1 ===========================================================
2 Linux* Base Driver for Intel(R) Ethernet Network Connection
3 ===========================================================
5 Intel Gigabit Linux driver.
6 Copyright(c) 1999 - 2013 Intel Corporation.
11 - Identifying Your Adapter
12 - Command Line Parameters
13 - Speed and Duplex Configuration
14 - Additional Configurations
17 Identifying Your Adapter
18 ========================
20 For more information on how to identify your adapter, go to the Adapter &
23 http://support.intel.com/support/go/network/adapter/idguide.htm
25 For the latest Intel network drivers for Linux, refer to the following
26 website. In the search field, enter your adapter name or type, or use the
27 networking link on the left to search for your adapter:
29 http://support.intel.com/support/go/network/adapter/home.htm
31 Command Line Parameters
32 =======================
34 The default value for each parameter is generally the recommended setting,
35 unless otherwise noted.
38 For more information about the AutoNeg, Duplex, and Speed
39 parameters, see the "Speed and Duplex Configuration" section in
42 For more information about the InterruptThrottleRate,
43 RxIntDelay, TxIntDelay, RxAbsIntDelay, and TxAbsIntDelay
44 parameters, see the application note at:
45 http://www.intel.com/design/network/applnots/ap450.htm
50 (Supported only on adapters with copper connections)
52 :Valid Range: 0x01-0x0F, 0x20-0x2F
55 This parameter is a bit-mask that specifies the speed and duplex settings
56 advertised by the adapter. When this parameter is used, the Speed and
57 Duplex parameters must not be specified.
60 Refer to the Speed and Duplex section of this readme for more
61 information on the AutoNeg parameter.
66 (Supported only on adapters with copper connections)
68 :Valid Range: 0-2 (0=auto-negotiate, 1=half, 2=full)
71 This defines the direction in which data is allowed to flow. Can be
72 either one or two-directional. If both Duplex and the link partner are
73 set to auto-negotiate, the board auto-detects the correct duplex. If the
74 link partner is forced (either full or half), Duplex defaults to half-
80 :Valid Range: 0-3 (0=none, 1=Rx only, 2=Tx only, 3=Rx&Tx)
81 :Default Value: Reads flow control settings from the EEPROM
83 This parameter controls the automatic generation(Tx) and response(Rx)
84 to Ethernet PAUSE frames.
89 (not supported on Intel(R) 82542, 82543 or 82544-based adapters)
92 0,1,3,4,100-100000 (0=off, 1=dynamic, 3=dynamic conservative,
93 4=simplified balancing)
96 The driver can limit the amount of interrupts per second that the adapter
97 will generate for incoming packets. It does this by writing a value to the
98 adapter that is based on the maximum amount of interrupts that the adapter
99 will generate per second.
101 Setting InterruptThrottleRate to a value greater or equal to 100
102 will program the adapter to send out a maximum of that many interrupts
103 per second, even if more packets have come in. This reduces interrupt
104 load on the system and can lower CPU utilization under heavy load,
105 but will increase latency as packets are not processed as quickly.
107 The default behaviour of the driver previously assumed a static
108 InterruptThrottleRate value of 8000, providing a good fallback value for
109 all traffic types,but lacking in small packet performance and latency.
110 The hardware can handle many more small packets per second however, and
111 for this reason an adaptive interrupt moderation algorithm was implemented.
113 Since 7.3.x, the driver has two adaptive modes (setting 1 or 3) in which
114 it dynamically adjusts the InterruptThrottleRate value based on the traffic
115 that it receives. After determining the type of incoming traffic in the last
116 timeframe, it will adjust the InterruptThrottleRate to an appropriate value
119 The algorithm classifies the incoming traffic every interval into
120 classes. Once the class is determined, the InterruptThrottleRate value is
121 adjusted to suit that traffic type the best. There are three classes defined:
122 "Bulk traffic", for large amounts of packets of normal size; "Low latency",
123 for small amounts of traffic and/or a significant percentage of small
124 packets; and "Lowest latency", for almost completely small packets or
127 In dynamic conservative mode, the InterruptThrottleRate value is set to 4000
128 for traffic that falls in class "Bulk traffic". If traffic falls in the "Low
129 latency" or "Lowest latency" class, the InterruptThrottleRate is increased
130 stepwise to 20000. This default mode is suitable for most applications.
