Sorry, this is not an article about how plastic bottles affect the environment. 😀 It’s about “bottleneck” issue when it comes to pairing of processor and graphic card.
Is it really THAT bad to pair the lowest end processor with a high-end graphic card? Many claim that it doesn’t make sense to pair the lowest-end processor with any of the high-range graphic card, I am one of them but I decided to find out the truth of the matter for myself.
In this article, we’ll have a closer look on the bottleneck issue.
What Is A Bottleneck?
No, it’s not part of a bottle. It’s a term to describe something being unable to cope with another, thus holding back performance. A choke-point if you would like another word for it.
The users who are concerned over this bottleneck matter are gamers. The thought of what a waste it would be to own a good graphic card yet having it under-perform due to lack of processing power.
Test Setup
Here’s the test setup used for this activity.
| Processor | Intel Pentium Dual-Core E2140 |
| Graphic Card | ASUS EAH2900XT 512MB |
| RAM | 2x 1GB Kingston DDR2 667 |
| Motherboard | ASUS P5K-E |
| Monitor | 20″ Dell Ultra-sharp Wide-screen LCD |
| Operating System | Windows XP (SP2) |
| Driver Version | Catalyst 7.11 |
The Intel Pentium Dual-Core E2140 is just the perfect processor to use for this setup. Simply because of it can be easily overclocked and downclocked.
| Configurations for Intel Pentium Dual-Core E2140 | |||
| Downclocked | Stock | Overclocked | |
| Speed | 1.2 Ghz | 1.6 Ghz | 3.2 Ghz |
| Bus | 800 FSB | 800 FSB | 1600 FSB |
| FSB : RAM | 1:1 | 1:1 | 1:1 |
| RAM Speed | 200 Mhz (DDR2 400) |
200 Mhz (DDR2 400) |
400 Mhz (DDR2 800) |
The stock speed of the E2140 is 1.6 Ghz. To understand how processor affects gaming experience, I’ve overclocked it to 3.2Ghz and downclocked it to 1.2 Ghz as well.
At 3.2 Ghz overclocked speed, the E2140’s performance comes very close to the Core2Duo E6850 processor. So basically the processing power used in this article covers from anything that’s a little slower than the E2140 and up to almost the E6850 range, which is more than enough to find out how processors work with the graphic card and how bad could the bottleneck be.
Test Results
3Dmark06
Default setting.
| Score | SM2.0 | HDR / SM3.0 | CPU | |
| Overclocked | 10780 | 4535 | 5074 | 2728 |
| Stock | 6708 | 2812 | 3656 | 1365 |
| Downclocked | 5354 | 2222 | 3042 | 1042 |
Thoughts : As you can see, the scores are heavily affected by available processing power. The overclocked setup has double the score of the downclocked setup.
F.E.A.R.
Graphic Detail : Maximum
| Processor Mode | Resolution | Average |
| Overclocked | 1680 x 1050 | 59 |
| Stock | 1680 x 1050 | 54 |
| Downclocked | 1680 x 1050 | 52 |
| Overclocked | 640 x 480 | 150 |
| Stock | 640 x 480 | 73 |
| Downclocked | 640 x 480 | 66 |
Thoughts : By the looks of it, you’re set to play this game regardless of what processor you’re using. At high resolution, the difference in gaming experience is negligible. At low resolution, the framerate variance is great but all are above 60 frames per second so you wouldn’t even feel a thing.
World In Conflict
Graphic Detail : Maximum
| Processor Mode | Resolution | Average |
| Overclocked | 1680 x 1050 | 24 |
| Stock | 1680 x 1050 | 14 |
| Downclocked | 1680 x 1050 | 13 |
| Overclocked | 800 x 600 | 41 |
| Stock | 800 x 600 | 14 |
| Downclocked | 800 x 600 | 12 |
Thoughts : World In Conflict paints a different picture. Regardless of resolution, it’s not playable with when the processing power is on the low end. Do note that this benchmark was done with MAXIMUM details. The game has no problem running at 1680×1050 on medium / low details with lower-end cards like 8600GT / 2600XT.
Check out the ASUS EAH2600Pro review. The card is priced at around RM 300 at the time of this article, far cheaper than the ASUS EAH2900XT but as you can see it provides playable framerates at medium detail.
