E4300 3.1Ghz vs. E2140 3.1Ghz

[steve258]

E4300 3.1Ghz vs. E2140 3.1Ghz - A benchmark review

Hi all:

I've done a little benchmarking comparing the 2 chips' performance at 3.1Ghz and I'd like to share my findings to my fellow ocers. The ultimate reason for this test is to help me deciding which chip worth keeping, but the results are interesting in that the E4300 has larger L2 but the E2140 has a higher fsb at the same clock. So another reason for this test is to find out which one has a more significant effect on the performance - higher fsb or larger L2 cache?

The rig

All testing was done on the same machine:

Both chips have passed 8 hours orthos at 3.1Ghz.

The results

Super PI:
E4300 - 18Secs
E2140 - 20Secs

Nuclearus Multi Core V2.00:
E4300 - 9701
E2140 - 9682

Cinebench 9.5:
E4300:
Rendering (Single CPU): 510 CB-CPU
Rendering (Multiple CPU): 940 CB-CPU
Multiprocessor Speedup: 1.84

E2140:
Rendering (Single CPU): 506 CB-CPU
Rendering (Multiple CPU): 941 CB-CPU
Multiprocessor Speedup: 1.86

dBpowerAMP Music Converter: Converting a 46:18 160Kbps MP3 to 128Kbps
E4300: 2:30
E2140: 2:31

Auto GK V2.40: Converting a 1:26:05 MPEG2 to Divx using the preset quality of 40% and 128Kbps MP3 stereo
E4300: 38:04
E2140: 39:06

I have also benchmarked the latest version of SiSoft Sandra, but as one would expect the scores come very close except the memory bandwidth test, in which E2140 aced with its faster FSB. I am only going to post the CPU tests in Everest Ultimate because the results are interesting (Not sure what most of these results stand for though..):

CPU Queen:
E4300: 6253
E2140: 6248

CPU PhotoWorxx:
E2140: 18070
E4300: 17832

CPU ZLib:
E4300: 40755kb/s
E2140: 40618kb/s

CPU AES:
E4300: 11079
E2140: 10957

FPU Julia:
E4300: 11736
E2140: 8860 - 24.5% performance hit!!

FPU Mandel:
E4300: 5690
E2140: 5588

FPU SinJulia:
E4300: 8567
E2140: 8552

Games: All games are tested at 1280*1024 apart from 3DMark 2001

3DMark 2001 - OK so it's hardly a graphical intensive benchmark in this day and age, but that's precisely why I chose this benchmark as the results reflect the overall system performance rather than the graphics card performance.

3DMark 2006 - The scores come very close here as anticipated, I also included the CPU score in the chart for obvious reason.

Fear - It's been aroud for a few years now but still a good benchmark due to the built in benchmark tool. I tested both chips with both high and low graphics settings, while all the computer settings were set to maximum.

Company of Heroes - Again I ran the in game benchmark tool with both high (well, it's actually the automatic option which set most details on high) and minimum details.

World in Conflict - A very new game that's very demanding in both CPU and graphics, and that's why I've only tested medium and low settings with the in game benchmark tool.

Far Cry - I used the benchmark tool provided by HardwareOC Benchmarks website. Again both ultra and minimum detail settings were used.

Prey - Also used the test tool from HOC Benchmarks.

SupComMark - This is quite a lengthy test using the built in benchmark which can be activated by the game shortcut with the " map /perftest" added to the Target field in the properties. Again both "high" and "low" preset fidelity settings.

Half-Life 2 - My all-time favourite. It's not exactly graphical intensive anymore but as we all know it the source engine relies heavily on overall system performance. I ran the video stress tests in both Couter-Strike Source and Lost Coast with all settings at absolute max. Strange results with the Lost Coast test as E4300 actually scored 17fps slower than E2140. I'm not sure what is responsible for this massive performance drop but I'm pretty sure the 172Mhz drop in FSB couldn't have done this kind of damage. I've run the test 3 times on E4300 and all resulted in the same fps figure.

N0w sees ma 1337 chart mak1ng skilz 1n akti0n

So it seems that having more L2 cache has more positive impact on the performance than higher FSB. As we can see in the chart that the extra 1Mb L2 has given the E4300 an edge in almost all tests, even in higher graphical settings where in theory the performance should have been limited by the graphics card and difference between the 2 chips should have been virtually none. And in demanding games like World in Conflict and Company of Heroes even 3fps difference in the minimum frame rate can determine whether the game is playable.

So there you go my first benchmark review so my apologies if there is any error or omission in this article.

Have a nice weekend peeps

 

[steve258]

TA!

 

[steve258]

Really need to add the 4mb core 2 duos as well in the mix mate

good stuff regardless

sid

Yeah I would really love to if I have one of them 4mb C2Ds.

Melbourne720 has already done a benchmark a couple of months ago comparing E6600 and E2160 here.

Both have the same multi so he was directly comparing 1mb vs 4mb. Whereas I am more interested in finding out whether higher fsb has more positive impact on performance than the extra 1mb cache.

I was going to just lower the cpu multi but the P5N-E Sli didn't like that, even with earlier BIOS.

It's interesting though that Melbourne720 actually got the same fps for both E6600 and E2160 in FEAR.

 

[steve258]

Thanks Melbourne720!

I've changed the charts now but still couldn't work out how to make the 3DMark2001 chart to display from 1 instead of 30000. But hopefully it's now less confusing.

----------------------------------------------------------

A slight update: I've clocked the E2140 upto 3.16Ghz and ran a quick test of COH, World in Conflict and FEAR. While there were improvements in all 3 games the results were still falling short of E4300 apart from COH even with 60Mhz increase in speed. But for the price the E2140 is a truly great chip.

Also I've added results from everest ultimate.

 

[steve258]

Why not try lowering the multi on the e4300 see if hoe theycompre at similar bus speeds.

I was going to do that but my mobo didn't like overclocking with lowered multi unfortunately.

 

[steve258]

Would switching to low quality be the same as installing a super powerful 'next-gen' graphics card? Hard to explain what I mean there but in essence what I am trying to say is once you take away the GPU bottleneck and the emphasis is placed onto the CPU, do you think the gains seen above would still hold true?

Thanks

I think that's exactly what happened there, with all graphics on minimum the games become CPU bound so the differences between the two chips were magnified, but it's not like you can tell the difference between 410 and 353fps without fraps. And yes I think I know what you meant that in essence it's the equivalent of installing a powerful graphics card where the bottleneck becomes the CPU.