TCP Chimney Offload
TCP chimney offload enables Windows to offload all TCP processing for a connection to a network adapter (with proper driver support). Offloads are initiated on a per-connection basis and reduce networking-related CPU overhead, theoretically enabling better overall system performance by freeing up CPU time for other tasks. TCP Chimney Offload controls what is referred to as the “TCP Offload Engine” (TOE) in Windows in general. It is a global setting that has to be enabled for many of the other offloads to work. Enabling this setting had some negative effects in the past because of buggy network adapter drivers, however its implementation has gotten much better with time. It is useful for CPU-bound client computers and very fast broadband connections, not recommended in some server environments.
Default: automatic
Recommended: disabled
To check the setting:
Get-NetOffloadGlobalSetting
netsh int tcp show global
To disable TCP Chimney Offload using netsh:
netsh int tcp set global chimney=disabled
To set using Powershell cmdlets:
Set-NetOffloadGlobalSetting -Chimney Disabled (recommended: disabled*)
Notes:
* There were issues with buggy Network adapter drivers and the implementation of offloading with older OSes (Windows 2003 Server), however, this has improved over time. One should be more careful using offloading in server environments, as there have been some reports of issues with TCP Chimney Offload and SQL servers under heavy load, affecting both application concurrency and throughput. See this MSDN reference and MS KB942861.
Chimney Offload does not work together with NetDMA (mute point with Windows 8/8.1 as NetDMA is not supported).
Chimney Offload may not be supported with NAT/ICS (reference: Technet TCP Chimney Offload Compatibility for Windows 2008 R2). However, this only applies if NAT/ICS is enabled on the computer running Windows itself, and that is rarely the case.
Setting Chimney Offload to disabled is recommended for VMWare servers, and the setting is now considered deprecated by Microsoft. See: Obsolete RFCs and Overview of TCP Timers
Setting MTU
netsh int ipv4 show subinterface
or, if you are using IP version 6:
netsh int ipv6 show subinterface
Alternative netsh syntax: netsh int ipv4 show interfaces
Alternative PowerShell syntax: Get-NetAdapter (to view all advanced properties: Get-NetAdapterAdvancedProperty)
netsh int ipv4 set subinterface “network interface name” mtu=#### store=persistent
Where “network interface name” is your specific network adapter name as obtained above (also viewable under Network adapters), and mtu=#### is the desired MTU value. For example, if the name of your network card is “Wi-Fi” and you’d like to set its MTU to 1500, you’d have to type:
netsh int ipv4 set subinterface “Wi-Fi” mtu=1500 store=persistent
Notes:
For IPv6, you can set the MTU using: netsh int ipv6 set subinterface “Wi-Fi” mtu=1500 store=persistent
The maximum MTU value is usually 1500, and up to 1492 for PPPoE connections.
The PowerShell cmdlet using Set-NetAdapterAdvancedProperty SetMTU doesn’t seem to work yet. See this nice attempt to set MTU using PowerShell at Serverfault.
Checksum Offload
The PowerShell Enable-NEtAdapterChecksumOffload cmdlet enables checksum offloads on the network adapter. Checksum offloading is also required for some other stateless offloads to work, including Receive Side Scaling (RSS), Receive Segment Coalescing (RSC), and Large Send Offload (LSO).
Default state: adapter-dependent
Recommended: enabled
Enable-NetAdapterChecksumOffload -Name * (use to enable checksum offload on all adapters that support it)
Disable-NetAdapterChecksumOffload -Name * (use to disable checksum offload for all network adapters)
Get-NetAdapterChecksumOffload (use to view network adapters that support checksum offload and their state)
Registry Tweaks
DefaultTTL
TTL does not directly affect speed, and can be safely left alone in many cases. It is a limit to the time and number of hops/routers a packet will travel before being discarded. A number that’s too small risks packets being discarded before reaching their destination. A number that’s too large (over 128) will cause delay in when lost IP packets are discarded. The default, when the setting is not present in the Windows 8/10/2012 registry the system assumes a value of 128.
(DWORD, not present by default denoting 128, valid range is 1-255, recommended: 64)
Host Resolution Priority Tweak
LocalPriority = 4 (DWORD, default 499, recommended 4)
HostPriority = 5 (DWORD, default 500, recommended 5)
DnsPriority = 6 (DWORD, default 2000, recommended 6)
NetbtPriority = 7 (DWORD, default 2001, recommended 7)
MaxUserPort and TCPTimedWaitDelay (port allocation)
Short lived (ephemeral) TCP/IP ports above 1024 are allocated as needed by the OS. The Windows 8/2012 defaults are usually sufficient under normal network load. However, under heavy network load it may be necessary to adjust these two registry settings to increase port availability and decrease the time to wait before reclaiming unused ports.
QoS Reserved Bandwidth
(DWORD, not present in the registry by default. When not present it reserves 20% of bandwidth. Recommended: 0 , possible values between 0 and 100) – indicates the percentage value of reserved bandwidth for QoS applications. Set to 0 to disable.
