mirror of
https://github.com/skywind3000/kcp.git
synced 2025-03-14 00:30:11 +08:00
new document
This commit is contained in:
parent
59f21870b7
commit
48c759ccd8
54
README.md
54
README.md
@ -103,24 +103,21 @@ TCP是为流量设计的(每秒内可以传输多少KB的数据),讲究的
|
||||
协议默认模式是一个标准的 ARQ,需要通过配置打开各项加速开关:
|
||||
|
||||
1. 工作模式:
|
||||
```cpp
|
||||
int ikcp_nodelay(ikcpcb *kcp, int nodelay, int interval, int resend, int nc)
|
||||
```
|
||||
|
||||
int ikcp_nodelay(ikcpcb *kcp, int nodelay, int interval, int resend, int nc)
|
||||
|
||||
nodelay :是否启用 nodelay模式,0不启用;1启用。
|
||||
|
||||
interval :协议内部工作的 interval,单位毫秒,比如 10ms或者 20ms
|
||||
|
||||
resend :快速重传模式,默认0关闭,可以设置2(2次ACK跨越将会直接重传)
|
||||
|
||||
nc :是否关闭流控,默认是0代表不关闭,1代表关闭。
|
||||
|
||||
普通模式:`ikcp_nodelay(kcp, 0, 40, 0, 0);
|
||||
|
||||
极速模式: ikcp_nodelay(kcp, 1, 10, 2, 1);
|
||||
> nodelay :是否启用 nodelay模式,0不启用;1启用。
|
||||
> interval :协议内部工作的 interval,单位毫秒,比如 10ms或者 20ms
|
||||
> resend :快速重传模式,默认0关闭,可以设置2(2次ACK跨越将会直接重传)
|
||||
> nc :是否关闭流控,默认是0代表不关闭,1代表关闭。
|
||||
> 普通模式:`ikcp_nodelay(kcp, 0, 40, 0, 0);
|
||||
> 极速模式: ikcp_nodelay(kcp, 1, 10, 2, 1);
|
||||
|
||||
2. 最大窗口:
|
||||
|
||||
int ikcp_wndsize(ikcpcb *kcp, int sndwnd, int rcvwnd);
|
||||
```cpp
|
||||
int ikcp_wndsize(ikcpcb *kcp, int sndwnd, int rcvwnd);
|
||||
```
|
||||
该调用将会设置协议的最大发送窗口和最大接收窗口大小,默认为32.
|
||||
|
||||
3. 最大传输单元:
|
||||
@ -133,9 +130,9 @@ TCP是为流量设计的(每秒内可以传输多少KB的数据),讲究的
|
||||
不管是 TCP还是 KCP计算 RTO时都有最小 RTO的限制,即便计算出来RTO为40ms,由
|
||||
于默认的 RTO是100ms,协议只有在100ms后才能检测到丢包,快速模式下为30ms,可
|
||||
以手动更改该值:
|
||||
```cpp
|
||||
kcp->rx_minrto = 10;
|
||||
```
|
||||
```cpp
|
||||
kcp->rx_minrto = 10;
|
||||
```
|
||||
|
||||
# 最佳实践
|
||||
#### 内存分配器
|
||||
@ -143,9 +140,7 @@ TCP是为流量设计的(每秒内可以传输多少KB的数据),讲究的
|
||||
默认KCP协议使用 malloc/free进行内存分配释放,如果应用层接管了内存分配,可以
|
||||
用ikcp_allocator来设置新的内存分配器,注意要在一开始设置:
|
||||
|
||||
```cpp
|
||||
ikcp_allocator(my_new_malloc, my_new_free);
|
||||
```
|
||||
> ikcp_allocator(my_new_malloc, my_new_free);
|
||||
|
||||
|
||||
#### 前向纠错注意
|
||||
@ -169,3 +164,20 @@ P3 = (3, 2, 1)
|
||||
这样几个包发送出去,接收方对于单个原始包都可能被解出3次来(后面两个包任然会重
|
||||
复该包内容),那么这里需要记录一下,一个下层数据包只会input给kcp一次,避免过
|
||||
多重复ack带来的浪费。
|
||||
|
||||
#### 管理大规模连接
|
||||
|
||||
如果需要同时管理大规模的 KCP连接(比如大于3000个),比如你正在实现一套类 epoll
|
||||
的机制,那么为了避免每秒钟对每个连接调用大量的调用 ikcp_update,我们可以使用
|
||||
ikcp_check来大大减少 ikcp_update调用的次数。 ikcp_check返回值会告诉你需要
|
||||
在什么时间点再次调用 ikcp_update(如果中途没有 ikcp_send, ikcp_input的话,
|
||||
否则中途调用了 ikcp_send, ikcp_input的话,需要在下一次interval时调用 update)
|
||||
|
||||
标准顺序是每次调用了 ikcp_update后,使用 ikcp_check决定下次什么时间点再次调用
|
||||
ikcp_update,而如果中途发生了 ikcp_send, ikcp_input的话,在下一轮 interval
|
||||
立马调用 ikcp_update和 ikcp_check。
|
||||
|
||||
原来在处理2000个 kcp连接且每个连接每10ms调用一次update,改为 check机制后,cpu
|
||||
从 60%降低到15%。
|
||||
|
||||
|
||||
|
BIN
doc/donate.png
Normal file
BIN
doc/donate.png
Normal file
Binary file not shown.
