How important is hyperthreading in gaming?

[wolfmcbeard]

So before I I upgraded to a 1050, I had been using an Asus Radeon HD 6570 1gb gpu oc'd to 800mhz, first on a pc that had an i5 660 dual core (with hyperthreading - 4 threads) in a sff Dell, I got honestly poor performance in many games but at least had a loosely playable experience in Skyrim SE on low.
I fast forward to about two-three months ago, I bought a newer Dell full tower pc with an i5 3570 quad core cpu (without hyperthreading - also 4 threads). When I first got it, I planned on buying a new gpu to go with it, but before the upgrade to the 1050, I popped in my hd6570 to get me by since the igpu was awful and almost instantly I noticed that some games seemed to run worse while some did improve.
Are there certain game loads that worked better on my hyperthreaded dual core that have a harder time on a non-hyperthreaded quad core?
Should I look into a cpu with this feature?

[minibois]

In this post I will abbreviate Hyperthreading to HT.

 

If they have the same Ghz and IPC (what the CPU can do per Hz) + all other factors are the same (so the only difference is the core amount and HT support):

4 core no HT > 2 core with HT.

 

Hyperthreaded threads are a bit weaker than a thread that has a core on its own, because by the way HT works, 2 threads will share components from 1 core. From what I have heard, an hyperthreaded thread has about 70% of a normal core performance.

 

Of course there are a lot of exceptions to that rule, if you're comparing an old HT CPU with a new non-HT CPU, you will of course get a different comparison..

[Jurrunio]

10 minutes ago, wolfmcbeard said:

some games seemed to run worse

that shouldn't happen... What about the rest of the system? Single channel memory on the newer i5?

[wolfmcbeard]

1 hour ago, Jurrunio said:

that shouldn't happen... What about the rest of the system? Single channel memory on the newer i5?

Pretty sure it's dual channel, 4x 2gb, same setup as the older one.
I though it was weird too.

[FellTheSky]

worse memory / bad PSU

[wolfmcbeard]

Just now, FellTheSky said:

worse memory / bad PSU

Memory from both pc's is identical.
The psu seems fine, to be honest, I think the older systems psu was giving out, but there's no what that somehow boosted performance.
The only difference in the computers is size, age, cpu, and psu output as far as I can tell,

[Mesterial]

2 hours ago, Minibois said:

an hyperthreaded thread has about 70% of a normal core performance.

No, it's not the 90's anymore! The performances of the logical CPUs attached to the same physical CPU are the same (1 or 2% difference) when using 1 load thread (game, compute, etc.).

 

When using 2 load threads on the same physical CPU, performance depends a lot, and especially it depends on the load itself (% compute cycles, % wait, etc.) and the OS scheduler (awake mode when 'idle' during a program, what CPU to bind to, etc.). Most games uses only 2 to 4 threads, with 1 or 2 being highly loaded. If the OS put them on the same physical CPU, then you have a performance hit for sure. The "easiest" way to get rid of this is to disable HT, so that the OS cannot do this. That's why people say it's "better without HT".

The games taking advantage of HT do so because:

- the game itself might use a lot of threads (hundreds of units to give orders to, etc.),

- game developers program an engine that produces multiple meaningful threads, and so that these threads are well treated by the OS.

 

[TechyBen]

Whoa. Double post ideas batman! (Relevant:  

) 

 

PS, lots of misinformation and incorrect ideas in this thread. Meserial is correct.

 

Most i5s 4 cores with no HT will perform 90% of an i7 4 cores with HT if same clock speed etc. BUT HT is helpful. It's just not 100% increase on performance.

Only like 15-30% even best case. Some games will benefit, some will not. I can help if you are running Chrome/music/streaming in the background while gaming.

[_Hustler_One_]

I just want to simply answer the topic title. It depends on how the game developers optimize their game engine.

 

A single example is what Wargaming did on their World of Tanks. Earlier they optimized the game for a 2 cores CPU and we cant get massive advantages by using more cored CPUs but less cores and high clock are more preferred.

 

But in 2019 Wargaming optimized their World of Tanks game engine to take advantages over higher core count/multicore or even multithreaded CPUs.

 

I play WoT on i7-8700K before vs after the optimization, I gained around 20fps than I used to get at the same CPU clock speed and same GPU.

[wolfmcbeard]

2 minutes ago, _Hustler_One_ said:

I just want to simply answer the topic title. It depends on how the game developers optimize their game engine.

 

A single example is what Wargaming did on their World of Tanks. Earlier they optimized the game for a 2 cores CPU and we cant get massive advantages by using more cored CPUs but less cores and high clock are more preferred.

 

But in 2019 Wargaming optimized their World of Tanks game engine to take advantages over higher core count/multicore or even multithreaded CPUs.

 

I play WoT on i7-8700K before vs after the optimization, I gained around 20fps than I used to get at the same CPU clock speed and same GPU.

I have noticed the weird performance thing most heavily in Crossout, but I haven't played since I upgraded due to bad internet speed lately.

Another thing I'm wondering is if I might be dealing with a  bottleneck between the i5-3570 and my 1050, and what type of performance hit that might result.
I don't really know what to look for.

[TechyBen]

Check on a performance graphing app/software. See which maxes out or which stutters/drops performance.