Content deleted Content added
m Task 18 (cosmetic): eval 21 templates: hyphenate params (14×); |
GreenC bot (talk | contribs) Rescued 2 archive links. Wayback Medic 2.5 per WP:URLREQ#anandtech.com |
||
(45 intermediate revisions by 26 users not shown) | |||
Line 1:
{{Short description|Computer memory architecture}}
{{multiple issues|
{{confusing |date=October 2011}}
{{refimprove|date=January 2014}}
{{expand German|Dual Channel|date=March 2018}}
}}
In the fields of [[digital electronics]] and [[computer hardware]], '''multi-channel memory architecture''' is a technology that increases the data transfer rate between the [[DRAM]] memory and the [[memory controller]] by adding more channels of communication between them. Theoretically, this multiplies the data rate by exactly the number of channels present. Dual-channel memory employs two channels. The technique goes back as far as the 1960s having been used in [[IBM System/360 Model 91]] and in [[CDC 6600]].<ref name="JacobNg2007">{{cite book| first1=Bruce | last1 = Jacob | first2 = Spencer | last2 = Ng | first3=David | last3 = Wang|title= Memory systems: cache, DRAM, disk|year = 2007 | publisher = Morgan Kaufmann|isbn=978-0-12-379751-3|page= 318}}</ref>
Modern high-end desktop and workstation processors
== Dual-channel architecture ==
Line 15 ⟶ 16:
=== Operation ===
Dual-channel architecture requires a dual-channel-capable motherboard and two or more
* Capacity (e.g. 1024 MB). Certain Intel chipsets support different capacity chips in what they call Flex Mode: the capacity that can be matched is run in dual-channel, while the remainder runs in single-channel.
* Speed (e.g. PC5300). If speed is not the same, the lower speed of the two modules will be used. Likewise, the higher latency of the two modules will be used.
*
* Number of chips and sides (e.g. two sides with four chips on each side).
*
With the introduction of [[DDR5]], each DDR5 DIMM has two independent sub-channels.
=== Performance ===
Theoretically, dual-channel configurations double the memory bandwidth when compared to single-channel configurations. This should not be confused with [[double data rate]] (DDR) memory, which doubles the usage of DRAM bus by transferring data both on the rising and falling edges of the memory bus clock signals.
A benchmark performed by [[TweakTown]], using [[SiSoftware]] Sandra, measured around 70% increase in performance of a quadruple-channel configuration, when compared to a dual-channel configuration.<ref name="tweaktown-banchmark">{{cite web▼
| url = http://www.tweaktown.com/articles/4416/intel_x79_quad_channel_and_z68_dual_channel_memory_performance_analysis/index.html▼
| title = Intel X79 Quad Channel and Z68 Dual Channel Memory Performance Analysis▼
| date = 2011-11-16 | access-date = 2013-11-30▼
| author = Shawn Baker | publisher = [[TweakTown]]▼
}}</ref>{{rp|p. 5}} Other tests performed by TweakTown on the same subject showed no significant differences in performance, leading to a conclusion that not all benchmark software is up to the task of exploiting increased parallelism offered by the multi-channel memory configurations.<ref name="tweaktown-banchmark" />{{rp|p. 6}}▼
=== {{Anchor|GANGED}}Ganged versus unganged ===
Line 64 ⟶ 60:
=== Operation ===
[[DDR3]] triple-channel architecture is used in the [[Intel]] [[Core i7]]-900 series (the Intel Core i7-800 series only support up to dual-channel). The LGA 1366 platform (e.g. Intel X58) supports DDR3 triple-channel, normally 1333 and 1600Mhz, but can run at higher clock speeds on certain motherboards. AMD Socket AM3 processors do not use the DDR3 triple-channel architecture but instead use dual-channel DDR3 memory. The same applies to the Intel Core i3, [[Core i5]] and Core i7-800 series, which are used on the [[LGA 1156]] platforms (e.g., [[Intel P55]]). According to Intel, a Core i7 with DDR3 operating at 1066 MHz will offer peak data transfer rates of 25.6 GB/s when operating in triple-channel [[interleaved memory|interleaved]] mode. This, Intel claims, leads to faster system performance as well as higher [[performance per watt]].<ref>{{Citation | publisher = Intel | title = X58 Product Brief |url= http://www.intel.com/Assets/PDF/prodbrief/x58-product-brief.pdf }}</ref>
When operating in triple-channel mode, [[memory latency]] is reduced due to interleaving, meaning that each module is accessed sequentially for smaller bits of data rather than completely filling up one module before accessing the next one. Data is spread amongst the modules in an alternating pattern, potentially tripling available memory bandwidth for the same amount of data, as opposed to storing it all on one module.
