Update 20/03/09

Now includes E2140 results from 1.6-3.6Ghz and E8500 E0 results from 2.4-4.5Ghz. All done with the exact same components and drivers to give true comparisons.

Hi guys, i know some of you have been waiting for this so here it is at last. Bear with me as this is my first review.

I set about doing this to try to offer some help to those who have been wondering if it is worth upgrading from what they have. I have done a full suite of benchies with a E2140, E4300, E5200, E6600 and E8500. I also have partial results from a Q6600. Benching was performed in my main rig to keep it all consistent. The specs are as follows:-

Asus P5E X38 with bios 903.
2x1Gb Crucial Ballistix PC2 8500.
Zotac 8800GT @701/1836/1950
Creative X-Fi Extreme Gamer
Zalman ZM750-HP Psu.
Seagate 7200.10 320Gb sata2 hdd
Samsung 20x DL DVDRW
TT Aluminium Armour.
High end watercooling setup.
178.13 WHQL driver.
8.09.04 Physx driver.
Windows XP Home SP3.

E2140 - 1.6Ghz - 8x200 - 1mb L2 - 65nm Conroe core - SLA93 - FPO/Batch # L732A988 - Vid 1.2125.

E4300 - 1.8Ghz - 9x200 - 2mb L2 - 65nm Allendale core - SL9TB - FPO/Batch # Q641A275 - Vid 1.3250.

E5200 - 2.5Ghz - 12.5x200 - 2mb L2 - 45nm Wolfdale core - SLAY7 - FPO/Batch # 3827A333 - Vid 1.1625.

E6600 - 2.4Ghz - 9x266 - 4mb L2 - 65nm Conroe core - SL9ZL - FPO/Batch # L702C486 - Vid 1.3500.

E8500 - 3.16Ghz - 9.5x333 - 6mb L2 - 45nm Wolfdale core - SLB9K - FPO/Batch # Q820A837 - Vid 1.212.

Q6600 - 2.4Ghz - 9x266 - 8mb L2 - 65nm Kentsfield core - SLACR - FPO/Batch # Can't remember - Vid 1.3500.



I tested at the following speeds:-

E2140 - Stock (1.6Ghz) - 8x200, 1.22v, ram @800mhz.
- 2.4Ghz - 8x300, 1.25v, ram @960mhz.
- 3Ghz - 8x375, 1.35v, ram @999mhz.
- 3.6Ghz - 8x450, 1.472v, ram@1081mhz.

E4300 - Stock (1.8Ghz) - 9x200, 1.3v, ram @800mhz.
- 2.4Ghz - 9x266, 1.3v, ram @1066mhz.
- 3Ghz - 9x333, 1.4v, ram @1066mhz.
- 3.4Ghz - 9x378, 1.475v, ram @1008mhz

E5200 - Stock (2.5Ghz) - 12.5x200, 1.16v, ram @800mhz.
- 3Ghz - 12.5x240, 1.16v, ram @960mhz.
- 3.6Ghz - 12.5x280, 1.325v, ram @960mhz.

E6600 - Stock (2.4Ghz) - 9x266, 1.31v, ram @1066mhz.
- 3Ghz - 9x333, 1.35v, ram @1066mhz.
- 3.6Ghz - 9x400, 1.45v, ram @1066.

E8500 - 2.4Ghz - 9x266, 1.2v, ram @1066mhz.
- 3Ghz - 9x333, 1.2v, ram @1066mhz.
- 3.6Ghz - 9x400, 1.2v, ram @1066mhz.
- 4Ghz - 1.285v, 9.5x421, ram @1058mhz.
- 4.5Ghz - 1.4v, 9x500, ram @1000mhz.

Q6600 - Stock (2.4Ghz) - 9x266, 1.31v, ram @1066mhz.
- 3Ghz - 9x333, 1.4v, ram @1066mhz.
- 3.6Ghz - 9x400, 1.456v, ram @1066.


The E5200 has a disadvatage when it comes to benching as it has a high multiplier but a low fsb. This means it has less bandwidth available and i could'nt run the ram as fast as i could with the other cpu's. Drop the multiplier and increase the fsb and it more than makes up the difference. More about this later. The benchies i used are:

3Dmark 01, 03, 05 and 06. Default settings.

Aquamark 3. Default settings.

Super Pi 1m, 16m and 32m.

Nuclear MC. Default settings.