QoS Policy
QoS Policy settings under Windows 7/8 can be edited using the Group Policy Editor (gpedit.msc): Computer Configuration -> Windows Settings -> Policy-based QoS
(you may have to create the QoS key)
“Do not use NLA”=”1” (REG_SZ string value, not DWORD, not present by default, recommended: 1 if you plan to edit DSCP values via gpedit.msc)
Notes:
gpedit.msc is only provided with the Professional/Ultimate and Server Windows variants.
This registry key may also be needed in systems with multiple network adapters present.
Network Memory Allocation (Event ID 2017 error)
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Session Manager\Memory Management
(DWORD, default value: 0, recommended value: 1 for LAN file transfers, 0 for gaming)
A value of zero establishes a cache of ~8 MB, a value of 1 allows the cache to expand up to the amount of physical memory minus 4 MB, as necessary. Reportedly some ATI video card drivers may have an issue with corrupt cache and degraded application performance when this is enabled, so we recommend to leave it off/zero for gaming.
(DWORD, default value: 1, recommended value: 3)
minimizes used memory
balance used memory
optimal setting for file sharing and network applications
Internet Explorer Web Browser Tweaks
MaxConnectionsPerServer / MaxConnectionsPer1_OServer Internet Explorer browser tweaks adjust the maximum number of concurrent connections your web browser can establish to any single website.
(DWORD, default: 4, recommended: 8, valid range: 2-128)
(DWORD, default: 2, recommended: 8, valid range: 2-128)
In addition to explorer.exe, you can also add a 32bit DWORD “iexplore.exe” (according to MSDN), and set it to the same corresponding value as “explorer.exe”
(DWORD, not present by default, recommended: 8, valid range: 2-128)
(DWORD, not present by default, recommended: 8, valid range: 2-128)
Note: Increasing this setting much over 10 can cause incomplete page loading and other issues with some pages.
Network Throttling Index (Gaming)
By default, Windows 8 continues to implement a network throttling mechanism to restrict the processing of non-multimedia network traffic to 10 packets per millisecond (a bit over 100 Mbits/second). The idea behind such throttling is that processing of network packets can be a resource-intensive task, and it may need to be throttled to give prioritized CPU access to multimedia programs. In some cases, such as Gigabit networks and some online games, for example, it is beneficial to turn off such throttling all together for achieving maximum throughput.
(DWORD, default: 10, recommended: 10 for media sharing, ffffffff for gaming and max throughput, valid range: 1 through 70 decimal or ffffffff to completely disable throttling)
It is only recommended to change this setting in saturated Gigabit LAN environments, where you do not want to give priority to multimedia playback. Reportedly, disabling throttling by using ffffffff can also help reduce ping spikes in some online games. Games that may be affected by this throttling: Source Engine games (TF2, Left 4 Dead, CS:S), HoN, CoD, Overlord series.
SystemResponsiveness (Gaming and Multimedia)
In Windows 8/8.1, just like with Windows 7, multimedia applications use the “Multimedia Class Scheduler” service (MMCSS) to ensure priritized access to CPU resources, without denying CPU resources to lower-priority background applications. However, this also reserves 20% of CPU by default for background processes, your multimedia streaming and some games can only utilize up to 80% of the CPU. This setting, in combination with the above “NetworkThrottlingIndex” can help some games and video streaming. We recommend reducing the reserved CPU for background processes from the default of 20%.
(DWORD, default is 20 denoting 20% of CPU reserved, recommended: decimal 10, or 0 for pure gaming/multimedia performance)
Notes: The number in this key is rounded by MMCSS to the nearest 10. In some server operating systems (Windows 2008 Server), the SystemResponsiveness may be set to 100, instead of 20 by default. This is by design, giving higher priority to background services over multimedia.
See also: MSDN ms684247
Disable Nagle’s algorithm (Gaming)
In the same location, add a new DWORD value:
(DWORD value, not present by default, 0 to enable Nagle’s algorithm, 1 to disable)
For Server Operating Systems that have Microsoft Message Queuing (MSMQ) installed, or if you have the MSMQ registry hive present, also add TCPNoDelay to:
(DWORD, not present by default, 0 to enable Nagle’s algorithm, 1 to disable)
See also: Gaming Tweaks article
SG TCP Optimizer (version 4 Beta)
The SG TCP Optimizer is a free program that allows for easy tweaking of all the above TCP/IP settings for broadband with a couple of clicks. Note that only version 4 of the prorgam supports Windows 8 (and newer) and works with PowerShell cmdlets. There program is available in our downloads section.
References
Technet – Network Adapter Cmdlets
TechNet set-NetTCPSetting
MSDN MSFT_NetTCPSetting class
Increasing TCP Initial congestion Window in Windows 2008 Server R2
MS Hotfix for netsh errors in Windows Server 2008 R2 – MSKB 2472264
See Also
Quality of Service – ToS DSCP WMM
Windows 8 FAQs
Windows 7, Vista, 2008 Tweaks
Firefox / Internet Explorer Browser Tweaks
Congestion Control Algorithm Comparison
Notes
Some templates/settings, as well as changing the default NetTransportFilter are no longer editable in client Windows versions (Windows 8, 8.1), only Server variants (Server 2008, Server 2012/2012 R2).
