Evidence
Evidence explorer
Historical source evidence (legacy, read-only): what each test measures, which GPUs it covers and how fresh it is. Upgrade decisions use UOS workload indices only; these records are not decision inputs. Results from different sources are never averaged together.
| Test | Domain · workload | Metric | GPUs | Freshness | Evidence | Legacy admission |
|---|---|---|---|---|---|---|
| LocalScore · repo bd6ffe9 · pp16+tg1536 | Local AI inference | prompt tokens per second (tok/s) | 2 | Latest measurement Mar 19, 2025 | Standardized test — The same benchmark suite and version run on each GPU. | Supporting only |
| LocalScore · repo bd6ffe9 · pp16+tg1536 | Local AI inference | generation tokens per second (tok/s) | 2 | Latest measurement Mar 19, 2025 | Standardized test — The same benchmark suite and version run on each GPU. | Supporting only |
| LocalScore · repo bd6ffe9 · pp16+tg1536 | Local AI inference | ttft ms (ms) | 2 | Latest measurement Mar 19, 2025 | Standardized test — The same benchmark suite and version run on each GPU. | Supporting only |
| LocalScore · repo bd6ffe9 · pp64+tg1024 | Local AI inference | prompt tokens per second (tok/s) | 2 | Latest measurement Mar 19, 2025 | Standardized test — The same benchmark suite and version run on each GPU. | Supporting only |
| LocalScore · repo bd6ffe9 · pp64+tg1024 | Local AI inference | generation tokens per second (tok/s) | 2 | Latest measurement Mar 19, 2025 | Standardized test — The same benchmark suite and version run on each GPU. | Supporting only |
| LocalScore · repo bd6ffe9 · pp64+tg1024 | Local AI inference | ttft ms (ms) | 2 | Latest measurement Mar 19, 2025 | Standardized test — The same benchmark suite and version run on each GPU. | Supporting only |
| LocalScore · repo bd6ffe9 · pp384+tg1152 | Local AI inference | prompt tokens per second (tok/s) | 2 | Latest measurement Mar 19, 2025 | Standardized test — The same benchmark suite and version run on each GPU. | Supporting only |
| LocalScore · repo bd6ffe9 · pp384+tg1152 | Local AI inference | generation tokens per second (tok/s) | 2 | Latest measurement Mar 19, 2025 | Standardized test — The same benchmark suite and version run on each GPU. | Supporting only |
| LocalScore · repo bd6ffe9 · pp384+tg1152 | Local AI inference | ttft ms (ms) | 2 | Latest measurement Mar 19, 2025 | Standardized test — The same benchmark suite and version run on each GPU. | Supporting only |
| LocalScore · repo bd6ffe9 · pp1024+tg16 | Local AI inference | prompt tokens per second (tok/s) | 2 | Latest measurement Mar 19, 2025 | Standardized test — The same benchmark suite and version run on each GPU. | Supporting only |
| LocalScore · repo bd6ffe9 · pp1024+tg16 | Local AI inference | generation tokens per second (tok/s) | 2 | Latest measurement Mar 19, 2025 | Standardized test — The same benchmark suite and version run on each GPU. | Supporting only |
| LocalScore · repo bd6ffe9 · pp1024+tg16 | Local AI inference | ttft ms (ms) | 2 | Latest measurement Mar 19, 2025 | Standardized test — The same benchmark suite and version run on each GPU. | Supporting only |
| LocalScore · repo bd6ffe9 · pp1024+tg1024 | Local AI inference | prompt tokens per second (tok/s) | 2 | Latest measurement Mar 19, 2025 | Standardized test — The same benchmark suite and version run on each GPU. | Supporting only |
| LocalScore · repo bd6ffe9 · pp1024+tg1024 | Local AI inference | generation tokens per second (tok/s) | 2 | Latest measurement Mar 19, 2025 | Standardized test — The same benchmark suite and version run on each GPU. | Supporting only |
