Please register or login. There are 0 registered and 587 anonymous users currently online. Current bandwidth usage: 326.30 kbit/s January 16 - 11:50pm EST 
Hardware Analysis
Forums Product Prices

  Latest Topics 

More >>


  You Are Here: 
/ Forums / Problem Products /

 Date Written 
Continue Reading on Page: 1, 2, Next >>
Berserker23 nick Jun 24, 2003, 04:56am EDT Report Abuse

my main configuration is:

P4T533-R mobo
P4/2,8 Ghz
ATI RADEON 9700 PRO card powered by 3CLUB
1 x 512 MB 32-bit PC1066 RAM chip (SAMSUNG)
2x120 GB MAXTOR HDDs in a stripped set

and I have the following problem:

If playing games ( all kinds of games ), each game shuts down after a random play-time. I mean, it looks like if you hit ALT + F4 to close it. I had mobo BIOS ver. 1005 and flashed it to the next version 1006. The problem persists, but now if the game stops and PC begins to read out the game data from the RAM, a window appears saying something about an exceptional gathering memory problem. The TURBO-RAM mode isnt activated in BIOS, and nothing is overclocked. The temperature of mobo and CPU is on an acceptible level too.

I tested the RDRAM with QAPlus/FE Version 5.60 and no errors were detected.

I think, its a mobo problem, but I dont know how to fix it.

Please tell me your expirience about that if you had similar problems.

Want to enjoy fewer advertisements and more features? Click here to become a Hardware Analysis registered user.
Charles Burnett Jul 24, 2003, 01:45pm EDT Report Abuse
I am having the exact same problem. I have a P4T533 board, though. It's got 1GB of memory and a 3.06 GHz CPU. Memory is Samsung.

All of the temperatures are ok, too. Nothing overclocked.

Another problem I have is running the Backup program. When it is verifying large files (10's to 100's of MB's), it always fails on at least one file. It's not always the same files. The 'bad' files can be on a RAID or a non-RAID drive in the machine. I have been able to run Backup on another machine (same version of Backup, same OS) and the files backup/verify fine. I have also been able to get it to work by disabling the L1 and L2 caches on the problem machine. Of course, that isn't an acceptable workaround as it performs like a 386.

I'm still hunting.

Can you run Backup and backup/verify large files?

Gray Jul 25, 2003, 10:40pm EDT Report Abuse
I don't know this for a fact, but before I build my computer I was told that any board which used a GB, or more, of NON-ECC RDRAM PC1066 would be unstable. I won't disclose the source, because the person who told me is affiliate w/ a business that builds and sells custom computers. He also mentioned that although they had put together numerous configurations, it only seemed to work when they used ECC RDRAM.

I have an ASUS P4T533-C w/ 1GB of ECC RDRAM PC1066 (4x256) and I haven't had any problems related to that. Actually my computer has been exceptionally stable (especially when you consider that amount of crap that I put it through). Anyway if you want any other info just let me know.

Charles Burnett Jul 25, 2003, 11:10pm EDT Report Abuse
Ok, that's very interesting about the ECC/non-ECC info. I originally intended to purchase ECC memory for my machine because I personally think the chance of a statistical memory error is getting fairly high when the memory gets that large. I talked to one memory dealer that told me that Samsung has had trouble making and/or designing the ECC version of the 32-bit RDRAM. So, I elected to go with non-ECC memory. Maybe that was a mistake.

The POST has never found a memory problem (full-blown memory test in POST - not the quick test). If I run a utility that rigorously and continuously tests the memory, do you think I will get errors? Any recommendation on a memory testing program?

I guess I could try running with 512MB of memory and see if the problem persists.

Gray Jul 26, 2003, 11:02am EDT Report Abuse
Well to test memory I have always used "memtest86", although to be honest I can't seem to find the file around right now. I remember it used to make a bootable diskette which booted directly into the test GUI. However, I am not sure that this will turn up any errors. Even if the problems you are having is caused by memory errors, I think that even "correctly" working PC1066 RDRAM causes the errors because of some very low level hardware problems with the way that the chipset handles the transfer of data. But I am way.... way out on a limb with that supposition. What I do know concretely is that even with known good Non-ECC PC1066 RDRAM can cause random memory errors.