Company Of Heroes : Opposing Fronts
Graphic Detail : Maximum
| Processor Mode | Resolution | Average |
| Overclocked | 1680 x 1050 | 56.1 |
| Stock | 1680 x 1050 | 48.6 |
| Downclocked | 1680 x 1050 | 43.1 |
| Overclocked | 800 x 600 | 121.9 |
| Stock | 800 x 600 | 61.8 |
| Downclocked | 800 x 600 | 50.7 |
Thoughts : While the game works fine with low-end processing, do note that you’re losing out quite a fair bit of framerate when running on high resolution. At low resolution, the game runs fine with everything above 50 frames per second.
Lost Planet : Extreme Condition
Graphic Detail : Maximum
| Processor Mode | Resolution | Average (Snow) | Average (Cave) |
| Overclocked | 1680 x 946 | 22.4 | 29.8 |
| Stock | 1680 x 946 | 22.6 | 23.2 |
| Downclocked | 1680 x 946 | 21.9 | 17.6 |
| Overclocked | 640 x 400 | 69.7 | 50.5 |
| Stock | 640 x 400 | 48.2 | 23.0 |
| Downclocked | 640 x 400 | 38.9 | 17.6 |
Thoughts : Looks like the processing power does determine whether the game is playable. Notice how the average framerate for Cave under 1680 x 946 resolution, notice how it scales with the processing power?
Need For Speed : Pro Street
Graphic Detail : Maximum
| Processor Mode | Resolution | Average |
| Overclocked | 1680 x 1050 | 32 |
| Stock | 1680 x 1050 | 31 |
| Downclocked | 1680 x 1050 | 31 |
| Overclocked | 640 x 480 | 56 |
| Stock | 640 x 480 | 49 |
| Downclocked | 640 x 480 | 42 |
Thoughts : Looks like this is less processor dependant. All games look playable regardless of resolution and processing power as long as you have a good graphic card with you.
Conclusion
As you can see, bottleneck happens BUT it’s not as huge a matter as what people say it is. Sure, some games are affected by processing power but that’s not always the case. Your gaming experience isn’t going to be THAT badly affected by processing power.
I mean look at that, an E2140 going against an overclocked E2140 and there’s not much of difference when it comes to framerates. So we’re looking at a whole lot of $$$ saved while having very negligible framerate difference.
Based on the benchmarks, there’s no denying that more processing power helps your graphic card perform faster. How much faster? Well that depends on what game we’re looking at. A faster processor would also mean you could probably shave a processing / loading time. 🙂
So is bottleneck something you should worry about? No, I don’t think so but you should take note of it. Bear in mind that having better processing power would mean that you’re less likely to face situations where games slowdown due to the lack of it.
Bottlenecks are inevitable. Higher processing power just reduces the bottleneck. Lower-end card increases the bottleneck. You will never be without bottlenecks as there’s no perfect way to match a processors and graphic card.
I also have bottlenecks problem. My cpu is low-end and my gpu is High-end.
My 3DMark06 never reach 10k…… hampeh betul la…….
btw here my spec
AMD x2 4000+ pair with Sapphire HD 3870….
3dmark06 is something you don’t have to treat seriously. 😀
more important is your gaming experience. Look at the benchies from my the article.
Now the X2 4000+ is a little faster than E2140, perhaps around the a E2180.
While 3dmark06 show scores that seem to double, actual gaming only shows varying degree in framerate difference. that’s it. 🙂 So you shouldn’t worry about 3dmark06 unless you’re doing it for bragging purpose.
Nice article. You always come out with something different lol. Nah, too bad i don’t have such high-end rig. Only running p4 1.7 with geforce2 mx400 lol
Nice write up, mate. 😀
Now I know more about Bottle Necks. 😉
My current rig is only a:
P4 3.0Ghz with HT Tech
2.5GB DDR2 RAM 533Mhz
Gigabyte 8600GT 256MB DDR3 RAM
(I noticed that you have done a lot of 8600GT reviews! :D)
So far I am contented with playing newer games on medium-low to medium settings.
A bit older games I can jack it all up! 😀
Cheers! 🙂
You need to try out different architecture . Besides , I believe faster CPU provide more stable FPS and better minimum . Overall , a smoother gameplay .
Yup. Faster CPU would give a better minimum. The faster loading time and all would help overall gaming experience.
i guess RTS games (CoH and WiC) require powerful cpu ….. T.T
i think it is bcoz of the AI…..
i think the most CPU hungry RTS is Supreme Commander.
