Intel X58 platform system versus Intel Z97 platform system in FPS

[Tech Tweaker]

So, I decided to do a test for how much my Z97 would improve on an X58 system for gaming in Battlefield 4 a while back. The results were... surprising to say the least.

I started with the i5 4690K at stock and compared it to an Xeon L5630 overclocked to 3.6GHz. The L5630 won pretty easily. Then overclocked the i5 4690K to 4.2GHz thinking that would beat the L5630, but nope, the 4690K still loses.

Then I thought, okay, maybe the L5630 is winning just because of having more threads (8 versus the 4 of the 4690K). So I swapped out the Xeon L5630 for a Xeon E5506 (overclocked to 3.45GHz) (4 cores/4 threads), but the E5506 still managed to outperform the 4690K even when said 4690K was running 750MHz faster. Although, the FPS scores did drop when going from a 4C/8T processor to a 4C/4T processor on the same platform.

These benchmarks were run multiple times on both systems for the sake of consistency just in case one run was unusually better or unusually worse than the rest.

This honestly doesn't make much sense to me, with the Z97+4690K platform being much newer, having faster SATA transfer rates, faster memory, and higher IPC on the CPU it probably should have been performing better in my mind than a platform from 8-9 years ago with the same GPU.

Wasn't sure whether to put this in the GPU's forum or the Games forum, hopefully I put this in the right one.

CPU: Intel Core i5 4690K @ stock
Motherboard: ASRock Z97 Extreme6
Memory: 2x8GB G.Skill RipjawsX F3-1866C9D-16GXM (9-10-9-28 1T)
GPU: Evga GTX 780 FTW
SSD: Samsung 850 Evo 500GB
HDD: WD Black 1TB
PSU: Corsair RM650i
CPU cooler: Corsair H100i

Min: 91 Max: 169 Average: 124.33


CPU: Intel Core i5 4690K @ 4.2GHz
Motherboard: ASRock Z97 Extreme6
Memory: 2x8GB G.Skill RipjawsX F3-1866C9D-16GXM (9-10-9-28 1T)
GPU: Evga GTX 780 FTW
SSD: Samsung 850 Evo 500GB
HDD: WD Black 1TB
PSU: Corsair RM650i
CPU cooler: Corsair H100i

Min: 106.44 Max: 175.67 Average: 139.56


CPU: Intel Xeon L5630 @ 3.6GHz
Motherboard: Asus Rampage III Formula
Memory: 2x8GB G.Skill Ripjaws F3-12800CL8T-6GBRM (8-8-8-24 1T)
GPU: Evga GTX 780 FTW
SSD: Samsung 840 Evo 120GB
HDD: WD Black 500GB
PSU: Corsair TX750
CPU cooler: Cooler Master Nepton 280L

Min: 126 Max: 186 Average: 150.928


CPU: Intel Xeon E5506 @ 3.45GHz
Motherboard: Asus Rampage III Formula
Memory: 2x8GB G.Skill Ripjaws F3-12800CL8T-6GBRM (8-8-8-24 1T)
GPU: Evga GTX 780 FTW
SSD: Samsung 840 Evo 120GB
HDD: WD Black 500GB
PSU: Corsair TX750
CPU cooler: Cooler Master Nepton 280L

Min: 108 Max: 182 Average: 138.156

 

[EarthDog]

Frankly, I'm not suprised at all. Firsthe, it depends on the game, settings, res, and gpu used as to how well/not well new cpus can help out in games. You are also taking low 2 ghz chips and overclocking the entire bus they are on versus the haswell just using the multi.

Second, techspot tested this years ago and in this tithe at their settings..see anything there?: http://www.techspot.com/review/734-battlefield-4-benchmarks/page6.html

Also, what part of the game did you test? Something repeatable? BF4 doesn't have a built in benchmark so it's a manual run through of wherever you did it. I assume single player?

You are also using a 2 gen old gpu. Strap a 980ti, 1070 or 1080 and those results change too.

SATA speeds are completely irrelevant in gaming when talking FPS. Memory only makes a difference in a rare couple of titles and isn't typically a bottleneck.


Moved to general hardware.

 

[Tech Tweaker]

I don't own a 980Ti, 1070, or 1080. Frankly the best card I have on hand is a GTX 780Ti Superclocked, although it performs roughly on par with a GTX 980 according to reviews I've read.

I ran a few tests at the Test Range, just repeating certain actions and capturing the results with FRAPS. Though I did most of my tests in multiplayer maps with 32p and 24p servers, to give it more load. Then averaged the two sets of results together.

 

[EarthDog]

Test range.. well, you have repeatable..lol, but there is absolutely nothing going on there is the problem. Its not a realistic scenario.

I have seen this tested before, but the results don't jive with yours for whatever reason. IF you only test a specific map in MP or a passage in SP, are your results still the same? What happens when you clock the CPUs to the same speed? That will tell you as best you can the difference. But again, its been tested and while the X58 is still a viable platform, in many titles and higher end cards, it does put a glass ceiling on performance.