Attention A8V Owners!!! Please read: important BIOS issue!

EDIT: This issue appears to affect every BIOS version since 1011. If you plan to use a memory divider other than [2:1 DDR400] with one of these versions, please read this thread carefully!

First of all, if your system is running fine and you haven't flashed to BIOS past version 1009 and are considering doing so, I strongly advise against it until this issue has been resolved.

I need your help in gathering as much information as possible about a potential "bug" in the BIOS affecting the memclock ratio settings ([5:3 DDR333], [3:2]). Initially, it was thought to be an isolated issue since no one else had reported any trouble. However, I was giving some help to kenekenny in this thread when it was discovered that BOTH of his A8V Deluxe boards, running BIOS 1014, are suffering from the same problem. Additionally, I think very few overclockers have flashed to 1014, and only some of those who have will actually be running the [5:3 DDR333] setting.

Description of issue:
- MEMCLK:CPU setting of [5:3 DDR333] does not function properly (very unstable, memory frequency not recognized by CPU-Z)
- MEMCLK:CPU setting of [3:2] operates as if it were [5:3 DDR333]
*NOTE: [2:1 DDR400] setting works properly, other settings not yet tested

Here is a brief overview of how I came to be aware of this problem. When I first set up my new rig (Scarlet), I was running BIOS 1009, and the [5:3 DDR333] setting worked fine. I was using this setting at 240x10, which put the memory right at its default speed of 200 MHz. The system was perfectly stable. For reasons related to my crappy memory, I decided to flash to the latest BIOS version, 1014, in the hopes of improving my overclock. After successfully flashing the board, I tried to boot up at 240x10 with the [5:3 DDR333] setting. However, when I hit F10 in BIOS setup to "save settings and exit", the system just hung there and I had to hold in the power button to shut it down. When I started it back up, I couldn't get into Windows. Instead, I got an automatic reboot. I lowered the FSB (HTT) some and I was able to get into Windows, but when I opened up CPU-Z, it could not read my memory frequency! When I tried to go back into setup to change settings, I was getting lock-ups and failures to reboot/save settings. BIOS itself was rather unstable.

I flashed my BIOS to 1013, but the issue was not resolved. Thinking that my memory might be at fault, I changed the MEMCLK:CPU setting to [3:2] and booted at 240x10. I opened up CPU-Z and to my surprise the memory frequency was being reported as 200 MHz! This is the exact frequency that should result when running 240x10 with the MEMCLK:CPU set at [5:3 DDR333], i.e. memory at 5/6 of CPU speed! Using this setting, the system is stable at 2600+ MHz (still testing capability of CPU and memory).

Here is what I would like to know from you, and I have suggested actions that you can take to help (only applies to people running 1012 or more recent BIOS versions):
- What BIOS version are you running?
- Does your [5:3 DDR333] setting appear to function properly? If unsure, could you please try running with this setting (stock CPU speed is fine) and check what CPU-Z reports for memory frequency? If it doesn't appear to be functioning properly, try running [3:2] and again check if CPU-Z gives an appropriate memory frequency.

At this point I believe the problem lies with the 1014 BIOS, although flashing to 1013 didn't help my situation. This is why I'm interested to know if people running 1012 or 1013 are experiencing the problem. I'm hoping that the "damage" is not permanent. Do you think Asus should be made aware of this problem? If so, wouldn't openly admitting to OCing void my warranty? Is there some way around this?

Ya i dont know what the hell ASUS were doing whilst testing this bios?...maybe in the middle of an office party?

Flashed mine the other day and hellllo wheres my OC gone

Was on 1010 b4 and have gone back to 1009 now as i can get a slightly better OC on my Ballistix DDR500 on that.

Think i'll stick by my old policy, if it aint broken and you DONT NEED to update then DONT!

KB,

I'd ask you to perform a few tests with a fresh installation of Windows. The fact that you could not revert back to stable settings, after you flashed to 1013 leads me to believe that Windows may have been corrupted. A test with a fresh install will confirm this theory.

Does DDR333 work in single channel mode? What about with a single stick?
If you are going to post up a new issue, might as well get all the basic tests out of the way.

It's always safer to set your BIOS back to the "default or optimized" settings before you flash. Then after flashing boot into Windows with the safe settings and shut down. Then you can go in and tweak the BIOS with much less chance of corrupting the Windows files. It's also more risky with SATA boot drive than with IDE.(See which ports/channels are locked, and use one of them.)
I know this is an obvious question but.........
Did you try clearing the CMOS jumper and re-starting?
Good luck.

Super Nade said:

KB,

I'd ask you to perform a few tests with a fresh installation of Windows. The fact that you could not revert back to stable settings, after you flashed to 1013 leads me to believe that Windows may have been corrupted. A test with a fresh install will confirm this theory.

Does DDR333 work in single channel mode? What about with a single stick?
If you are going to post up a new issue, might as well get all the basic tests out of the way.

Click to expand...

