Core2 E8400 -> i5 3570K for gaming w/results

[Avalon]

Yes, I know there are quite a few discussion and overclocking threads out there now for Ivy Bridge, but this is a little bit different and I thought it deserved its own thread.

I used to game on an E8400 overclocked to 3.8Ghz, paired with an MSI GTX 460 Hawk clocked at 830Mhz core, 1900Mhz mem. My monitor is a 22" Dell 2209WA 1680x1050 LCD. For a lot of you a few years ago, this was a very common setup, and I had wondered what upgrading my CPU alone would net me.

I ran some benchmarks of my old system overclocked, my new system at stock, and my new system overclocked to 4.4Ghz. Common between all three setups is the video card, hard drive (1TB WD Black), the drivers (285.62), and high quality texture filtering.

Old system (min/avg/max fps): E8400 @ 3.8, 4GB DDR2 @ 840Mhz, GTX 460, Win7 64
SC2 - 50/64.5/91
NS2 - 33/49.8/60
Civ5 - 27/50.2/64
Skyrim - 31/48.9/61

New system stock: i5 3570K @ stock, 8GB DDR3 @ 1600 9-10-9-28, GTX 460, Win7 64
SC2 - 68/98.1/130
NS2 - 57/65.5/71
Civ5 - 39/53.4/64
Skyrim - 45/56.9/62

New system OC'd: i5 3570K @ 4.4, 8GB DDR3 @ 1866 9-10-9-28, GTX 460, Win7 64
SC2 - 69/105.6/132
NS2 - 55/72/95
Civ5 - 46/57.7/63
Skyrim - 44/57/62

About the tests:
SC2 - Single player all in mission, average of 2x repeatable run of sending all troops to zerg base. All graphical settings set to Ultra, no vsynch.
NS2 - Natural Selection 2 beta 207, all settings set to max. Mineshaft map, 2x repeatable run w/environment and gun fire.
Civ5 - Max settings w/4x MSAA. Running a few troops from one city to the other, 2x repeatable run. FPS seems capped around 62FPS.
Skyrim - Max settings, 2x MSAA, 8x AF. 2x repeatable run in Eldergleam caverns. FPS seems capped around 62FPS.

Thoughts:

Now is a good time to upgrade if you're still sitting on a Core2 Duo or Quad. The i5 3570k brought huge benefits to all my games, and I'd suspect even at 1920x1080/1200 if your graphics card is up to task, you would still see likewise.

I had thought maybe some of these gains were from gaining two cores, but that's not always the case. SC2 and Skyrim are both limited to roughly 2 cores, while Civ5 is able to take advantage of all 4, but utilization is low. NS2 saw the biggest use of 4 cores, but even the games that were stuck running off 2 cores saw big gains going from my E8400.

My biggest surprise was the lack of noticeable gains from overclocking. With a nearly 30% overclock, I typically only saw gains of around 5% on average.

 

[khurios2000]

simply the 3570k, the result kills all Oc'ed e8400

great job and congrats

 

[Ferzerp]

My biggest surprise was the lack of noticeable gains from overclocking. With a nearly 30% overclock, I typically only saw gains of around 5% on average.

Why? Once you reach the GPU's limit, no amount of increase will give more FPS. At that point, you cease to show meaningful info about your processor and instead are benchmarking the video card.

 

[Don Karnage]

Talk about a weak overclock on the E8400

 

[Avalon]

@Ferzerp - Didn't think I'd reach the limit of it, especially at 1680x1050.

@Don - lol well...It does do 4.2 but I ran it at 3.8 at low volts daily.

 

[jor8888]

like he said ur video card is out dated.

 

[althaz]

I am about to experience something similar to this (once my new PC arrives) as I'm running a Core 2 Duo E6300, it has been running at 3.29Ghz but my motherboard doesn't like a 470 FSB anymore so it's running at stock.

