A8N-SLI Premium TwinX 1024-3200XLPT

1 -- My mobo autodetects and sets timings as 2.5-3-3-5 2T instead of the proper 2-2-2-5 1T.

2 -- Manually set to 1T without changing the other defaults, the system ran at 400 MHz in dual-channel mode very stably for months.

3 -- After a glitch (possibly caused by memory timings, I guess), I had to restore the system with a Ghost image that was 12 weeks old. NOW, the system gives me several BSODs per day unless I restore the defaults, which places the command rate at 2T.

4 -- Why in the world would the system run at 1T for four months, and now, only at 2T??

I've tried setting the voltage at default (I don't know what that might be) and at 2.75 without it making any difference. How do I get back to 1T, and what kind of hit will I take with performance by running at 2.5-3-3-5 2T if I can get to any other timings?

Ron

THANKS for the reply!!

OK, I upped the voltage to 2.8 and set the command rate to 1T. I left the other settings as per default. I am still seeing BSODs. When everything is at default, the computer's stable. When I set 1T and 2.75v, I got a lot of BSODs. At 2.8 and 1T the problem's better, but the BSODs continue. It must be a DRAM problem. Is it safe to try this memory at 2.9? Is that the next logical troubleshooting step?

Ron

It's interesting that you mention the BIOS needing to be mature. My motherboard's only 4½ months old, but its BIOS is 1004, and Asus has released four upgrades since then. Might it not be a better troubleshooting step to simply upgrade the BIOS before going to the trouble of running one stick at a time?

Ron

In my opinion, it's usually better to test the sticks with Memtest86 one at a time before updating the bios. If one of the sticks is bad and you try to update the bios with the bad stick in there's a chance you can end up with a bad bios flash. You would then have to go through the additional hassle of resetting the board back to defaults so that you could boot up to run Memtest86 anyways.

The link for Ram Guy's instructions for downloading and using Memtest86:

http://www.houseofhelp.com/forums/showthread.php?t=38146

Usually there is an optimal order for doing any troubleshooting. If the sticks pass Memtest86 individually, then I would update the bios with only one good stick on the board, then shut down, add the second stick in the slot dictated by your motherboard manual for dual channel operation and fire the system up to see if your dual channel and timing woes have been corrected by the updated bios. You'll also need to reset the correct timing and voltage in the updated bios as mentioned in previous posts.

milleron,

Please note that there are two different E revisions of your 3500+: Rev. E3 and Rev. E6. The A8N-SLI Premium does not officially support the E3 version of the 3500+ (or any other Rev. E3 processor), or Rev. E4 of the A64 3700+ and 4000+ processors - even with the latest BIOS version (1009, which came out early this month). And if you try to use a CPU on a motherboard which does not officially support that CPU, then who knows? Anything can happen, in this case. Thus, if your 3500+ is Rev. E3, then you are risking compatibility issues with the A8N-SLI Premium motherboard.

If on the other hand your 3500+ is actually Rev. E6, then you will need to update your motherboard's BIOS to at least 1006 in order to work correctly.

In my opinion, it's usually better to test the sticks with Memtest86 one at a time before updating the bios. If one of the sticks is bad and you try to update the bios with the bad stick in there's a chance you can end up with a bad bios flash. You would then have to go through the additional hassle of resetting the board back to defaults so that you could boot up to run Memtest86 anyways.

 

The link for Ram Guy's instructions for downloading and using Memtest86:

 

http://www.houseofhelp.com/forums/showthread.php?t=38146

 

Usually there is an optimal order for doing any troubleshooting. If the sticks pass Memtest86 individually, then I would update the bios with only one good stick on the board, then shut down, add the second stick in the slot dictated by your motherboard manual for dual channel operation and fire the system up to see if your dual channel and timing woes have been corrected by the updated bios. You'll also need to reset the correct timing and voltage in the updated bios as mentioned in previous posts.

Thanks for the info. I've run Memtest86 for over eleven hours without a single error. Is it important to run it individually if the pair has passed?

This is very important because I can't get one stick out without removing my heatsink.