Cinebench R10. Default settings.

Fluidmark v1.00. 1280x1024, Default settings.

Lightsmark 2007. 1280x1024 Default settings.

Company of Heroes Opposing fronts in built test. 1280x1024, all set to high or
Ultra where available and 8x AA.

World in Conflict. Built in test. 1280x1024, high no AA.

Crysis. Crysis benchmark tool, all high no AA, 1280x1024.

Far Cry. Hardware OC Benchmark. 1280x1024, default settings.

Prey. Hardware OC Benchmark. 1280x1024, default settings.

All tests run 3 times and average scores taken. One of the reasons it took so long. Anyway, here's the first lot of synthetic benchies:-



As you can see, it's a bit of a mixed bunch. Sometimes the E5200 is beaten by the E4300 and sometimes it's the other way around. What is impressive is that it beats the E6600 in 06. It get's a hiding in Aquamark 3 though. This is mostly due to the low fsb. As a experiment i dropped the multiplier down to 9x and increased the fsb to 266 to get the same 2.4Ghz as the E4300 and E6600 and it beat the pair of them in everything but Aquamark. The E6600, Q6600 and E8500 were all still out in front. Probably the extra L2 Cache was having an effect here. The E2140 is hopelessly overwhelmed and the lack of cache (only 1mb remember) is hurting it.

On to the next batch.



I was extremely impressed by the E5200 here. In NuclearMC, it beats even the E6600 in everything but the Alu test and it is'nt far behind in that. Most impressive. Do the same with dropping the multiplier and increasing the fsb and it does beat the E6600. The poor E2140 and E4300 get left behind quite considerably although they do respond very well to clocking. The E5200 is'nt far behind the E6600 in Super Pi either. The E8500 is all powerful which is to be expected.

Update, 20/03/09

The next batch.



Why did i include Fluidmark i hear some of you ask? Well i included it because you can make large gains by clocking your cpu. Just look at the gains for the E4300 from stock to 3Ghz and 3.4Ghz. Same with the E2140. The same thing applies to some games as you will see later. Once again the E5200 shows what it can do in Cinebench R10. It beats the other two in everything but the OpenGL test where the E6600 has a slight lead once more. Bear in mind that the E6600 has the ram running at 1066mhz and more bandwidth available to it. Do the business with the multiplier/fsb on the E5200 and it pulls ahead again. Nothing can keep up with the mighty E8500 though.

Onto the games.



The poor little E2140 is hopeless here although there are massive gains to be had from clocking it it still lags behind the other's. This is the E5200's achilles heal. It get's beaten by the E4300 and E6600 here. Even the E4300 at 3.4Ghz beat the E5200 at 3.6Ghz and the fsb/multiplier buisiness did'nt work here. Even clocking to 4Ghz made very little difference. An extra 1.8fps in prey and an extra 4fps avg in WiC. I just could'nt get any more out of it. It can't be cache as the E4300 has the same amount. It still does good for a budget cpu though. Wic, CoH and Crysis are GPU dependant but you can clearly see that clocking the cpu does make a big difference to your minimum fps. The average hardly changes but the minimum does, especially in CoH and WiC, making for smoother gameplay. Prey seems to like the extra cache of the E6600 as that starts off well even at stock. The E8500 show's it's dominance here. Nothing at all can keep up with it. The Q6600 although i only benched it with the two games i did play most of my games on it while i had it and my experience was dissapointing to say the least. It was wasted on me as my pc is just for gaming so it was sold off rather quickly.

Overclocking. I could get into windows at 4Ghz (12.5x320) no problem with 1.4v. I could even run some of the benchies at that speed. However, i could not get it prime stable even with 1.43v. I certainly was'nt going to give it any more than that after my E8500 experience. I also could'nt get it prime stable at 3.8Ghz (12.5x304) with the same voltage. Prime would fail after a couple of hours. However, it was rock solid at 3.6Ghz (12.5x280) at only 1.325v with load temps in prime of 38 degrees (my rad is on a windowsill). I have yet to find it's max clock but expect it to be around 3.7-3.75Ghz. Not bad for a cheap cpu. I will let you know how high i can get it. My second pc has a P35 mobo so it clocks well. If you are interested the scores i got at 4Ghz are:-

NuclearMC.
Alu score - 7298.
FPU score - 9419.
MT score 2T - 17648.
Total score - 14211.

