best fsb speed for gaming

[bcham]

i have my quad6600 at 3gig,it runs nice and cool and stable at default voltage,
if i increase the fsb to say 3,6gig will it make any difference in games.from 3gig.will i see higher fps,

 

[uv]

yes, you will.

 

[Guest2]

Yes. I very much doubt you will be able to keep default voltage though

Someone posted benchmarks of Cryis running Q6600@stock, Q6600@3ghz and [email protected] amongst other CPUs. Results showed that with similar setup a faster CPU increases FPS by quite a lot (if not bottlenecked by other hardway, the GPU)

 

[Marine-RX179]

Someone posted benchmarks of Cryis running Q6600@stock, Q6600@3ghz and [email protected] amongst other CPUs. Results showed that with similar setup a faster CPU increases FPS by quite a lot (if not bottlenecked by other hardway, the GPU)

It was me who posted that I was copying the results from Custom PC magazine. But I later found that their website (bit-tech) actually got the results online:
http://www.bit-tech.net/hardware/cpus/2010/03/03/overclocking-intel-core-i3-530/8

The bottleneck happens in this situation because it is a 5870, which a lot faster than a 9800GTX+. I think the GPU bound of 9800GTX+ would actually kick in earlier than a Q6600 reaching bottleneck if it was overclock to 3.2GHz or above.

 

[alexp999]

Depending on your mobo you might not needed to up any voltages.

On a P45 it can do 400 FSB on stock volts. Drop the multi and you are running 3.2 GHz with nice 400 FSB to alleviate any potential bottlenecks.

Course if you are talking about cpu clock speed then thats another matter

 

[drunkenmaster]

You might see marginally higher FPS, but not much. That review shows the difference between 2.4Ghz, and 3.7Ghz, due to diminishing returns and GPU limits, the first 600Mhz would give massively more performance than that last 700Mhz, as you can see by the results of similar chips but in dual core, at 3Ghz being noticeably faster.

But the point of CPU REVIEWS is they ALWAYS show at less than gpu limited resolution/situations on purpose to highlight the difference. Its a 5870, at 1920x1200(almost standard resolution for that power of card) those results would be hugely closer and a 3Ghz Q6600 is likely to be very close to the 4.15Ghz i5 result.

A 9800gtx would also be showing far closer results across the board due to less power and being far more gpu limited. Thing is, its less about number of cores, the threads aren't all equal load/power requiring, the heaviest loads will run on only one core and need a minimum speed to run, once you get to that speed(clearly a little above 2.4Ghz, but not much) then you're basically above the cpu limits required.

You should be able to go quite a bit higher on the CPU though, maybe with a little voltage, and a little extra voltage even on the stock cooler, is absolutely fine, if on 3rd party cooling, its a non issue.

IIRC some chips would do around 3.2Ghz before needing a voltage bump, and they tended to scale well from 3-3.8Ghz with voltage increases, though the higher end of that would often require VERY good air or water cooling to keep temps stable. 3.4Ghz is very easy on basically al Q6600's with a minor bump in voltage, frankly theres no reason not to run a little faster.

 

[Guest2]

On a P45 it can do 400 FSB on stock volts. Drop the multi and you are running 3.2 GHz with nice 400 FSB to alleviate any potential bottlenecks.

Course if you are talking about cpu clock speed then thats another matter

Didnt think about that. I run 400 x 9 = 3.6ghz but I seem to recall testing at 500 x 7 = 3.5ghz and it worked.

I havent tried using 515 x 7 = 3.6ghz instead of 400 x 9 but I imagine the FPS increase in games is very little if any. Not worth it for the extra messing around and testing that its stable.

If I get some time when im bored I might benchmark Crysis at 400 x 9 and then try at 515 x 7

 

[Marine-RX179]

You might see marginally higher FPS, but not much. That review shows the difference between 2.4Ghz, and 3.7Ghz, due to diminishing returns and GPU limits, the first 600Mhz would give massively more performance than that last 700Mhz, as you can see by the results of similar chips but in dual core, at 3Ghz being noticeably faster.

