Not really, changing timings from 5-5-5-15 to 4-4-4-12 doesn't do squat to my superPi. Anyway, here it is: 17secs with stock vcore. Once I am able to modify the vcore, I think I could go beyond 3ghz.

This post has been edited by jarofclay: Aug 2 2006, 10:39 PM


Attached thumbnail(s)

However, I do see a difference in terms of amount of L2 cache. 2.88ghz with 4MB L2 can get lower times compared to a 3ghz ++ 2MB L2 cache E6300. If the E6600 can run at 3-3.3ghz and above, I think 14s is not impossible since every 100MHz increase roughly translates to a second off superPI.

Problem is the Intel D975XBX doesn't allow adjusting the Vcore without any modding. Fear not though, X6800 is on the way and with multipliers enabled upwards on 975X chipset, I think 3.5 and above might be achievable.

Ehhh, why the stock voltage is shown as 1.31V?

If you check my screenshot, it was 1.175V stock. Let me check some documents on what's the correct default voltage.

34K in 3dmark01se with just an 8 pipeline card! Unprecedented!!

The most important thing for this is to have a mobo that can have high MCH voltage so that you could push the FSB to 420 and above. Just for comparison, Intel Bad Axe has upto 1.725V for MCH. Most other boards has upto 1.6++V for MCH voltage IIRC.

Just curious, did you guys enable Core Multiplexing in your testings?

It's supposed to enable the whole L2 cache to a single CPU when you run any single threaded software such as SuperPI. Supposed to boost some performance. My tests were done with this enabled; hmmm I should try disabling it to see how it goes.

@Irenic, I think it was 1.5V or 1.4V. I'll confirm it out for you later.

@gengstapo, one thing that most overlooked is that E6600 has 4MB L2 cache and that could give you roughly 10-15% increment in performance. That's why I was able to get sub 18s with 2.88ghz only while others get more than 18 sec even running at 3.15ghz.

Most important criteria to reach high FSB is a good mobo that gives you a wide range of FSB and MCH voltage. With that, you could get lower multiplier chips and save up extra money for other stuffs such as fatbody Microns.

Guys, you all might want to post this in the superPi post as well? I think it's time we post the times there The link is here.

This post has been edited by jarofclay: Aug 4 2006, 03:51 PM

From what I heard from the Tech Summit, it seems it does more than that. However, let me just check to see what you said is true...

Perhaps waiting for the 775 RTM would enable you hit higher?!

Your water cooling retention module.

Cool, as long as there's a thread. Didn't notice that thread there.

Actually, he is kinda right. Clock speed yields more benefits than memory timings.

Not yet, been playing with my Atom the whole weekend now... will try it later.... am curious to see that as well.

Yeah, confirmed that the OS shows only a single CPU once this feature is disabled. When enabled, I see two cores just as you suspected.

However, it's info is very scarce as this is something Intel doesn't really wanna market out. Not sure why.

I am getting my X6800 next week or so. Do you have the link to that? They raised the FSB only or did it with upping the multiplier as well?

Sometimes, I just begin to wonder about this obsession of priming for hours on end. I've had my X2 failing prime at 2.6ghz but never encountered a single hang with FEAR, Q4, Most Wanted, HL2, Oblivion and any other games I thrown at it. To me, as long as it can pass the benchmark; even if it fails prime for 5 or 6 hours, I consider that a successful benching speed.

Just my thought... unless you're planning to use your rig to build rockets that must have absolute precision.