132 For situations where low latency is vital such as cluster or
133 grid computing, the algorithm can reduce latency even more when
134 InterruptThrottleRate is set to mode 1. In this mode, which operates
135 the same as mode 3, the InterruptThrottleRate will be increased stepwise to
136 70000 for traffic in class "Lowest latency".
138 In simplified mode the interrupt rate is based on the ratio of TX and
139 RX traffic. If the bytes per second rate is approximately equal, the
140 interrupt rate will drop as low as 2000 interrupts per second. If the
141 traffic is mostly transmit or mostly receive, the interrupt rate could
144 Setting InterruptThrottleRate to 0 turns off any interrupt moderation
145 and may improve small packet latency, but is generally not suitable
146 for bulk throughput traffic.
149 InterruptThrottleRate takes precedence over the TxAbsIntDelay and
150 RxAbsIntDelay parameters. In other words, minimizing the receive
151 and/or transmit absolute delays does not force the controller to
152 generate more interrupts than what the Interrupt Throttle Rate
156 If you are using the Intel(R) PRO/1000 CT Network Connection
157 (controller 82547), setting InterruptThrottleRate to a value
158 greater than 75,000, may hang (stop transmitting) adapters
159 under certain network conditions. If this occurs a NETDEV
160 WATCHDOG message is logged in the system event log. In
161 addition, the controller is automatically reset, restoring
162 the network connection. To eliminate the potential for the
163 hang, ensure that InterruptThrottleRate is set no greater
164 than 75,000 and is not set to 0.
167 When e1000 is loaded with default settings and multiple adapters
168 are in use simultaneously, the CPU utilization may increase non-
169 linearly. In order to limit the CPU utilization without impacting
170 the overall throughput, we recommend that you load the driver as
173 modprobe e1000 InterruptThrottleRate=3000,3000,3000
175 This sets the InterruptThrottleRate to 3000 interrupts/sec for
176 the first, second, and third instances of the driver. The range
177 of 2000 to 3000 interrupts per second works on a majority of
178 systems and is a good starting point, but the optimal value will
179 be platform-specific. If CPU utilization is not a concern, use
180 RX_POLLING (NAPI) and default driver settings.
186 - 48-256 for 82542 and 82543-based adapters
187 - 48-4096 for all other supported adapters
190 This value specifies the number of receive buffer descriptors allocated
191 by the driver. Increasing this value allows the driver to buffer more
192 incoming packets, at the expense of increased system memory utilization.
194 Each descriptor is 16 bytes. A receive buffer is also allocated for each
195 descriptor and can be either 2048, 4096, 8192, or 16384 bytes, depending
196 on the MTU setting. The maximum MTU size is 16110.
199 MTU designates the frame size. It only needs to be set for Jumbo
200 Frames. Depending on the available system resources, the request
201 for a higher number of receive descriptors may be denied. In this
202 case, use a lower number.
207 :Valid Range: 0-65535 (0=off)
210 This value delays the generation of receive interrupts in units of 1.024
211 microseconds. Receive interrupt reduction can improve CPU efficiency if
212 properly tuned for specific network traffic. Increasing this value adds
213 extra latency to frame reception and can end up decreasing the throughput
214 of TCP traffic. If the system is reporting dropped receives, this value
215 may be set too high, causing the driver to run out of available receive
219 When setting RxIntDelay to a value other than 0, adapters may
220 hang (stop transmitting) under certain network conditions. If
221 this occurs a NETDEV WATCHDOG message is logged in the system
222 event log. In addition, the controller is automatically reset,
223 restoring the network connection. To eliminate the potential
224 for the hang ensure that RxIntDelay is set to 0.
