Editing Hyper-threading (section)

== Performance claims ==
According to Intel, the first hyper-threading implementation used only 5% more [[Die (integrated circuit)|die area]] than the comparable non-hyperthreaded processor, but the performance was 15–30% better.<ref>
{{cite journal
 |url=http://www.intel.com/technology/itj/2002/volume06issue01/vol6iss1_hyper_threading_technology.pdf
 |title=Hyper-Threading Technology
 |journal=Intel Technology Journal
 |volume=06
 |issue=1
 |date=14 February 2012
 |issn=1535-766X
 |archive-url=https://web.archive.org/web/20121019025809/http://www.intel.com/technology/itj/2002/volume06issue01/vol6iss1_hyper_threading_technology.pdf
 |url-status=dead
 |archive-date=19 October 2012
}}</ref><ref>
{{cite web
 |url=https://software.intel.com/en-us/articles/how-to-determine-the-effectiveness-of-hyper-threading-technology-with-an-application
 |title=How to Determine the Effectiveness of Hyper-Threading Technology with an Application
 |date=28 April 2011
 |url-status=dead
 |archive-url=https://web.archive.org/web/20100202034723/http://software.intel.com/en-us/articles/how-to-determine-the-effectiveness-of-hyper-threading-technology-with-an-application
 |archive-date=2 February 2010
 |website=software.intel.com}}</ref> Intel claims up to a 30% performance improvement compared with an otherwise identical, non-simultaneous multithreading Pentium&nbsp;4. [[Tom's Hardware]] states: "In some cases a P4 running at 3.0&nbsp;GHz with HT on can even beat a P4 running at 3.6&nbsp;GHz with HT turned off."<ref>{{cite web|url=http://www.tomshardware.com/reviews/single-cpu-dual-operation,549-25.html |title=Summary: In Some Cases The P4&nbsp;3.0HT Can Even Beat The 3.6&nbsp;GHz Version : Single CPU in Dual Operation: P4&nbsp;3.06&nbsp;GHz with Hyper-Threading Technology |publisher=Tomshardware.com |date=2002-11-14 |access-date=2011-04-05}}</ref> Intel also claims significant performance improvements with a hyper-threading-enabled Pentium&nbsp;4 processor in some artificial-intelligence algorithms.

Overall the performance history of hyper-threading was a mixed one in the beginning. As one commentary on high-performance computing from November 2002 notes:<ref>{{cite web|title=A Study of Hyper-Threading in High-Performance Computing Clusters|url=https://ahelp.com/wp-content/uploads/2025/03/4q02-Len.pdf|publisher=Dell|access-date=12 November 2012|author=Tau Leng|author2=Rizwan Ali |author3=Jenwei Hsieh |author4=Christopher Stanton |page=4|date=November 2002}}</ref>
<blockquote>
Hyper-Threading can improve the performance of some [[Message Passing Interface|MPI]] applications, but not all. Depending on the cluster configuration and, most importantly, the nature of the application running on the cluster, performance gains can vary or even be negative. The next step is to use performance tools to understand what areas contribute to performance gains and what areas contribute to performance degradation.
</blockquote>

As a result, performance improvements are very application-dependent;<ref name="extremetech-133121">{{cite web
 | url = http://www.extremetech.com/computing/133121-maximized-performance-comparing-the-effects-of-hyper-threading-software-updates
 | title = Maximized performance: Comparing the effects of Hyper-Threading, software updates
 | date = 24 July 2012 | access-date = 2 March 2015
 | author = Joel Hruska | website = extremetech.com
}}</ref> however, when running two programs that require full attention of the processor, it can actually seem like one or both of the programs slows down slightly when Hyper-Threading Technology is turned on.<ref>{{cite web|url=http://users.telenet.be/nicvroom/performanceP4.htm|title=CPU Performance Evaluation - Benchmark - Pentium 4 2.8 and 3.0|website=users.telenet.be|access-date=12 April 2011|archive-date=24 February 2021|archive-url=https://web.archive.org/web/20210224131422/http://users.telenet.be/nicvroom/performanceP4.htm|url-status=dead}}</ref> This is due to the [[replay system]] of the Pentium&nbsp;4 tying up valuable execution resources, equalizing the processor resources between the two programs, which adds a varying amount of execution time. The Pentium&nbsp;4 "Prescott" and the Xeon "Nocona" processors received a replay queue that reduces execution time needed for the replay system and completely overcomes the performance penalty.<ref>{{cite web|title=Replay: Unknown Features of the NetBurst Core. Page 15|url=http://www.xbitlabs.com/articles/cpu/display/replay_15.html#sect0|website=Replay: Unknown Features of the NetBurst Core.|publisher=Xbitlabs|access-date=24 April 2011|url-status=dead|archive-url=https://web.archive.org/web/20110514180659/http://www.xbitlabs.com/articles/cpu/display/replay_15.html#sect0|archive-date=14 May 2011}}</ref>

According to a November 2009 analysis by Intel, performance impacts of hyper-threading result in increased overall latency in case the execution of threads does not result in significant overall throughput gains, which vary<ref name="extremetech-133121" /> by the application.  In other words, overall processing latency is significantly increased due to hyper-threading, with the negative effects becoming smaller as there are more simultaneous threads that can effectively use the additional hardware resource utilization provided by hyper-threading.<ref>{{cite web
 |url         = https://software.intel.com/en-us/articles/performance-insights-to-intel-hyper-threading-technology
 |title       = Performance Insights to Intel Hyper-Threading Technology
 |date        = 20 November 2009
 |access-date  = 26 February 2015
 |first       = Antonio |last = Valles
 |publisher   = [[Intel]]
 |url-status  = dead
 |archive-url  = https://web.archive.org/web/20150217050949/https://software.intel.com/en-us/articles/performance-insights-to-intel-hyper-threading-technology/
 |archive-date = 17 February 2015}}</ref> A similar performance analysis is available for the effects of hyper-threading when used to handle tasks related to managing network traffic, such as for processing [[Interrupt request (PC architecture)|interrupt requests]] generated by [[network interface controller]]s (NICs).<ref>{{cite web
 | url = https://calomel.org/network_performance.html
 | title = Network Tuning and Performance
 | date = 12 November 2013 | access-date = 26 February 2015
 | website = calomel.org
}}</ref> Another paper claims no performance improvements when hyper-threading is used for interrupt handling.<ref>{{cite web
 | url = https://www.kernel.org/doc/Documentation/networking/scaling.txt
 | title = Linux kernel documentation: Scaling in the Linux Networking Stack
 | date = 1 December 2014 | access-date = 2 March 2015
 | publisher = [[kernel.org]]
 | quote = Per-cpu load can be observed using the mpstat utility, but note that on processors with hyperthreading (HT), each hyperthread is represented as a separate CPU.  For interrupt handling, HT has shown no benefit in initial tests, so limit the number of queues to the number of CPU cores in the system.
}}</ref>