After Width: | Height: | Size: 5.2 KiB |
17
ikcp.c
17
ikcp.c
@ -1042,7 +1042,9 @@ void ikcp_flush(ikcpcb *kcp)
|
||||
|
||||
|
||||
//---------------------------------------------------------------------
|
||||
// input update
|
||||
// update state (call it repeatedly, every 10ms-100ms), or you can ask
|
||||
// ikcp_check when to call it again (without ikcp_input/_send calling).
|
||||
// 'current' - current timestamp in millisec.
|
||||
//---------------------------------------------------------------------
|
||||
void ikcp_update(ikcpcb *kcp, IUINT32 current)
|
||||
{
|
||||
@ -1074,13 +1076,12 @@ void ikcp_update(ikcpcb *kcp, IUINT32 current)
|
||||
|
||||
//---------------------------------------------------------------------
|
||||
// Determine when should you invoke ikcp_update:
|
||||
// if there is no ikcp_input/_send calling, you can call ikcp_update
|
||||
// after millisecs ikcp_check returns, instead of call update repeatly.
|
||||
// It is important to reduce unnacessary ikcp_update calling. you can
|
||||
// just call ikcp_update in a very small interval, or you can use it to
|
||||
// schedule ikcp_update calling (eg. when you are implementing an epoll
|
||||
// like mechanism, or optimize ikcp_update when handling massive kcp
|
||||
// connections)
|
||||
// returns when you should invoke ikcp_update in millisec, if there
|
||||
// is no ikcp_input/_send calling. you can call ikcp_update in that
|
||||
// time, instead of call update repeatly.
|
||||
// Important to reduce unnacessary ikcp_update invoking. use it to
|
||||
// schedule ikcp_update (eg. implementing an epoll-like mechanism,
|
||||
// or optimize ikcp_update when handling massive kcp connections)
|
||||
//---------------------------------------------------------------------
|
||||
IUINT32 ikcp_check(const ikcpcb *kcp, IUINT32 current)
|
||||
{
|
||||
|
17
ikcp.h
17
ikcp.h
@ -342,18 +342,17 @@ int ikcp_recv(ikcpcb *kcp, char *buffer, int len);
|
||||
int ikcp_send(ikcpcb *kcp, const char *buffer, int len);
|
||||
|
||||
// update state (call it repeatedly, every 10ms-100ms), or you can ask
|
||||
// ikcp_check when to call it again (without low level packet input).
|
||||
// 'current' - current timestamp in millisec
|
||||
// ikcp_check when to call it again (without ikcp_input/_send calling).
|
||||
// 'current' - current timestamp in millisec.
|
||||
void ikcp_update(ikcpcb *kcp, IUINT32 current);
|
||||
|
||||
// Determine when should you invoke ikcp_update:
|
||||
// if there is no ikcp_input/_send calling, you can call ikcp_update
|
||||
// after millisecs ikcp_check returns, instead of call update repeatly.
|
||||
// It is important to reduce unnacessary ikcp_update calling. you can
|
||||
// just call ikcp_update in a very small interval, or you can use it to
|
||||
// schedule ikcp_update calling (eg. when you are implementing an epoll
|
||||
// like mechanism, or optimize ikcp_update when handling massive kcp
|
||||
// connections)
|
||||
// returns when you should invoke ikcp_update in millisec, if there
|
||||
// is no ikcp_input/_send calling. you can call ikcp_update in that
|
||||
// time, instead of call update repeatly.
|
||||
// Important to reduce unnacessary ikcp_update invoking. use it to
|
||||
// schedule ikcp_update (eg. implementing an epoll-like mechanism,
|
||||
// or optimize ikcp_update when handling massive kcp connections)
|
||||
IUINT32 ikcp_check(const ikcpcb *kcp, IUINT32 current);
|
||||
|
||||
// when you received a low level packet (eg. UDP packet), call it
|
||||
|
Loading…
Reference in New Issue
Block a user