229 (This parameter is supported only on 82540, 82545 and later adapters.)
231 :Valid Range: 0-65535 (0=off)
234 This value, in units of 1.024 microseconds, limits the delay in which a
235 receive interrupt is generated. Useful only if RxIntDelay is non-zero,
236 this value ensures that an interrupt is generated after the initial
237 packet is received within the set amount of time. Proper tuning,
238 along with RxIntDelay, may improve traffic throughput in specific network
244 (This parameter is supported only on adapters with copper connections.)
246 :Valid Settings: 0, 10, 100, 1000
247 :Default Value: 0 (auto-negotiate at all supported speeds)
249 Speed forces the line speed to the specified value in megabits per second
250 (Mbps). If this parameter is not specified or is set to 0 and the link
251 partner is set to auto-negotiate, the board will auto-detect the correct
252 speed. Duplex should also be set when Speed is set to either 10 or 100.
258 - 48-256 for 82542 and 82543-based adapters
259 - 48-4096 for all other supported adapters
262 This value is the number of transmit descriptors allocated by the driver.
263 Increasing this value allows the driver to queue more transmits. Each
264 descriptor is 16 bytes.
267 Depending on the available system resources, the request for a
268 higher number of transmit descriptors may be denied. In this case,
274 :Valid Range: 0-65535 (0=off)
277 This value delays the generation of transmit interrupts in units of
278 1.024 microseconds. Transmit interrupt reduction can improve CPU
279 efficiency if properly tuned for specific network traffic. If the
280 system is reporting dropped transmits, this value may be set too high
281 causing the driver to run out of available transmit descriptors.
286 (This parameter is supported only on 82540, 82545 and later adapters.)
288 :Valid Range: 0-65535 (0=off)
291 This value, in units of 1.024 microseconds, limits the delay in which a
292 transmit interrupt is generated. Useful only if TxIntDelay is non-zero,
293 this value ensures that an interrupt is generated after the initial
294 packet is sent on the wire within the set amount of time. Proper tuning,
295 along with TxIntDelay, may improve traffic throughput in specific
301 (This parameter is NOT supported on the 82542-based adapter.)
306 A value of '1' indicates that the driver should enable IP checksum
307 offload for received packets (both UDP and TCP) to the adapter hardware.
312 :Valid Range: 0-xxxxxxx (0=off)
314 :Usage: modprobe e1000.ko copybreak=128
316 Driver copies all packets below or equaling this size to a fresh RX
317 buffer before handing it up the stack.
319 This parameter is different than other parameters, in that it is a
320 single (not 1,1,1 etc.) parameter applied to all driver instances and
321 it is also available during runtime at
322 /sys/module/e1000/parameters/copybreak
328 :Default Value: 0 (disabled)
330 Allows PHY to turn off in lower power states. The user can turn off
331 this parameter in supported chipsets.
333 Speed and Duplex Configuration
334 ==============================
336 Three keywords are used to control the speed and duplex configuration.
337 These keywords are Speed, Duplex, and AutoNeg.
339 If the board uses a fiber interface, these keywords are ignored, and the
340 fiber interface board only links at 1000 Mbps full-duplex.
342 For copper-based boards, the keywords interact as follows:
344 - The default operation is auto-negotiate. The board advertises all
345 supported speed and duplex combinations, and it links at the highest
346 common speed and duplex mode IF the link partner is set to auto-negotiate.
348 - If Speed = 1000, limited auto-negotiation is enabled and only 1000 Mbps
349 is advertised (The 1000BaseT spec requires auto-negotiation.)
351 - If Speed = 10 or 100, then both Speed and Duplex should be set. Auto-
352 negotiation is disabled, and the AutoNeg parameter is ignored. Partner
353 SHOULD also be forced.
355 The AutoNeg parameter is used when more control is required over the
356 auto-negotiation process. It should be used when you wish to control which
357 speed and duplex combinations are advertised during the auto-negotiation
360 The parameter may be specified as either a decimal or hexadecimal value as
361 determined by the bitmap below.