Ron

milleron,

 

Please note that there are two different E revisions of your 3500+: Rev. E3 and Rev. E6. The A8N-SLI Premium does not officially support the E3 version of the 3500+ (or any other Rev. E3 processor), or Rev. E4 of the A64 3700+ and 4000+ processors - even with the latest BIOS version (1009, which came out early this month). And if you try to use a CPU on a motherboard which does not officially support that CPU, then who knows? Anything can happen, in this case. Thus, if your 3500+ is Rev. E3, then you are risking compatibility issues with the A8N-SLI Premium motherboard.

 

If on the other hand your 3500+ is actually Rev. E6, then you will need to update your motherboard's BIOS to at least 1006 in order to work correctly.

I have an AMD utility for identifying CPUs. Perhaps it will tell me if this Venice-core chip is E3 or E6. I "hang out" on the Asus Usenet group regularly, and I'm surprised to learn that some Rev-E CPUs aren't compatible with this board because I've never seen it reported there. I've certainly never read that Rev-E chips need BIOS 1006. Can you please tell me where this information came from?

Thanks again.

Ron

Garvin,

In principle I would agree with you, but in this case the CPU he has will need the latest bios to run properly. So yes please update the bios before you do anything.

Will update today and report results tomorrow. Thanks very much. I've searched all over the 'Net for help on this issue, but the help I've received on this forum is by far the best I've run across. You guys are great!

Ron.

I have an AMD utility for identifying CPUs. Perhaps it will tell me if this Venice-core chip is E3 or E6.

That may tell you. Also, you can download CPU-Z, and it will also tell you if you have an E3 or E6 Venice CPU.

I "hang out" on the Asus Usenet group regularly, and I'm surprised to learn that some Rev-E CPUs aren't compatible with this board because I've never seen it reported there. I've certainly never read that Rev-E chips need BIOS 1006. Can you please tell me where this information came from?

That information came from the CPU support page for your particular motherboard:

A8N-SLI Premium CPU Support Page

Please note that the Rev. E3 CPUs and the Rev. E4 3700+ and 4000+ CPUs are completely missing from that compatibility list.

In addition, the compatibility list implies that your BIOS version 1004 was the first BIOS release for that motherboard - and as such, it supports only the Clawhammer, Winchester and Newcastle-core CPUs.

RJ,

I'd been to that page before but not since I actually assembled my rig. It's changed, and clearly indicates that I need to upgrade the BIOS. That's a delicate process that I'm going to leave until I have some time to devote to it in a day or so.

Big Question: How does one determine for sure if he has an E3 or an E6.

The AMD CPU ID utility does not give this information, nor does Lavalys Everest or PC Wizard. The AMD utility does ID my CPU, however as "Family.Model.Step = 15.47.0. The part number on the shipping invoice makes me almost certain that this is an E3. I'm so disappointed about the apparent incompatibility because in June, this page simply listed 3500+ Rev E processors, not the E3, E4, and E6 variations that are there now. It makes me think that the Asus page in stating "E6" means all Rev-E CPUs "UP TO AND INCLUDING E6. It's almost impossible to believe that a motherboard that supports E6 would not support E3 processors. In addition, the AMD page lists the Asus A8N series as compatible with E3, E4, and E6 CPUs. The computer ran perfectly stable for four months, so I'm hopeful that I'll be able to get it stable again without tearing it apart to install a new CPU.

Ron

If the pair ran Memtest86 for 11 hours without error then there's no need to run them individually. A link to download a utility that should give you ALL the info on your AMD processor including revision subset:

http://downloads.guru3d.com/download.php?det=1148

ps: I just reread the entire thread. If the system ran fine for four months at the relaxed timings and 1T there's something else going on here. I'd still update the bios as long as it supports the processor as Ram Guy suggested so you can get the tighter Corsair spec timings but I have a suspicion there's a driver problem somewhere as well.

That may tell you. Also, you can download CPU-Z, and it will also tell you if you have an E3 or E6 Venice CPU.

 

 

 

That information came from the CPU support page for your particular motherboard:

 

A8N-SLI Premium CPU Support Page

 

Please note that the Rev. E3 CPUs and the Rev. E4 3700+ and 4000+ CPUs are completely missing from that compatibility list.

