Was it worth it?

Associate
Joined
12 Sep 2009
Posts
659
Location
Harrogate
Hi, I'm a bit of a newbie here and as this is my first post please do not be offended by my lack of o/c knowledge.
After researching on your forums I decided to get my feet wet and try clocking my Q6600. I guess the info I gleened was all well and good as the process went fine and my cpu now runs at 3ghz, doesn't even go above 40c on prime!

I ran Vantage and went from 11128 to 12158. Quite pleased with that :)

THEN, I ran Crysis Warhead, Stalker (cop) and Ungine Heaven benchy's.

NO difference in fps for Crysis
+2 fps for Stalker
NO difference in fps for Heaven.

Are these score's to be expected??
I was kinda expecting a little more!!:(
 
Sounds like something else was the bottlneck in the first place. :-)

2fps can be awesome or crap depending on what you started from. On my laptop I'd kill for an extra 2fps, on my desktop it just pushes it from "fine" into "fine+2fps"

Overclocking generally doesn't give that much improvement in gaming fps since that's almost all offloaded to your GPU. You might notice the difference more in strategy games than shooters since they're much more CPU intensive (keeping track of 10,000 units rather then you and the 4 you're shooting).

The main place you'll notice the difference is in applications. You'll tend to find that if you use photoshop, maya or other "artsy" programs that's where you get the benefits.
 
Are these score's to be expected??
I was kinda expecting a little more!!:(
Hello Mangochutney!

Welcome to OcUK forums! :)

Please read and understand the forum rules and enjoy your stay!

What resolution were you running the gaming tests at? . . . if it was high resolution (i.e 1920x1200) the GPU becomes a bottleneck but if your tests were at lower resolution (1280x1024) you should see some good gains with an extra 600MHz processor frequency. You would also see some nice gains if you cranked your memory frequency high, I hope you don't have it running at DDR2-666 or DDR2-800, hopefully not! :cool:
 
The main place you'll notice the difference is in applications
I agree with that, a good result from overclocking a quad core from stock speeds up a lot higher would be the reduction in time of heavy duty encoding, something like one hour quicker with the overclock etc! :cool:
 
Thanks for the replies guys.

I ran the benchy's at my native resolution 1680x1050.

My RAM is DDR2-800 and I purposely set the freq to 800 in the BIOS, I was concerned that the CPU adjustment would push the RAM speed beyond its 'safe' speed and damage it. Is that wrong, should I leave it on AUTO?

Do you think I should get some 1066 RAM?
 
Personally, I don't notice much real world difference between 800 & 1066 RAM. Certainly not enough to justify the stupendous price of upgrading DDR2 at the moment.
But you can try getting a little more out of you RAM by altering the dividers in conjunction with your BUS speed, you won't damage your RAM but you may make your system unstable or refuse to boot if you go too far, that's all. AFAIK it's only additional voltage that will maybe damage the RAM.
 
If you got DDR2-800 then your all good! :)

If you could swap it for DDR2-1066 that would be great but probably a hassle and expense! ££

I've just swapped out some older DDR2-800 on a friends system, got a used 4GB set of Corsair DDR2-1066 and cranked the memory frequency as high as the older motherboard would allow (ASUS P5B-Deluxe/Intel P965) . . .

fastddr2.gif


If you get time you could maybe try some overclocking your sticks, something they won't give much more but sometimes they have a few hundred MHz hiding, probably worth a try! :)

Even with the chip just overclocked a little to 3.0GHz the whole system *flies* :cool:
 
Thanks for the replies guys.

I ran the benchy's at my native resolution 1680x1050.

My RAM is DDR2-800 and I purposely set the freq to 800 in the BIOS, I was concerned that the CPU adjustment would push the RAM speed beyond its 'safe' speed and damage it. Is that wrong, should I leave it on AUTO?

Do you think I should get some 1066 RAM?

Not really, chances of needing 1066 ram on a Q6600 is pretty slim, but setting 800 will actually be 800 + whatever % overclock you have.

I've always found very minimal gain in having memory clocked higher than the CPU, so at stock cpu FSB 1066, memory dual channel 266mhz (533 ddr2). To actually "NEED" 1066 ram, you would need to overclock the FSB to over 2Ghz, and realistically thats not gonna happen.
 
I hope the people who are saying faster memory isn't all that have run a system at some point with the memory running at:

  • DDR2-1000 4-4-4-10
  • DDR2-1200 5-5-5-15
On the Intel® Core™2 LGA775 platform it's not just about Memory Bandwidth but also about System Latency, if you can get the memory running at increased frequency (inc good timings!) the Northbridge will lower the tRD setting (aka Performance level), the lower tRD is the *snappier* every single memory transaction is. Once you get used to a low latency system its hard to go back . . .

