CCIE-16-PIM

目录

  • 实验条件
    • 网络拓朴
    • 实验环境
    • 实验目的
  • 开始实验
    • 实验1:PIM-DM
      • 配置PIM域中的路由,开启PIM-DM组播路由功能,
      • 验证组播情况
    • 实验2:PIM-SM(静态RP)
      • 配置PIM域中的路由,开启PIM-SM组播路由功能,并指定RP
    • 实验3:PIM-SM(auto-RP)
      • 开启PIM-SM组播路由功能
      • 验证结果
    • 实验4:PIM-SM(auto-RP)
      • 开始配置
      • 验证结果

实验条件

网络拓朴

在这里插入图片描述

实验环境

在这里插入图片描述

实验目的

实验组播功能

开始实验

实验1:PIM-DM

  1. R1/R2/R3/R5在一个PIM域中,这4台路由器需要开启组播路由功能,同时相应接口需要运行PIM-SM,从而为组播流量进行路由;
  2. R7/R8/R9模拟主机,分别模拟加入到不同的组播组当中
  3. R5模拟组播末节路由器,需要与主机R7/R8/R9之间通过IGMP进行通信

查询配置发现R7/R8/R9已经通过igmp-join分别加入了不同或相同的组播组,查看加入的组播情况

R7#show ip igmp groups
IGMP Connected Group Membership
Group Address    Interface                Uptime    Expires   Last Reporter   Group Accounted
239.1.1.1        Ethernet0/0              00:06:23  never     192.168.2.1     
239.1.1.3        Ethernet0/0              00:06:23  never     192.168.2.1     R8#show ip igmp groups
IGMP Connected Group Membership
Group Address    Interface                Uptime    Expires   Last Reporter   Group Accounted
239.1.1.1        Ethernet0/0              00:07:33  never     192.168.2.1     
239.1.1.3        Ethernet0/0              00:07:33  never     192.168.2.1     
239.1.1.2        Ethernet0/0              00:07:33  never     192.168.2.2 R9#show ip igmp groups
IGMP Connected Group Membership
Group Address    Interface                Uptime    Expires   Last Reporter   Group Accounted
239.1.1.3        Ethernet0/0              00:07:57  never     192.168.2.1     
239.1.1.2        Ethernet0/0              00:07:57  never     192.168.2.3     

三组组播:
Group1:239.1.1.1 R7、R8
Group1:239.1.1.2 R8、R9
Group1:239.1.1.3 R7、R8、R9

配置PIM域中的路由,开启PIM-DM组播路由功能,

R1:
conf t
ip multicast-routing
interface range e0/0-2ip pim dense-modeR2:
conf t
ip multicast-routing
interface range e0/0-1ip pim dense-modeR3:
conf t
ip multicast-routing
interface range e0/1-2ip pim dense-modeR5:
conf t
ip multicast-routing
interface range e0/0, e0/2ip pim dense-mode