363 ============== ====== ====== ======= ======= ====== ====== ======= ======
364 Bit position 7 6 5 4 3 2 1 0
365 Decimal Value 128 64 32 16 8 4 2 1
366 Hex value 80 40 20 10 8 4 2 1
367 Speed (Mbps) N/A N/A 1000 N/A 100 100 10 10
368 Duplex Full Full Half Full Half
369 ============== ====== ====== ======= ======= ====== ====== ======= ======
371 Some examples of using AutoNeg::
373 modprobe e1000 AutoNeg=0x01 (Restricts autonegotiation to 10 Half)
374 modprobe e1000 AutoNeg=1 (Same as above)
375 modprobe e1000 AutoNeg=0x02 (Restricts autonegotiation to 10 Full)
376 modprobe e1000 AutoNeg=0x03 (Restricts autonegotiation to 10 Half or 10 Full)
377 modprobe e1000 AutoNeg=0x04 (Restricts autonegotiation to 100 Half)
378 modprobe e1000 AutoNeg=0x05 (Restricts autonegotiation to 10 Half or 100
380 modprobe e1000 AutoNeg=0x020 (Restricts autonegotiation to 1000 Full)
381 modprobe e1000 AutoNeg=32 (Same as above)
383 Note that when this parameter is used, Speed and Duplex must not be specified.
385 If the link partner is forced to a specific speed and duplex, then this
386 parameter should not be used. Instead, use the Speed and Duplex parameters
387 previously mentioned to force the adapter to the same speed and duplex.
389 Additional Configurations
390 =========================
395 Jumbo Frames support is enabled by changing the MTU to a value larger than
396 the default of 1500. Use the ifconfig command to increase the MTU size.
399 ifconfig eth<x> mtu 9000 up
401 This setting is not saved across reboots. It can be made permanent if
406 to the file /etc/sysconfig/network-scripts/ifcfg-eth<x>. This example
407 applies to the Red Hat distributions; other distributions may store this
408 setting in a different location.
411 Degradation in throughput performance may be observed in some Jumbo frames
412 environments. If this is observed, increasing the application's socket buffer
413 size and/or increasing the /proc/sys/net/ipv4/tcp_*mem entry values may help.
414 See the specific application manual and /usr/src/linux*/Documentation/
415 networking/ip-sysctl.txt for more details.
417 - The maximum MTU setting for Jumbo Frames is 16110. This value coincides
418 with the maximum Jumbo Frames size of 16128.
420 - Using Jumbo frames at 10 or 100 Mbps is not supported and may result in
421 poor performance or loss of link.
423 - Adapters based on the Intel(R) 82542 and 82573V/E controller do not
424 support Jumbo Frames. These correspond to the following product names::
426 Intel(R) PRO/1000 Gigabit Server Adapter
427 Intel(R) PRO/1000 PM Network Connection
432 The driver utilizes the ethtool interface for driver configuration and
433 diagnostics, as well as displaying statistical information. The ethtool
434 version 1.6 or later is required for this functionality.
436 The latest release of ethtool can be found from
437 https://www.kernel.org/pub/software/network/ethtool/
439 Enabling Wake on LAN* (WoL)
440 ---------------------------
442 WoL is configured through the ethtool* utility.
444 WoL will be enabled on the system during the next shut down or reboot.
445 For this driver version, in order to enable WoL, the e1000 driver must be
446 loaded when shutting down or rebooting the system.
451 For general information, go to the Intel support website at:
453 http://support.intel.com
455 or the Intel Wired Networking project hosted by Sourceforge at:
457 http://sourceforge.net/projects/e1000
459 If an issue is identified with the released source code on the supported
460 kernel with a supported adapter, email the specific information related
461 to the issue to e1000-devel@lists.sf.net