Effect of .5 multiplier on E8400?

[sswilson]

I've done some preliminary googling, and will continue to do so, but would greatly appreciate it if anybody knows of a discussion with benchmarks regarding the effects of the .5 multiplier (other than the obvious of raising the cpu speed... :) ).

S939 chips had .5 multis on some boards, but they messed up memory bandwidth or something like that so it wasn't worth using. Has anybody seen anything like that for the E8400? (Currently running FSB 500 mem 1:1 @ 8.5 for 4250).

[1Tanker]

I can't imagine there being any issues... the E8500 uses a half multi.. as well as the Q9xxx quads.

[BALISTX]

The new E7200 also uses a half multiplier.

[sswilson]

So far I haven't found anything which suggests that it's a bad thing to do.

Unfortunately my chip isn't behaving quite as well as I'd like it to. In order for me to get 4G stable I've got the voltage up to 1.44 which is more than I'd hoped to have to use.

I thought I had it at 1.43 but that crapped out after 4 hours of orthos. 1.425 crapped out after 2 hours so I'm seeing a trend here... I suspect 1.44 is going to give me the 8 hour prime that I'm looking for.

I've had it posting over 4400 with vcore at 1.45, but that craps out within a minute of prime. Still trying to decide what I'm going to be comfortable with as far as vcore is concerned.... I know conventional wisdom these days is for no higher than 1.4, but I've always lived on the edge when it comes to vcore, and the chip only cost me $200.... :)

Gonna get this 4G stable first and then we'll tinker.... :)

[1Tanker]

Originally Posted by sswilson

So far I haven't found anything which suggests that it's a bad thing to do.

Unfortunately my chip isn't behaving quite as well as I'd like it to. In order for me to get 4G stable I've got the voltage up to 1.44 which is more than I'd hoped to have to use.

I thought I had it at 1.43 but that crapped out after 4 hours of orthos. 1.425 crapped out after 2 hours so I'm seeing a trend here... I suspect 1.44 is going to give me the 8 hour prime that I'm looking for.

I've had it posting over 4400 with vcore at 1.45, but that craps out within a minute of prime. Still trying to decide what I'm going to be comfortable with as far as vcore is concerned.... I know conventional wisdom these days is for no higher than 1.4, but I've always lived on the edge when it comes to vcore, and the chip only cost me $200.... :)

Gonna get this 4G stable first and then we'll tinker.... :)

Yeah, without crazy vCore/VTT levels, 4005-4050 seems to be the "sweet-spot", without fancy water/phase. That's nothing to sneer at...40xx delivers beautiful performance...especially with good RAM running @>500@5-5-5-x or tighter with exceptional RAM. I do find that once i get over 4.3, it hits another level.. dammmed FAST and super-responsive which is great for benching... but unfortunately not long-term...due to needed vCore.

[sswilson]

Originally Posted by 1Tanker

Yeah, without crazy vCore/VTT levels, 4005-4050 seems to be the "sweet-spot", without fancy water/phase. That's nothing to sneer at...40xx delivers beautiful performance...especially with good RAM running @>500@5-5-5-x or tighter with exceptional RAM. I do find that once i get over 4.3, it hits another level.. dammmed FAST and super-responsive which is great for benching... but unfortunately not long-term...due to needed vCore.

So far I've got my ballistix (older sticks) running @ 4-4-4-10 1:1 for 1000. I'm pretty happy with that (especially considering I'm running 4 sticks), and if I do decide to go with a higher Vcore I can bump up .5 to give me 4250 without messing with the memory at all. :)

Another happy discovery is that the newest bios for the DQ6 has a setting to stabilize Vdroop and it seems to work quite well.

[sswilson]

Well...... I did some stock checking around the house and decided that I have enough backup kit to do me if I fry this chip. Currently running orthos @ 4250 w/ 1.475 Vcore. We'll see how that does. If it seems to run fine I might stick with that, but I'm itching to throw some more Vs at this thing to make it bow down to my will....... :)

[3.0charlie]

My will was 4.5 GHz with my ex-E8400. It took 1.49V Vcore Bios (1.45V Windows) to make it. My personal limit was 1.5V Vcore, under water with a max of 65C. As for the .5 multis, I played around with 6.5 and 7.5 - saw no obvious difference from the prime multis. But I do agree with 1Tanker - passed 4.3GHz, that chip enters another world of performance.

If it wasn't for the folding and the Photoshopping that the wife does, I would replace the Quad with one of those Duals.

[sswilson]

Originally Posted by 3.0charlie

My will was 4.5 GHz with my ex-E8400. It took 1.49V Vcore Bios (1.45V Windows) to make it. My personal limit was 1.5V Vcore, under water with a max of 65C. As for the .5 multis, I played around with 6.5 and 7.5 - saw no obvious difference from the prime multis. But I do agree with 1Tanker - passed 4.3GHz, that chip enters another world of performance.

If it wasn't for the folding and the Photoshopping that the wife does, I would replace the Quad with one of those Duals.

Yeah, my quad's out of production for a few weeks while I wait for my 1 computer change / month limit to reset for Napster.

I'm looking forward to seeing what I'll get out of the 8400 for folding, and what the quad/GPU folding combination will get me once I slap the Q6600 onto the DFI board.