验证组播情况

在R6模拟发出组播流量之前,观察各组播路由器的组播路由表情况。

  1. 发起组播流量前:R1/R2/R3/R5组播路由表暂没有形成(S,G)的表项:
    下图是R5上查看的结果,只有组成员发送IGMP report后形成的(*,G)项,无任何(S,G)表项,其他R1/R2/R3也都不会形成
R5#show ip mroute   
IP Multicast Routing Table
Flags: D - Dense, S - Sparse, B - Bidir Group, s - SSM Group, C - Connected,L - Local, P - Pruned, R - RP-bit set, F - Register flag,T - SPT-bit set, J - Join SPT, M - MSDP created entry, E - Extranet,X - Proxy Join Timer Running, A - Candidate for MSDP Advertisement,U - URD, I - Received Source Specific Host Report, Z - Multicast Tunnel, z - MDT-data group sender, Y - Joined MDT-data group, y - Sending to MDT-data group, G - Received BGP C-Mroute, g - Sent BGP C-Mroute, N - Received BGP Shared-Tree Prune, n - BGP C-Mroute suppressed, Q - Received BGP S-A Route, q - Sent BGP S-A Route, V - RD & Vector, v - Vector, p - PIM Joins on route, x - VxLAN group
Outgoing interface flags: H - Hardware switched, A - Assert winner, p - PIM JoinTimers: Uptime/ExpiresInterface state: Interface, Next-Hop or VCD, State/Mode(*, 239.1.1.1), 00:43:35/00:02:27, RP 0.0.0.0, flags: DCIncoming interface: Null, RPF nbr 0.0.0.0Outgoing interface list:Ethernet0/2, Forward/Dense, 00:43:35/stoppedEthernet0/0, Forward/Dense, 00:43:35/stopped(*, 239.1.1.3), 00:43:35/00:02:25, RP 0.0.0.0, flags: DCIncoming interface: Null, RPF nbr 0.0.0.0Outgoing interface list:Ethernet0/2, Forward/Dense, 00:43:35/stoppedEthernet0/0, Forward/Dense, 00:43:35/stopped(*, 239.1.1.2), 00:43:35/00:02:25, RP 0.0.0.0, flags: DCIncoming interface: Null, RPF nbr 0.0.0.0Outgoing interface list:Ethernet0/2, Forward/Dense, 00:43:35/stoppedEthernet0/0, Forward/Dense, 00:43:35/stopped(*, 224.0.1.40), 00:43:36/00:02:46, RP 0.0.0.0, flags: DCLIncoming interface: Null, RPF nbr 0.0.0.0Outgoing interface list:Ethernet0/2, Forward/Dense, 00:43:35/stoppedEthernet0/0, Forward/Dense, 00:43:36/stopped
  1. R6模拟发出组播流量
R6#ping 239.1.1.1
Type escape sequence to abort.
Sending 1, 100-byte ICMP Echos to 239.1.1.1, timeout is 2 seconds:Reply to request 0 from 192.168.2.1, 26 ms
Reply to request 0 from 192.168.2.1, 47 ms
Reply to request 0 from 192.168.2.2, 47 ms
Reply to request 0 from 192.168.2.2, 26 ms
R6#ping 239.1.1.1
Type escape sequence to abort.
Sending 1, 100-byte ICMP Echos to 239.1.1.1, timeout is 2 seconds:Reply to request 0 from 192.168.2.1, 1 ms
Reply to request 0 from 192.168.2.2, 1 ms
R6#

再次观察R1/R2/R3/R5组播路由表,已形成(S,G)表项

R1#show ip mroute
......
(*, 239.1.1.3), 00:01:39/stopped, RP 0.0.0.0, flags: DIncoming interface: Null, RPF nbr 0.0.0.0Outgoing interface list:Ethernet0/2, Forward/Dense, 00:01:39/stoppedEthernet0/1, Forward/Dense, 00:01:39/stopped(192.168.1.1, 239.1.1.3), 00:01:39/00:01:20, flags: TIncoming interface: Ethernet0/0, RPF nbr 0.0.0.0Outgoing interface list:Ethernet0/1, Prune/Dense, 00:01:39/00:01:20Ethernet0/2, Forward/Dense, 00:01:39/stoppedR2#show ip mroute
......
(*, 239.1.1.3), 00:02:03/stopped, RP 0.0.0.0, flags: DIncoming interface: Null, RPF nbr 0.0.0.0Outgoing interface list:Ethernet0/1, Forward/Dense, 00:02:03/stoppedEthernet0/0, Forward/Dense, 00:02:03/stopped(192.168.1.1, 239.1.1.3), 00:02:03/00:00:56, flags: PTIncoming interface: Ethernet0/1, RPF nbr 12.1.1.1Outgoing interface list:Ethernet0/0, Prune/Dense, 00:02:03/00:00:58......

上面模拟向239.1.1.1组中发起1个包的流量,第一次发起时由R7/R8组成员分别回复,但是会发现每个组成员均回复了2次,
原因分析:R2、R3作为运行了PIM-DM的路由器,处在一个MA网络中,当有第一次组播流量触发后,会发assert声明报文进行DR选举,但第一次的流量仍然会下发,所以R7、R8会各自接收到R2和R3共2份包,所以就各回复1份;当DR选举完后(R3成为assert winner),所以后续该组播流量就只有R3进行转发,R2会发prune报文进行修剪!
在这里插入图片描述
实现结果表明:

  1. PIM-DM模式是先进行组播流量的push
  2. 组播流量触发后,各运行PIM-DM的路由器才会形成组播路由表(S,G),形成SPT源树!
  3. 在MA网络中, Assert声明选举DR失败的一方,将发送prune进行修剪(还有其他可能造成prune的情况,本次实验没有涉及)。

实验2:PIM-SM(静态RP)

