SMT on and off.

[WARburton]

so after having a chat on here about intel dropping HT, and been made aware that in can make the CPU faster in games if you have enough core to power the game without SMT.

ive decided to have a play but im struggling to work out the best way to bench test it.
the only program i own is 3D Mark, what would be your suggestions and what's the best program to log FPS and time in games.

 

[Disco_P]

I think the latest "Nvidia" software can do performance logging or MSI afterburner + rivatuner is a classic staple of benchmarking games

 

[Rroff]

When I messed about with CP2077 with my 14700K I didn't get great results with HT off - the max FPS slightly increased but it seemed to shift the load a little to the E cores with resulting slight reduction in min FPS as well, with the average within margin of error and if anything a slightly worse feel to the game.

Interestingly other testers have seen a moderate increase in 0.1% lows but decrease in 1% lows in similar circumstances but I didn't see that.

 

[ToOo]

The ancient lore from the P4 days did see benefits from disabling HT on Intel chips, could yield a reasonable gain in single thread heavy situations but was still not recommended to disable HT even in those days, unless you were going for an LN2 OC record. With how multi-threadded most applications are now it's almost always a regression.

 

[WARburton]

Soni did turn it off and on a few times and tbf in squads and very badly threaded game I did see an uplift and over all fps but didn’t have anything to log lows…

But in ever other game I play there zero difference. So just decided to leave it o also windows didn’t feel as snappy with it off