The tweaks were tested and confirmed to work with Windows 10 Technical Preview build 10041.
Feedback and suggestions are always appreciated.
Receive-Side Scaling State (RSS)
The receive-side scaling setting enables parallelized processing of received packets on multiple processors, while avoiding packet reordering. It avoids packet reordering by separating packets into “flows”, and using a single processor for processing all the packets for a given flow. Packets are separated into flows by computing a hash value based on specific fields in each packet, and the resulting hash values are used to select a processor for processing the flow. This approach ensures that all packets belonging to a given TCP connection will be queued to the same processor, in the same order that they were received by the network adapter.
Default: enabled
Recommended: enabled (provided you have 2 or more processor cores and a NIC that can handle RSS)
To check current RSS setting:
Get-NetOffloadGlobalSetting (view global setting)
Get-NetAdapterRss -Name * (use cmdlet to view RSS capable Network Adapters)
netsh int tcp show global
To change RSS using netsh:
netsh int tcp set global rss=enabled
To change using PowerShell cmdlets:
Enable-NetAdapterRss -Name * (use this to enable RSS for all adapters that support it)
Disable-NetAdapterRss -Name * (use this to disables RSS for all adapters)
Set-NetOffloadGlobalSetting -ReceiveSideScaling Enabled (alternate global command for all adapters)
Notes:
Needs Checksum Offload to be enabled. Only supported by some network adapters.
NetDMA State
NetDMA (TCPA) enables support for advanced direct memory access. In essence, it provides the ability to more efficiently move network data by minimizing CPU usage. NetDMA frees the CPU from handling memory data transfers between network card data buffers and application buffers by using a DMA engine.
Default: disabled
Recommended: leave alone, don’t bother (setting not supported in Windows 8 and later, according to MS)
You can still edit it using:
netsh int tcp set global netdma=disabled
Notes:
According to Microsoft, NetDMA is not supported in Windows 8/8.1 and changing this setting will have no effect. MSDN NetDMA
NetDMA (TCPA) does not work together with Chimney Offload either, you have to choose one or the other. For NetDMA to work, it must be enabled/supported by your BIOS and your CPU must support Intel I/O Acceleration Technology (I/OAT).
Direct Cache Access (DCA)
Direct Cache Access (DCA) allows a capable I/O device, such as a network controller, to deliver data directly into a CPU cache. The objective of DCA is to reduce memory latency and the memory bandwidth requirement in high bandwidth (Gigabit) environments. DCA requires support from the I/O device, system chipset, and CPU(s).
Possible states are: enabled, disabled
Default state: disabled
Recommended: enabled (provided the CPU/Chipset/NIC support it)
To set DCA using netsh:
netsh int tcp set global dca=enabled
Note: The impact of DCA is more significant with older CPUs. Not present in Windows 10 Creators’ update.
Add-On Congestion Control Provider (CTCP)
The traditional slow-start and congestion avoidance algorithms in TCP help avoid network congestion by gradually increasing the TCP window at the beginning of transfers until the TCP Receive Window boundary is reached, or packet loss occurs. For broadband internet connections that combine high TCP Window with higher latency (high BDP), these algorithms do not increase the TCP windows fast enough to fully utilize the bandwidth of the connection.
Compound TCP (CTCP) is a newer congestion control method that increases the TCP Send Window more aggressively for broadband connections (with large RWIN and BDP). CTCP attempts to maximize throughput by monitoring delay variations and packet loss. It also ensures that its behavior does not impact other TCP connections negatively. CTCP is available in Windows 8/10 and Server editions. Using CTCP can significantly increase throughput and packet loss recovery.
Possible settings are: none, ctcp, dctcp (Datacenter TCP), also CUBIC and New-Reno in newer Windows 10 variants.
CTCP – Compound TCP increases the receive window and amount of data sent. It improves throughput on higher latency/broadband internet connections.
DCTCP – Data Center TCP adjusts the TCP Window based on network congestion feedback based on Explicit Congestion Notification (ECN) signaling, it is designed to improve throughput on low latency/local links.
CUBIC – New default setting as of Windows 10 Creators update, default in Linux kernels 2.6 through 3.2. Uses a cubic TCP congestion window growth function. The algorithm uses the amount of time since the last congestion event (instead of ACK timing) to advance the TCP congestion window. It is designed for high-speed TCP transfers. The only downside is that it can possibly cause some bufferbloat in large BDP networks (like the Internet) with unmanaged queues, such as ADSL and LTE. Theoretically it performs as well as CTCP.
NewReno – RFC 6582 (old RFC 3782) implementation for fast retransmit and fast recovery algorithm. There is an issue with more than 3 reordered packets where New Reno enters fast recovery. Older congestion control algorithm, not recommended.
Unfortunately, the PowerShell cmdlet allows for changing this only in some Windows builds, and the netsh is deprecated and may be bugged in some Windows 10 builds, so it could be challenging to change the congestion provider. Yay for Microsoft! Fortunately, the default setting is already CTCP for earlier Windows 10 builds, and CUBIC for newer ones (as per the PowerShell Get-NetTCPSetting cmdlet.