PIM域为R1/R2/R3/R5,静态指定R3为RP
R1/R2/R3/R5开启组播路由功能,并配置PIM-SM,每台PIM路由器均手工指定R3为RP
(R3上也要指定自己为RP)

重置实验LAB,再来配置实验,以免之前的配置干扰

配置PIM域中的路由,开启PIM-SM组播路由功能,并指定RP

R1:
conf t
ip multicast-routing
interface range e0/0-2ip pim sparse-mode
ip pim rp-address 3.3.3.3 overrideR2:
conf t
ip multicast-routing
interface range e0/0-1ip pim sparse-mode
ip pim rp-address 3.3.3.3 overrideR3:
conf t
ip multicast-routing
interface range e0/1-2ip pim sparse-mode
ip pim rp-address 3.3.3.3 overrideR5:
conf t
ip multicast-routing
interface range e0/0, e0/2ip pim sparse-mode
ip pim rp-address 3.3.3.3 override

这里不指定override会让静态指定的RP无效。因为思科路由器会默认开启其它类型的RP协议
这里指定R3为RP(其lo0已经在IGP里通告,可以被其他PIM路由器学习到,该lo0无需接收组播流量,可以不用开启PIM-SM)

  1. R6模拟发出组播流量前:R5会接受R7/R8/R9的IGMP-join的report报文,本身会形成(,G)表项,同时R5会从RPF接口沿着RP方向发送(,G)join报文形成共享树!
    分析:对于本拓扑来说,由于R3被静态指定成RP,所以R5会从RPF接口e0/2发出(*,G)join,被R3-e0/1接收到

R5上观察组播路由表:静态RP为3.3.3.3

R5#show ip mroute
IP Multicast Routing Table
Flags: D - Dense, S - Sparse, B - Bidir Group, s - SSM Group, C - Connected,L - Local, P - Pruned, R - RP-bit set, F - Register flag,T - SPT-bit set, J - Join SPT, M - MSDP created entry, E - Extranet,X - Proxy Join Timer Running, A - Candidate for MSDP Advertisement,U - URD, I - Received Source Specific Host Report, Z - Multicast Tunnel, z - MDT-data group sender, Y - Joined MDT-data group, y - Sending to MDT-data group, G - Received BGP C-Mroute, g - Sent BGP C-Mroute, N - Received BGP Shared-Tree Prune, n - BGP C-Mroute suppressed, Q - Received BGP S-A Route, q - Sent BGP S-A Route, V - RD & Vector, v - Vector, p - PIM Joins on route, x - VxLAN group
Outgoing interface flags: H - Hardware switched, A - Assert winner, p - PIM JoinTimers: Uptime/ExpiresInterface state: Interface, Next-Hop or VCD, State/Mode(*, 239.1.1.1), 00:04:14/00:02:45, RP 3.3.3.3, flags: SJCIncoming interface: Ethernet0/2, RPF nbr 2.3.5.3Outgoing interface list:Ethernet0/0, Forward/Sparse, 00:04:13/00:02:45(*, 239.1.1.3), 00:04:14/00:02:46, RP 3.3.3.3, flags: SJCIncoming interface: Ethernet0/2, RPF nbr 2.3.5.3Outgoing interface list:Ethernet0/0, Forward/Sparse, 00:04:13/00:02:46(*, 239.1.1.2), 00:04:13/00:02:46, RP 3.3.3.3, flags: SJCIncoming interface: Ethernet0/2, RPF nbr 2.3.5.3Outgoing interface list:Ethernet0/0, Forward/Sparse, 00:04:13/00:02:46