The architecture can only be used when all three, or a multiple of three, memory modules are identical in capacity and speed, and are placed in three-channel slots. When two memory modules are installed, the architecture will operate in dual-channel architecture mode.<ref name="support.intel.com">{{Citation|title=Desktop Boards – Triple Memory Modules|url=http://support.intel.com/support/motherboards/desktop/sb/CS-011965.htm#triple|publisher=Intel|access-date=2011-10-01|archive-date=2009-03-08|archive-url=https://web.archive.org/web/20090308093852/http://support.intel.com/support/motherboards/desktop/sb/CS-011965.htm#triple|url-status=dead}}, [https://www.intel.com/content/www/us/en/support/articles/000005657/boards-and-kits.html Single- and Multichannel Memory Modes]</ref>
=== Supporting processors ===
Line 100 ⟶ 96:
=== Operation ===
Quad-channel memory debuted on Intel's Nehalem-EX [[
The architecture can be used only when all four memory modules (or a multiple of four) are identical in capacity and speed, and are placed in quad-channel slots.
=== Performance ===
▲A benchmark performed by [[TweakTown]], using
▲ | url = http://www.tweaktown.com/articles/4416/intel_x79_quad_channel_and_z68_dual_channel_memory_performance_analysis/index.html
▲ | title = Intel X79 Quad Channel and Z68 Dual Channel Memory Performance Analysis
▲ | date = 2011-11-16 | access-date = 2013-11-30
▲ | author = Shawn Baker | publisher = [[TweakTown]]
▲}}</ref>{{rp|p. 5}} Other tests performed by TweakTown on the same subject showed no significant differences in performance, leading to a conclusion that not all benchmark software is up to the task of exploiting increased parallelism offered by the multi-channel memory configurations.<ref name="tweaktown-banchmark" />{{rp|p. 6}}
=== Supporting processors ===
Line 108 ⟶ 113:
{{Col-break}}
'''AMD Threadripper:'''
* AMD Ryzen Threadripper 2nd gen 2990WX▼
* AMD Ryzen Threadripper 3rd gen 3960X▼
* AMD Ryzen Threadripper 3rd gen 3970X▼
* AMD Ryzen Threadripper 3rd gen 3990X
▲* AMD Ryzen Threadripper 3rd gen 3970X
▲* AMD Ryzen Threadripper 3rd gen 3960X
▲* AMD Ryzen Threadripper 2nd gen 2990WX
* AMD Ryzen Threadripper 2nd gen 2970WX
* AMD Ryzen Threadripper 2nd gen 2950X
Line 118 ⟶ 123:
* AMD Ryzen Threadripper 1920X
* AMD Ryzen Threadripper 1900X
'''AMD Epyc:'''
* Epyc 7003 series processors
* Epyc 7002 series processors
* Epyc 7001 series processors
'''AMD Opteron:'''
* Opteron 6100-series "Magny-Cours" (45 nm)<ref name="Opteron6100"/>▼
* Opteron 6200-series "Interlagos" (32 nm)<ref name=Opteron6200>{{cite web | title = AMD Opteron 6200 Series Processor Quick Reference Guide | url = https://www.amd.com/us/Documents/Opteron_6000_QRG.pdf | access-date = 2012-10-15 }}</ref>▼
* Opteron 6300-series "Abu Dhabi" (32 nm)<ref name=Opteron6300>{{cite web | title = AMD Opteron 6300 Series processor Quick Reference Guide | url = https://www.amd.com/us/Documents/Opteron_6300_QRG.pdf | access-date = 2013-12-11 }}</ref>