Default setting: CUBIC or CTCP ** (netsh incorrectly shows “none” by default in Windows 8 and early Windows 10 builds, while PowerShell Get-NetTCPSetting shows “CTCP” or “CUBIC” depending on build)
Recommended setting: CTCP for gaming/latency/speed if network congestion/packet loss is expected, CUBIC for pure throughput on stable networks with no congestion/packet loss
To check the current Congestion provider, use:
To change using PowerShell cmdlet (does not work in some older Windows builds):
Set-NetTCPSetting -SettingName InternetCustom -CongestionProvider CTCP
(“InternetCustom” template is for Windows 8.1 and 10, “Custom” for Windows 8, “Internet” for newer Windows builds since “Creators Update”. Above command may may be read-only in some Windows versions, use the alternate netsh command instead)
To change using netsh in elevated command prompt:
netsh int tcp set supplemental Internet congestionprovider=ctcp
Notes: For Windows 10 Creators Update and later, it is also possible to set congestionprovider=cubic. The “Internet” template is for later Windows 10 builds, for Windows 8 it may be “InternetCustom”.
Notes:
Supposedly the netsh congestionprovider setting is deprecated, and one should use the PowerShell cmdlets when possible. Changing CTCP directly with netsh is not possible by default under Windows 8, the commands are still listed below for reference, and other OSes:
netsh int tcp show supplemental (see the currently used supplemental template)
netsh int tcp set supplemental custom 300 10 ctcp disabled 50 (minrto=300ms, initial congestion window icw=10 packets, congestionprovider=ctcp, enablecwndrestart=disabled, delayedacktimeout=100 ms) netsh int tcp set supplemental custom (tell Windows to use the custom template)
1. Open notepad and copy this code into it:
Windows Registry Editor Version 5.00
[HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Nsi\{eb004a03-9b1a-11d4-9123-0050047759bc}\0] “0200”=hex:00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,\
00,00,00,00,00,00,02,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,\
00,00,00,ff,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,\
ff,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00
“1700”=hex:00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,\
00,00,00,00,00,00,02,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,\
00,00,00,ff,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,\
ff,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00
Here is a screenshot from Windows 8.1 after merging the above with the registry and rebooting:
You can also just copy all the text to the appropriate registry hive directly. Merging the above with the registry and rebooting will show CTCP as the addon congestion control algorithm. Still, as per Microsoft, netsh is deprecated and you should be using Powershell.
Note that the above registry hack will change not only CTCP, but other related parameters as well, here is some additional info:
The binary values above affect many parameters: “Chimney Offload State” (offset 04, 00=disabled,01=enabled,02=auto), RSS (offset 05, 00=default/enabled,01=disabled), ECN Capability (offset 06, 00=disabled/default, 01=enabled), CTCP (offset 28, 00=none, 02 = ctcp), etc.
See Also: Windows 10 Manual Registry Tweaks
Windows PowerShell
The Windows PowerShell is a tool developed by Microsoft designed to replace and extend the functionality of the command prompt. It allows you to run all DOS commands similarly to command prompt, however, it also provides additional functionality in the form of many customizable commands called cmdlets. We will be using PowerShell to tweak TCP/IP settings, as some of the parameters are no longer tweakable using command prompt netsh in Windows 8 and newer OSes.
The PowerShell commands below are an overview of what we can use to tweak TCP/IP settings under Windows 10:
Get-NetTCPSetting – this can be used to view TCP parameters
Set-NetTCPSetting – this can be used to set TCP parameters
netsh int tcp show global – you can still see and change some (not all) parameters using netsh
There are a few different networking profiles called “templates” in Windows 8 PowerShell under Get-NetTCPSetting.
In Windows 8 (and 2012 Server): Automatic, Internet, Datacenter, Compat
In Windows 8.1 (also Windows 10, Windows 2012 Server R2): Automatic, Internet, InternetCustom, Datacenter, DataCenterCustom, Compat
Theoretically, only one of the above templates can be modified by default – “Custom” under Windows 8, “InternetCustom” under Windows 8.1 (and Windows 10). The “DatacenterCustom” template can only be modified if Windows is installed on a server. If you try to modify one of the other templates, you will get an error, something like:
Set-NetTCPSetting : Only the custom templates InternetCustom and DatacenterCustom can be modified under Windows 8 and older Windows 10 versions.
Even though theoretically only the “Custom” templates can be modified, many of the commands below (both netsh and PowerShell TCP cmdlets) are global and modify all templates simultaneously.
By default, the “Internet” template/profile is applied to TCP connections. To find the currently used template
or, using netsh:
netsh int tcp show supplemental
To view the current template settings, use:
Get-NetTCPSetting -SettingName “Internet” (or your current template name)
See also: MS TechNet Set-NetTCPSetting
Netsh
It is still possible to check the current status of Windows TCP/IP parameters with netsh, both in PowerShell and elevated command prompt. We recommend using the PowerShell for changes. While most settings can still be changed using the “netsh” tool, others require the Windows PowerShell interface and cmdlets. Netsh (and the command prompt) is being deprecated by Microsoft, and some of the netsh commands are bugged in Windows 8/8.1 and Server 2012/2012 R2.
