what's your Q9300 overcloking results?

[nyker96]

I'm helping a relative building a new SLI gaming machine. currently COnsidering Q9300. But its stock performance is just not up to par for gaming definitely need to OC it. So need to know your OCing exp on this chip. I'd probably guy the OEM version since I will get a decent cooler like Vendetta 2 for this rig. Thanks.

 

[GuitarDaddy]

I don't have a Q9300, I have an x3350(Q9450) but from what I've seen 3.0-3.2ghz is pretty easy and 3.4ghz/450fsb is the practical max on the Q9300. Around 450fsb seems to be the practical max for the 45nm quads, I can actually get mine prime stable at 470mhz but it takes way more volts than I'm willing to risk for 24/7

 

[Foxery]

Run a search - there have been a few posts, (see sig) though it doesn't seem to be very popular with ATers.

The Vendetta cooler is a wise choice. 45nm or not, these things cook eggs!

 

[nyker96]

@GuitarDaddy so I guess that's 450FSB x 7.5 multipler = 3.38, not too shabby but I guess the physical limit of that chip is probably 4ghz. Sigh too bad that multiplier is sooo limiting. Like you, no way I'm gonna pump too much volt, seen plenty reports that people fry them at 1.45+ voltages. I ain't taking that chance on an OEM chip, maybe if it's retail I'd be little more bold cause the warranty.
@Foxery yeah I think the new vendetta 2 is probably one of the best, right up there with TRUE even beat it in some setups.

 

[AmberClad]

Originally posted by: nyker96
I ain't taking that chance on an OEM chip, maybe if it's retail I'd be little more bold cause the warranty.

Overclocking damage is not covered by the warranty...OCing will void it immediately.

 

[nyker96]

Originally posted by: AmberClad

Originally posted by: nyker96
I ain't taking that chance on an OEM chip, maybe if it's retail I'd be little more bold cause the warranty.

Overclocking damage is not covered by the warranty...OCing will void it immediately.

all the more reason not to overvolt this chip. :]

 

[error8]

Originally posted by: AmberClad

Overclocking damage is not covered by the warranty...OCing will void it immediately.

Well that's what they say, but can they really find out the truth about the torture you've put your chip through? Unless, of course, you'll tell them about it....

 

[AmberClad]

Yeah, let's not go there . I like the chip prices just fine where they are. Methinks they won't stay quite so affordable if we all felt free to overvolt our CPUs into oblivion, because we could just keep running back to Intel or AMD for replacements every time they died.

Originally posted by: error8
but can they really find out the truth about the torture you've put your chip through?

I have no idea. Aigomorla apparently thinks that it's possible that they could determine that if they chose.

 

[nyker96]

Personally I never overvolted a single chip, I always shoot for a practical OC not max OC. The latter is for bragging rights not realistic benefit. But if it's not for a few vanguards amongst us, how we ever find out the max safe voltage on those chips? I'd say most of such reports came straight out of 'practical' experiences :] someone's gotta do the dirty work for the rest of us. my hat off to those who does.

 

[GuitarDaddy]

Originally posted by: nyker96
@GuitarDaddy so I guess that's 450FSB x 7.5 multipler = 3.38, not too shabby but I guess the physical limit of that chip is probably 4ghz. Sigh too bad that multiplier is sooo limiting. Like you, no way I'm gonna pump too much volt, seen plenty reports that people fry them at 1.45+ voltages. I ain't taking that chance on an OEM chip, maybe if it's retail I'd be little more bold cause the warranty.
@Foxery yeah I think the new vendetta 2 is probably one of the best, right up there with TRUE even beat it in some setups.

People assume that since the QX penryns are pretty successful in hitting 4ghz that any Yorkie has the potential to hit 4ghz and I'm not so sure. With as many speed grades as there are for Yorkies 93,94,95,96,97 I find it very likely that the QX chips are highly tested and binned and are capable of the best clocks. I suspect many 93's and 94's couldn't hit 4ghz stable even if they had an unlocked multi.

But you are correct that the lower multi's on the Penryns are a bummer for overclockers, the higher FSB required to overclock them stresses mobo and ram much more and makes overclocking a definate challenge

 

[BTRY B 529th FA BN]

I can't get beyond 3ghz with my 9300... wish i was a little more familiar with Overclocking on an Intel board

Load vcore: 1.248
RAM: 1.8v even with 9.9.9.24 but i run 7.7.7.18
NB: auto
pretty much the rest is on auto

 

[nyker96]

Originally posted by: BTRY B 529th FA BN
I can't get beyond 3ghz with my 9300... wish i was a little more familiar with Overclocking on an Intel board

Load vcore: 1.248
RAM: 1.8v even with 9.9.9.24 but i run 7.7.7.18
NB: auto
pretty much the rest is on auto

really? I think that chip's potential should be higher than 3 considering 65nm Q6600 already can do 3.2+. However, considering the lack of info here. I take not too many bought this chip on this forum at least. So not sure what the average is for this chip.