But hey, give it a shot you never know. Perhaps it will turn out that you do have a defective stick. Maybe I'm wrong. I do remember that the MemTest app had an extended test which was particularly effective at diagnosing errors. Best of luck

Charles Burnett Jul 29, 2003, 04:53pm EDT Report Abuse
I tried running with 512MB of memory and got interesting results:

module A = Samsung 512MB 4200 non-ECC RIMM
module B = Samsung 512MB 4200 non-ECC RIMM
module C = Samsung 512MB 4200 non-ECC RIMM
module D = Samsung 512MB 4200 non-ECC RIMM

Config (1024MB total) - FAILS on a Backup
Memory slot 1: module A
Memory slot 2: module B

Config (1024MB total) - FAILS on a Backup
Memory slot 1: module C
Memory slot 2: module D

Config (512MB total) - WORKS on a Backup
Memory slot 1: module A
Memory slot 2: CRIMM

Config (512MB total) - WORKS on a Backup
Memory slot 1: module B
Memory slot 2: CRIMM

Thus it appears I get problems whenever I put real memory in slot 2.

Since I have rev. 1.03 of the P4T533 board which has a fan on the north bridge chip, I think I have the latest ASUS hardware (BIOS is 1006). This board must have a manufacturing defect instead of a design defect as others seem to be able to use 1GB of memory with this rev.

Gray Jul 29, 2003, 07:20pm EDT Report Abuse
Well, first of all I have to say that I'm a little confused by your descriptions of the modules. You seem to say that each one is a 512 Non-ECC RDRAM PC1066, however when all slots are filled you say that you only have 1GB. If each module was a 512, then you would have 2GB. But I'm going to assume that each one is actually 256. However, there have been quite a few problems with this board as I found in this thread

So you may well have a defective board. But then again it may still be the memory. For your info, when sloting RDRAM, you must populate slots 1 and 3 before filling 2 and then 4. Don't ask me why, I barely remember and I would screw up the explanation horribly. I would suggest that you try running modules D and C in the 1 and 3 slots to see if any of the modules are defective. If none of the modules are defective then you can establish that it is the levle of RAM which your board cannot handle. Also I believe that the BIOS rev. is 1009 at this point, I have a 1008 anyway. Flashing your BIOS is an option, but as always, be careful. As an aside I was curious if ASUS ever released any kind of information along w/ their BIOS revs. I haven't been able to find any release notes of any kind and I was wondering if I was missing anything.

Charles Burnett Jul 29, 2003, 10:07pm EDT Report Abuse

I have a P4T533 board which has 2 memory slots. It sounds like you have a P4T533-C board which has 4 memory slots. These boards are definitely different as mine uses 32-bit RDRAM while the -C uses 16-bit RDRAM.

A 32-bit RDRAM module is basically two 16-bit RDRAM modules combined into one physical package. They did this in hopes of achieving lower cost for the motherboard since you need half as many slots for the same memory configuration plus it takes less board space. Every penny is a big deal to PC manufacturers. I know from designing motherboards myself that space is always a premium.

So if I use 512MB modules (I haven't seen 1GB modules, but I think the spec. exists for somebody to make them), then I can have 1GB of memory maximum on the P4T533 board.

Charles Burnett Jul 29, 2003, 10:13pm EDT Report Abuse

I have not seen any release notes for the P4T533 BIOS versions. It's too bad as I would like to know where they tinker as it could assist in resolving problems.

Gray Jul 29, 2003, 10:38pm EDT Report Abuse
Yeah it really would be nice to know that kind of information, especially when trouble shooting. As for the whole module issue, sorry about the misunderstanding.

