Jump to content

So with intel doing away with ht. Is it worth disabling for current cpus? I've read it can lower power draw and improve games. But after testing my self. I don't notice any less power draw, at least non I can measure beyond looking at stats in xtu. I might have gained like 4fps in calisto protocol which is actually like 5% but in ki3 it seems to test around 300 points less (it doesn't show fps for its benchmark) but that game seems to heavily rely on only 1 core. And since intels latest patches using their new performance profile I can't actually seems to get the cpu to boost higher than 5.3ghz anyway to benifit from a single core being allowed more voltage or what ever. Even setting it to go to 5.4 5.5ghz it just sticks at 5.3 regardless

 

Anyone else done extensive testing to see if there's any benefits over time for disabling ht on modern cpus 12 13 14th gen intel cpus?

Link to comment
https://linustechtips.com/topic/1586602-disable-hyperthreading-really-worth-it/
Share on other sites

Link to post
Share on other sites

For most users it really isn't worth it. Disabling hyperthreading on 12/13/14th gen does decrease power consumption in high load scenarios (think Cinebench), but it also drops performance in those units by a similar amount. For things like games, it has the same performance. Because of the decreased power consumption, if you're doing manual OC you can usually get a few extra MHz, though it's something like 200-300MHz at most, and when the chip is running at well over 5GHz it's not going to be too big of a performance uplift either way. 

 

As for increasing performance in games, it really depends on the specific game and how the scheduler handles it. In general it does boost performance slightly, but it's in the low single digit range rather than anything actually worthwhile. 

Link to post
Share on other sites

there's a difference between disabling hyperthread, or designing a cpu without hyperthread.

 

it might be that at the given core counts, it might make more sense to allocate die space to cache as opposed to hyperthreading logic, for example.

Link to post
Share on other sites

image.thumb.png.402b22ab5d835b6717f387a230cf9955.png

Not a direct answer but indirect. This is Watch Dogs Legion running on a 7920X with 3070, 1080p low. I disabled cores/HT to see how this game scaled. First take away point is that unless you know you have an abundance of CPU perf, you might not want to be lowering its maximum potential.

 

I had that CPU in 2019. For time context, Zen 2 (Ryzen 3000) series had only just released and that was AMD's first gen to match/beat Skylake in all areas. At the time, people had up to 8 cores as 12 and 16 core consumer tier only just came out. For this game at least, 12c12t was higher performance than 8c16t.

 

For quite a time I ran that system HT off since it was comparable or better than anything 8 core at the time. The theory was that HT means if you get 2 threads on one core, they WILL slow each other down. You gain throughput at the cost of latency. With HT off, then providing you still have sufficient overall CPU time, that doesn't happen any more. I later had to turn HT back on since I was also using it for other things than could actually use it.

 

Note a difference between designing a product without HT vs one with. You could use that "saving" in other areas that you wont get from turning it off on an older one. 

Gaming system: R7 7800X3D, Asus ROG Strix B650E-F Gaming Wifi, Thermalright Phantom Spirit 120 SE ARGB, Corsair Vengeance 2x 32GB 6000C30, MSI Ventus 3x OC RTX 5070 Ti, MSI MPG A850G, Fractal Design North, Samsung 990 Pro 2TB, Alienware AW3225QF (32" 240 Hz OLED)
Productivity system: i9-7980XE, Asus X299 TUF mark 2, Noctua D15, 64GB ram (mixed), RTX 4070 FE, NZXT E850, GameMax Abyss, Samsung 980 Pro 2TB, iiyama ProLite XU2793QSU-B6 (27" 1440p 100 Hz)
Gaming laptop: Lenovo Legion 5, 5800H, RTX 3070, Kingston DDR4 3200C22 2x16GB 2Rx8, Kingston Fury Renegade 1TB + Crucial P1 1TB SSD, 165 Hz IPS 1080p G-Sync Compatible

Link to post
Share on other sites

Intel removed HT on their 9th gen and then bought them back on their 10th gen. The upcoming Ultra 200s, they seem to be removing HT again, but I would rather have them give us that feature, and let us decide whether we want to use it or not. Who knows maybe they're testing this no HT again, and may again bring it back with their series after the Ultra 200s.

On their server side, they kept HT but only for CPUs with P cores, and no HT for CPU using E cores.

Granite Rapids only uses P cores so they get HT.

Sierra Forest uses E cores and they do not get HT.

 

 

 

Link to post
Share on other sites

in my experience using CPU-z, disabling HT did almost nothing for single core speeds, but disabling Virtualization and VT-D in the bios gave me a massive boost, about 30-40 points (910 to 945-950 single thread score)

14900K, RTX 5090 (360mm AiO), 64GB RAM, 4K 144Hz OLED

Link to post
Share on other sites

  • 9 months later...

I currently have an older Alienware R11 with i5-10400f with 64GB RAM and RTX 2070 Super. I've been experiencing major slowdowns and stuttering in Rust with the current patch and disabling hyperthreading solved it! My graphics card was previously bottlenecked and stuck around 115W power usage, and after disabling HT it's still bottlenecked but more like 165W

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×