To see current tcp parameters in netsh, use:
netsh int tcp show global
There are two extra settings in Windows 8.1 and Windows 10 (right image above)
Explanation of all the tweakable TCP/IP parameters, possible ways to change them, as well as their default/recommended states for broadband internet connections are listed below. We recommend using the Windows PowerShell for all changes. The parameters are loosely sorted in order of importance with the parameters that generally have more significant impact listed first.
Checksum Offload
The PowerShell Enable-NEtAdapterChecksumOffload cmdlet enables checksum offloads on the network adapter. Checksum offloading is also required for some other stateless offloads to work, including Receive Side Scaling (RSS), Receive Segment Coalescing (RSC), and Large Send Offload (LSO).
Default state: adapter-dependent
Recommended: enabled
Enable-NetAdapterChecksumOffload -Name * (use to enable checksum offload on all adapters that support it)
Disable-NetAdapterChecksumOffload -Name * (use to disable checksum offload for all network adapters)
Get-NetAdapterChecksumOffload (use to view network adapters that support checksum offload and their state)
Add-On Congestion Control Provider (CTCP)
The traditional slow-start and congestion avoidance algorithms in TCP help avoid network congestion by gradually increasing the TCP window at the beginning of transfers until the TCP Receive Window boundary is reached, or packet loss occurs. For broadband internet connections that combine high TCP Window with higher latency (high BDP), these algorithms do not increase the TCP windows fast enough to fully utilize the bandwidth of the connection.
Compound TCP (CTCP) is a newer congestion control method that increases the TCP Send Window more aggressively for broadband connections (with large RWIN and BDP). CTCP attempts to maximize throughput by monitoring delay variations and packet loss. It also ensures that its behavior does not impact other TCP connections negatively. CTCP is available in Windows 8/10 and Server editions. Using CTCP can significantly increase throughput and packet loss recovery.
Possible settings are: none, ctcp, dctcp (Datacenter TCP), also CUBIC and New-Reno in newer Windows 10 variants.
CTCP – Compound TCP increases the receive window and amount of data sent. It improves throughput on higher latency/broadband internet connections.
DCTCP – Data Center TCP adjusts the TCP Window based on network congestion feedback based on Explicit Congestion Notification (ECN) signaling, it is designed to improve throughput on low latency/local links.
CUBIC – New default setting as of Windows 10 Creators update, default in Linux kernels 2.6 through 3.2. Uses a cubic TCP congestion window growth function. The algorithm uses the amount of time since the last congestion event (instead of ACK timing) to advance the TCP congestion window. It is designed for high-speed TCP transfers. The only downside is that it can possibly cause some bufferbloat in large BDP networks (like the Internet) with unmanaged queues, such as ADSL and LTE. Theoretically it performs as well as CTCP.
NewReno – RFC 6582 (old RFC 3782) implementation for fast retransmit and fast recovery algorithm. There is an issue with more than 3 reordered packets where New Reno enters fast recovery. Older congestion control algorithm, not recommended.
Unfortunately, the PowerShell cmdlet allows for changing this only in some Windows builds, and the netsh is deprecated and may be bugged in some Windows 10 builds, so it could be challenging to change the congestion provider. Yay for Microsoft! Fortunately, the default setting is already CTCP for earlier Windows 10 builds, and CUBIC for newer ones (as per the PowerShell Get-NetTCPSetting cmdlet.
Default setting: CUBIC or CTCP ** (netsh incorrectly shows “none” by default in Windows 8 and early Windows 10 builds, while PowerShell Get-NetTCPSetting shows “CTCP” or “CUBIC” depending on build)
Recommended setting: CTCP for gaming/latency/speed if network congestion/packet loss is expected, CUBIC for pure throughput on stable networks with no congestion/packet loss
To check the current Congestion provider, use:
To change using PowerShell cmdlet (does not work in some older Windows builds):
Set-NetTCPSetting -SettingName InternetCustom -CongestionProvider CTCP
(“InternetCustom” template is for Windows 8.1 and 10, “Custom” for Windows 8, “Internet” for newer Windows builds since “Creators Update”. Above command may may be read-only in some Windows versions, use the alternate netsh command instead)
To change using netsh in elevated command prompt:
netsh int tcp set supplemental Internet congestionprovider=ctcp
Notes: For Windows 10 Creators Update and later, it is also possible to set congestionprovider=cubic. The “Internet” template is for later Windows 10 builds, for Windows 8 it may be “InternetCustom”.
