Memory Timing

[Guest simjunkie]

Final try at OCCT, it ran for about 20 minutes and had an error on Core 0. The system seems to be stable and running ok.... Do you think this could be an issue with OCCT?

It looks like with all those memory multipliers you showed that you have the 965. So you shouldn't even be raising you Bclock at all. Just use the 133MHz Bclock and DDR3 1600, that's it. Believe me, I have this chip. I've tried quite a few different combinations of Bclock and multiplier and all the higher bclocks were hard to get stable even at 1.40 vcore. That's using OCCT. Finally I settled on the standard 133MHz and 30x mult. Everest 5.00 shows this as one of the better performing settings too.Also you should be aware that on all the current revisions of core i7 chips, 920, 940, and 965, there are manufacturing differences in the Integrated Memory Controller that cause instability at higher bclocks and frequencies. This is according to an article Anandtech wrote a couple months ago. They tested different i7 920 chips, but advised that even the i7 965 can fall flat on it's face past 4GHz and no safe amount of voltage increase seems to help. But you gotta understand all this clocking stuff first. Hope you get it nailed down. DDR3 1600 and 4.0GHz should be attainable and rock solid stable on the 965.-jk

[falk781 0]

It looks like with all those memory multipliers you showed that you have the 965. So you shouldn't even be raising you Bclock at all. Just use the 133MHz Bclock and DDR3 1600, that's it. Believe me, I have this chip. I've tried quite a few different combinations of Bclock and multiplier and all the higher bclocks were hard to get stable even at 1.40 vcore. That's using OCCT. Finally I settled on the standard 133MHz and 30x mult. Everest 5.00 shows this as one of the better performing settings too.Also you should be aware that on all the current revisions of core i7 chips, 920, 940, and 965, there are manufacturing differences in the Integrated Memory Controller that cause instability at higher bclocks and frequencies. This is according to an article Anandtech wrote a couple months ago. They tested different i7 920 chips, but advised that even the i7 965 can fall flat on it's face past 4GHz and no safe amount of voltage increase seems to help. But you gotta understand all this clocking stuff first. Hope you get it nailed down. DDR3 1600 and 4.0GHz should be attainable and rock solid stable on the 965.-jk

I did exactly as you and Nick N said. 133 Bclock with the multiplier to get to 4.00. Adjusted the voltages as outlined above and FSX runs fine, so does windows. Thje problem is is that it either halts the OCCT test or it finishes with an error in one of the cores. I just got the latest revisions of Real Temp and OCCT and I think i'll try it again. The only other thing I guess I can do is start reducing the multiplier.

[Guest simjunkie]

I did exactly as you and Nick N said. 133 Bclock with the multiplier to get to 4.00. Adjusted the voltages as outlined above and FSX runs fine, so does windows. Thje problem is is that it either halts the OCCT test or it finishes with an error in one of the cores. I just got the latest revisions of Real Temp and OCCT and I think i'll try it again. The only other thing I guess I can do is start reducing the multiplier.

OCCT was updated for the i7 but I never got any errors with either version. Nick's the guy you want to talk to to figure this out, he's just super busy with software goodies for us simmers otherwise he'd spend more time, he's spread pretty thin. I have the P6T Deluxe and the 965 and I'm happy to help if I can but I don't know the R2E very well. Let's see what the updated OCCT does.-jk

[falk781 0]

OCCT was updated for the i7 but I never got any errors with either version. Nick's the guy you want to talk to to figure this out, he's just super busy with software goodies for us simmers otherwise he'd spend more time, he's spread pretty thin. I have the P6T Deluxe and the 965 and I'm happy to help if I can but I don't know the R2E very well. Let's see what the updated OCCT does.-jk

Here's another update... Since another test with the newest OCCT v3.0.1 ended in a Core 1 error, again, I reset the bios back to the factory config and the only adjustments I made were setting teh DDR3 to 1600 as well as the memory timing to 77724. I reran the test and it completed sucessfully without any errors over 1 hour. What does this tell us? Is it a voltage setting in the bios?

[Guest simjunkie]

Here's another update... Since another test with the newest OCCT v3.0.1 ended in a Core 1 error, again, I reset the bios back to the factory config and the only adjustments I made were setting teh DDR3 to 1600 as well as the memory timing to 77724. I reran the test and it completed sucessfully without any errors over 1 hour. What does this tell us? Is it a voltage setting in the bios?