| LocalScore · repo bd6ffe9 · pp1024+tg1024 | Local AI inference | ttft ms (ms) | 2 | Latest measurement Mar 19, 2025 | Standardized test — The same benchmark suite and version run on each GPU. | Supporting only |
| LocalScore · repo bd6ffe9 · pp1280+tg3072 | Local AI inference | prompt tokens per second (tok/s) | 2 | Latest measurement Mar 19, 2025 | Standardized test — The same benchmark suite and version run on each GPU. | Supporting only |
| LocalScore · repo bd6ffe9 · pp1280+tg3072 | Local AI inference | generation tokens per second (tok/s) | 2 | Latest measurement Mar 19, 2025 | Standardized test — The same benchmark suite and version run on each GPU. | Supporting only |
| LocalScore · repo bd6ffe9 · pp1280+tg3072 | Local AI inference | ttft ms (ms) | 2 | Latest measurement Mar 19, 2025 | Standardized test — The same benchmark suite and version run on each GPU. | Supporting only |
| LocalScore · repo bd6ffe9 · pp2048+tg256 | Local AI inference | prompt tokens per second (tok/s) | 2 | Latest measurement Mar 19, 2025 | Standardized test — The same benchmark suite and version run on each GPU. | Supporting only |
| LocalScore · repo bd6ffe9 · pp2048+tg256 | Local AI inference | generation tokens per second (tok/s) | 2 | Latest measurement Mar 19, 2025 | Standardized test — The same benchmark suite and version run on each GPU. | Supporting only |
| LocalScore · repo bd6ffe9 · pp2048+tg256 | Local AI inference | ttft ms (ms) | 2 | Latest measurement Mar 19, 2025 | Standardized test — The same benchmark suite and version run on each GPU. | Supporting only |
| LocalScore · repo bd6ffe9 · pp2048+tg768 | Local AI inference | prompt tokens per second (tok/s) | 2 | Latest measurement Mar 19, 2025 | Standardized test — The same benchmark suite and version run on each GPU. | Supporting only |
| LocalScore · repo bd6ffe9 · pp2048+tg768 | Local AI inference | generation tokens per second (tok/s) | 2 | Latest measurement Mar 19, 2025 | Standardized test — The same benchmark suite and version run on each GPU. | Supporting only |
| LocalScore · repo bd6ffe9 · pp2048+tg768 | Local AI inference | ttft ms (ms) | 2 | Latest measurement Mar 19, 2025 | Standardized test — The same benchmark suite and version run on each GPU. | Supporting only |
| LocalScore · repo bd6ffe9 · pp4096+tg256 | Local AI inference | ttft ms (ms) | 2 | Latest measurement Mar 19, 2025 | Standardized test — The same benchmark suite and version run on each GPU. | Supporting only |
| LocalScore · repo bd6ffe9 · pp4096+tg256 | Local AI inference | prompt tokens per second (tok/s) | 2 | Latest measurement Mar 19, 2025 | Standardized test — The same benchmark suite and version run on each GPU. | Supporting only |
| LocalScore · repo bd6ffe9 · pp4096+tg256 | Local AI inference | generation tokens per second (tok/s) | 2 | Latest measurement Mar 19, 2025 | Standardized test — The same benchmark suite and version run on each GPU. | Supporting only |
| Q6_K_S · 8K · ROCm · llama-bench | Local AI inference | generation tokens per second (tok/s) | 1 | Latest measurement Sep 25, 2026 | Measured — Measured on the exact workload (model, settings, and runtime) being compared. | Met legacy decision rules |
| Q6_K_S · 8K · ROCm · llama-bench | Local AI inference | prompt tokens per second (tok/s) | 1 | Latest measurement Sep 25, 2026 | Measured — Measured on the exact workload (model, settings, and runtime) being compared. | Met legacy decision rules |
| Q6_K_S · 32K · ROCm · llama-bench | Local AI inference | generation tokens per second (tok/s) | 1 | Latest measurement Sep 25, 2026 | Measured — Measured on the exact workload (model, settings, and runtime) being compared. | Met legacy decision rules |