But the point of CPU REVIEWS is they ALWAYS show at less than gpu limited resolution/situations on purpose to highlight the difference. Its a 5870, at 1920x1200(almost standard resolution for that power of card) those results would be hugely closer and a 3Ghz Q6600 is likely to be very close to the 4.15Ghz i5 result.

Actually the reason why Q6600 overclocked to 3.7GHz is bottlenecking the 5870 in Crysis is because the game was not optimised for Quad (you can see i3 530 at 4.07GHz got identical frame rates to Phenom II 965BE at 3.99GHz), not because of resolution being at 1680x1050 instead of 1920x1200. The frame rate of 33min, 57average is the GPU bound of 5870 at those settings.

It's not listing there, but a i7 980X at 4.4GHz and 4.72GHz (watercooled) on a Asus P6DT Deluxe with a 5870 is 40min, 57average for those settings in Crysis. The average frame rate for GPU bound is pretty determined by the graphic card, whereas the minimum frame is more affected by the CPU's speed and architecture design, as well as the speed of motherboard (a i7 980X at 4.4GHz on a Gigabyte GA-X58A-UD3R with a 5870 is 42min, 57 average, which is 2fpsmin higher than the Asus P6DT Deluxe set up).

In games like Battlefield Bad Company 2, an overclocked Q6600 would be able to use a 5870 to 100% capability with no bottleneck, just like its i5 750 or i7 920 big brothers.

 

[drunkenmaster]

Can't say I'm sure what your point was, I didn't say why it was bottlenecked, nor did I say it was a bottleneck.

I said in CPU reviews they always run tests at lower res, all I said was, at a higher resolution, one appropriate for the card being tested, the results between top and bottom would be MUCH closer.

LIkewise, in 1920x1200, with af/aa on the I7 at 4.72Ghz will be much closer to a 4Ghz P2, than it would be at 1680x1050 and no aa/af.

it doesn't matter if this one specific game(remember the OP didn't ask about Crysis, but gaming) isn't quad core optimised, or if none are, a Q6600 still has one, and two cores, and more speed = more speed if its a quad, a dual or a single core.

The simple fact is the overclock from 2.4Ghz to 3Ghz will yield far higher increases in performance than 3Ghz to 3.7Ghz, and remove the bottleneck in the VAST majority of games.

 

[Marine-RX179]

My point was the op shouldn't have to worry too much about CPU bottlenecking the 9800GTX+with the Q6600 overclocked to 3.0GHz. He shouldn't have to concern himself with overclocking Q6600 to 3.6GHz unless he's upgrading to graphic card of 4870x2 or 5850 and above, and playing games that are not optimised for Quad (well, basically all games pre-date Windows 7 I guess). A Q6600 overclocked to 3.2GHz should pretty much remove any or most bottleneck for graphic cards up to HD4870/GTX260, even for games that runs pretty much on the 2 main cores.

drunkenmaster, sorry if I sounded like accusing you are wrong or something...I was merely quoting what you said for the sake of providing additional info, not really directing at what I was saying to you.

I think the Crysis example serves as a good example of showing how Quad-cores perform with faster cards such as 5850 and above in games that are light threaded/not optimised for Quad (most games that pre-date Windows 7) that don't really use the 3rd and 4th core much. It illustrates that Q6600 overclocked to 3.7Ghz and Phenom II 965BE overclock to 3.99GHz would bottleneck cards such as 5870 if their 3rd and 4th core are not being put into use.

I think the reason why they haven't bother to show test results for 1920x1200 at Crysis is because even with a stock speed 5850, the minimum frame rate would drop to sub-25fps even when without AA/AF on, and that's considered as below playable smooth level already, so 1680x1050 is better balance between quality and performance than 1920x1200. Playing Crysis at 1920x1200 with AA/AF switch on would have to sacrifice too much frame frame and performance in exchange, which is probably not worth it.