R3上观察组播路由表:R3接收到(*,G)join报文后,会在组播路由表形成(*,G)表项

R3(config)#do show ip mroute
IP Multicast Routing Table
Flags: D - Dense, S - Sparse, B - Bidir Group, s - SSM Group, C - Connected,L - Local, P - Pruned, R - RP-bit set, F - Register flag,T - SPT-bit set, J - Join SPT, M - MSDP created entry, E - Extranet,X - Proxy Join Timer Running, A - Candidate for MSDP Advertisement,U - URD, I - Received Source Specific Host Report, Z - Multicast Tunnel, z - MDT-data group sender, Y - Joined MDT-data group, y - Sending to MDT-data group, G - Received BGP C-Mroute, g - Sent BGP C-Mroute, N - Received BGP Shared-Tree Prune, n - BGP C-Mroute suppressed, Q - Received BGP S-A Route, q - Sent BGP S-A Route, V - RD & Vector, v - Vector, p - PIM Joins on route, x - VxLAN group
Outgoing interface flags: H - Hardware switched, A - Assert winner, p - PIM JoinTimers: Uptime/ExpiresInterface state: Interface, Next-Hop or VCD, State/Mode(*, 239.1.1.1), 00:02:47/00:02:37, RP 3.3.3.3, flags: SIncoming interface: Null, RPF nbr 0.0.0.0Outgoing interface list:Ethernet0/1, Forward/Sparse, 00:02:47/00:02:37(*, 239.1.1.3), 00:02:47/00:02:39, RP 3.3.3.3, flags: SIncoming interface: Null, RPF nbr 0.0.0.0Outgoing interface list:Ethernet0/1, Forward/Sparse, 00:02:47/00:02:39(*, 239.1.1.2), 00:02:47/00:02:39, RP 3.3.3.3, flags: SIncoming interface: Null, RPF nbr 0.0.0.0Outgoing interface list:Ethernet0/1, Forward/Sparse, 00:02:47/00:02:39

以上从R5的RPF接口开始沿着RP方向到R3之间会形成共享树!

R2上观察组播路由表:由于R2不是RP,同时也不在R5的RPF口沿着RP方向的链路上,所以R2上不会接收到(,G)join报文,不会形成(,G)表项:

R2(config)#do show ip mroute
IP Multicast Routing Table
Flags: D - Dense, S - Sparse, B - Bidir Group, s - SSM Group, C - Connected,L - Local, P - Pruned, R - RP-bit set, F - Register flag,T - SPT-bit set, J - Join SPT, M - MSDP created entry, E - Extranet,X - Proxy Join Timer Running, A - Candidate for MSDP Advertisement,U - URD, I - Received Source Specific Host Report, Z - Multicast Tunnel, z - MDT-data group sender, Y - Joined MDT-data group, y - Sending to MDT-data group, G - Received BGP C-Mroute, g - Sent BGP C-Mroute, N - Received BGP Shared-Tree Prune, n - BGP C-Mroute suppressed, Q - Received BGP S-A Route, q - Sent BGP S-A Route, V - RD & Vector, v - Vector, p - PIM Joins on route, x - VxLAN group
Outgoing interface flags: H - Hardware switched, A - Assert winner, p - PIM JoinTimers: Uptime/ExpiresInterface state: Interface, Next-Hop or VCD, State/Mode(*, 224.0.1.40), 00:02:55/00:02:12, RP 3.3.3.3, flags: SJPLIncoming interface: Ethernet0/0, RPF nbr 2.3.5.3Outgoing interface list: NullR2(config)#
  1. R6模拟发出组播流量:R1与R3形成GRE-tunnel,然后将组播报文封装成单播发送一个register报文,沿着GRE隧道发送给RP进行注册,RP将报文拆包还原成组播沿着共享树的方向发给R5,同时RP发送register-stop报文给R1,R1-R3之间形成SPT源树!
    以下是R3上的组播路由表:
    在这里插入图片描述
R1(config)#do show ip int br
Interface                  IP-Address      OK? Method Status                
......   
Loopback0                  1.1.1.1         YES TFTP   up                    up      
Tunnel0                    13.1.1.1        YES unset  up                    up      
R1(config)#do show int tunnel0
Tunnel0 is up, line protocol is up Hardware is TunnelDescription: Pim Register Tunnel (Encap) for RP 3.3.3.3Interface is unnumbered. Using address of Ethernet0/2 (13.1.1.1)MTU 17912 bytes, BW 100 Kbit/sec, DLY 50000 usec, reliability 255/255, txload 1/255, rxload 1/255Encapsulation TUNNEL, loopback not setKeepalive not setTunnel linestate evaluation upTunnel source 13.1.1.1 (Ethernet0/2), destination 3.3.3.3Tunnel Subblocks:src-track:Tunnel0 source tracking subblock associated with Ethernet0/2Set of tunnels with source Ethernet0/2, 1 member (includes iterators), on interface <OK>Tunnel protocol/transport PIM/IPv4Tunnel TTL 255Tunnel transport MTU 1472 bytesTunnel is transmit only
......
R1(config)#

以上实验表明:

