716566815,
912
+ "active_payload_basis": "active param count x 0.5 GiB/B x 1.10 (Q4 + 10% parsed overhead), ESTIMADO",
913
+ "theoretical_tok_s_conventional_ram": 47.431,
914
+ "ssd_mne_upper_tok_s": [
915
+ {
916
+ "ssd_cache_miss": 0.1,
917
+ "cache_label": "90%",
918
+ "upper Tok_s": 7.905
919
+ },
920
+ {
921
+ "ssd_cache_miss": 0.05,
922
+ "cache_label": "95%",
923
+ "upper Tok_s": 15.81
924
+ },
925
+ {
926
+ "ssd_cache_miss": 0.01,
927
+ "cache_label": "99%",
928
+ "upper Tok_s": 79.051
929
+ },
930
+ {
931
+ "ssd_cache_miss": 0.005,
932
+ "cache_label": "99.5%",
933
+ "upper Tok_s": 158.103
934
+ },
935
+ {
936
+ "ssd_cache_miss": 0.001,
937
+ "cache_label": "99.9%",
938
+ "upper Tok_s": 790.514
939
+ }
940
+ ],
941
+ "kv_cache_bytes_est": 3523215360,
942
+ "kv_cache_status": "ESTIMADO no config auditable",
943
+ "conventional_resident_total_gb_est": 6.201205,
944
+ "mne_resident_total_gb_est": 8.614687,
945
+ "ssd_required_bytes": 3009448896,
946
+ "feasible_conventional": "VIABLE con margen de memoria (límite superior de presión, no medición)",
947
+ "feasible_mne": "ESCENARIO VIRTUOSO DE RAM (cache capado; no prueba de throughput)",
948
+ "throughput_status": "ESTIMADO (techo de ancho de banda, no benchmark)",
949
+ "throughput_note": "No real benchmark is stored for MNE; upper bounds only."
950
+ }
951
+ ],
952
+ "charts": [
953
+ "tokens-per-second.png",
954
+ "ram-vs-mne.png",
955
+ "context-performance.png",
956
+ "memory-breakdown.png",
957
+ "model-size-vs-ram.png",
958
+ "cache-hit-analysis.png",
959
+ "prefill-vs-decode.png"
960
+ ]
961
+ }
performance/prefill-vs-decode.png ADDED
performance/ram-vs-mne.png ADDED
performance/simulation-results.csv ADDED
@@ -0,0 +1,17 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ repository,model,hardware,ram_gb,ram_bandwidth_gbps,context,context_supported,artifact_size_bytes,active_weight_payload_bytes,token_upper_bound_conventional_tok_s,ssd_90_miss_upper,ssd_95_miss_upper,ssd_99_miss_upper,ssd_99_5_miss_upper,ssd_99_9_miss_upper,kv_cache_bytes_estimated,conventional_resident_gb_est,mne_resident_gb_est,throughput_status,ram_feasible,mne_feasible,notes
2
+ ModernoTech/Nova-Core,Nova Core,MacBook Air 2014,4.0,25.6,4K,true,3009448896,2716566815,10.119,7.905,15.81,79.051,158.103,790.514,110100480,3.022494,5.435977,"ESTIMADO (techo de ancho de banda, no benchmark)",NO VIABLE en RAM para pesos completos (límite teórico),ESCENARIO NO FAVORABLE EN RAM (estimado),none
3
+ ModernoTech/Nova-Core,Nova Core,Lenovo Yoga 720,8.0,34.1,4K,true,3009448896,2716566815,13.478,7.905,15.81,79.051,158.103,790.514,110100480,3.022494,5.435977,"ESTIMADO (techo de ancho de banda, no benchmark)","VIABLE con margen de memoria (límite superior de presión, no medición)",ESCENARIO VIRTUOSO DE RAM (cache capado; no prueba de throughput),none
4
+ ModernoTech/Nova-Core,Nova Core,MacBook Neo A18 Pro,8.0,60.0,4K,true,3009448896,2716566815,23.715,7.905,15.81,79.051,158.103,790.514,110100480,3.022494,5.435977,"ESTIMADO (techo de ancho de banda, no benchmark)","VIABLE con margen de memoria (límite superior de presión, no medición)",ESCENARIO VIRTUOSO DE RAM (cache capado; no prueba de throughput),none
5
+ ModernoTech/Nova-Core,Nova Core,MacBook Air M4,16.0,120.0,4K,true,3009448896,2716566815,47.431,7.905,15.81,79.051,158.103,790.514,110100480,3.022494,5.435977,"ESTIMADO (techo de ancho de banda, no benchmark)","VIABLE con margen de memoria (límite superior de presión, no medición)",ESCENARIO VIRTUOSO DE RAM (cache capado; no prueba de throughput),none
6
+ ModernoTech/Nova-Core,Nova Core,MacBook Air 2014,4.0,25.6,8K,true,3009448896,2716566815,10.119,7.905,15.81,79.051,158.103,790.514,220200960,3.125033,5.538516,"ESTIMADO (techo de ancho de banda, no benchmark)",NO VIABLE en RAM para pesos completos (límite teórico),ESCENARIO NO FAVORABLE EN RAM (estimado),none