I'll quote some results from their magazine instead:
Crysis, DX10, HIGH
1920x1200 0xAA, 16xAF
HD5850 min22fps, average 38fps
HD5830 min 16fps, average 31fps
HD5770 min 15fps, average 27fps

1920x1200 4xAA, 16xAF
HD5850 min17fps, average 32fps
HD5830 min 11fps, average 24fps
HD5770 min 11fps, average 22fps

 

[mrk]

CPU increases don't make much difference for me in Bad Company 2 because my GTX260 is the bottleneck at 1920x1200 high settings. I've played at 3.2GHz and 3.6GHz, both give the same numbers. Overclocking the GFX card though gives a few extra frames and a higher minimum fps.

 

[bcham]

thanks for the info,ill try clock it to 3.2GHz,at stock volts,just buy increasing my fsb,and putting the ram at 667mhz,
will 360mhz give me 3.2GHz,thanks.

 

[Marine-RX179]

thanks for the info,ill try clock it to 3.2GHz,at stock volts,just buy increasing my fsb,and putting the ram at 667mhz,
will 360mhz give me 3.2GHz,thanks.

Is you memory 667MHz or 800Mhz? If it is 800MHz, you could try lowering the multipler to x8 and use 400FSB (if you motherboard let you), that way you would have Q6600 at 3.2GHz and RAMs at 800MHz at 1:1.

You might need to raise the vcore by 1 step up at a time if you can't stable it on stock voltage. Run Prime95 for an hour or something (and if the voltage isn't enough, you would get BSOD and PC reboot in under 15mins anyway).
My crappy nforce board need +0.05v to get my Q6600 stable at 3.15GHz at 9x350, ram 800MHz, FSB: DRAM 1:1. My board would become unstable at over 350FSB even if I add extra voltage...it must be the low FSB wall imposed nforce chipset that everyone's talked about.

What motherboard you got by the way?

Also, rather than just overclocking your CPU, you should try overclocking your 9800GTX+ as well. MSI Afterburner is good and easy to use.

 

[Guest2]

will 360mhz give me 3.2GHz,thanks.

Yes it will, but 400 x 8 is more commonly used and will be better because of the higher fsb. Then just up the multi to 9

 

[Sukh]

What are stock volts for Q6600?

 

[mrk]

Depends on what VID the Q6600 in question has.

 

[Vince]

Vids range from 1.2 to 1.34 for a Go afaik.

 

[Rroff]

Q6600 is pretty similiar to the E6600 in respect to overclocking, tho it does put more strain on the North Bridge and power circuits, but on an ideal motherboard both do 3150MHz +/- 50MHz on stock voltage and very very very few will do anything near 3.6gig on stock voltage.

 

[bcham]

Is you memory 667MHz or 800Mhz? If it is 800MHz, you could try lowering the multipler to x8 and use 400FSB (if you motherboard let you), that way you would have Q6600 at 3.2GHz and RAMs at 800MHz at 1:1.

You might need to raise the vcore by 1 step up at a time if you can't stable it on stock voltage. Run Prime95 for an hour or something (and if the voltage isn't enough, you would get BSOD and PC reboot in under 15mins anyway).
My crappy nforce board need +0.05v to get my Q6600 stable at 3.15GHz at 9x350, ram 800MHz, FSB: DRAM 1:1. My board would become unstable at over 350FSB even if I add extra voltage...it must be the low FSB wall imposed nforce chipset that everyone's talked about.

What motherboard you got by the way?

Also, rather than just overclocking your CPU, you should try overclocking your 9800GTX+ as well. MSI Afterburner is good and easy to use.

mother board is an asus p5kpl ,ill try 8x400 and setting my ram to 800mhz,
i can use 1066 ,800,or a667mhz my ram is ddr2 800,
i tried 360mhz fsb .but pc wont boot
in the cpu ratio setting this is set on auto ,what do i put in the box ,i have to use four digits,is it 0800.or 08x0.

 

[Marine-RX179]

Does your bios interface look like this:
http://www.youtube.com/watch?v=xHbf_tA3kUM

or different?