S-N, I firmly believe this issue to be independent of the state of the OS. To verify, I just ran memtest86 from a bootable floppy with various settings to see what it would report for memory frequency. The results:
- 240x10 using the 3:2 setting yields a memory speed of 200 MHz, which corresponds to the 5:3 DDR333 setting
- 200x10 using the 5:3 setting yields a memory speed of 200 MHz, which corresponds to the 2:1 DDR400 setting!
- 220x10 using the 5:3 setting yields 220 MHz for the memory, also 2:1!
- 230x10 using the 4:3 setting yields 153 MHz, which is correct for 4:3

This explains some things. I was trying to run with a 5/6 divider when in fact the system was running my memory 1:1, resulting in complete instability. This helps to explain the lockups in BIOS, CPU-Z failing to report the memory speed, etc.

I wouldn't have started this thread unless I felt strongly that something wasn't right with one of the BIOS releases. As I have mentioned previously, my system worked fine with the 5/6 divider before flashing to 1014. My OS has been fine (I have an image of a clean Windows install that I restored after I had my previous corruption problems), and further the memtest results show that Windows is not influencing the improper mapping/reporting of memory frequency. I was willing to accept that this may have been an isolated issue, possibly related to a defective CMOS chip, but after discovering that someone else has 2 of these boards, both running 1014, both with the same incorrect memory divider problems, I think it's improbable that this can be attributed to coincidence.

The purpose of this thread was 1) to warn people of a potential issue with a BIOS version so they could make an informed decision before flashing, and 2) to gather information from as many users as possible to determine things like "is this an isolated issue", "what BIOS releases are affected", "what can we do about it", etc. It's very possible that this bug exists in 1013 as well, which would explain why my flash from 1014 to 1013 didn't fix the problem. Honestly, I have seen only 1 or 2 people who are using BIOS version 1013 or later. I'd like to hear from more people.

I agree that I should do more testing. Unfortunately, I haven't had much time lately and I don't see that changing anytime soon (most of my posts on this forum are while I'm at work). I will keep you informed of my progress.

This is my guess:-

Somehow, ASUS tried to implement support for the new up-dividers supported by the Rev E's and goofed up with the mem tables. What happens when you flash to the first BIOS which supports the Rev E?

Super Nade said:

This is my guess:-

Somehow, ASUS tried to implement support for the new up-dividers supported by the Rev E's and goofed up with the mem tables. What happens when you flash to the first BIOS which supports the Rev E?

Click to expand...

Hmm, well I flashed to 1011 (first BIOS that supported Rev E) and cleared the CMOS memory, and the problem is not solved. I left everything on auto/default settings except for the MEMCLK:CPU setting and the necessary on-board devices, and the [5:3 DDR333] setting still resulted in a [2:1 DDR400] memory frequency according to memtest86. The [3:2] setting still sets the memory at [5:3 DDR333]... I think at this point you are right about my CMOS chip being screwed up. The question is: what caused it? I believe it was related to the severe instability that the system was experiencing when I first flashed to 1014. I had lockups in BIOS setup and the computer would freeze as soon as I saved the settings. That BIOS didn't seem right from the first time I booted up with it.

What would be really helpful is if we could collect more data on the issue from other users. There must be others out there using 1014. If not, surely somebody is curious enough to volunteer as a guinea pig, right?

I'm not sure what to do at this point. I can continue to use the [3:2] setting to achieve the desired [5:3] divider since it seems to work okay, or I can request a new CMOS chip from Asus. The way the board is now, I do not have a working [3:2] setting, which sucks because my memory looks like it hits a wall around 220 MHz, making 2650 MHz the highest speed I can achieve.

Try procuring another BIOS chip. Maybe your CMOS is leaky/screwy ?

I am current using 1014 and find that it is better than 1009 for me.

Due to the limitation of my poor CPU(CBBID 0451), the highest memory frequency I can stably run now is 250MHz 2.5-3-3-6. The memory can pass memtest at 260MHz, but can not stably run in window due to CPU I think.

jeeplover said:

I am current using 1014 and find that it is better than 1009 for me.

Due to the limitation of my poor CPU(CBBID 0451), the highest memory frequency I can stably run now is 250MHz 2.5-3-3-6. The memory can pass memtest at 260MHz, but can not stably run in window due to CPU I think.

Click to expand...

jeeplover, are you running your memory 1:1 with the HTT? If so, could you please try using the [5:3 DDR333] setting and report the memory speed that the board sets using memtest86 or CPU-Z? Also, could you test the [3:2] setting if it turns out that the [5:3 DDR333] setting is wrong. I'd like to see if this is a problem with BIOS 1014 since I'm now aware of 3 motherboards that have improperly mapped memory dividers since flashing to 1014. Thanks a lot.

Well, well, well. I posted my issue on the A8V forum at Asus' website yesterday. Today, somebody starts a new thread stating that he has the SAME PROBLEM SINCE FLASHING TO 1014. That's now 3 individuals reporting the problem with 1014 (one person with 2 boards). I am now convinced that this is not a coincidence. Asus has screwed something up big time with 1014.