Super pi 1m - 13.515 Sec.

Prey - 185.5 fps.

World in Conflict (Min/Max/Avg) - 34/110/62.

I did'nt bother with any of the others as it was not stable. My E4300 can boot into windows at 3.6Ghz (100% overclock) and the E6600 at 4Ghz but neither are stable so i did'nt count them. It would seem that the temp sensors are stuck on quite a few E5200's. Mine reads 32/34 degrees at idle with Realtemp and Core Temp which is wrong as my E6600 does'nt hit that on load. It seems to be ok for reading load temps though. Well, as long as they are over 32/34 degrees anyway.

Update, 20/03/09. I cannot get the E5200 stable beyond 3.6Ghz. It does'nt matter what motherboard it's in (tried 3 different ones), what cooling (air or water), or what voltage. I have tried up to 1.52v which is too much for a 45nm cpu. I am not going to try any more as it is'nt worth frying it for a couple of hundred mhz. I have settled on a permanent overclock of 3.33Ghz with 1.26v. Considering it needed 1.325v for 3.6Ghz this is a good "half way house" and is more than enough for the job it is doing, general purpose/internet pc. I have become a little "disillusioned" with the E5200 though. I was expecting more from it. Maybe it's just that i got used to massive overclocks from the Core2Duo's and i expected too much in the first place. Whatever it is there are a lot of very poor clocking E5200's around. There are quite a few people with very good motherboards that cannot get their E5200's past 3.2Ghz for whatever reason. Most top out at 3.6-3.7Ghz while some very rare ones will hit 4Ghz. These are extremely rare though. From my experience they hate high fsb's. Anything over 300 and you are very lucky indeed.

So, after all this testing what have i learnt? Benching is really boring. Seriously, it is. I hope i don't have to bench anything else for the rest of this year. Back on subject. What did i think of the E5200? Well, for the price i paid (can't say how little as it's against forum rules) it is an unbeatable cpu. It has excellent performance and overclocks well. It is also cooler running and more energy efficient. It is now sat in my second pc and the whole pc is drawing 84w compared to the 87w it did with the E4300. bear in mind that the E5200 is also running 700mhz faster than the E4300!!

Is it worth upgrading too. This is a tough one and not as clear cut as you may think. If you have a E2000 series then yes. It is considerably faster, especially if the extra cache is needed and the E5200 runs cooler and uses less power. If you have a E4000 series cpu then no not really. Performance is similar in games, the E4300 beats it in some synthetic benchies and loses to it in others. The E4000 series are still good cpu's, especially when clocked. Just look at my results to see the gains. If you have one at stock then give yourself a slap and get clocking.

If you have a E6000 series then no. It beats the E5200 in most things and has the extra 2mb of L2 cache which will help in programs that make use of it. Again, if you are running one at stock, get it clocked and reap the rewards.

What if you are building a new budget rig? The E2000 series was a good budget cpu to go for but not now. The E5200 has double the cache, is more power efficient and is cooler running. It is faster clock for clock by quite a bit. It can also be had for a similar price. What about the E4000 series? These should'nt even be considered for a new rig. They are overpriced and use an older 65nm process. There is no question. The E5200 is the new budget king.

The biggest surprise i got from all this was the E2140. I only paid £18 for it off the bay and it is a full retail one. At stock it is next to useless and is only fit to work in a internet pc. Get it clocked though and it shows large gains. Wow, did this thing clock!! 3.6Ghz from a 1.6Ghz cpu!! A massive 2Ghz, 125% overclock. It certainly put's the other's to shame from a clocking point of view even it struggles performance wise. I was going to stop at 3Ghz as i did'nt think it would go any further. How surprised was i when it booted at 3.6Ghz and then passed a quick 10 rounds of linx stress test and then did'nt even crash once during benching. Here's a screenie to prove it:-



As you can also see from the E8500 scores, pretty much anything over 3.6Ghz left me GPU limited as the gains from clocking got smaller and smaller. This is evident from some of the game and 3DMark benchies. I will stick my 4870 back in to see if it free's up any more performance as i have never benched it before.

I hope this answers some of your questions and helps some of you decide what to go for. If you have any questions post away.

