Sprocket sashays...

[clocker]

I sold the e8400.

Now have to decide between a q6600 (excellent price/performance ratio with the GO stepping) or the newer 45nm q9450 ( approx. 10% performance increase over the 6600 w/ lower voltage but significantly higher price).

At this point, despite the price, I'm leaning towards the 9450.
I already know that my motherboard will easily run a 500FSB, so the chip's lower multiplier shouldn't be a problem and the lower voltage (and therefore, heat output) is very attractive.

The sale itself was a Marx brothers slapstick comedy involving a drug addled internet buddy from Florida and a desperate New Zealander but worked out fine in the end and for the first time ever I actually made a small profit on the chip.
Usually I buy stuff and see the price plummet within weeks, this time I lucked into a widespread shortage of a highly desirable product and sold just in time.

Fortunately, I have backup machines so I can disable Sprocket and ponder my choices for a few days without being completely cut off.
I received notice that my copy of Vista 64 will be shipping early next week and I have a few other irons in the fire, so there is no rush to decide anything right away.

Rest assured, when she comes back online Sprocket will be bigger and badder than before.

[j2k4]

I sold the e8400.

Lordy, you're moving pretty fast.

Given Newegg is usually sold out of the E8400, the market for "pre-owned" versions must be fairly healthy.

How'd you do?

[clocker]

Purchased for $189, sold for $275.
'Twas fun while it lasted but I really wanted to try a quad.

Pondered which quad to get, was primarily thinking of the older 65nm q6600- a proven overclocker (at least the GO steppings)- and the brand new q9450.

The 9450 was really the chip I wanted but it's priced at $350 and that was a bit rich compared to the 6600's $200 tag.
So, I ended up compromising and got the q9300 instead.

It's a 45nm Yorkfield and only sacrifices some L3 cache compared to the 9450.
Microcenter had a deal going...the chip and 2 x 1 GB of OCZ RAM for $299.
So now I have 6GB of memory, even though it can't all be used till Vista 64 shows up (which should be in the next week or two) and a low wattage (1.0875 VID) quad chip to play with.

Stock speed is 2.5GHz and I'm already at 3.3GHz (un-stresstested).
Going to hit 3.5 (fingers crossed) and see what she'll take to run OTTC stable.
Next week, if all goes as planned, I should have another 250GB Seagate to add to the array which will up my disk I/O by another 33%.

I still haven't even touched the timing on the vid card yet.
That will be a whole 'nother adventure.
Life is good.

[Detale]

Damn! You're always one step ahead of me aren't you? I was just looking at the q9450 the other day eh don't have the money right now anyway still haven't gotten the new CPU fan/loop yet. Im saving for a new TV 50" I think

[clocker]

I'm not sure this was a step "ahead" yet...may turn out to be a sideways move.

I'm discovering that quads are not just bigger dual cores and clocking them- or at least, this one- isn't going to be the piece of cake that the Wolfie was.

The 9300 runs a max divider of 7.5, which I knew going in but discounted as a problem since my board/RAM had already proven capable of a 500FSB, so 3.5+ GHz should have been within reach.

So far....no.

Going to have to study the quad threads and see what successful people are doing.
It's already apparent that I'll have to learn about/tweak voltage settings that I previously left on "Auto" and at this point, I have no clue what "safe" vCore settings might be.

Also, I've run into a problem with the video.
After rebooting, many times the PCI-e bus width shows up as 1x instead of the full 16x.
This translates into horrible vid benchmark scores- which is how I noticed this to begin with, I couldn't figure out how K.'s vid scores were almost double mine with the same card.
Still not sure what's going on.

Oh well, if it was easy everybody would do it...

Edit: Hmmm, may have found something to try.
At least I know I'm not the only one with this issue.

Remarkably, I found discussion that deals with my specific hardware- eVGA 8800GT and the Corsair 620PSU- and a Corsair rep was participating.
This is a "known issue" and Corsair is laying the blame squarely on nVidia...no surprise there, eh?
Apparently, the problem lies with the nVidia cards low power draw- which I'd have thought to be a good thing- and the way the card and the motherboard interact at boot.
Somehow there is a transactional disconnect and the card believes it is not getting full juice, so it downclocks itself.
There is no known 100% fix but many have solved the problem by reflashing to an OLDER BIOS...seems odd to me but I have updated my BIOS and even though I can't precisely pin my problem on that, it's the only thing I can think to try.

Off to give it a whirl, wish me luck.

Edit part2: That didn't work.
Damn.

[kaiweiler]

Try bumping the MCH voltage a couple notches, that was the only other voltage setting I really needed to play with the get a good overclock on my quad.
I'm using a Corsair HX520, was that problem specific to the 620? or Corsair PSU's in general? I have not experienced it yet...
Also try turning down some of the visual settings in Vista if you want some better benchmark scores, some of them eat up a lot of video resources and can be disabled for the benchmark.

[clocker]

The problem was not Corsair specific, they were able to duplicate it with a variety of units- it just so happened that a Corsair rep responded in the thread.
It was not a consistently reproducible error either, cards that did it once might be fine after a reboot and vice-versa.

I'm going nuts now.
I have tried everything I can think of...reseated card, tried another PCI 6- pin cable, flashed the BIOS, gone back to all stock settings, broke the RAID array and reinstalled Pro 64 to a single drive- and still the problem persists. There is no obvious manifestation, the screen looks fine, plays movies fine, does everything fine except benchmark. If I didn't know it should be working better, I'd never think twice about it.

Time to contact eVga, methinks.

[Detale]

After reading this I had to run and check but my card is fine. Oh Oh yeah time to contact Evga, I understand they have pretty good customer service though.

On a lighter note what is the max OC you've successfully had with the 9300 so far?

[kaiweiler]

Hmm where do you see that it is only running at 1x? When it runs at 1x does it still use 3.3v on GPU Vcc? Or is there a drop in voltage as well?
I'm wondering if maybe manually setting your PCI-e voltage might help...
Aside from that I can't really think of any other things to try either.

[clocker]

Look at the "motherboard" tab of CPUID...at the bottom it shows the PCI-e link width.
Setting voltages in this Gigabyte is tough...they don't show what the voltage actually is, just allow you to add/subtract from some unknown default value.
I don't trust their defaults either...it shows the CPU voltage as "normally" 1.2v when everything else tells me it's much lower.
I'll have to see where I can find a value for GPU voltage...any suggestions?

@Detale: Last night I was at 3.2 GHz but throttled back to 3 even for stability testing.
I need to go back to basics here and see what my lowest possible vCore is and then decide how high I'm willing to go in search of clocks.
Currently running stock (7.5 x 333= 2.5GHz) at 1.075v (that's the VID shown in CoreTemp, so I decided to start there).

Just like the Wolfdale, one of the sensors was stuck but a round of OCCT freed it up and all sensors now show a movement of 5 in RealTemp, so at least that is going well.
Running temps seem comparable to the Wolfdale also...@ 24-28° idle and high 30's load.
Very little variation between cores, so that's good too.