ASUS AMI BIOS - Won't Set Memory Timings

Recommended by Wim's BIOS...

BIOS Gurus,

I have been pulling my hair out for several weeks trying to troubleshoot an intermittent POSTing issue with an ASUS P4C800-E based system. Thanks to CPUZ and a little help from an ASUS forum I believe that I have found the the root cause of the POST problems.

CPUZ shows the RAM timings to be set for 2.5-3-3-6. However, by SPD information the RAM is only rated for 3-3-3-8. Realizing this difference I then tried to set the BIOS for these settings to see if the POST problem would go away. First I tried setting the memory timing by SPD in the BIOS. After saving and rebooting I ran CPUZ and the timings were still set at 2.5-3-3-6. So apparently the BIOS couldn't read the SPD settings and defaulted to its own incorrect settings.

Next I went back into the BIOS and set the memory timings manually to the specified 3-3-3-8, saved and then rebooted. CPUZ still showed the timings as 2.5-3-3-6 the same as before. As a sanity check I went back into the BIOS and the manual timings that I had previously set of 3-3-3-8 were still displayed.

By now it appears that this BIOS version is FUBAR. Unfortunately, it is the last one released for this MB that is not a Beta so there is no way to upgrade. I am wondering if this issue exists only in this particular version of this BIOS or is common to all versions.

Do any of you know how to fix this BIOS or know where I might be able to obtain older versions that might not have this issue?

BIOS ID = 63-2301-000106-00101111-062005-I875P-P4CED106-Y2KC


Thanks,

Steve

[edwin]

What version of CPUZ are you using and are you 100% sure it is reading the settings in the right place anyway? What is the brand/model of the memory sticks?

edwin wrote:

What version of CPUZ are you using and are you 100% sure it is reading the settings in the right place anyway?

The CPUZ version is the latest at 1.58. Of course, I have no idea if it is reading the right place. It appears to get the SPD data correctly. I am assuming that the rest is coming from reading the chipset. I would think that once you have figured out how to read an 875P chipset it should be the same on all motherboards. I have also used this version of CPUZ on two different systems of the same vintage as the one in question and the actual memory timings have matched the SPD values.

Do you know of any other utilities that could be used to verify CPUZ's results?

edwin wrote:

What is the brand/model of the memory sticks?

The memory sticks are currently Crucial/Micron 8VDDT6464AY-40BF4. 512mB single sided PC3200 200MHz

Thanks,

Steve

[edwin]

Were these bought specific for this board? The Micron part number is not on the QVL from Asus. What is the Crucial CTxxxxxxx number for these sticks? Can be they are simply not compatible with this board, if they were bought from Crucial for this board, send them a message stating your troubles and ask for an RMA, they should send you a replacement set. Oh and ofcourse you already replaced your CMOS battery with a fresh new one + checked for bad capacitors on the board (see http://www.badcaps.net for info), right?

[Denniss]

You may want to test other system info software - they may or may not be able to read the memory timings. Some examples are hwinfo, sandra, everest/aida, etc.

edwin wrote:

Were these bought specific for this board? The Micron part number is not on the QVL from Asus. What is the Crucial CTxxxxxxx number for these sticks? Can be they are simply not compatible with this board, if they were bought from Crucial for this board, send them a message stating your troubles and ask for an RMA, they should send you a replacement set. Oh and ofcourse you already replaced your CMOS battery with a fresh new one + checked for bad capacitors on the board (see http://www.badcaps.net for info), right?

Edwin,

Sorry, but you just asked for the long gory details! I was hoping to spare folks from paths that I had already closed but here goes. The system has worked very reliably for quite a few years with 2 sets of memory. The original set was 2-512Mb double sided Crucial CT6464Z40B.M16FKY. Around 2008 I added 2-1Gb Corsair ValuSelect VS1GB400C3 sticks. Everyone was happy until several months ago when the system developed a noisy video problem.

The video problem was repaired with a new adapter card. While doing the repair I noticed the start of bulging caps on the MB. I also noticed more bulging caps in the output section of the Antec Truepower 430 PSU. I replaced the MB caps with Nichicon HZ series. I kept the old PSU while researching a new one. The system ran reliably for several more months AFTER replacing the MB caps AND AFTER replacing the video adapter..

For the new PSU I decided on a Corsair by Seasonic TX650 V2. This is a well reviewed quality supply. The recent set of problems began immediately upon installing this new PSU. However I don't believe that this is a PSU issue. The output voltages measure near perfect with a good digital meter as well as SpeedFan. I even went as far as looking at the loaded output voltages on an oscilloscope and they all met the ATX power supply specification. The connection between the current symptoms and the PSU is still unclear. My speculation is that the new PSU being much less noisy and having slightly different DC voltages has affected the memory operating voltage on the MB for the worse.