Notes:
Supposedly the netsh congestionprovider setting is deprecated, and one should use the PowerShell cmdlets when possible. Changing CTCP directly with netsh is not possible by default under Windows 8, the commands are still listed below for reference, and other OSes:
netsh int tcp show supplemental (see the currently used supplemental template)
netsh int tcp set supplemental custom 300 10 ctcp disabled 50 (minrto=300ms, initial congestion window icw=10 packets, congestionprovider=ctcp, enablecwndrestart=disabled, delayedacktimeout=100 ms) netsh int tcp set supplemental custom (tell Windows to use the custom template)
1. Open notepad and copy this code into it:
Windows Registry Editor Version 5.00
[HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Nsi\{eb004a03-9b1a-11d4-9123-0050047759bc}\0] “0200”=hex:00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,\
00,00,00,00,00,00,02,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,\
00,00,00,ff,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,\
ff,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00
“1700”=hex:00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,\
00,00,00,00,00,00,02,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,\
00,00,00,ff,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00,\
ff,00,00,00,00,00,00,00,00,00,00,00,00,00,00,00
Here is a screenshot from Windows 8.1 after merging the above with the registry and rebooting:
You can also just copy all the text to the appropriate registry hive directly. Merging the above with the registry and rebooting will show CTCP as the addon congestion control algorithm. Still, as per Microsoft, netsh is deprecated and you should be using Powershell.
Note that the above registry hack will change not only CTCP, but other related parameters as well, here is some additional info:
The binary values above affect many parameters: “Chimney Offload State” (offset 04, 00=disabled,01=enabled,02=auto), RSS (offset 05, 00=default/enabled,01=disabled), ECN Capability (offset 06, 00=disabled/default, 01=enabled), CTCP (offset 28, 00=none, 02 = ctcp), etc.
See Also: Windows 10 Manual Registry Tweaks
Receive-Side Scaling State (RSS)
The receive-side scaling setting enables parallelized processing of received packets on multiple processors, while avoiding packet reordering. It avoids packet reordering by separating packets into “flows”, and using a single processor for processing all the packets for a given flow. Packets are separated into flows by computing a hash value based on specific fields in each packet, and the resulting hash values are used to select a processor for processing the flow. This approach ensures that all packets belonging to a given TCP connection will be queued to the same processor, in the same order that they were received by the network adapter.
Default: enabled
Recommended: enabled (provided you have 2 or more processor cores and a NIC that can handle RSS)
To check current RSS setting:
Get-NetOffloadGlobalSetting (view global setting)
Get-NetAdapterRss -Name * (use cmdlet to view RSS capable Network Adapters)
netsh int tcp show global
To change RSS using netsh:
netsh int tcp set global rss=enabled
To change using PowerShell cmdlets:
Enable-NetAdapterRss -Name * (use this to enable RSS for all adapters that support it)
Disable-NetAdapterRss -Name * (use this to disables RSS for all adapters)
Set-NetOffloadGlobalSetting -ReceiveSideScaling Enabled (alternate global command for all adapters)
Notes:
Needs Checksum Offload to be enabled. Only supported by some network adapters.
Netsh
It is still possible to check the current status of Windows TCP/IP parameters with netsh, both in PowerShell and elevated command prompt. We recommend using the PowerShell for changes. While most settings can still be changed using the “netsh” tool, others require the Windows PowerShell interface and cmdlets. Netsh (and the command prompt) is being deprecated by Microsoft, and some of the netsh commands are bugged in Windows 8/8.1 and Server 2012/2012 R2.
To see current tcp parameters in netsh, use:
netsh int tcp show global
There are two extra settings in Windows 8.1 and Windows 10 (right image above)
Explanation of all the tweakable TCP/IP parameters, possible ways to change them, as well as their default/recommended states for broadband internet connections are listed below. We recommend using the Windows PowerShell for all changes. The parameters are loosely sorted in order of importance with the parameters that generally have more significant impact listed first.
Non Sack Rtt Resiliency (Windows 8
Enables/Disables RTT resiliency for non SACK clients. This can help slow clients/connections as it makes TCP/IP less aggressive in retransmitting packets when enabled.
Possible states: enabled,disabled,default
Default state: disabled
Recommended: “disabled” for stable connections without much latency variations, “enabled” for connections with fluctuating ping and in the presence of packet loss.
To set using netsh:
netsh int tcp set global nonsackrttresiliency=disabled
To set using PowerShell cmdlets:
Set-NetTCPSetting -SettingName InternetCustom -NonSackRttResiliency disabled
Note: This property only supported in Windows 8.1 and Windows Server 2012 R2.
Windows PowerShell
The Windows PowerShell is a tool developed by Microsoft designed to replace and extend the functionality of the command prompt. It allows you to run all DOS commands similarly to command prompt, however, it also provides additional functionality in the form of many customizable commands called cmdlets. We will be using PowerShell to tweak TCP/IP settings, as some of the parameters are no longer tweakable using command prompt netsh in Windows 8 and newer OSes.
The PowerShell commands below are an overview of what we can use to tweak TCP/IP settings under Windows 10:
Get-NetTCPSetting – this can be used to view TCP parameters
Set-NetTCPSetting – this can be used to set TCP parameters
netsh int tcp show global – you can still see and change some (not all) parameters using netsh
There are a few different networking profiles called “templates” in Windows 8 PowerShell under Get-NetTCPSetting.
In Windows 8 (and 2012 Server): Automatic, Internet, Datacenter, Compat
In Windows 8.1 (also Windows 10, Windows 2012 Server R2): Automatic, Internet, InternetCustom, Datacenter, DataCenterCustom, Compat
Theoretically, only one of the above templates can be modified by default – “Custom” under Windows 8, “InternetCustom” under Windows 8.1 (and Windows 10). The “DatacenterCustom” template can only be modified if Windows is installed on a server. If you try to modify one of the other templates, you will get an error, something like:
Set-NetTCPSetting : Only the custom templates InternetCustom and DatacenterCustom can be modified under Windows 8 and older Windows 10 versions.