▲* Opteron 6200-series "Interlagos" (32 nm)<ref name=Opteron6200>{{cite web | title = AMD Opteron 6200 Series Processor Quick Reference Guide | url = https://www.amd.com/us/Documents/Opteron_6000_QRG.pdf | access-date = 2012-10-15 }}</ref>
▲* Opteron 6100-series "Magny-Cours" (45 nm)<ref name="Opteron6100"/>
{{Col-break}}
'''Intel Core:'''
* Intel Core i9-10900X
* Intel Core i7-9800X
* Intel Core i9-7980XE
* Intel Core i9-7940X
* Intel Core i9-7900X
* Intel Core i7-7820X
Line 154 ⟶ 169:
{{col-end}}
==
Supported by [[Qualcomm Centriq]] server processors,<ref>{{cite news|last1=Kennedy|first1=Patrick|title=Qualcomm Centriq 2400 ARM CPU from Hot Chips 29|url=https://www.servethehome.com/qualcomm-centriq-2400-arm-cpu-hot-chips-29/|access-date=14 November 2017|publisher=Serve The Home|date=23 August 2017}}</ref> and processors from the Intel Xeon Scalable
==
[[File:HP Z6 (极客湾Geekerwan) 029.png|thumb|alt=The image shows 4 RAM slots on the left and 4 on the right. The image center shows the socket and the AMD Ryzen Threadripper Pro 7995WX|Octa-channel setup with an AMD Ryzen Threadripper Pro 7995WX]]
Supported by [[Cavium#ThunderX2_SoCs|Cavium ThunderX2]] server processors, AMD's server processors from their [[Epyc]] platform, and the [[
== Dodeca-channel architecture ==
▲Supported by [[AMD Epyc]] and [[Cavium#ThunderX2_SoCs|Cavium ThunderX2]] server processors.<ref>{{cite news|last1=Cutress|first1=Ian|title=AMD Prepares 32-Core Naples CPUs for 1P and 2P Servers: Coming in Q2|url=http://www.anandtech.com/show/11183/amd-prepares-32-core-naples-cpus-for-1p-and-2p-servers-coming-in-q2|access-date=7 March 2017|publisher=Anandtech|date=7 March 2017}}</ref><ref>{{cite news|last1=Kennedy|first1=Patrick|title=Cavium ThunderX2 and OCP Platform Details|url=https://www.servethehome.com/cavium-thunderx2-ocp-platform-details/|access-date=14 November 2017|publisher=Serve the Home|date=9 November 2017}}</ref>
[[File:AMD Epyc 9754 in Dual-CPU-Socket-Serversystem (极客湾Geekerwan) 07.png|thumb|Server system containing two AMD Epyc CPUs with one dodeca-channel per CPU (768 GiB RAM in total)]]
Dodeca-channel or 12-channel memory architecture is introduced with AMD's server processors from their [[Epyc#Fourth generation Epyc (Genoa, Bergamo and Siena)|Epyc 9004]] platform released in 2022, using [[DDR5]] memory.<ref>{{Cite web |last=Goetting |first=Chris |date=2022-11-10 |title=AMD 4th Gen EPYC 9004 Series Launched: Genoa Tested In A Data Center Benchmark Gauntlet |url=https://hothardware.com/reviews/amd-genoa-data-center-cpu-launch |access-date=2023-12-07 |website=HotHardware |language=en-us}}</ref>
== See also ==
* [[List of
* [[Lockstep (computing)]]
Line 185 ⟶ 204:
[[de:Dual Channel]]
[[de:Triple Channel]]
[[es:Arquitectura de memoria multicanal]]
[[it:Dual channel]]
[[ja:デュアルチャネル]]
|