Comparison Counted
Which engine runs Ternary Bonsai 2 27B best on M5 Max · 64 GiB?
- Device
- M5 Max · 64 GiB (Apple silicon Macs)
- OS
- macOS 27
- Model
- Ternary Bonsai 2 27B, 27B parameters
- Weights
- 2-bit
- Workload
- Archive 2048 input / 128 output: 2,048 tokens in, 128 out
- Decoding
- Greedy, batch 1, context 4,096
- Power
- Plugged in
Ranking
Speed: Decode speed: tokens generated per second.
| # | Engine | Decode and First token |
|---|---|---|
| 1 | Ollama†, an approximate comparison, see details | Decode: 44.6 tok/s 2,672 ms first token |
| 2 | MLX LM†, an approximate comparison, see details | Decode: 43.4 tok/s 3,291 ms first token |
| 3 | llama.cpp†, an approximate comparison, see details | Decode: 42.7 tok/s 2,814 ms first token |
| Not ranked in this view | ||
| Not ranked | Magnitude | Unsupported Unsupported in 1 comparison. Magnitude 0.2.17 reports this GGUF as Unsupported (magnitude catalog inspect) |
details
Speed Decode speed: tokens generated per second.
Score: 100 is the best result in every comparison an engine entered; 80 is 80% of the best, on average. Only complete experiments with a passed output check count. How rankings work
† Ollama: approximate comparison. The experiments use different effective KV cache precision: f16, model-dtype.
† MLX LM: approximate comparison. The experiments use different effective KV cache precision: f16, model-dtype. Its weights are 2-bit in another scheme: mlx-affine-2bit-g128, where most use Q2_0.
† llama.cpp: approximate comparison. The experiments use different effective KV cache precision: f16, model-dtype.
Every experiment
Medians over valid trials, with the lowest and highest beneath. Eligible experiments are complete and passed their output check. The ranking above selects each engine's configuration for this objective when its required metrics are measured.
| Experiment | Ranking eligibility | Decode tok/s, higher is better | First token ms, lower is better | Memory GiB, lower is better | Prefill tok/s, higher is better | Valid trials |
|---|---|---|---|---|---|---|
| Ollama 0.40.13 · Metal · Q2_0 | Eligible, approximately: The experiments use different effective KV cache precision: f16, model-dtype. | 44.6 range44.2–44.6 | 2,672 range2,572–2,941 | 8.12 range8.12–8.12 | 767.8 range697.4–798.1 | 6of 6 planned |
| MLX LM 0.32.2 · Metal · mlx-affine-2bit-g128 | Eligible, approximately: The experiments use different effective KV cache precision: f16, model-dtype. Its weights are 2-bit in another scheme: mlx-affine-2bit-g128, where most use Q2_0. | 43.4 range40.4–43.7 | 3,291 range2,966–3,482 | 14.45 range14.45–14.46 | 622.5 range588.5–690.9 | 6of 6 planned |
| llama.cpp b12110 · Metal · Q2_0 | Eligible, approximately: The experiments use different effective KV cache precision: f16, model-dtype. | 42.7 range42.2–42.8 | 2,814 range2,575–3,078 | 8.12 range8.12–8.12 | 728.1 range665.6–795.8 | 6of 6 planned |
| Magnitude 0.2.17 · Metal · Q2_0 | Unsupported Magnitude 0.2.17 reports this GGUF as Unsupported (magnitude catalog inspect) | – | – | – | – | 0of 6 planned |
Every trial, metric by metric
The line runs from an experiment's lowest to its highest valid trial, the dot is the median and each tick is one trial. Every scale starts at zero.
Caveats
- Publisher cards identify the base model but do not always pin the exact export source checkpoint revision.
- Engines may differ in quantization scheme, effective KV cache precision, kernels, and activations.
About this comparison
Worked out from its experiments: every experiment here shares the device, the OS major version, the power state, the base model, the quantization class and the workload. Measured under methodology v0.6. Comparison id glenn-m5-max-64gb--macos-27--ternary-bonsai-2-27b--2-bit--archive-2048-128.