Legit Reviews

Well, for the past few days I've been playing around with dual core and so far it's been an interesting ride. At the lowest multiplier (14) I can max out the system at a 30% FSB overclock and use the most radical memory divider (1:2) to reach 260FSB and 1GHz+ Memory scores. At this speed the CPU is run at 3.7Ghz.

Intel told me from the day I got this processor 4GHz is possible on air and with water cooling 4.2GHz is possible. I've lost sleep trying to match what they said should be possible.

With water cooling the best I can pull off is this:



It was stable enough to benchmark and validate at 3.8GHz, but anything over 3.8GHz was totally unstable and the system simply turns off.

This is with watercooling, fans over the voltage regulators, on the open test bench with an OCZ PowerStream 620W PSU feeding the beast plenty of power. I've tried maxing out the voltage at 1.6V CPU and pushed the n/b and mem to their max with no luck of going higher. 20x200 will post and then blue screen on the XP Pro splash screen.

How are performance numbers at 19x200 on Dual Core? Well this is why I haven't done an overclocking article!

This is a STOCK Intel 840 Extreme Edition at 4-4-4-7 timings:



here it is at 3.8GHz:



Interesting eh?

wow... something's gotta be way outta spec... have you tried any testing utilities? Like, I see that the FSB is at 200 there, but have you tried memtest86 at 19x200? my guess is that some other part of the system is out of spec, causing the performance to fall on its face... but if all you did was up the multiplier, this makes no sense.

yeah something isn't right... I tried low voltage on everything, default voltages, max voltages and nothing helps... everything else is set up the same. Anything above 3.7GHz on this system gets really strange. It honestly just turns off and wont restart till the entire system is drained of power then you can restart the system again. Very strange indeed.

If you click the validator link you can see memory timings, dividers, and all that jazz for that side of things.

are you using a SATA drive or PATA, because that could have something to do with it, some people still report better overclocks with PATA.

I'm using a Seagate SATA 7200.7 drive with NCQ... It's the same drive that went 4+ GHz on this board at 294MHz FSB, so i don't think it's a drive issue.

That's very strange indeed, Tom's got the Pentium D up to 4 gig on water
cooling with 20x200. they did say however that the system is not
very stable above 3.8

but i have just realised how they bugged the scores,
they used terrible timings on their RAM!
at 667 they had timings of 5.0-5-5-10, which must have quite an impact
in some of the tests. anyway, just me ranting.

I am by no means saying that this was done, but often many sites take a processor score and use simple math to get the next processors performance levels based off the previous ones. Something else to think about.

you cant oc the bus higher right? From the way it seems, you have an Intel board without some of the more advanced options? I really want to see people overclock the FSB of the Pentium D. I was talking to my friend about HT about a year or two ago when the 2.4C was an extremely popular and he was proud of his fathers Shuttle SFF. He was saying that the reason why HT decreases performance while video editing is because there is so much data being sent through the front side bus that HT will actually bottleneck that. So, based on that, I would like to see people overclock the front side bus of the Pentium D really high. In theory, performance would increase tremendously because right now, it looks lie the FSB is being shared by two cores. So in theory, each core can get say 3.2 GB/sec which is that of the original Pentium 4. When they increased the FSB, performance dramatically increased. This just shows that with multi cores, the world needs faster FSBs and faster ram and integrated memory controller. If Intel had used the integrated memory controller, its DDR2 performance should be a bit better. They would also be able to give each core more bandwith and you would most likely see much greater performance.

hey have u tried a vapor cooling system that they are selling at http://www.crazypc.com. i have 1 and the get my overclocked Intel EE into the -10 Farienheit

Capper owns one

I've been running the Pentium D 840 EE at 3.84 24/7 (Idles at -48, load temp -42). I was running at 4GHz, but it would crash during benchmarking. I'll try maxing things out again today and posting some results

I can run windows okay at 200x20, but cannot run any Futuremark benches or Aquamark 3....Sandra and Everest run fine though. system locks up, single beep, and reboots after 10-15 minutes. I really have a hard time believing anyone who says they can run it over 3.8 stable on an unmodded board without extreme cooling.

if intel really wanted to keep up with amd they would actually support dual channel memory to get tru dual core benefits.