CoH / WiC # of unit is nothing compared to that of SC (not StarCraft), furthermore SC you can really way point / queue tasks and many other things.
yea so far i know
its nt really diff for games at the time, if u tested it
but how abt future?
my graphic card is 8800gt, i want to know whether there is bottle necks if im using amd x2 4800 or 5000 or c2d e4500? is it these processor r good for gaming performance or any recommendation?
mic, bottlenecks are always there. it’s just a matter of how severe is the bottleneck.
in your case, I’d say the X2 4800 / 5000 and C2D E4500 would lie somewhere between the STOCK and the OCed speed as in the benchmarks taken.
the 8800GT is pretty powerful (I’m using one now) and it’s similar in power to the 2900XT in this test. So those processors are generally fine for gaming, it’s not a problem at all.
so, x2 4800 or e4500 fine for gaming? is there any lagging or poor performance if im using these processor?
right now im using 8800gt, is it fine if im using PSU cooler master extreme power 650w? i dont have any knowledge on choosing a right PSu. thanks in advance goldfries
Those processors are fine for gaming. the PSU is alright too.
thanks for the answer, i want to know more about PSU. is cooler master extreme 650w suitable for SLI or crossfire? i dont know what is the difference between this PSU and TRUE POWER PSU. thanks
Hey bro. This is my rig,
Amd FX-60 (939)
512mb 8800gt
2gb ram
coolermaster realpower 450W psu.
The question is, i dun think my cpu is bottlenecking the gpu, but lets say if i want to SLI. what do you think?? is it worth it???do u think my psu can tahan?? thanks
i don’t know what resolution you play games at but I’m sure it’s bottlenecking.
bottlenecks can never be eliminated, just reduced. in your case, it should be far less a hold back that many medium / low range processors.
as to SLI – I’m not sure, I use the RealPower with 2900XT before. It worked alright but I think it’s at nearly maximum usage. A little overclocking and the thing powers down when on load.
It might stand, I just can’t guarantee. 🙂 I’ll be crossfiring my rig soon with the same PSU as yours.
hmm…
nice article, i faced that before when my 2.4Ghz P4 478pins paired with a GeForce 7600.
The game i was playing is Need For Speed Carbon. as i was playing that game everything runs smoothly, but sometimes when comes to certain parts of the game, some of the graphics may disappear. i think the GPU is producing enough speed, but the processor cant process it so some graphics are left empty.
that happens the most when i was playing GTA San Andreas, the game turns out really like a piece if sh**. well, there are holes everywhere. graphics failed to produce and the game crash lots of time. but when i chnage my graphic card back to GeForce 2 mx 32MB, all problems settles. although i cant play carbon any more and my GTA San Andreas become more laggy.
Doesn’t sound like bottleneck issue though.
Err…just want to know.. Is my processor Athlon XP 3000+ bottleneck my GPU 7600GS Sonic DDR3?
Either way bottlenecks are there. either one could be a bottleneck, depending on the application. 🙂 nevertheless I think it’s a pretty close match.
well here’s my setup
E2140
DFI P965 Dark
2x 1gb Mushkin Value Enhanced DDR667
1950gt 512mb
OCA Powerstream 520w
i overclocked it to 2.6ghz and did it with the help of bryanyeo on memory tweakings
seriously goldfries, i need help on the trial and error for the trcd etc settings…cuz i think the proc can go further
anyways, i never thought that the cikai ram like mushkin value could go up to 800mhz (i downed the 667 to 553)…
🙁
Use slower timings yet? btw this isn’t the OCing article. 🙂
i’ll help u add another set of numbers
ASUS EAH 4850 + Intel D820 2.8/3.2
cod 4 all max @ 1280×1024 – 20-30 fps / 25-39
crysis all max @ 1280×1024 – 22-31 / 28-43
this is the set up i have on a workstation (previously main comp) so the reso sux.
decided to do another test
ASUS EAH 4850 + Q9550 @ Stock 2.83 1280 x 1024 because of the previously used reso
cod 4 online narakka hellish spam server = 150-340 and 615 max i’ve seen whilst lookin at the sky
k. 🙂 others could find these numbers useful. that.
Erm, will my Pentium D 3.0 + 4850 will bottlenecks alot??
Not so clear on the bottlenecks thing, but this is a great thing to add up my knowledge.
Thxs.
I think it’ll bottleneck.
In articles as such, we’re saying that the processor is lacking power to feed the graphic card, hence not running at full potential.
Ideally to maximize the graphic card performance, the bottleneck has to be the graphic card. 😀
I thought just playing the game is hard enough. This tech stuff gives me a headache. I wish I was born smart instead of goodlooking. Wait…maybe my looks are overrated too
Thanks~! I was so worried about that. (Was getting a new gpu for my mega crappy computer.)
I did not know it is dangerous! OMG!
Great post on bottleneck! Bottleneck ensures better gaming.