Update 20/03/09. This has become more of a general cpu comparison guide now rather than just for the E5200. It's more than what i orginally set out to do and has taken over 100 hours of benching and tweaking to compile all of this info. Each test was carried out 3 times at each cpu speed and then i took the average from the three so you can see it all adds up. The Super Pi runs took the longest, especially with the E2140 at stock. Poor little thing. At least anybody that has a dual core and wants to upgrade to a faster one can look at my charts and see if the upgrade is actually worthwhile doing so.

If you want a gaming rig then the mighty E8000 series is the way to go. Most E8400's will go past 4Ghz easily and if you notice, i tested the E8500 at E8400 speed on purpose. I went down to 2.4Ghz to give a direct comparison with the other cpu's and even then it is faster at the same speed. It just shows the improvements that Intel has made over the E6000 series. The 45nm core cpu's are a big improvement over their 65nm older cousins.

Since i started this the E5200 has risen from £53 to nearly £70 on some retailers sites. This is too much in my opinion and if you really want one you should shop around for the best deal's. A quick check around reveal's it still for sale at £54-57 so it pays to look. The E5300 can also be had for the same price or just a couple of pounds more than the E5200 and the extra .5x multiplier may come in handy if it does'nt like a lot of fsb.

Really nice work. Thanks for taking the time to do all that.

Great work, thanks for sharing it.

As I expected, the 5200 seems to have replaced the 21*0 as the budget overclocking chip.

You're a legend, albeit a legend with far too much time on his hands!! Doesn't look like I'll be replacing my E4500 with a 5200 anytime soon though

Great work pastymuncher, got there in the end .

Your hard work is much appreciated.

Very impressed, thanks for sharing

(Given you some stars)

Well done! The question is do i upgrade!!!? (well, you did already answer that)

stunning work and much appreciated, quick Q I have an e6600 which I can only really run at 3.0 due to my mobo limiting my fsb to 333, would there be any gain in real terms with a e5200 @ 3.6 ish ? obviously given your experiences I value your input.

Thanks in advance

Thanks Pasty! Much appreciated.

Very interesting to see the min frame rate jump up on all CPUs when clocked. That does indeed make a huge difference to gaming as you stated and not a statement one normally associates with increasing fps.

This is a really good thread and shows a lot of interesting results, not least of which the huge impact on minimum framerate the CPU speed shows, which was an above-linear increase in some circumstances. That said, it makes my [email protected] seem really crippled in comparison to either of the CPUs shown. OCing my CPU made several games go from 'crap' to 'very playable' for me. That said, the Q6600 seems to give crappy average and minimum framerates at higher clock speeds when compared with the dual cores which is certainly interesting.

I was thinking of getting an E5200 sooner rather than later, or waiting for 45nm quads to hit the second hand market when Nehalem becomes a bit more popular, I think this makes my decision a little easier - the only other factor being Deneb. Great thread.

Pasty i take my hat off to you
Superb post,thanks for doing that

And..... relax.

Interesting post, I'd completely overlooked the E5200 as OCUK list it with 1MB cache. With 2MB and a 12.5x multi ... it has to be a pure winner.

Good effort

AD

I've decided I'm going to pass my e6600 on via auction and pick up an e5200 as there's going to be little if any cost to consider, I'm limited to 333 fsb there seem's to small gains over the 3.0 I can get with my e6 to the 3.6-3.8 the e5200.

Thanks again

cpatain-chaos said:

stunning work and much appreciated, quick Q I have an e6600 which I can only really run at 3.0 due to my mobo limiting my fsb to 333, would there be any gain in real terms with a e5200 @ 3.6 ish ? obviously given your experiences I value your input.

Thanks in advance

Click to expand...

Sorry for the delay. In your case it would be worth it. It is easy to compare on my charts as well as the E5200 @3.6Ghz results are diectly above the E6600 @3Ghz results. You would see a nice boost. However, what is holding back your E6600? Is it the cpu or an Asrock mobo? If it's the mobo you would see a bigger gain by changing that and clocking the E6600 further as the extra cache appears to help it pull away in games. It is the more expensive option though and the 333 cap will not hold the E5200 back at all. You just want something new to play with.

decto said:

And..... relax.

Interesting post, I'd completely overlooked the E5200 as OCUK list it with 1MB cache. With 2MB and a 12.5x multi ... it has to be a pure winner.

Good effort

AD

Click to expand...

OCUK still has the wrong description after all this time. I think that a lot of people are overlooking it due to the fact that they state it only has 1mb of L2 cache. Pity because it's a fantastic little chip.