| Q6_K_S · 32K · ROCm · llama-bench | Local AI inference | prompt tokens per second (tok/s) | 1 | Latest measurement Sep 25, 2026 | Measured — Measured on the exact workload (model, settings, and runtime) being compared. | Met legacy decision rules |
| Q6_K_S · 64K · ROCm · llama-bench | Local AI inference | generation tokens per second (tok/s) | 1 | Latest measurement Sep 25, 2026 | Measured — Measured on the exact workload (model, settings, and runtime) being compared. | Met legacy decision rules |
| Q6_K_S · 64K · ROCm · llama-bench | Local AI inference | prompt tokens per second (tok/s) | 1 | Latest measurement Sep 25, 2026 | Measured — Measured on the exact workload (model, settings, and runtime) being compared. | Met legacy decision rules |
| Qwen3.8-27B-Q6_K_S.gguf · Q6_K_S · 8K · ROCm · llama-bench | Local AI inference | generation tokens per second (tok/s) | 1 | Latest measurement Sep 25, 2026 | Measured — Measured on the exact workload (model, settings, and runtime) being compared. | Supporting only |
| Qwen3.8-27B-Q6_K_S.gguf · Q6_K_S · 8K · ROCm · llama-bench | Local AI inference | prompt tokens per second (tok/s) | 1 | Latest measurement Sep 25, 2026 | Measured — Measured on the exact workload (model, settings, and runtime) being compared. | Supporting only |
| Qwen3.8-27B-Q6_K_S.gguf · Q6_K_S · 8K · ROCm · llama-server | Local AI inference | ttft ms (ms) | 1 | Latest measurement Sep 25, 2026 | Measured — Measured on the exact workload (model, settings, and runtime) being compared. | Supporting only |
| Qwen3.8-27B-Q6_K_S.gguf · Q6_K_S · 8K · ROCm · llama-server | Local AI inference | generation tokens per second (tok/s) | 1 | Latest measurement Sep 25, 2026 | Measured — Measured on the exact workload (model, settings, and runtime) being compared. | Supporting only |
| Qwen3.8-27B-Q6_K_S.gguf · Q6_K_S · 8K · ROCm · llama-server | Local AI inference | prompt tokens per second (tok/s) | 1 | Latest measurement Sep 25, 2026 | Measured — Measured on the exact workload (model, settings, and runtime) being compared. | Supporting only |
LocalScore · repo bd6ffe9 · pp16+tg1536
Standardized test — The same benchmark suite and version run on each GPU.Supporting only
LocalScore · repo bd6ffe9 · pp16+tg1536
Standardized test — The same benchmark suite and version run on each GPU.Supporting only
LocalScore · repo bd6ffe9 · pp16+tg1536
Standardized test — The same benchmark suite and version run on each GPU.Supporting only
LocalScore · repo bd6ffe9 · pp64+tg1024
Standardized test — The same benchmark suite and version run on each GPU.Supporting only
LocalScore · repo bd6ffe9 · pp64+tg1024
Standardized test — The same benchmark suite and version run on each GPU.Supporting only
LocalScore · repo bd6ffe9 · pp64+tg1024
Standardized test — The same benchmark suite and version run on each GPU.Supporting only
LocalScore · repo bd6ffe9 · pp384+tg1152
Standardized test — The same benchmark suite and version run on each GPU.Supporting only
LocalScore · repo bd6ffe9 · pp384+tg1152
Standardized test — The same benchmark suite and version run on each GPU.Supporting only
LocalScore · repo bd6ffe9 · pp384+tg1152
Standardized test — The same benchmark suite and version run on each GPU.Supporting only
LocalScore · repo bd6ffe9 · pp1024+tg16
Standardized test — The same benchmark suite and version run on each GPU.Supporting only
LocalScore · repo bd6ffe9 · pp1024+tg16
Standardized test — The same benchmark suite and version run on each GPU.Supporting only
LocalScore · repo bd6ffe9 · pp1024+tg16
Standardized test — The same benchmark suite and version run on each GPU.Supporting only