 

[aaksheytalwar]

Your experience shows a difference >50% and <100%, but with a modern GPU like 7850 or GTX 580 or even more powerful GPUs like 7950 OC or 670, the difference in performance will be >100%. And a 2500k will be essentially the same difference as well. And a i7 860 will be a little behind your system but even a i5 750 will blow away that E8400 OC, even at stock, with a modern GPU.

 

[Termie]

OP, thanks for this info.

A couple of ideas for you:

(1) averaging your results may not make sense, because in two games you are limited by some type of vsync cap. Maybe just state the improvement on the two games that weren't vsync capped: 42%. Also look at the minimums - in Skyrim you went up nearly 50%, which would yield substantial improvements in playability.

(2) when you are overclocking, you are probably starting to hit a GPU limitation (if you hadn't already). I'm sure the 30% CPU overclock will come in handy eventually, but at present, I'd probably skip it, at least for gaming purposes. The benefit you'd get would mostly be in the minimums, but interestingly only Civ5 shows an improvement here: 18%.

 

[TakeNoPrisoners]

My guess is that with a modern GPU you would see even more of a difference between the two CPUs.

Looks like it is time for a new monitor and GPU.

 

[Ferzerp]

My guess is that with a modern GPU you would see even more of a difference between the two CPUs.

Looks like it is time for a new monitor and GPU.

Probably, but they still aren't going to see much if any difference once overclocked because (almost all) games at the resolutions we play at typically have only one value when trying to measure processors.

All they can tell you is if your processor is so old that it really, really sucks.

Once you reach a specific (pretty low) level of performance, all processors, when paired with the same video card, perform nearly equally in *most* games.

 

[BTRY B 529th FA BN]

This is a great example of 'GPU Bottleneck'

 

[RaistlinZ]

Throw a 670 in there and you'll be in gaming bliss.

 

[richierich1212]

Throw a 670 in there and you'll be in gaming bliss.

He's playing at 1680x1050. I don't think he needs to spend $400 right now lol.

 

[aaksheytalwar]

But he still needs a 7850 nevertheless

 

[redstratus]

Thanks for the post Avalon, I have been reading forums and reviews for the past week or so deciding if I should pull the trigger on an upgrade.

I currently run a similar setup to the one Avalon upgraded from, so I was really interested to see the results from the CPU upgrade.

Current Rig:
Q6600 GO @ 3.0
MSI P7N SLI Plat (not great OC mobo)
4 GB OCZ DDR2
Gigabyte 6950 1GB
240GB OCZ Vertex 2
I game at 1920x1080

Anyways last night I ordered this:
3570k
Gigabyte Z77X-UD3H
8GB Kingston DDR3

I hope this upgrade will be really solid for a few years. I thought a lot about waiting for Haswell, but I felt like it might be a pretty long year (or more) to wait without much competition from AMD these days. Also a buddy of mine that really got me into PC gaming is going to buy my old setup so hopefully I can get him back into it again. So that was my final decision maker. Get another PC Gamer back in the game!!

 

[blckgrffn]

Nice improvements and good write up :thumbsup: I would have expected even better minimum frames with the OC, but you still can't complain.

It would have been interesting to see the difference in power draw, too. I am guessing it stayed the same or even decreased...

Did you celebrate with a beer?

 

[Gryz]

I also upgraded to a i5-3570K. From a E8500.
My E8500 was not overclocked.
I have a gtx680. (Also not overclocked). I guess my system was not gpu-limited.

I ran a bunch of tests in Skyrim.
Note, I had Skyrim running fine on my E8500+gtx680. Of course I was CPU-limited. But I kept enabling all the gpu-features I could think of. Each new feature hardly had an impact on my framerates.
I play with:
1920x1200 resolution. 8xMSAA (via launcher), 4xSSAA transparency (via nvidia control panel), very high SSAO (via nvidia inspector), 16xAF. Official high-res texture pack. A bunch of addons (Vurt's, water, lush grass, SMIM, better snow, faces, etc).

