GPUs / NVIDIA

GTX 1080 Ti

Memory
GDDR5X
Bus
352-bit
Published peak
484 GB/s
Capacity
11 GB

Measured

Read
425.8GB/s
Write
443.2GB/s
Copy
416.8GB/s
Random
27.3GB/s
Latency
278ns
Sustain
416.9GB/s

Medians of 3 verified runs from 3 operators. Read reaches 88 % of the published peak.

77.33 verified runs, median 77.377.4

Local AI speed

ModelParameters4-bit, tokens/s8-bit, tokens/s
Llama 3.1 8B8.03B≤ 106≤ 53
Qwen 2.5 14B14.7B≤ 58Does not fit
gpt-oss-20b21B, 3.6B active≤ 237Does not fit
Mistral Small 24B24BDoes not fitDoes not fit
Qwen 2.5 32B32.5BDoes not fitDoes not fit
Llama 3.3 70B70.6BDoes not fitDoes not fit
gpt-oss-120b117B, 5.1B activeDoes not fitDoes not fit
Upper bounds: median read ÷ bytes read per token. Mixture-of-experts models read only their active experts. Fit checks weights against the 11 GB configuration; the KV cache needs room on top.

Top operators

#OperatorScoreReadDate
1OP-A821F477.4428.8 GB/s2026-09-24
2OP-91C77D77.3425.8 GB/s2026-09-27
3OP-C7DCC977.3425.6 GB/s2026-09-27

Full GTX 1080 Ti leaderboard