Overclock.net › Articles › Intel Hyper Threading Performance Scaling With Cpu Frequency

Intel Hyper Threading - Performance Scaling with CPU Frequency

Test System

[CPU] Intel Core i7 4790K
[GPU] Gigabyte GTX 970 mini-ITX
[MOB] Gigabyte Z97-D3H (F7 bios)
[RAM] G. Skill Trident X 4GBx2 DDR3 @ 2400 MHz 10-12-12-31 2T
[HSF] Noctua NH-U12S

Windows 7 64-bit with Service Pack 1
GeForce 355.82



Test Methodology

For productivity testing, I only used HandBrake. The short movie (about 10 minutes long) Big Buck Bunny was encoded from AVI to MP4 using High Profile preset. Video resolution is 1080p. Only the first 2 minutes of the video was encoded to shorten the benchmarking time. Big Buck Bunny is an open movie which was created using open software. You can download the movie here https://peach.blender.org/about/

Gaming performance will be measured by using FRAPS to record the frame times. For those who are new to frame times, it is the measure of time taken to render a frame and its unit is milliseconds (ms). The lower the frame time, the better. The 99th percentile frame time is a measure of the overall "smoothness" of game play. For example, a 99th percentile frame time of 20ms means that 99% of the frames were rendered within 20ms. A more detailed explanation of frame time benchmarking can be found here http://techreport.com/review/21516/inside-the-second-a-new-look-at-game-benchmarking

Only Crysis 3 was used because it is more CPU-intensive than Witcher 3 and Grand Theft Auto V. For those who want to replicate my gaming benchmark, check the YouTube link below to see what part of the game I benchmarked. The settings used for each game are also indicated below. The game is updated to the latest version as of September 11, 2015.

Crysis 3 https://www.youtube.com/watch?v=aX1S4aSJ3lU
Resolution: 1920 x 1080
Texture Resolution: Very High
Anti-aliasing: SMAA T2X
System Spec: Very High



Performance Results

The results below are the average of 3 runs

H6IUaB7.png
9y563U5.png



Conclusion

In HandBrake, Hyper Threading gives about 20% more performance. It scales linearly. However, it is very different when it comes to a CPU-intensive game. With Hyper Threading enabled, the 99th percentile frame time increased by 39.5% when going from 4.4 GHz to 3.2 GHz. Disabling Hyper Threading increased the 99th percentile frame time by 61.8% when going from 4.4 GHz to 3.2 GHz. This just shows that thread count matters more than CPU frequency in CPU intensive games like Crysis 3.

Comments (4)

Have you tested other games? Most games actually gain a few fps by disabling HT. Crysis 3 gained fps in my testing, but most other "CPU intensive" games did not benefit at all.
@ BradleyW

Crysis 3 was the most CPU-intensive game I have tested. Click the YouTube link I provided to see what part of the game I benchmarked then you can replicate it
"This just shows that thread count matters more than CPU frequency in CPU intensive games like Crysis 3."

This is also because Crysis is one of few games that can utilize more than 4 threads. There are only a small handful that can do this. You are still correct though, but it might be best to emphasize 4+ threaded games, not just CPU intensive. CPU intensive can mean a game like an RTS that barely uses any GPU power.
CSGO gains FPS and tightens frame times when you:

keep HT enabled

-threads 4

and then set affinity to 0 2 4 6

priority high

i've gotten down to an average of 1.5-2.75ms frame times at this point

also dude, ur not proving anything really here... you cherrypicked an ENCODER... AND CRYSIS 3... two things that will just show the love for hyperthreading.

do a test on a game that can barely use 2 cores but can eat up gpu like crazy for more fps... the big thing with hyperthreading for me is that it in some cases disrupts frame timing, which i consider more important then raw fps.

i'm currently gathering data for a youtube video i'm going to do focusing on this with counter strike global offensive. record a demo of a deathmatch or a causal, not sure yet. DM might be less prone to people pointing out that its not realistic or something. then change driver settings for my 390x. change clock speeds. use benchstudio to graph and compile my findings. should be interesting with what i've found so far, ^_^
Overclock.net › Articles › Intel Hyper Threading Performance Scaling With Cpu Frequency