What hardware affects frametime?

[ezio313]

Hi, bought a gaming PC 2 weeks ago:

 

i510400

 

Hikvision 16gb 3200hz cl16 (dual channel)

 

Msi 6600xt

 

Hikvision SSD 

 

Biostar h510 

 

Cooler master gold 700w

 

Case with 4 fans

 

I noticed games like warzone, elden ring and bf1 suffered from stuttering. Among further inspections. I noticed that although frame rate is high(above 100 at 1080p) the frame time is so bad, going from 8 to 20 and even 40ms(especially in warzone). CPU usage was maxing out in task manager but 70% on msi afterburner?! Other games like tomb raider and God of war didn't suffer like this. 

 

 I tried many online solutions (capping fps at 80, lowering resolution, reinstall windows and GPU drivers, installing c++ packages) and none worked. I got so frustrated that I sold the pc at a 150$ loss. 

I tried warzone at a gaming lounge with i7 8700 3060 and it worked fine.

 

Now I want to build a new PC, I'm looking into keeping the same parts except for replacement of CPU,GPU and Mobo. I'm getting i5 12400,3060ti and biostar b660 Mobo. My question is whether the crappy biostar mother board or the hikvision ram could have played a role in this and thus need replacement also? I have msi b660m-e for Mobo and thermaltake m one 3200hz cl16 for ram

[Levent]

Everything can. It also may not be hardware, graphics card drivers are also a culprit when it comes to unstable frame times.

[ezio313]

11 minutes ago, Levent said:

Everything can. It also may not be hardware, graphics card drivers are also a culprit when it comes to unstable frame times.

That's why I reinstalled windows. I uninstalled them using an app that makes sure they are removed completely. Unfortunately both options didn't work.

[191x7]

The Biostar H510 probably had a bad VRM so the CPU probably had load spikes, voltage spikes and maybe some short term throttling.

 

Frametime is mostly affected by the CPU and GPU, but also RAM and Storage play a role.

If you want it smooth, the CPU absolutely has to be strong enough not to bottleneck the GPU and the GPU has to be limited never to reach 100% (for example, a stable in-game frame limit that wouldn't let the GPU go above 95% in the worst situation) and there absolutely have to be no limits by the RAM amount and storage speed.

 

That means a strong CPU, more than 16GB RAM, game on an SSD and a GPU that can do high framerates while not being pushed to the limits.

[ezio313]

26 minutes ago, 191x7 said:

The Biostar H510 probably had a bad VRM so the CPU probably had load spikes, voltage spikes and maybe some short term throttling.

 

Frametime is mostly affected by the CPU and GPU, but also RAM and Storage play a role.

If you want it smooth, the CPU absolutely has to be strong enough not to bottleneck the GPU and the GPU has to be limited never to reach 100% (for example, a stable in-game frame limit that wouldn't let the GPU go above 95% in the worst situation) and there absolutely have to be no limits by the RAM amount and storage speed.

 

That means a strong CPU, more than 16GB RAM, game on an SSD and a GPU that can do high framerates while not being pushed to the limits.

On xtu 2.0, the CPU showed normal results with no thermal throttling, didn't pay attention to power spikes though. Yes there was a bottleneck especially when I run the game at low settings,CPU usage on task manager spiked to 100% every now and then. Especially in warzone and bf1. The ram usage never went above 14gb so I don't think its an issue. And yes it was installed on an SSD.

[Fasauceome]

1 hour ago, ezio313 said:

On xtu 2.0, the CPU showed normal results with no thermal throttling, didn't pay attention to power spikes though. Yes there was a bottleneck especially when I run the game at low settings,CPU usage on task manager spiked to 100% every now and then. Especially in warzone and bf1. The ram usage never went above 14gb so I don't think its an issue. And yes it was installed on an SSD.

The thermals are not really the concern. Rather, the motherboard's weak power delivery can affect clock speeds.

[ezio313]

6 hours ago, Fasauceome said:

The thermals are not really the concern. Rather, the motherboard's weak power delivery can affect clock speeds.

On msi afterburner,clock speeds were consistent at 4000hz.or do u mean other thing?

[MadAnt250]

10 hours ago, ezio313 said:

noticed games like warzone, elden ring

You picked the wrong games. (not really)

These games are known to have some issues like stuttering and whatnot, just do a quick search on their performance issues and you will see.

[IkeaGnome]

1 hour ago, ezio313 said:

On msi afterburner,clock speeds were consistent at 4000hz.or do u mean other thing?

This is what they're referring to with VRMs.

They regulate the power going into the CPU to give it clean, stable power. They create heat and can overheat(oversimplified). The temperatures can be found with HWinfo64 in sensors only mode under the motherboard section.

https://www.hwinfo.com/download/

[Fasauceome]

2 hours ago, ezio313 said:

On msi afterburner,clock speeds were consistent at 4000hz.or do u mean other thing?

I'm not sure afterburner is being 100% accurate, it disagrees with task manager and while task manager is not know to be super accurate in general, the motherboard's uncovered VRM and the high CPU usage feels like you might have some clock speed discrepancy to look into.

 

HWInfo64 might be a good alternative to verify CPU clocks and usage.

[ezio313]

Hi so the ram is hikvision u10 (8gb 3200hz cl16), I want to buy 2 of them, each stick cost 28$ while the next cheapest model is thermaltake m-one 8gb cl16 3200hz for 40$, my question is whether going for thee cheaper brand with same specs could affect my frame time in-game.

I'm on a tight budget,that's why I'm asking

[Eigenvektor]

If the specs are the same, the performance should be the same. You're unlikely to notice small differences outside of benchmarks.

[ezio313]

2 minutes ago, Eigenvektor said:

If the specs are the same, the performance should be the same. You're unlikely to notice small differences outside of benchmarks.

Another guy told me that the timings of the two kits could play a role in frame TIME

[Origami Cactus]

28 minutes ago, ezio313 said:

Another guy told me that the timings of the two kits could play a role in frame TIME

That is true, but that depends on your cpu etc, and it doesn't make such a difference to pay 40% more, especially because i don't expect fantastic timings out of that thermaltake ram either.

So just enjoy your cheap ram, and put the money saved towards an ssd or something

[Somerandomtechyboi]

The performance will be mostly the same since small diff in subtimings shouldnt affect performance much

 

Btw you may be able to get cheap bare pcb samsung 2133-2666 and overclock to 4000 or higher, samsung b die is what you should look for but d die is more accesible (the ending letter of the ram ic model is the die so b die is k4a8g085wb and d die is k4a8g085wd), 4000+ is pretty easy on decent samsung ics like 8gbit b, d, and 4gbit e die though over 4000mhz may not be very benificial due to bad timings (excl b die), i can actually get em for ~20$ locally here in indo

[Eigenvektor]

4 hours ago, ezio313 said:

Another guy told me that the timings of the two kits could play a role in frame TIME

Sure, but since the specs are identical their timings should be identical, so there should be no difference between them worth paying more.