example_performance.toml•1.65 kB
# Example configuration for different performance modes
# High Accuracy Mode - Best quality, slower processing
[performance_high_accuracy]
mode = "high_accuracy"
[performance_high_accuracy.index]
index_type = "Flat"
nprobe = 50
nlist = 200
m = 48
ef_construction = 500
ef_search = 200
use_gpu = false
[performance_high_accuracy.cache]
enabled = true
max_size_mb = 512
ttl_secs = 7200
eviction_policy = "lru"
preload_common = true
[performance_high_accuracy.processing]
batch_size = 16
parallel_workers = 4
chunk_size = 256
overlap_size = 100
max_queue_size = 500
timeout_secs = 60
# High Speed Mode - Fast processing with good accuracy
[performance_high_speed]
mode = "high_speed"
[performance_high_speed.index]
index_type = "IVFPQ"
nprobe = 8
nlist = 50
m = 16
ef_construction = 100
ef_search = 32
use_gpu = false
quantization = "PQ8"
[performance_high_speed.cache]
enabled = true
max_size_mb = 128
ttl_secs = 1800
eviction_policy = "lfu"
preload_common = false
[performance_high_speed.processing]
batch_size = 64
parallel_workers = 16
chunk_size = 1024
overlap_size = 25
max_queue_size = 2000
timeout_secs = 15
# Ultra Fast Mode - Minimal latency, edge deployment
[performance_ultra_fast]
mode = "ultra_fast"
[performance_ultra_fast.index]
index_type = "IVFPQ"
nprobe = 4
nlist = 25
m = 8
ef_construction = 50
ef_search = 16
use_gpu = false
quantization = "PQ4"
[performance_ultra_fast.cache]
enabled = true
max_size_mb = 64
ttl_secs = 900
eviction_policy = "fifo"
preload_common = false
[performance_ultra_fast.processing]
batch_size = 128
parallel_workers = 24
chunk_size = 2048
overlap_size = 0
max_queue_size = 5000
timeout_secs = 10