Jump to content

Guide to Q: What Performance to expect from upgrading CPU?


Recommended Posts

Same as many of you, I am always looking at how I can get more FPS, run the sim at better visual quality etc, so there is always this question, what if I do an upgrade and mostly this is concerning getting faster CPU and/or video card.

 

So I was trying to figure out what sort of performance increase (if any) would I get if I was tp upgrade my CPU to something of newer generation and is it worth it?

 

It's hard to quantatise this performace difference as it's other factors in your system that affect it too but to (like different motherboard or memory do give slight difference in performance even with exact same CPU/VGA) but it is still the CPU and video card that are most determing factors.

 

So I thought of how about if I can find some specs of my current CPU and newer one, something I could compare well to give me this ratio so I know what I can expect if I do this or that CPU upgrade.... and there is a way to do this... there is this thing called FLOPS (floating point calculations) that every CPU can be rated at, which is actual proccessing power the CPU has.

 

Here how you do it:

Download IntelBurnTest utility (download page) that is mostly used when overclocking to check your overclock stability and run the standard test. This will give you processor speed/performance in GFlops and you can then use this to compare with other CPU's so you know what to expect if you did a CPU upgrade.

 

Here is a thread at TechPowerUp forums where they have done exactly that:

http://www.techpowerup.com/forums/showthread.php?t=94721

 

I've just run my i7 950 that is overclocked to 4.2GHz and I got little over 60GFlops which pretty much matches the i7 950 overclocked to 4.2GHz found on this table.

 

Now, since my CPU has 4 cores, I devide this by 4 and I get gFlops per 1 core (this is important as DCS runs on 1 core at the moment so having CPU with 6 cores will give more GFlops but those are not usefull as still only 1/6th of that will be used when running DCS

 

I ran this test with 1 thread only (so to use 1 CPU core only) and I get 15.8GFlops on 1 core

 

Now I look at i7 3930K clocked at 5GHz (that's big overclock by the way) getting 104GFlops, I take into account that i7 3930 has 6 cores, so Ihave to divide the 104 by 6 and get GFlops of i7 3930K and I get 17.4GFlops on 1 core

 

Now I cam see that if I was to upgrade to i7 3930K, I would imporve my FPS from around 15-16fps to 17-18fps... I can see now this is not worth the expense.

 

So anyway, hope this helps so many others always asking what they should upgrade to... simply run IntelBurnTest to get your CPU performance and compare to this chart... or alternatively we can have our own database of CPU's in this thread? :book:


Edited by Kuky

No longer active in DCS...

Link to comment
Share on other sites

Looks like a useful tool.

Question: why did you assume a linear relationship between gflops and FPS in DCS?

Spoiler

AMD Ryzen 9 5900X, MSI MEG X570 UNIFY (AM4, AMD X570, ATX), Noctua NH-DH14, EVGA GeForce RTX 3070 Ti XC3 ULTRA, Seasonic Focus PX (850W), Kingston HyperX 240GB, Samsung 970 EVO Plus (1000GB, M.2 2280), 32GB G.Skill Trident Z Neo DDR4-3600 DIMM CL16, Cooler Master 932 HAF, Samsung Odyssey G5; 34", Win 10 X64 Pro, Track IR, TM Warthog, TM MFDs, Saitek Pro Flight Rudders

 

Link to comment
Share on other sites

I am assuming linear relationship because these FLOPS are the same type of calculations showing actual processing power of the CPU.

 

I am aware that newer CPU's do have more types of instructions to use which does matter if aplication is coded to use them, but I think in DCS this doesn't matter.

 

Before I got i7 950 I had Core2Duo E8500 and looking at that chard Core 2 Duo E8500 @ 5.1GHz gets 34.7GFlops. As I know that I did get quite a bost in performance when I did that upgrade comparing just the GFlops between what I had and what I have now does match quite well... so I think my assumption that this linear perfromance in game and GFlops is valid, same as there is pretty much linear performance relationship between same CPU GHz values (10% higher clock yields around 10% performance increase)

 

Just take into account number of cores and the number of cores application uses.

No longer active in DCS...

Link to comment
Share on other sites

I think also in most of the cases it is so (linear) and it should work well when comparing processors with at least same architecture (like your example with the clock). But for different CPU´s you might encounter irregularities. For example a 64 bit processor can process larger numbers then a 32 bit processor. *- I think that is why they recommend to use 64bit systems.

Spoiler

AMD Ryzen 9 5900X, MSI MEG X570 UNIFY (AM4, AMD X570, ATX), Noctua NH-DH14, EVGA GeForce RTX 3070 Ti XC3 ULTRA, Seasonic Focus PX (850W), Kingston HyperX 240GB, Samsung 970 EVO Plus (1000GB, M.2 2280), 32GB G.Skill Trident Z Neo DDR4-3600 DIMM CL16, Cooler Master 932 HAF, Samsung Odyssey G5; 34", Win 10 X64 Pro, Track IR, TM Warthog, TM MFDs, Saitek Pro Flight Rudders

 

Link to comment
Share on other sites

They stopped making 32bit CPU's years ago, all of the CPU's on that table (some more than 5 years old) are 64bit CPU's. Maybe if you run 32bit OS compared to 64bit SO on came CPU it does mate a difference (and it should)

No longer active in DCS...

Link to comment
Share on other sites

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...