  1. PIM-SM是last-hop路由器先沿着RP方向发送join报文形成(*,G)表项,首先构造出一个共享树;
  2. first-hop到RP之间通过通过GRE隧道发送register报文和register-stop,RP通过反向溯源的方式,成功构造出一个源树;
  3. 当R5溯源成功后,从RPF口沿着到source的方向沿途形成(S,G),于是完成SPT切换;
  4. 后续组播流量沿着SPT源树进行下发

实验3:PIM-SM(auto-RP)

将R1作为MA,R2/R3作为C-RP,为保证RP相关报文能顺利从PIM-SM接口泛洪,本实验主要是要让R5也能接收,则R2和R3开启autorp listen功能,使得在还没有选出RP的情况下,有关RP选举报文能泛洪出去被R5接收到!

开启PIM-SM组播路由功能

注:由于R1要配置成MA,并配置其lo0要发送rp-discovery报文而都要配置C-RP,并配置其lo0要能发送rp-announce报文

R1:
conf t
ip multicast-routing
interface range e0/0-2ip pim sparse-mode
interface lo 0ip pim sparse-mode
ip pim send-rp-discovery loopback 0 scope 255
ip pim autorp listenerR2:
conf t
ip multicast-routing
interface range e0/0-1ip pim sparse-mode
interface lo 0ip pim sparse-mode
ip pim send-rp-announce loopback 0 scope 255
ip pim autorp listenerR3:
conf t
ip multicast-routing
interface range e0/1-2ip pim sparse-mode
interface lo 0ip pim sparse-mode
ip pim send-rp-announce loopback 0 scope 255R5:
conf t
ip multicast-routing
interface range e0/0, e0/2ip pim sparse-mode

验证结果

  1. 查看RP选举情况:
    在MA上查看RP选举情况:R3被选举为RP(选举IP地址大的成为RP)
R1#show ip pim rp mapping in-use 
PIM Group-to-RP Mappings
This system is an RP-mapping agent (Loopback0)Group(s) 224.0.0.0/4RP 3.3.3.3 (?), v2v1Info source: 3.3.3.3 (?), elected via Auto-RPUptime: 00:00:54, expires: 00:02:01RP 2.2.2.2 (?), v2v1Info source: 2.2.2.2 (?), via Auto-RPUptime: 00:01:04, expires: 00:02:52Dynamic (Auto-RP or BSR) RPs in cache that are in use:
R1#

在R5上查看:RP选举结果已经泛洪被R5接收

R5#show ip pim rp mapping in-use 
PIM Group-to-RP MappingsGroup(s) 224.0.0.0/4RP 3.3.3.3 (?), v2v1Info source: 1.1.1.1 (?), elected via Auto-RPUptime: 00:01:42, expires: 00:02:14Dynamic (Auto-RP or BSR) RPs in cache that are in use:
Group(s): 224.0.0.0/4, RP: 3.3.3.3, expires: 00:00:57
R5#

现在,整个PIM域中,选举出R3作为RP

实验4:PIM-SM(auto-RP)

将R1设置为C-BSR(最终成为BSR),R2/R3设置为C-RP

开始配置

R1:
conf t
ip multicast-routing
interface range e0/0-2ip pim sparse-mode
interface lo 0ip pim sparse-mode
ip pim bsr-candidate lo0 0 0R2:
conf t
ip multicast-routing
interface range e0/0-1ip pim sparse-mode
interface lo 0ip pim sparse-mode
ip pim rp-candidate lo0 priority 0R3:
conf t
ip multicast-routing
interface range e0/1-2ip pim sparse-mode
interface lo 0ip pim sparse-mode
ip pim rp-candidate lo0 priority 0R5:
conf t
ip multicast-routing
interface range e0/0, e0/2ip pim sparse-mode