LocalScore · repo bd6ffe9 · pp1024+tg1024
Standardized test — The same benchmark suite and version run on each GPU.Supporting only
LocalScore · repo bd6ffe9 · pp1024+tg1024
Standardized test — The same benchmark suite and version run on each GPU.Supporting only
LocalScore · repo bd6ffe9 · pp1024+tg1024
Standardized test — The same benchmark suite and version run on each GPU.Supporting only
LocalScore · repo bd6ffe9 · pp1280+tg3072
Standardized test — The same benchmark suite and version run on each GPU.Supporting only
LocalScore · repo bd6ffe9 · pp1280+tg3072
Standardized test — The same benchmark suite and version run on each GPU.Supporting only
LocalScore · repo bd6ffe9 · pp1280+tg3072
Standardized test — The same benchmark suite and version run on each GPU.Supporting only
LocalScore · repo bd6ffe9 · pp2048+tg256
Standardized test — The same benchmark suite and version run on each GPU.Supporting only
LocalScore · repo bd6ffe9 · pp2048+tg256
Standardized test — The same benchmark suite and version run on each GPU.Supporting only
LocalScore · repo bd6ffe9 · pp2048+tg256
Standardized test — The same benchmark suite and version run on each GPU.Supporting only
LocalScore · repo bd6ffe9 · pp2048+tg768
Standardized test — The same benchmark suite and version run on each GPU.Supporting only
LocalScore · repo bd6ffe9 · pp2048+tg768
Standardized test — The same benchmark suite and version run on each GPU.Supporting only
LocalScore · repo bd6ffe9 · pp2048+tg768
Standardized test — The same benchmark suite and version run on each GPU.Supporting only
LocalScore · repo bd6ffe9 · pp4096+tg256
Standardized test — The same benchmark suite and version run on each GPU.Supporting only
LocalScore · repo bd6ffe9 · pp4096+tg256
Standardized test — The same benchmark suite and version run on each GPU.Supporting only
LocalScore · repo bd6ffe9 · pp4096+tg256
Standardized test — The same benchmark suite and version run on each GPU.Supporting only
Q6_K_S · 8K · ROCm · llama-bench
Measured — Measured on the exact workload (model, settings, and runtime) being compared.Met legacy decision rules
Q6_K_S · 8K · ROCm · llama-bench
Measured — Measured on the exact workload (model, settings, and runtime) being compared.Met legacy decision rules
Q6_K_S · 32K · ROCm · llama-bench
Measured — Measured on the exact workload (model, settings, and runtime) being compared.Met legacy decision rules
Q6_K_S · 32K · ROCm · llama-bench
Measured — Measured on the exact workload (model, settings, and runtime) being compared.Met legacy decision rules
Q6_K_S · 64K · ROCm · llama-bench
Measured — Measured on the exact workload (model, settings, and runtime) being compared.Met legacy decision rules
Q6_K_S · 64K · ROCm · llama-bench
Measured — Measured on the exact workload (model, settings, and runtime) being compared.Met legacy decision rules
Qwen3.8-27B-Q6_K_S.gguf · Q6_K_S · 8K · ROCm · llama-bench
Measured — Measured on the exact workload (model, settings, and runtime) being compared.Supporting only
Qwen3.8-27B-Q6_K_S.gguf · Q6_K_S · 8K · ROCm · llama-bench
Measured — Measured on the exact workload (model, settings, and runtime) being compared.Supporting only
Qwen3.8-27B-Q6_K_S.gguf · Q6_K_S · 8K · ROCm · llama-server
Measured — Measured on the exact workload (model, settings, and runtime) being compared.Supporting only
Qwen3.8-27B-Q6_K_S.gguf · Q6_K_S · 8K · ROCm · llama-server
Measured — Measured on the exact workload (model, settings, and runtime) being compared.Supporting only
Qwen3.8-27B-Q6_K_S.gguf · Q6_K_S · 8K · ROCm · llama-server
Measured — Measured on the exact workload (model, settings, and runtime) being compared.Supporting only