I can't give a direct link to the thread due to the way Asus has their message boards set up, but here is the forum and you can look for this thread with the title "5:3 memory divider problems", and my thread with the title "Corrupted BIOS after flash to 1014?"
http://vip.asus.com/forum/bbs.aspx?board_id=1&SLanguage=en-us

If you don't feel like searching, here is the text of the other person's post:
"I have an A8v DeLuxe rev.2.0, using bios 1014 and running a Winchester AMD64 3200+. I used to run this rig using 2 Kingston CL2 512Mb sticks at 400MHz (2-2-2-5)with bios version 1009 in Dualchannel mode with no problems. The CPu was running @2.4GHz with the HTT set to 240Mhz and the divider set to 5:3 resulting in a memory speed of exactly 200MHZ.
I recently added another 2 512Mb Kingston sticks, knowing that this might result in some speed problems. I upgraded the bios to version 1014, and know I'm having the following problem:

The 5:3 divider does not work properly. HTT is again set to 240MHz, which should result in a memory speed of 200Mhz. But the resulting memory speed is 220MHZ?!. If I set the HTT to 220MHz, the 5:3 divider should result in a memory speed of 183MHz...However the resulting speed is 220Mhz?!

Anyone observed the same behaviour, or perhaps has a solution?

Any help is much appreciated! "

Revivalist said:

Hey Buckeye, I see some folks talking about the same thing in this thread over at ABX Zone forums. One person mentions, "Biosses from 1011 on ( to support the newer Venice and San Diego core CPU's ) all have memory divider problems this includes also the 1014 release."

Have you tried reverting back to a BIOS before the 1011 version? If not, then maybe your BIOS chip is still fine. Just use 1010 or earlier (since you said that 1009 was supporting your chip anways) and see if your dividers work correctly again.

Click to expand...

Wow, I thought for sure this was a 1014 issue since there have been absolutely no reports of the problem on this forum. Surely there are plenty of overclockers out there who are using the [5:3] divider on BIOS versions 1011 and up, right? Maybe not... Amazing, I will try flashing back to 1009 when I get home. I can't believe I need to use a BIOS that technically doesn't support my CPU to fix the problem . Imagine how happy I'll be if the board doesn't POST when I flash back to 1009...

Thanks a lot for digging that up Revivalist. This is the type of information I was hoping to get by starting this thread.

PS: The link you provided also explains why 1T isn't working when I disable [2T]. I'm running 1011 right now, but the fix for this was implemented in 1014. Now I can either a) run 1014 at 1T with screwed up dividers, or b) run 1009 at 2T with correct dividers. This is almost laughable.

Revivalist said:

Why can't you run 1T with the 1009 BIOS? I'm using the 1010 and it runs 1T just fine.

Click to expand...

Do you have a Venice CPU? In the thread you linked, somebody claimed that BIOS 1014 fixes a problem in which Venice CPUs could not run 1T timing. I'm currently running BIOS 1011, and I have verified that if I disable [2T] in BIOS, it doesn't actually do anything. The setting is saved in BIOS, but if I open A64 Tweaker in Windows, it shows 2T as being enabled. If I disable it via A64 Tweaker and run Prime95, I quickly get a failure, so clearly 2T is not being disabled via the BIOS option.

Revivalist said:

No, I have a Newcastle. I didn't realize that the 1T problem had to do with the Venice core. . . .

That's interesting about using A64 Tweaker to enable 1T. I'm pretty sure that's the same as setting it in the BIOS, which means that if Prime95 fails with 1T set by A64 Tweaker then it probably would fail even if it was set in the BIOS, right? What HTT are using? When I use anything above 250MHz, I can't run 1T, even if the timings are totally loosened, the CPU is underclocked, and voltages are pumped up. I've read about others hitting the same wall.

Click to expand...

LOL, I'd kill for 250 MHz with 1T. I guess you haven't seen the thread I started about my Ballistix not being able to run 1T even at the stock settings? I've contacted Crucial about getting these duds replaced, but I'm still waiting for a response. Even at 2T, these modules aren't stable above 215 MHz regardless of timings.

Revivalist said:

I must admit I'm a bit confused. You mentioned . . .But I suppose that option a) is not an option since the RAM can't do 1T. In that case option b) seems good. . .

Click to expand...

Sorry, I know that was confusing. I made that statement assuming that I would get new memory modules that can handle 1T.

If you get your RAM replaced with some that can do 1T, then Prime95 shouldn't have a problem when you set 1T using A64 Tweaker. Then you can just have A64 Tweaker set the system to 1T right at startup. (I have A64 Tweaker setting a bunch of tweaks at startup that boosted the RAM performance by about 5-10%.)

Click to expand...

Cool, that's a good idea that would allow me to run 1T using BIOS 1009, at least until Asus releases 1015 with fixes to the memory divider problems (I hope).

Well it is confirmed--flashing to BIOS 1009 restores the memory dividers to the proper functionality. I am so relieved to know that my BIOS chip is not corrupted. Now if only Asus would release 1015 with a fix to this silly problem.

Super Nade, would you mind adding a blurb about this in your A8V overclocking guide? It might save some new A8V Deluxe owners from going through similar frustrations. BTW, congrats on your "blue" status!