验证结果

  1. 查看BSR:R1成为BSR
R1#show ip pim bsr-router 
PIMv2 Bootstrap information
This system is the Bootstrap Router (BSR)BSR address: 1.1.1.1 (?)Uptime:      00:01:05, BSR Priority: 0, Hash mask length: 0Next bootstrap message in 00:00:56
R1#
  1. 查看RP:R2成为RP
R2#show ip pim rp mapping in-use 
PIM Group-to-RP Mappings
This system is a candidate RP (v2)Group(s) 224.0.0.0/4RP 2.2.2.2 (?), v2Info source: 1.1.1.1 (?), via bootstrap, priority 0, holdtime 150Uptime: 00:00:47, expires: 00:01:40RP 3.3.3.3 (?), v2Info source: 1.1.1.1 (?), via bootstrap, priority 0, holdtime 150Uptime: 00:00:48, expires: 00:01:40Dynamic (Auto-RP or BSR) RPs in cache that are in use:
Group(s): 224.0.0.0/4, RP: 2.2.2.2, expires: 00:00:58
R2#show ip pim rp-hash 239.1.1.1RP 2.2.2.2 (?), v2Info source: 1.1.1.1 (?), via bootstrap, priority 0, holdtime 150Uptime: 00:01:42, expires: 00:01:45PIMv2 Hash Value (mask 0.0.0.0)RP 2.2.2.2, via bootstrap, priority 0, hash value 1524600152RP 3.3.3.3, via bootstrap, priority 0, hash value 450145259
R2#

注:这里通过RP选举规则第三条:较大的HASH值成为RP(前两条规则比不出来)R2运算得到的hash值更大:

    RP 2.2.2.2, via bootstrap, priority 0, hash value 1524600152RP 3.3.3.3, via bootstrap, priority 0, hash value 450145259
  1. 查看组播路由表:R5上查看组播路由表,当有组播流量后,完成了SPT切换
R5#show ip mroute 
IP Multicast Routing Table
Flags: D - Dense, S - Sparse, B - Bidir Group, s - SSM Group, C - Connected,L - Local, P - Pruned, R - RP-bit set, F - Register flag,T - SPT-bit set, J - Join SPT, M - MSDP created entry, E - Extranet,X - Proxy Join Timer Running, A - Candidate for MSDP Advertisement,U - URD, I - Received Source Specific Host Report, Z - Multicast Tunnel, z - MDT-data group sender, Y - Joined MDT-data group, y - Sending to MDT-data group, G - Received BGP C-Mroute, g - Sent BGP C-Mroute, N - Received BGP Shared-Tree Prune, n - BGP C-Mroute suppressed, Q - Received BGP S-A Route, q - Sent BGP S-A Route, V - RD & Vector, v - Vector, p - PIM Joins on route, x - VxLAN group
Outgoing interface flags: H - Hardware switched, A - Assert winner, p - PIM JoinTimers: Uptime/ExpiresInterface state: Interface, Next-Hop or VCD, State/Mode(*, 239.1.1.1), 00:38:10/00:02:49, RP 2.2.2.2, flags: SJCIncoming interface: Ethernet0/2, RPF nbr 2.3.5.2Outgoing interface list:Ethernet0/0, Forward/Sparse, 00:38:08/00:02:49(*, 239.1.1.3), 00:38:10/00:02:52, RP 2.2.2.2, flags: SJCIncoming interface: Ethernet0/2, RPF nbr 2.3.5.2Outgoing interface list:Ethernet0/0, Forward/Sparse, 00:38:08/00:02:52(*, 239.1.1.2), 00:38:10/00:02:51, RP 2.2.2.2, flags: SJCIncoming interface: Ethernet0/2, RPF nbr 2.3.5.2Outgoing interface list:Ethernet0/0, Forward/Sparse, 00:38:08/00:02:51

R6模拟发出组播流量,再查看R5路由表(*, 239.1.1.1),变成了从(192.168.1.1,239.1.1.1)从R3转发,完成了SPT切换。
在这里插入图片描述

本文来自互联网用户投稿,该文观点仅代表作者本人,不代表本站立场。本站仅提供信息存储空间服务,不拥有所有权,不承担相关法律责任。如若转载,请注明出处:http://www.hqwc.cn/news/636160.html

如若内容造成侵权/违法违规/事实不符,请联系编程知识网进行投诉反馈email:809451989@qq.com,一经查实,立即删除!

相关文章

什么是大语言模型以及如何构建自己的大型语言模型?