Thank you all for your comments. They are much appreciated. As a side note, i already have this back to 3.6Ghz prime stable in my second rig which uses a £35 Biostar TP35D2-A7 P35 motherboard, AC Freezer 7 Pro, 2x1Gb Patriot PC2 5300 and a ATI X1600XT (being replaced with a brand new, very cheap, HD3850). Going to try for 3.7Ghz next and then when i find the final max i will post a screenie of it passing prime for 8+ hours along with the settings so that others may use them.

mame said:

Well done! The question is do i upgrade!!!? (well, you did already answer that)

Click to expand...

Here's something that may help you decide. A while back my son wanted an upgrade from his 3Ghz P4. I got hold of a cheap E2180 and put together a rig from my spare parts and compared it to the E4300. The rig was:-

E2180 (10x240)/E4300 (9x266) both at 2.4Ghz (his mobo restricted his clock as it had no voltage options).
Gigabyte P31 DS3L (one of the best boards i have ever used).
2x1Gb Geil Ultra PC2 6400.
ATI X1900XT.
Samsung Spinpoint 320Gb sata2 hdd.
Samsung 20X DL DVDRW.

I only tested using 3DMark etc but my results were as follows:-

3DMark06
E4300 - Total - 5556 / SM 2.0 - 2104 / HDR/SM3.0 - 2394 / CPU - 2082.

E2180 - Total - 5507 / SM 2.0 - 2092 / HDR/SM 3.0 - 2376 / CPU - 2041.

3DMark05
E4300 - Total - 10979 / CPU - 7747.

E2180 -Total - 10590 / CPU -6792

3DMark03
E4300 - Total - 18338 / CPU Test 1 - 167.6fps / CPU Test 2 - 22.6fps.

E2180 - Total - 17664 / CPU Test 1 - 134.3fps / CPU Test 2 - 20.3fps.

3DMark01
E4300 - Total - 31176.

E2180 - Total - 27559.

The E2180 did go on to hit 3.6Ghz but the gains after 3Ghz were minimal and the extra heat was not worth the minimal performance gains so we settled on 3Ghz for it.

Now while these scores are not comparable to my scores above it does show the E4300 beating the E2180 in every benchmark. The E5200 beats the E4300 in most of the synthetic benchies, sometimes by a considerable margin (Cinebench and Nuclear MC especially) but is let down a little in games. The E6600 is the clear winner here and probably due to having double the L2 cache. Well the E5200 has double the cache of your E2180 so logically it should be better in games than the E2180. Basically the E4300 is better than the E2180 and the E5200 is (mostly) better than the E4300. The E2180 also drew more power than the E4300 for some reason so the E5200 would be a good choice for power reduction too. Of course, it runs cooler too.

I would go for it if i was in your position and the E2000 series are getting good prices on the bay at the moment as well which would offset the cost of changing.

Hi pastymuncher,

Thanks for that it's not an Asrock mobo it s a gigabyte Gigabyte GA-73PVM-S2H matx mobo, it'll run a flaky 360 fsb but is rock solid at 333-350. obviously with a higher multi I can do some damage limitation with regards to overclocking.

I'll pop the e6600 up tonight will be sad to see it go as it'll be cracking but onwards and upwards eh

Thanks again for the time and patience displayed, and all the help given to date

cpatain-chaos said:

Hi pastymuncher,

Thanks for that it's not an Asrock mobo it s a gigabyte Gigabyte GA-73PVM-S2H matx mobo, it'll run a flaky 360 fsb but is rock solid at 333-350. obviously with a higher multi I can do some damage limitation with regards to overclocking.

I'll pop the e6600 up tonight will be sad to see it go as it'll be cracking but onwards and upwards eh

Thanks again for the time and patience displayed, and all the help given to date

Click to expand...

With the E5200 you only need 280 for 3.6Ghz. If you play around with the multiplier though, say drop to 11x and increase the fsb to 333 (still 3.66Ghz) and increase the ram speed you will see what i meant about the high multiplier/low fsb holding it back. One thing's for sure though. You will not be dissapointed.

nice work thanks

Any thoughts as to why the quad appears to bench significantly lower in Games compared to E6600 at the same clocks?

Always thought the extra cache made a small difference in framerates especially at the (relatively) low res of 1280x1024 in favour of the C2Q....