 

In addition, the compatibility list implies that your BIOS version 1004 was the first BIOS release for that motherboard - and as such, it supports only the Clawhammer, Winchester and Newcastle-core CPUs.

I think that E3 designates single-core Rev-E chips while E6 designates dual-core Rev-E chips. If that's correct, then I'd interpret the Asus support page to mean that the board supports Rev-E CPUs *UP TO* E6 stepping, rather than that it supports only E6 Rev-E CPUs.

Ron

I upgraded my BIOS to 1008 and set the command rate to 1T. The computer seemed better, but this may have been because I finally uninstalled a problematic installation of Norton Ghost 9. Then, less than 24 hours into the test, I got another PFN_LIST_CORRUPT blue screen. I'm not terribly discouraged yet because this occurred immediately after defragging one of my drives. However, it was a drive that did not have a page file on it, and defragging shouldn't cause a BSOD. Most of the guesses about this stop error place RAM problems at the top of the list.

My options now are to

1. Increase DRAM voltage further to 2.9
   
2. Upgrade to the just-released BIOS 1009.
   
3. Just give up and go back to "Auto" timings.
   

I hate to flash a BIOS unnecessarily, so I thought I'd ask which is the best troubleshooting step? (It's an important question because each step takes at least a day.)

Ron

That offer is VERY kind, since I'm not yet certain that the problem is RAM.

I think a fresh install of the OS is not wise because I DID that prior to resorting to the Ghost backup, and I had worse problems with that than I'm having with the current setup, which is not far from being acceptable.

I'll try another BIOS upgrade.

If that doesn't work, what would you think of trying the 2.9-volt setting?

Ron

I flashed the BIOS up to 1009, the most recent non-beta version, but I still got BSODs with various messages. I then took the following steps:

1 -- I uninstalled and reinstalled sound drivers since I thought there might be a conflict between Realtek and nVidia. Now, only nVidia software is installed.

2 -- I upgraded the firmware in my router.

3 -- I disabled Cool and Quiet, thinking that it might have been erroneously undervolting my CPU.

4 -- I opened the side of the case. Even though my reported temps were very low -- nothing over 40° and CPU actually less than 30° -- I wondered if something not being reported might be a little hot. The RAM is cool to the touch.

5 -- I relaxed RAM timings totally to BIOS "Auto" -- 2.5-3-3-5, 2T, which is WAY below spec for these 3200XLPT sticks.

So far, the computer's stable for about 20 hours, although without running a burn-in program.

If a BSOD or reboot occurs, I'll uninstall and reinstall Catalyst drivers.

If all that doesn't rectify the situation, I'll take you up on your offer of an RMA.

I'll post results in 48 hours, assuming you still follow threads that old.

Thanks for all the help.

Ron

Help, RAM Guy,

After doing all that, I thought it was OK to try a stress test. I ran the SANDRA Burn-In applet and returned to a black (not blue) screen and an unresponsive computer. I then ran Prime95. On the "Blend" option under Torture Test, it halted with errors after about twelve minutes -- "rounding error," if that makes any difference. On the "Short FFT" option which stresses the CPU preferentially, it ran for an 93 minutes without a problem. On the "Long FFT" option, which stresses RAM disproportionately, it crapped out after 90 seconds the first time and 12 minutes the second time with the same "rounding error." I think it's down to RAM and motherboard in that order, don't you?

May I please take you up on your offer to exchange this DRAM?

Thanks,

Ron

It actually turns out that Newegg will replace the two sticks, even though I bought it almost 6 months ago, so I won't need an RMA from Corsair.

I'd still like to know if in your opinion, given these Prime95 results, replacement of the RAM is the logical next step.

Thanks

Ron

Roger that. I've found that Newegg will replace them, so I'm going to RMA them to Newegg. I'll let you know if that makes a difference.

It's going to take me a couple of weeks to get the exchange made, and I'll update this thread with the results.

I guess I'm wanting to think it's RAM for a variety of reasons, not the least of which is that stability improved significantly when I took the timings all the way back to BIOS "Automatic" defaults. Doesn't that seem pretty significant? Could a problem with the motherboard also be partly alleviated by relaxing these timings??

Thanks again.

Ron