I think my fps are very low compared to my friend build

iNqus

Commendable
Jan 12, 2017
7
0
1,510
Intel I5-4690k overclocked to 4.5GHz
Cooler Master Hyper 212 Evo
2x8gb Kingston HyperFury, Crucial Ballistix 1x4gb = 20GB
EVGA GTX 980TI FTW Edition
256 GB SSD Samsung 850 evo, 640 GB HDD Seagate, 1 TB WD BLUE
Asus Z97 Pro wifi ac
EVGA 600w 80+ bronze

I mostly play h1z1, which is cpu game, and I'm getting 70-90fps when I'm streaming with obs at towns, which are very cpu intensive, where my friend with 6600k clocked to 4.4GHz and 1060 has 120-130 in the same places, we do have same settings, and he streams too!

I know that i5 is not good for streaming but I use nvenc for streaming encoder.
Maybe its a problem with dual channel ram and that 1x4gb that I do have ?
 
Solution
The IPC (Instructions Per Clock) between these CPU's are not that different.

UPDATE: apparently IPC is a 10% advantage to the i5-6600K in Cinebench (so Passmark apparently sucks?), though probably not quite as high with gaming (due to how buffers etc work). The frequency difference is only 2% (4.5/44) so at BEST the i5-6600K in these two setups 8% faster assuming 100% CPU bottleneck...

So probably in this game, with EXACT SAME MEMORY BANDWIDTH, no more than 5% or so FPS advantage such as 105FPS vs 100FPS.

The memory bandwidth is by far the most significant. The IPC might only make a few percent difference, whereas running single channel (what frequency?) might lose as much as 30% or so performance. Really hard to say.

Rapidly...

Karadjgne

Titan
Ambassador
For games, you aren't getting any benefit at all from the 1x4Gb Crucial, and depending on the setup you could be running all 3x sticks in single channel, which would show a small fps drop in some games. Add to the fact that the 6600k has a faster IPC than the 4690k, it's not surprising you are getting lower fps. H1Z1 is very cpu dependent, so every cpu discrepancy is going to count as both the 980ti and 1060 are more than capable at 1080p.

You are also streaming. That's an extra stressor on your internet connection. If your friend happens to live closer to the internet node than you, he'll get a stronger connection. It's a good bet you are using higher cpu% usage than he is. His pc might also have different abilities, his ssd might be faster, ram better organized, there's quite a few things that can knock off a measly 4-5 fps each. Add them up and you are getting a much larger drop.
 
That extra 4GB stick is definitely a problem. Not necessarily the only problem but it might be.

1) shut down PC and remove the 4GB stick
2) verify via motherboard MANUAL that your two sticks are in the recommended locations

3) choose "XMP" for the memory for the optimal profile then save
4) run MEMTEST86 for a full pass to validate the memory www.memtest86.com
- may need to change boot order, or run USB/DVD/CD as appropriate via a quick boot option in BIOS to start MEMTEST86

5) find one or more repeatable benchmarks to test and compare to your friends PC

OTHER:
NVENC adds very little CPU processing and generally doesn't change the FPS by more than 5% with a setup like yours
 
The IPC (Instructions Per Clock) between these CPU's are not that different.

UPDATE: apparently IPC is a 10% advantage to the i5-6600K in Cinebench (so Passmark apparently sucks?), though probably not quite as high with gaming (due to how buffers etc work). The frequency difference is only 2% (4.5/44) so at BEST the i5-6600K in these two setups 8% faster assuming 100% CPU bottleneck...

So probably in this game, with EXACT SAME MEMORY BANDWIDTH, no more than 5% or so FPS advantage such as 105FPS vs 100FPS.

The memory bandwidth is by far the most significant. The IPC might only make a few percent difference, whereas running single channel (what frequency?) might lose as much as 30% or so performance. Really hard to say.

Rapidly diminishing returns, but for a very CPU intensive game the maximum bandwidth is with DDR3 2400MHz (DDR4 is rated differently). Many games are fine with only DDR3 1600MHz (Dual Channel).

He might be running effectively 1600MHz Single Channel which we'll just call "800MHz" to avoid confusion. In other words the CPU is probably waiting on the system memory.

In DUAL CHANNEL the data is split between two modules to double the available bandwidth, however you can't do that when you have an ODD number of sticks.
 
Solution

Karadjgne

Titan
Ambassador
Yep. It's all added up. @10% for IPC, upto @20% for single channel, ?% difference with whatever ram the friend is running etc,and you are starting put at a 30% loss in ability. Impossible to say exactly how that translates to fps, but it'll be a chunk. This is also just the beginning, you've got other apps running, which also rely on the ram, it's also possible with all you have running that you are seeing 100% cpu usage and further tanking the cpu. You say your friend has the same settings, that's not entirely accurate, he might be using physX through the gpu instead of cpu, might have a better net connection, definitely could have better mobo drivers etc.
 

iNqus

Commendable
Jan 12, 2017
7
0
1,510
Ok, so I took off that 1x4 ram, I saw an improvement in fps a bit, I set default settings in bios, and overclocked the cpu again, but now to 4.6GHz with 1.27v.
https://www.twitch.tv/sielu28 my friends channel
https://www.twitch.tv/iNqu_ my channel
I added them both if u want to look for our specs.
I'm using physX through the gpu. Thinking about selling my i5 4690k and add 100-120$ for i7 4790k, do you think it is worth it right now?