These are framerates I got with 1) E8500+gtx680 and 2) i5-3570K+gtx680.
At the stairs in Whiterun, looking down at tree and city: 27 fps (E8500) - 49 fps (i5-3570k)
Markath, at the entrance gate: 24 fps - 44 fps
Markath, above smith, looking down at lower area, at night: 20 fps - 38 fps
Markath, same spot, but with morning light: 29 fps - 42 fps
Soltitude, at the entrance gate: 25 fps - 38 fps
Soltitude, middle, looking at the castle at the end of the street, at night: 37 fps - 40 fps
Riften, square, at night: 33 fps - 43 fps
High Rothgar, outside, front, looking at the steps, daylight: 27 fps - 27 fps
High Rothgar, garden, back, daylight: 45 fps - 45 fps
Riverwood: 35 fps - 41 fps.

As you can see, getting a faster CPU did improve my framerates. Alsmost double in some situations. But in some spots, it didn't make a difference at all ! The weirdest thing was High Rothgar. (A temple on a high mountain top). There isn't that much too see, but still framerates were only 27 fps ! And the CPU didn't matter at all ! Something weird going on there. Maybe the snow blowing over the steps ? Maybe lots of stuff you don't see that is still rendered ?

 

[havuk]

I have a Q6600 setup and a 295gtx co-op, would my gpu be a bottleneck if i upgrade to a new i5-3570k, new mb, and 8gb ram??

 

[itzbiff]

I have a Q6600 setup and a 295gtx co-op, would my gpu be a bottleneck if i upgrade to a new i5-3570k, new mb, and 8gb ram??

I dont think so. I think that it would still hold its own compared to current gen video cards.

 

[Avalon]

But he still needs a 7850 nevertheless

Nah, I'm good for now. My GTX 460 Hawk is overclocked to roughly GTX 560 speeds.

http://www.anandtech.com/bench/Product/549?vs=543

So there's a pretty close comparison. I don't play Crysis, Metro, or Batman...everything else and I'm pretty much looking at 60+fps average at my resolution. Not worth dropping $200+ for, IMO.

I honestly don't play much right now besides those 4 games I posted benches for. I'm going to go back and play Mass Effect 1/2 soon, and possibly pick up Path of Exile, but those aren't very demanding anyway.

I may do a GPU upgrade next year. For now I got what I need.

 

[sm625]

So if you tally up the results, it comes out to about 30% faster overall minimum framerates. Probably a third of that is due to the nearly double memory bandwidth and extremely improved memory latency. Had you "upgraded" to a G860, I bet you would have gotten the majority of the benefit.

 

[Denithor]

So if you tally up the results, it comes out to about 30% faster overall minimum framerates. Probably a third of that is due to the nearly double memory bandwidth and extremely improved memory latency. Had you "upgraded" to a G860, I bet you would have gotten the majority of the benefit.

Possibly. But honestly, these days, I wouldn't recommend anyone upgrade to a dual core. Quads are seeing much higher utilization and that is going to continue to grow.

 

[Avalon]

So if you tally up the results, it comes out to about 30% faster overall minimum framerates. Probably a third of that is due to the nearly double memory bandwidth and extremely improved memory latency. Had you "upgraded" to a G860, I bet you would have gotten the majority of the benefit.

But then I couldn't overclock and wouldn't have had anything fun to de-lid

Don't get me wrong, I do other things besides gaming on my PC, so it's alright, I'm pretty happy.

 

[Puppies04]

The weirdest thing was High Rothgar. (A temple on a high mountain top). There isn't that much too see, but still framerates were only 27 fps ! And the CPU didn't matter at all ! Something weird going on there. Maybe the snow blowing over the steps ? Maybe lots of stuff you don't see that is still rendered ?

Without knowing exactly where you are talking about as I haven't played skyrim much, maybe the game renders a set diameter from your character and being on a high mountain means that there is litterally "more" ground to render as the slopes around you would work out to a much greater square meterage than being on flat ground.