Even though theoretically only the “Custom” templates can be modified, many of the commands below (both netsh and PowerShell TCP cmdlets) are global and modify all templates simultaneously.
By default, the “Internet” template/profile is applied to TCP connections. To find the currently used template
or, using netsh:
netsh int tcp show supplemental
To view the current template settings, use:
Get-NetTCPSetting -SettingName “Internet” (or your current template name)
See also: MS TechNet Set-NetTCPSetting
Receive Segment Coalescing State (RSC)
Receive Segment Coalescing (RSC) allows the NIC to coalesce multiple TCP/IP packets that arrive within a single interrupt into a single larger packet (up to 64KB) so that the network stack has to process fewer headers, resulting in a 10% to 30% reduction in I/O overhead depending on the workload, thereby improving performance. Receive Segment Coalescing (RCS) is able to collect packets that are received during the same interrupt cycle and put them together so that they can be more efficiently delivered to the network stack. This can significantly increase the amount of traffic that can be handled without severely impacting the CPU.
Possible states: enabled, disabled, default
Default state: enabled in Windows 10, disabled in some older versions.
Recommended: disabled for lower latency and gaming, or when using Wi-Fi adapters. Enable for slightly higher throughput when lower CPU utilization is important. Microsoft recommends disabling it for Web, Mail, or Database Servers, enabling it for File, FTP or Media servers.
To check current RSC state:
netsh int tcp show global (Receive Segment Coalescing State)
Get-NetOffloadGlobalSetting (view global state)
GetNetAdapterRsc -Name * (view RSC setting for all adapters)
To disable using netsh:
netsh int tcp set global rsc=disabled
To disable using PowerShell cmdlets globally:
Set-NetOffloadGlobalSetting -ReceiveSegmentCoalescing disabled
To disable using PowerShell cmdlets for a single adapter:
Disable-NetAdapterRsc -Name * (use to disable RSC for all adapters)
Enable-NetAdapterRsc -Name * (use to enables RSC for all adapters that support it)
Set-NetOffloadGlobalSetting -ReceiveSegmentCoalescing Disabled (alternate syntax for globally changing all adapters)
Notes:
Needs Checksum Offload to be enabled. Only supported by some network adapters.
For gaming, where latency is more important than pure throughput, any type of packet/memory coalescing should be disabled or used very sparingly. Packet/memory coalescing reduces CPU utilization and increases throughput, however, it causes the network adapter to combine packets before interacting with other hardware, which may slightly increase latency. Also see our gaming tweaks article.
Receive Window Auto-Tuning Level
disabled: uses a fixed value for the tcp receive window. Limits it to 64KB (limited at 65535).
highlyrestricted: allows the receive window to grow beyond its default value, very conservatively
restricted: somewhat restricted growth of the tcp receive window beyond its default value
normal: default value, allows the receive window to grow to accommodate most conditions
experimental: allows the receive window to grow to accommodate extreme scenarios (not recommended, it can degrade performance in common scenarios, only intended for research purposes. It enables RWIN values of over 16 MB)
Default setting: normal
Our recommendation: normal (unless you’re experiencing problems)
To modify, in PowerShell (or elevated command prompt) type:
netsh int tcp set global autotuninglevel=normal
(where “normal” is one of the above listed possible states)
Same can also be accomplished using PowerShell cmdlets, just type:
Set-NetTCPSetting -SettingName InternetCustom -AutoTuningLevelLocal Normal
Note: You can also try “highlyrestricted” autotuninglevel for up to 10Mbps connections, as it actually uses a higher unscaled RWIN value (16384 vs. 256 bytes), and still scales up to 262140 bytes.
TCP 1323 Timestamps
RFC 1323 ads many useful options, most notably scaling of the TCP Window. “Timestamps” (TSOpt) is a less commonly used 1323 option that is intended to increase transmission reliability by retransmitting segments that are not acknowledged within some retransmission timeout (RTO) interval. The problem with timestamps is that they add 12 bytes to the 20-byte TCP header of each packet, so turning them on causes considerable overhead.
Possible states: enabled,disabled,default
Default state: disabled
Recommended: disabled
To set using netsh:
netsh int tcp set global timestamps=disabled
To set using PowerShell cmdlets:
Set-NetTCPSetting -SettingName InternetCustom –Timestamps Disabled
Non Sack Rtt Resiliency (Windows 8
Enables/Disables RTT resiliency for non SACK clients. This can help slow clients/connections as it makes TCP/IP less aggressive in retransmitting packets when enabled.
Possible states: enabled,disabled,default
Default state: disabled
Recommended: “disabled” for stable connections without much latency variations, “enabled” for connections with fluctuating ping and in the presence of packet loss.
To set using netsh:
netsh int tcp set global nonsackrttresiliency=disabled
To set using PowerShell cmdlets:
Set-NetTCPSetting -SettingName InternetCustom -NonSackRttResiliency disabled
Note: This property only supported in Windows 8.1 and Windows Server 2012 R2.