So back to the Crucial memory. Since it was the original set and a process generation or 2 behind the Corsair it was naturally slower than the newer Corsair sticks. The Crucial memory exhibited much worse symptoms than the newer Corsair. I have already RMA'd this memory and Crucial has sent the latest (current generation) single sided (2X density + unknown die shrinks) as the replacement. It's part number is CT6464Z40B.8TFY.

So now I have 2 sets of RAM that exhibit approximately the same symptoms. I am currently only running one set at a time. The only potential problem that I currently see is that the memory timing as set by the BIOS is faster than the SPD specification of the memory and that the BIOS is so far refusing to allow me to change it. I will agree to a memory problem ONLY when the memory timings match the SPD values and the POST still doesn't execute properly.

BTW, The CMOS battery has been changed and reads 3.07 volts after the system has been unplugged for 24 hours. While plugged in the CMOS is maintained by the +5VSB supply from the PSU so this is really a moot point . Also the Asus QVL for memory has not been maintained. As I recall the original Crucial parts were on the QVL.

Thanks for your indulgence!

Steve

[edwin]

My immediate question is, did you replace ALL the capacitors on the MB including the smaller ones around the memory slots? Even though not visibly bulging, these have the same problem as the larger ones and can account for unstable operation if not replaced.

edwin wrote:

My immediate question is, did you replace ALL the capacitors on the MB including the smaller ones around the memory slots? Even though not visibly bulging, these have the same problem as the larger ones and can account for unstable operation if not replaced.

Yes, those caps have all been replaced.

[edwin]

Right. I am not sure if you need to set the memory timings to "Manual" elsewhere in the bios as the bios still seems to be trying to read SPD values from somewhere.

You can try of course one of the older bioses available here:
http://support.asus.com/Download.aspx?S ... xe&p=1&os=

And take a note, I sincerely hope this is a retail Asus board, not one from HP/Compaq.
http://h10025.www1.hp.com/ewfrf/wc/soft ... pv-18264-1

edwin wrote:

You can try of course one of the older bioses available here:
http://support.asus.com/Download.aspx?S ... xe&p=1&os=

Yes, I have tried both 1023 and 1024 Beta. I could go back further but I am beginning to have the suspicion that the POST is broken and probably has been for some time. The post seems to be acting on uninitialized memory or registers and taking the path down the rabbit hole.

edwin wrote:

And take a note, I sincerely hope this is a retail Asus board, not one from HP/Compaq.
http://h10025.www1.hp.com/ewfrf/wc/soft ... pv-18264-1

Yes, this is the retail board. I am aware of OEM boards being de-featured with hardware removal and other methods to lower cost.

Thanks,

Steve

BIOS Gurus,

Well, It was a good theory... Overly aggressive memory timings causing POST failures when cold with multiple symptoms.

After the machine had been on for several hours I flashed the 1024 Beta BIOS, immediately reset CMOS, and then went into setup and reconfigured everything. I then booted into Windows 2K and ran Memset. Memset showed that the memory timing had been changed to match the SPD values (3,3,3,8), Performance Mode and CpC were now disabled as expected.

Initially this appeared to fix the cold POST issues. However, within a 1 or 2 cold boots the issue had returned. Now Memset shows that Performance Mode has been re-enabled either by the POST (BIOS) or something within Windows. At this point I am not sure if this is the real issue any more and have not bothered to pursue it any further. I have also installed the other set of memory which used to work fine and still does if the MB can get past the POST.

I have done some more troubleshooting at the next few cold boots and have found an interesting correlation. At the last cold boot I received the bogus "Overclocking Failed! from the onboard POST reporter. Without turning the power supply completely off (the +5V standby power was still applied)I then carefully unplugged everything from the MB - video card, sound card, both SATA hard drives, floppy drive, CD ROM drives. I just now realized that I had forgotten to unplug the 9 in 1 USB based flash card reader but I don't think it matters. With all of this unplugged I still received the bogus "Overclocking Failed! from the onboard POST reporter.

I next turned the power supply completely off (No +5V_SB) and plugged everything back in which took about a minute. Turned on the PSU switch and hit the on/off button and surprisingly POSTed first time cold!

This seems to indicate that some memory or register that the POST looks at is being maintained by the +5volt standby power supply and appears to be causing the POST to get "lost" during the next cold boot. I can't explain why this same occurrence doesn't happen with a reboot when the system is warm unless the temperature causes whatever memory or resister to come up in a state that allows the POST to proceed normally.

Does any of this sound familiar to any of you? Anything else that I should check?


Thanks,

Steve

[edwin]

Try a known good (preferrably new) power supply. I have a gut feeling the voltages on this one are not entirely stable and that too can cause all sorts of issues. The other option would be a bad bios chip or indeed some sort of tune-up utility from Asus that is setting it back to performance mode. However your bogus overclocking failed message can indicate a power supply problem. Due to bad capacitors again. These were not used on mainboards only, I fear.