一、关于大语言模型 LLM 对于无数的应用程序非常有用&#xff0c;如果我们自己从头开始构建一个&#xff0c;那我们可以了解底层的ML技术&#xff0c;并可以根据特定需求定制LLM&#xff0c;但是对资源的需求巨大。大型语言模型是一种 ML 模型&#xff0c;可以执行各种自然语言…

性能直逼GPT4,Llama3的三种在线体验方式

周五凌晨&#xff0c;Meta发布了其最新的开源大语言模型Llama-3&#xff0c;据说性能直逼GPT-4。 前两天百度老板刚刚批了一下开源模型&#xff0c;说开源模型打不过闭源模型&#xff0c;没想到这么快就被打脸了。 自从Sora之后&#xff0c;OpenAI也好久没有发布震撼人心的能…

idea 将项目上传到gitee远程仓库具体操作

目录标题 一、新建仓库二、初始化项目三、addcommit四、配置远程仓库五、拉取远程仓库内容六、push代码到仓库七、如果是私有仓库可能会拉取失败&#xff08;一&#xff09;需要增加SSH 公钥&#xff08;二&#xff09;把远程仓库地址换成ssh的连接八、如果是私有仓库&#xff…

centos7搭建git服务器

1.centos7安装git yum install -y git yum install -y git-daemon 2.初始化空目录仓库 mkdir /usr/local/git mkdir /usr/local/git/projects mkdir /usr/local/git/projects/test-projects.git cd test-projects.git git --bare init 3.修改目录权限 cd .. chmod 775 tes…

【linux运维】系统常见管理命令

系列综述&#xff1a; &#x1f49e;目的&#xff1a;本系列是个人整理为了学习基本的shell编程和linux命令&#xff0c;整理期间苛求每个知识点&#xff0c;平衡理解简易度与深入程度。 &#x1f970;来源&#xff1a;材料主要源于b站大学——linux运维课程进行的&#xff0c;…

Parade Series - CoreAudio Loopback

Scenario 鉴于业务场景需要&#xff0c; 经过技术路径探索&#xff0c; 发现 comtypes 兼容性过于混乱&#xff0c;故而考虑整合一个 CoreAudio 的轮子dll来解决实际问题&#xff01;std::StringStream ⇒ std::ios::binary ⇒ std::ofstream Loopback.dll #ifndef _DLL_C…

Git - 在PyCharm/Idea中集成使用Git

文章目录 Git - 在PyCharm/Idea中集成使用Git1.新建GitHub仓库2.将仓库与项目绑定3.在PyCharm中使用Git4.新建Gitee仓库5.将仓库与项目绑定6.在IDEA中使用Git Git - 在PyCharm/Idea中集成使用Git 本文详细讲解了如何在 PyCharm 或 Idea 中配置 Gitee 或 GitHub 仓库&#xff0…

*Linux系统的进程和计划任务管理

目录 一、查看进程 1、程序和进程的关系 *2、ps查看静态进程信息 1&#xff09;ps aux 2&#xff09;ps -elf *3、top查看动态进程信息 4、pgrep查看进程信息 5、pstree查看进程树 二、控制进程 1、进程启动方式 2、进程的前后台调度 3、终止进程的运行 三、计划任…

HCF-Net:用于红外小目标检测的分层上下文融合网络

摘要 红外小目标检测是一项重要的计算机视觉任务&#xff0c;涉及在红外图像中识别和定位微小物体&#xff0c;这些物体通常仅包含几个像素。然而&#xff0c;由于物体尺寸极小以及红外图像中通常复杂的背景&#xff0c;这项任务面临困难。在本文中&#xff0c;我们提出了一种…

【Unity学习笔记】第十三 · tag与layer(运行时创建tag和layer)

参考&#xff1a; Unity手册 标签Unity手册 LayersIs it possible to create a tag programmatically?脚本自动添加tag和Layer 注&#xff1a;本文使用Unity版本是2022.3.23f1 转载引用请注明出处&#xff1a;&#x1f517;https://blog.csdn.net/weixin_44013533/article/de…

政安晨:【深度学习神经网络基础】(十二)—— 深度学习概要

目录 概述 深度学习的概况 深度学习的组成部分 部分标记的数据 修正线性单元 卷积神经网络 神经元Dropout GPU训练 政安晨的个人主页&#xff1a;政安晨 欢迎 &#x1f44d;点赞✍评论⭐收藏 收录专栏: 政安晨的机器学习笔记 希望政安晨的博客能够对您有所裨益&#xf…

NVIDIA安装程序失败-Nsight Visual Studio Edition失败解决办法

博主是要升级cuda版本&#xff0c;那么在安装新版本之前需要卸载以前的版本。 博主一溜卸载下去&#xff0c;最后有这么个东西卸载不掉&#xff0c;Nsight Visual Studio Edition 不管是电脑系统卸载还是360卸载&#xff0c;都卸载不掉。 此时安装新的cuda也遇到了这个问题 由…