ECN Capability
Possible settings are: enabled, disabled, default (restores the state to the system default).
Default state: disabled
Recommendation: “enabled” only for short-lived, interactive connections and HTTP requests with routers that support it, in the presence of congestion/packet loss, “disabled” otherwise (for pure bulk throughput with large TCP Window, no regular congestion/packet loss, or outdated routers without ECN support). May be worth trying “enabled” for gaming with unstable connections.
To change using netsh:
netsh int tcp set global ecncapability=default
To change using PowerShell cmdlets:
Set-NetTCPSetting -SettingName InternetCustom -EcnCapability Disabled
Notes:
Alternative netsh syntax is: netsh int tcp set global ecn=default
ECN is only effective in combination with AQM (Active Queue Management) router policy. It has more noticeable effect on performance with interactive connections, online games, and HTTP requests, in the presence of router congestion/packet loss. Its effect on bulk throughput with large TCP Window are less clear. Currently, we do not recommend enabling this setting, as reportedly it has negative impact on throughput with some residential US ISPs. EA multiplayer games that require a profile logon do not support ECN as well (you will not be able to logon). However, it can also reduce latency in some games with ECN-capable routers in the presence of packet loss (dropped packets).
See also: Wikipedia – ECN, RFC 3168
Max SYN Retransmissions (Windows 8
Sets the number of times to attempt to reestablish a connection with SYN packets.
Possible values: 2 to 8
Default value: 2
Recommended: leave at 2
To set using netsh:
netsh int tsp set global maxsynretransmissions=2
To set using PowerShell cmdlets:
Set-NetTCPSetting -SettingName InternetCustom -MaxSynRetransmissions 2
To view the current setting under PowerShell:
Get-NetTCPSetting -SettingName InternetCustom
Or, to view only the current MaxSynRetransmissions setting as a number:
(Get-NetTCPSetting -SettingName InternetCustom).MaxSynRetransmissions
Note: This property only supported in Windows 8.1 and Windows Server 2012 R2. Windows 7 hotfix displays it in netsh as a read-only setting.
Packet Coalescing
This setting defines the grouping of network packets in general, to limit the number of receive interrupt and reduce the amount of required processing. In network adapters using Network Driver Interface Specification (NDIS) versions 6.3 and later, packet coalescing is used to group both random and multicast traffic for efficiency. This should be left enabled for pure throughput and efficiency, disabled for gaming and where lower latency is desired at the expense of a bit higher CPU utilization, and a bit more multicast traffic. It must be enabled if using RSC (Receive Segment Coealescing).
Possible states: enabled, disabled, default
Default state under Windows 10: enabled
Recommended: disabled (for gaming and slightly lower latency at the expense of higher CPU usage and more multicast traffic, and when using Wi-Fi adapters), enabled (for pure throughput when lower CPU utilization is important)
To check in Powershell: Get-NetOffloadGlobalSetting
Set-NetOffloadGlobalSetting -PacketCoalescingFilter enabled (or disabled)
Receive Segment Coalescing State (RSC)
Receive Segment Coalescing (RSC) allows the NIC to coalesce multiple TCP/IP packets that arrive within a single interrupt into a single larger packet (up to 64KB) so that the network stack has to process fewer headers, resulting in a 10% to 30% reduction in I/O overhead depending on the workload, thereby improving performance. Receive Segment Coalescing (RCS) is able to collect packets that are received during the same interrupt cycle and put them together so that they can be more efficiently delivered to the network stack. This can significantly increase the amount of traffic that can be handled without severely impacting the CPU.
Possible states: enabled, disabled, default
Default state: enabled in Windows 10, disabled in some older versions.
Recommended: disabled for lower latency and gaming, or when using Wi-Fi adapters. Enable for slightly higher throughput when lower CPU utilization is important. Microsoft recommends disabling it for Web, Mail, or Database Servers, enabling it for File, FTP or Media servers.
To check current RSC state:
netsh int tcp show global (Receive Segment Coalescing State)
Get-NetOffloadGlobalSetting (view global state)
GetNetAdapterRsc -Name * (view RSC setting for all adapters)
To disable using netsh:
netsh int tcp set global rsc=disabled
To disable using PowerShell cmdlets globally:
Set-NetOffloadGlobalSetting -ReceiveSegmentCoalescing disabled
To disable using PowerShell cmdlets for a single adapter:
Disable-NetAdapterRsc -Name * (use to disable RSC for all adapters)
Enable-NetAdapterRsc -Name * (use to enables RSC for all adapters that support it)
Set-NetOffloadGlobalSetting -ReceiveSegmentCoalescing Disabled (alternate syntax for globally changing all adapters)
Notes:
Needs Checksum Offload to be enabled. Only supported by some network adapters.
For gaming, where latency is more important than pure throughput, any type of packet/memory coalescing should be disabled or used very sparingly. Packet/memory coalescing reduces CPU utilization and increases throughput, however, it causes the network adapter to combine packets before interacting with other hardware, which may slightly increase latency. Also see our gaming tweaks article.