Just did some googling, some people hit about 3,5 (FSB 475) max with 1.27-1.28v vcore (CPUZ readings). I think either you can do better.

First, what type RAM you running? That's important to know. Making sure it isn't bottlenecking your OC.

 

[BTRY B 529th FA BN]

I thought id get higher too.

Well, I have the DDR3 1600, and my fsb at 400: so I am not pushing the ram much at all. What stumps me is I can barely go over 3Ghz. Even when the memory timings are set to SPD which is 9.9.9.24 ... i tried setting the cpu to 1.350v and still no go. With that, I tried bumping the NB to 1.25v and still no go...

 

[nyker96]

Hey listen, I don't know much about DDR3 boards but try this, see if you can lower the memory divider so the RAM is doing below DDR1600, then slowly up the FSB to beyond 400. Lower your CPU voltage to 1.248 load like before, see if you get further. Careful with the CPU voltage, don't set it above 1.3v. I don't know what is the max voltage but I think most can do 3.5 without going over 1.3v, do don't go over that for safety reasons. Let me know how it goes with this memory divider adjustments.

 

[BTRY B 529th FA BN]

so set the strap to 333, and the fsb to 333 and then push the fsb beyond 400?

i do have the strap at 400 btw

 

[nyker96]

Originally posted by: BTRY B 529th FA BN
so set the strap to 333, and the fsb to 333 and then push the fsb beyond 400?

i do have the strap at 400 btw

I don't have a strap on my board, just a mem divider that multiplies the FSB by a ratio to run the memory at certain speed. I guess this strap is the divider for you, if so set that lower than 400 and at boot up, you can see mem speed. make sure it's below 1600. Then slowly raise the FSB up past 400 see if you a better result.

 

[error8]

Originally posted by: BTRY B 529th FA BN
I thought id get higher too.

Well, I have the DDR3 1600, and my fsb at 400: so I am not pushing the ram much at all. What stumps me is I can barely go over 3Ghz. Even when the memory timings are set to SPD which is 9.9.9.24 ... i tried setting the cpu to 1.350v and still no go. With that, I tried bumping the NB to 1.25v and still no go...

Increase the North Bridge voltage , VTT voltage and south bridge voltage further.
1.25 V for NB is pretty low for a quad.

 

[BTRY B 529th FA BN]

Thinking about dropping this and getting a Q6600 if i can't clock this any higher: link

please read topic title

I will try those recommendations. Thanks.

 

[nyker96]

Originally posted by: BTRY B 529th FA BN
Thinking about dropping this and getting a Q6600 if i can't clock this any higher: link

please read topic title

I will try those recommendations. Thanks.

hey good luck but I still think your RAM is hitting a wall and that might be what's holding ya back. Of course, your m.b. FSB might have a limitation as well, although the newest chipsets like the one you got should probably be able to reach agood 500 before quiting. Hope everything works out for ya.

 

[BTRY B 529th FA BN]

vcore 1.3625v
dram 1.8v
NB 1.4
timings on ram 9.9.9.24
fails at 420fsb prime95 within 1min (3150mhz)

runs like a champ @ 3Ghz without any over-volting

nb auto
mem at stated 1.8v
cpu auto
all basically auto and can run Prime95 for hours stable at 3ghz & 7.7.7.18 1N

 

[BTRY B 529th FA BN]

Well, so far I'm stable at 3300Mhz with a 440fsb

 

[aigomorla]

Originally posted by: AmberClad
Yeah, let's not go there . I like the chip prices just fine where they are. Methinks they won't stay quite so affordable if we all felt free to overvolt our CPUs into oblivion, because we could just keep running back to Intel or AMD for replacements every time they died.

Originally posted by: error8
but can they really find out the truth about the torture you've put your chip through?

I have no idea. Aigomorla apparently thinks that it's possible that they could determine that if they chose.

i have no idea how they do it.

but i know they do check.

I believe they have a bench where they can put the chip on to debug it.

Anyhow, serveral QX RMA's were denied under the ground of excessive voltage.

There not nice with 45nm, and they know most of the time when it does die, its cuz you overclocked it.

If you want my honest advice.

45nm DO NOT LIKE HIGH FSB.

Do yourself the justice and upgrade to a higher Q9XXX. Or wait it out for the higher models to drop in price.

If it was me, theres only 2 Quads you should look at for yorkie. The QX and the Q9650, but those 2 procs are uber expensive.

But i gaurentee those will be the closest low gamble bet you have on getting a 4ghz quadcore.

Put it short, it took me 3 days to debug this stupid board and cpu combo to get this stable.
http://i125.photobucket.com/al.../aigomorla/prime95.jpg

I cant get 450fsb stable. Its just too hard. I think theres a wall there on my chip and board.

And its still tripping the OCCT temp sensor alarm. So i need to watercool it now. :\

 

[BTRY B 529th FA BN]

3400Mhz so far, 2hrs prime In-place large FFT's (max heat, power consumption, some ram) I just bumped up the vcore to 1.275 with load line calibration enabled gave me 1.248v vcore 3405Mhz 454-fsb