本帖最後由 harryytm 於 2020-3-31 17:57 編輯
而家果 D 係指邊 D 可唔可以舉一兩隻出黎做例子
我見過兩隻軟件都係用 PIECE SIZE黎做寫入單位
...
doggiestyle 發表於 2020-3-31 01:51 
qBittorrent 下載完成嘅 Piece 除咗由 Write Cache 寫落碟仲會保留喺 Read Cache
部份隨機寫入會比 Windows 個 Write-Cache 合拼
同埋會
Cache Set 得夠多嘅話下載緊嘅時候
大部份磁碟讀取都比 Read Cache 同 Windows 個 SuperPrefetch 頂咗
另外 qBittorrent 進階進項入面有啲設定可以進一步減少讀寫
Coalesce reads & writes - 合拼連續嘅讀寫減少交比 OS 嘅讀寫要求
Use piece extend affinity - piece size 太細嘅種會下載相鄰 4 MiB 嘅 pieces 去減少隨機寫入
Send upload piece suggestions - 建議其他 Peer 先下載 Read Cache 入面最新 Cache 咗嘅 Piece
| name | type | default | | coalesce_reads | bool | false | | coalesce_writes | bool | false | allocate separate, contiguous, buffers for read and write calls. Only used where writev/readv cannot be used will use more RAM but may improve performance
| name | type | default | | piece_extent_affinity | bool | false | when this is true, create an affinity for downloading 4 MiB extents of adjacent pieces. This is an attempt to achieve better disk I/O throughput by downloading larger extents of bytes, for torrents with small piece sizes
| name | type | default | | suggest_mode | int | settings_pack::no_piece_suggestions | suggest_mode controls whether or not libtorrent will send out suggest messages to create a bias of its peers to request certain pieces. The modes are:- no_piece_suggestions which will not send out suggest messages.
- suggest_read_cache which will send out suggest messages for the most recent pieces that are in the read cache.
|