• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

QX9650 Results and Tips Thread

Associate
Joined
25 Nov 2002
Posts
2,218
Location
Somerset
Arrived at last, the long awaited .45nm Quad core :)

If you thought the cost a bit steep, lets just look at what you get.

1. A very big box... the box is twice as big as a normal C2D box
2. A very big Heatsink... still useless cos who used the stock HS
3. A tiny piece of silcone.

Anyone else mad enough to have brought one of these, please post your findings in here :)

Let me start...
My aim is to be able to run day to day at 4ghz with a nice quiet system. But to be able to crank up the setup for some killer benchmarks. Within half an hour i think i have already achieved my aims..

Currently running Prime (30 minutes in so far) Small FFT at 10x400 = 4Ghz with 1.3v true (so thats 0.5 over volts - std is 1.25). Temps are at 58c full load... This with very poor water, probably no better than standard air cos i'm water cooling the PU, Chipset and 2x GPU's in one loop.

Super-Pi 1M at 4Ghz gave me circa 11.6 score... thats a full 1 second lower than the equivilent clock on my q6600 - so this shows the effect the extra L2 Cache has on this benchmark.

Idle temps are sat at 27c - 30c


I had a quick go at a bigger clock with these results...
To keep things simple i just increased the Multi to 11 leaving everything else the same. To get into windows i had to increase the vcore to 1.4v This was still not prime stable, but i dont want to go higher just yet.

In this I ran a 3dmark2006 test and scored 21.5k beating my previous best by 1.5k

I then increased the FSB to 410 while in windows so 4.5Ghz and ran SuperPi = 10.33

So early days and i really need to sort my water cooling out first but initial impressions are FANTATIC and despite the cost, well worth the money.

Latest Screen Shots last updated 4/12/07 5:50PM
First 4ghz low voltage Prime Stable for 1 Hour
primeqx9650@4ghz.jpg


Next - Max Overclock with Sub 10sec SuperPi
 
Last edited:
:p

Seem to run a lot cooler than the current quads despite the clock increase. Dunno if i would fork out £500+ notes on one though. :eek:

Here is my thinking...

When Penryn is fully launched i dont think we are going to see very high clocks because the multi's are goign to be set too low.

Penryns will be set to run default at 1333 or 1600 FSB (333 or 400) and with the multi locked (on non Extreme Editions) you will struggle to get a high clock speed on the CPU unless motherboards get significantly better - which is not on the cards as i see it.

For example a stock 2.4 chip will be stock 333x7 or a 2.66 333x8 and so on.
Now to get the really good clocks i.e. 4Ghz you will need to run your motherboard at say 571x7 or 500x8 which means the board will fall over long before the CPU does.

So... by my logic, it may be expensive but it should last a long time and hold its value because its going to be very sort after in the overclocking comunity.

Unless Intel bring out a 266FSB version or motherboards cope with 500FSB + all day long then i think this is the only way to go.
 
Great results. Make sure you enter it in the superpi table and Quad database!

will do, but fancy getting sub 10 seconds first :)

I have some more substantial water cooling kit in front of me, but am waiting for some 3/8 splitters before i can set it up... they are due tomorrow.
 
But it's only 1.5k on 3dMark faster than his old system, which is an increase of 7.5% for a processor that costs £700 - hardly a low price! Forgive me for being facetious, but that's £100 per % increase!

3DMark06 is a Graphics benchmark, not a CPU performance test... yes it does get affected by the CPU, not least because Futuremark put 2 CPU tests in a GPU benchmark program??? But the CPU plays a small part in the overall score...

Lets say that 80% of the benchmark is GPU related that leaves 20% CPU related. Now a 7.5% increase in total score purly down to the CPU is quite impressive... More like a 30% increase. I know i'm guessing the percentages, but the principle is right.
 
Guys this mobo is hard to work with. I had a little go at 8x Multi and it would not post at all :( even at 400FSB with the same settings as run perfectly at x10

I will try and re-visit soon.
 
Ok some updates for you...

Firstly Lower multi results on this board.
I tried a few combinations and basically the board is good for 435 FSB with default settings, that was within a few FSB for 7x or 8x Multi.

With a few tweaks to the NB and PLL volts I got up to 455 fsb again with both multi's. After that stability was not there, although i did not try too hard.

I would say that the max stable setups will be at circa 450FSB although you should remember that i have water cooled NB and the chipset does get hot.

Therefore an 8x Multi CPU will do 3.6Ghz. I still say that the chips will probably do more than that, but its the boards that will limit things.

---------------------------------------------------------

My overclocking on the chip is still going well.

I have managed to do..
SuperPi 1 at 4.7Ghz
Prime 8hr Small FFT at 4.3Ghz
3dMark06 and 01 at 4.5Ghz

Does no one else have one of these yet? Feeling very alone :rolleyes:
 
Are you sure its the motherboard ?, 520+ FSB seems quite common with that motherboard. Ive read the Penryns need lots more vcore to achieve high FSB, maybe its true.

Reading ox XS and chatting to the guys over there i have not seen any one able to post reliable and stable bios settings for greater than 450fsb, some people are close at 475 and some people claim 500 - but these always disapear when you start asking how.

I'm sure its tweakable to high 400's and 500's for suiside shots etc. Its the NB temp that starts to get out of control even with water.
 
Back
Top Bottom