I ran the benchy's at my native resolution 1680x1050.
Ok so that makes sense right, your scores are not increasing much with the CPU overclock as the GPU will be the bottleneck at that resolution . . . .

If you reset your system back to stock and run some fresh benchmarks at Medium/Low quality at 800x600 and then apply your overclock and re-run the tests you will see the results scale much better than before, this is obviously due to the low spec graphic details, it removes the GPU from the equation and the strain is placed on your computers subsystem (CPU, FSB, MEM).

Running the low graphic spec benches is a useful way to test bits of your overclock and although your not likely to ever run a game at that res it shows you what would happen when you install a much more capable GPU, the scores would scale higher on the overclock machne . . . :cool:
 
Thanks for the advice, I can just about get my head around the CPU & GPU o/c'ing but I do struggle with understanding the RAM aspect. I will check out the memory forum for further pearls of wisdom.
Sorry for my retardation!
Oh, and I'll compare the benchmarks at a lower resolution like you suggested BW, at least then I should be able to see the difference and give some much needed confidence in this exciting new world of overclocking!
 
I can just about get my head around the CPU & GPU o/c'ing but I do struggle with understanding the RAM aspect.
CPU overclocking is the most popular, FSB overclocking is the next most popular and the last and mostly overlooked aspect is the Memory clocking/tweaking! :D

To recite an old proverb . . .

"A chain is only as strong as its weakest link"

I'll let you off for the moment as your just getting to grips with lots of new theory but I'm convinced a system that has a good all-round balance of speed (CPU + FSB + MEM) is preferable to an imbalanced system . . . just my pennies worth . . . good luck! :cool:
 
CPU overclocking is the most popular, FSB overclocking is the next most popular and the last and mostly overlooked aspect is the Memory clocking/tweaking! :D

I agree with that order is the most popular, but I believe FSB overclocking is more important as long as its going along with the core speeds going up as well.

8x450 is faster than 9x400 just due to the extra 100mhz NB frequency

Also, I don't think his 5850 @ 1680x1050 res is going to be the bottle neck here.
 
I've looked into it and I did overclock the FSB, the biostar bios calls it the ' CPU frequency setting'. I did some maths and changed it from 266mhz to 333mhz to achieve the 3ghz overall.
I also gave it a little voltage boost +0.125v, no idea why but just went along with some posts!
Apart from setting the RAM speed, I made no other adjustments.
 
I agree with that order is the most popular, but I believe FSB overclocking is more important as long as its going along with the core speeds going up as well
I think they all should be tweaked, my point is most people only really care about CPU MHz, it's a common talking point, it's not often to hear someone braggin about having ram that runs at 2.5GHz :)

8x450 is faster than 9x400 just due to the extra 100mhz NB frequency
Are you talking about the NBCC? i.e 8x450 on a chip with default multi of 9 results in a 506.25MHz NBCC, 106.25Mhz faster than the same chip running at 9x400?

If so I agree however I didn't say anything to the contary? I'm just pointing out that memory Mhz needs a bit more loving! :D


Also, I don't think his 5850 @ 1680x1050 res is going to be the bottle neck here.
You wouldn't think a HD 5850 would have too much trouble running at 1680x1050 but if a game doesn't yeild any extra frames after a 600MHz CPU clock increase what else could it be?
 
You wouldn't think a HD 5850 would have too much trouble running at 1680x1050 but if a game doesn't yeild any extra frames after a 600MHz CPU clock increase what else could it be?

VSync? :)

If you've got vsync turned on your FPS won't go above your monitors refresh rate. eg. my 5850 varies between 100-300fps in CS:S at 1920x1200 without vsync, 60fps constant with.
 
With most games, my system does not struggle at all. MW2 sits on the 80fps multiplayer cap all the time and averages at around 120fps on the single player.
Its the Crysis and Stalker benchy's I've seen only a 1-2 fps gain on and was expecting a bit more!
I'm not such a fan of Crysis, its Stalker I'd like to play on full settings, I bloody love that game!! :D
When I tried o/c'ing the 5850, I noticed that on Vantage the graphics score went up but the CPU score went down! So maybe I should try o/c'ing my cpu more?:rolleyes:

P.S I have noticed a big decrease in the time it takes to encode video.
 
Last edited:
Back
Top Bottom