Tightest timings/best divider on IP35-E

[SerpentRoyal]

I simply advocate that you kill these services if you want to compare 12 vs 14 BIOS, or any small changes in system performance from tweaking RAMs. If the change is much smaller than the noise floor, then your data is probably not useable.

Calculting Pi is a real task. Under the right condition, Super Pi can be a useful tool to analyse small changes in system performance.

I use the hardware firewall from the router. No AV program on all of my PCs. LAN is also disabled in BIOS (Abit Wi-Fi PCI-E card). The most effective FW/AV is the user.

 

[brencat]

Originally posted by: MadScientist
Edit: I bumped the MCH voltage up one notch from 1.25V to 1.29V and was finally able to get my 4 sticks of ram up to 832 mhz, Memtest86+ stable, at it's rated settings of 4-4-4-12T, 2.2V.

Good to hear MadSci - I know you were having difficulty doing this with your 4 x 1 sticks. Make sure you run Orthos too, as I had a setting(I think it was 4-4-3-9 or 4-3-4-9 @ 1:1.25) that passed 16 hrs memtest but failed Orthos in < 10 mins. More vMCH didn't help.

My view on this topic is that yes, there may not be a significant real-world difference in running your RAM above 1:1, but there still is a difference. In my case, it was ~ 0.8 seconds higher in SP1M with 3-3-3-6 2T @ DDR 650 1:1 versus my current setting of 4-4-4-8 2T @ DDR 812 1:1.25. Real benefits could likely be realized on a 680i board where you can run a true 1T command rate, as well as having access to settings like tRC that we don't have on the IP35-E. But I won't be parting with this terrific mobo anytime soon!

Finally, to us overclockers, we push our hardware because we can -- not necessarily because we have to. Frankly, I don't give a rats ass about 1 - 2% extra performance in synthetics. I do it because it's fun to do -- as well as the challenge of learning something new with each new generation of hardware.

 

[MadScientist]

Originally posted by: brencat

Originally posted by: MadScientist
Edit: I bumped the MCH voltage up one notch from 1.25V to 1.29V and was finally able to get my 4 sticks of ram up to 832 mhz, Memtest86+ stable, at it's rated settings of 4-4-4-12T, 2.2V.

Good to hear MadSci - I know you were having difficulty doing this with your 4 x 1 sticks. Make sure you run Orthos too, as I had a setting(I think it was 4-4-3-9 or 4-3-4-9 @ 1:1.25) that passed 16 hrs memtest but failed Orthos in < 10 mins. More vMCH didn't help.

Yeah, I had about 5 or 6 ram settings that looked promising and passed Memtest86+ but failed Prime95 25.5 blend. I use the new Prime95 version for Vista 64 because it uses all 4 cores immediately, with Orthos don't you have to set the affinity for 4 cores? I'm still on the fence on whether to go 1:1 or 1:1.25 with my ram. I got my ram up to 770 mhz at 3-3-3-8-2T, 2.2V, Memtest86+ stable without having to increase the MCH voltage. This equates to a 3.465 Ghz CPU speed at a 1:1 divider which I can do with a vcore of 1.3175, but I have not Prime95 tested it yet. If it works it should give me load temps in the low 50s.

 

[brencat]

Originally posted by: MadScientist
I use the new Prime95 version for Vista 64 because it uses all 4 core immediately, with Orthos don't you have to set the affinity for 4 cores?

Yes, if you have a quad, definitely use the new Prime95. I have a dual core E2180 so Orthos is fine for me.