In the measurement queue
NVIDIA GeForce RTX 4090
24 GB · Ada sm_89
We have not measured this card yet.
It is in the queue. Until then we show its spec and an estimate from our formula, clearly labelled.
An estimate, not a measurement
≈26–32t/s
before, estimate · Qwen3.8-27B UD-Q6_K · after: soon
0 60 t/s
- Memory
- 24GB
- Bandwidth
- 1,008GB/s
- Architecture
- Ada sm_89
An estimate, not a measurement
What fits
The formula η · bandwidth / bytes per token, for stock llama.cpp. Once measured, real numbers replace it here.
The formula η · bandwidth / bytes per token, for stock llama.cpp. Once measured, real numbers replace it here.app/landing/src/lib/estimator.ts: decodeRange (η·BW/B, raport 04 §2.2.4), memoryNeed (raport 04 §2.2.4)app/landing/src/lib/data/gpus.jsonapp/landing/src/lib/data/models.json- Qwen3.8-27B UD-Q6_K24 GB
memory needed, estimate ≈ 93–95 % of the card's memory
before, estimate ≈ 26–32 t/s
after soon
- LFM2.5-24B-A2B Q5_K_M24 GB
memory needed, estimate ≈ 71–72 % of the card's memory
before, estimate ≈ 176–311 t/s
after soon