I believe that if you tweak a bunch of stuff trying to get stable and it keeps failing then at some point going back and resetting everything is the right thing to do. In fact, reset the CMOS, take out the battery for a few minutes, put it back in, boot up and load BIOS setup defaults, then reboot again and your starting off from a clean slate. Go in and set all the onboard devices and disable all the BS you need to disable for overclocking (turbo, speedstep, vanderpool, etc) and make sure to keep the HW Prefetcher and Adjacent Cache Line Prefetch enabled. Reboot and then start tweaking:This is how it is on my P6T: Set the overclock tuner to XMP profile and disable the two spread spectrums, now reboot. Get back into the bios and make sure the memory shows DDR3 1600, if not set to 1600. Don't mess with the cpu yet. Make sure DRAM voltage is 1.65v (I have to use 1.66v, there is no 1.65) and the QPI is 1.35v. This should automatically be set by the XMP profile. If not then set to 1600 and reboot.Go back into the bios and check the DRAM timing control and make sure you have 7-7-7-24. If not then set it manually and reboot. Finally, set the cpu multiplier to 30x and don't touch the Bclock. Set the vcore to 1.375v. Reboot. Now run OCCT and let's see if it passes. Check temps with Realtemp (i7 realtemp version). You may need 1.40 vcore but 1.375 should do it. I usually run memtest when I first clock my RAM up to speed just to make sure I can boot into windows without corrupting anything. Assuming your RAM's stable then just run OCCT.-jkBy the way, I have that swiftech system too, it's a beauty isn't it? I got mine from Gary at Sidewinder.

[falk781 0]

I believe that if you tweak a bunch of stuff trying to get stable and it keeps failing then at some point going back and resetting everything is the right thing to do. In fact, reset the CMOS, take out the battery for a few minutes, put it back in, boot up and load BIOS setup defaults, then reboot again and your starting off from a clean slate. Go in and set all the onboard devices and disable all the BS you need to disable for overclocking (turbo, speedstep, vanderpool, etc) and make sure to keep the HW Prefetcher and Adjacent Cache Line Prefetch enabled. Reboot and then start tweaking:This is how it is on my P6T: Set the overclock tuner to XMP profile and disable the two spread spectrums, now reboot. Get back into the bios and make sure the memory shows DDR3 1600, if not set to 1600. Don't mess with the cpu yet. Make sure DRAM voltage is 1.65v (I have to use 1.66v, there is no 1.65) and the QPI is 1.35v. This should automatically be set by the XMP profile. If not then set to 1600 and reboot.Go back into the bios and check the DRAM timing control and make sure you have 7-7-7-24. If not then set it manually and reboot. Finally, set the cpu multiplier to 30x and don't touch the Bclock. Set the vcore to 1.375v. Reboot. Now run OCCT and let's see if it passes. Check temps with Realtemp (i7 realtemp version). You may need 1.40 vcore but 1.375 should do it. I usually run memtest when I first clock my RAM up to speed just to make sure I can boot into windows without corrupting anything. Assuming your RAM's stable then just run OCCT.-jkBy the way, I have that swiftech system too, it's a beauty isn't it? I got mine from Gary at Sidewinder.

Can I reset the CMOS by just pushing the button on the back of the Rampage? I think that's what its there for?

[Guest simjunkie]

Can I reset the CMOS by just pushing the button on the back of the Rampage? I think that's what its there for?

You're very right. I just like to take that extra step to stop those gremlins, ya know. :( In your case it's not a bad idea. That CMOS bttn. is mostly a convience in case you lock up and can't restart. For starting completely new and chasing down bugs I like to take that extra step, even though the button is probably effectively doing the same thing.-jkHere's another good source if you haven't discovered it yet:http://support.asus.com/faq/faq.aspx?SLanguage=en-usFind the R2E forums. Read thru and see what people are doing with that board, what bugs they're having, what helps, what doesn't, what BIOS is working best. By the way, How's your BIOS revision? Up to date?

[falk781 0]

You're very right. I just like to take that extra step to stop those gremlins, ya know. :( In your case it's not a bad idea. That CMOS bttn. is mostly a convience in case you lock up and can't restart. For starting completely new and chasing down bugs I like to take that extra step, even though the button is probably effectively doing the same thing.-jkHere's another good source if you haven't discovered it yet:http://support.asus.com/faq/faq.aspx?SLanguage=en-usFind the R2E forums. Read thru and see what people are doing with that board, what bugs they're having, what helps, what doesn't, what BIOS is working best. By the way, How's your BIOS revision? Up to date?

I'm a member on that forum but I don't recall seeing that thread. I'll have to check it out. Bios rev should be up to date although I'll check.