7
+ ModernoTech/Nova-Core,Nova Core,Lenovo Yoga 720,8.0,34.1,8K,true,3009448896,2716566815,13.478,7.905,15.81,79.051,158.103,790.514,220200960,3.125033,5.538516,"ESTIMADO (techo de ancho de banda, no benchmark)","VIABLE con margen de memoria (límite superior de presión, no medición)",ESCENARIO VIRTUOSO DE RAM (cache capado; no prueba de throughput),none
8
+ ModernoTech/Nova-Core,Nova Core,MacBook Neo A18 Pro,8.0,60.0,8K,true,3009448896,2716566815,23.715,7.905,15.81,79.051,158.103,790.514,220200960,3.125033,5.538516,"ESTIMADO (techo de ancho de banda, no benchmark)","VIABLE con margen de memoria (límite superior de presión, no medición)",ESCENARIO VIRTUOSO DE RAM (cache capado; no prueba de throughput),none
9
+ ModernoTech/Nova-Core,Nova Core,MacBook Air M4,16.0,120.0,8K,true,3009448896,2716566815,47.431,7.905,15.81,79.051,158.103,790.514,220200960,3.125033,5.538516,"ESTIMADO (techo de ancho de banda, no benchmark)","VIABLE con margen de memoria (límite superior de presión, no medición)",ESCENARIO VIRTUOSO DE RAM (cache capado; no prueba de throughput),none
10
+ ModernoTech/Nova-Core,Nova Core,MacBook Air 2014,4.0,25.6,64K,true,3009448896,2716566815,10.119,7.905,15.81,79.051,158.103,790.514,1761607680,4.56058,6.974062,"ESTIMADO (techo de ancho de banda, no benchmark)",NO VIABLE en RAM para pesos completos (límite teórico),ESCENARIO NO FAVORABLE EN RAM (estimado),none
11
+ ModernoTech/Nova-Core,Nova Core,Lenovo Yoga 720,8.0,34.1,64K,true,3009448896,2716566815,13.478,7.905,15.81,79.051,158.103,790.514,1761607680,4.56058,6.974062,"ESTIMADO (techo de ancho de banda, no benchmark)","VIABLE con margen de memoria (límite superior de presión, no medición)",ESCENARIO VIRTUOSO DE RAM (cache capado; no prueba de throughput),none
12
+ ModernoTech/Nova-Core,Nova Core,MacBook Neo A18 Pro,8.0,60.0,64K,true,3009448896,2716566815,23.715,7.905,15.81,79.051,158.103,790.514,1761607680,4.56058,6.974062,"ESTIMADO (techo de ancho de banda, no benchmark)","VIABLE con margen de memoria (límite superior de presión, no medición)",ESCENARIO VIRTUOSO DE RAM (cache capado; no prueba de throughput),none
13
+ ModernoTech/Nova-Core,Nova Core,MacBook Air M4,16.0,120.0,64K,true,3009448896,2716566815,47.431,7.905,15.81,79.051,158.103,790.514,1761607680,4.56058,6.974062,"ESTIMADO (techo de ancho de banda, no benchmark)","VIABLE con margen de memoria (límite superior de presión, no medición)",ESCENARIO VIRTUOSO DE RAM (cache capado; no prueba de throughput),none
14
+ ModernoTech/Nova-Core,Nova Core,MacBook Air 2014,4.0,25.6,128K,true,3009448896,2716566815,10.119,7.905,15.81,79.051,158.103,790.514,3523215360,6.201205,8.614687,"ESTIMADO (techo de ancho de banda, no benchmark)",NO VIABLE en RAM para pesos completos (límite teórico),ESCENARIO NO FAVORABLE EN RAM (estimado),none
15
+ ModernoTech/Nova-Core,Nova Core,Lenovo Yoga 720,8.0,34.1,128K,true,3009448896,2716566815,13.478,7.905,15.81,79.051,158.103,790.514,3523215360,6.201205,8.614687,"ESTIMADO (techo de ancho de banda, no benchmark)",VIABLE con alta presión de memoria (estimado),ESCENARIO NO FAVORABLE EN RAM (estimado),none
16
+ ModernoTech/Nova-Core,Nova Core,MacBook Neo A18 Pro,8.0,60.0,128K,true,3009448896,2716566815,23.715,7.905,15.81,79.051,158.103,790.514,3523215360,6.201205,8.614687,"ESTIMADO (techo de ancho de banda, no benchmark)",VIABLE con alta presión de memoria (estimado),ESCENARIO NO FAVORABLE EN RAM (estimado),none
17
+ ModernoTech/Nova-Core,Nova Core,MacBook Air M4,16.0,120.0,128K,true,3009448896,2716566815,47.431,7.905,15.81,79.051,158.103,790.514,3523215360,6.201205,8.614687,"ESTIMADO (techo de ancho de banda, no benchmark)","VIABLE con margen de memoria (límite superior de presión, no medición)",ESCENARIO VIRTUOSO DE RAM (cache capado; no prueba de throughput),none
performance/tokens-per-second.png ADDED