Now to be confused on a whole different level, I mean, if the board is fine with 512 all of the modules have been shown to be good, then it must mean that the board is having a problem supporting 1GB. Considering that you aren't having any other problems with the board, it sounds an issue that could be best handled by flashing the BIOS. Except, there is one other possibility as you mentioned, that the second slot might be bad. I don't suppose you happen to a couple 256 meg sticks of RDRAM lying around? Perhaps even that is not necessary..... I don't remember off hand, but is it possible to populate only the second slot on the P4T533 boards? If so, try it w/ 512 in only the second slot.

Gray Jul 29, 2003, 10:39pm EDT Report Abuse
On a completely seperate note, you ever do any water cooling?

Charles Burnett Jul 30, 2003, 12:50pm EDT Report Abuse

Yes, it appears I either:

1) have a bad slot 2, or
2) have a board that does not work with more than 512MB of memory

In either case, it appears that the motherboard is the problem.

I do not have any 256MB modules.

The ASUS manual states to always fill slot 1 with memory first if only 1 real memory module is inserted. I could try a single memory module in slot 2 with a CRIMM in slot 1 just to see what happens, though.

I don't do any water cooling, but I am sorely tempted. I think the next generation of high-end PC's is going to come with liquid cooling as a standard feature. There is simply too much heat to extract out of the case using fans. I don't care about overclocking as I want a reliable system more than extra speed, but I have spent a great deal of effort building systems with quiet fans, my own speed regulator circuit board, etc, before they were available on the market. I have ended up using multiple, very large case fans spinning at low rpm to get heat out and not make noise. It has worked well, except now the CPU fan itself is way to noisy when the system is tasked (it's fine if it's not working hard since the ASUS BIOS regulates the fan speed to a much slower rate). I'm looking to replace it possibly with liquid cooling.

Charles Burnett Jul 30, 2003, 12:52pm EDT Report Abuse
My apologies, Gray!

I have been typing "Gary" instead. Sorry!!!

Gray Jul 30, 2003, 02:16pm EDT Report Abuse
Don't worry about the name thing, no biggie. Second, I had gone the same route w/ air cooling for this system. 5 Vantec 80mm fans, not to mention the stock intel cooler and the video card is an Abit Ti4200 x8 OTES. That damn card weighs about as much as the hard drive. Getting back to the point, I have recently switched over to water cooling. I am using a Koolance EXOS; very nice system, but a little hard to put together for a water cooling noobie such as myself. But I think I've figured out the problems I was having w/ it. First, that I had air bubbles in the lines; apparently this is common and should work itself out after a few days. Second that my temps as still relatively high, about 32 idle on the CPU and 33 idle on the MB. This is somewhat disappointing as the fans seemed to do a better job. However, the thermal monitor on the EXOS reports quite different than does the Asus Probe. Have you tried Motherboard Monitor? I've heard that the Asus Probe can be unreliable, but not gotten much info about it. Also I just got some rounded cables in the mail (literally, USPS just rang the bell about 15mins ago). I intend to install them shortly and I'll report back on the effect.

Getting back to our original issue w/ the P4T533, there seems to be a few other people who have had similar issues. Check out the thread here:

Sorry for the line break, but it couldn't be helped. Warning the thread is currently 127 pages long, I am working my way through it now, but am only up to page 30. I'll post links to the really interesting stuff and/or summarize. Also what is your system setup like as far as the other components are concerned? There have been some significant questions regarding voltage regularity, video cards and other variables. ttyl

Gray Jul 31, 2003, 12:02am EDT Report Abuse
All I can is WOW! What holy hell of a lot of information. Although this is really just an overall sketch, basically people were having problems with the board being unstable. Underclocked, stock or overclocked, nothing seemed to help. From what I understand there have been several revisions of that particular board. If your serial # on your motherboard starts with 27XXXXX or something along those lines, it appears you may have one of the bad batch. I think any serial # which has a letter in the other than 'A' as the second character, is from the revised boards. Also I have heard that some people who RMA'd their boards had them returned with modifications made, i.e. some capacitors soldered on and active cooling added to the northbridge.

1. Do you have a fan on your chipset or just a heatsink?
2. What is the serial #?
3. When did you purchase the board ?
4. Have you done any stress tests?
5. Do you run the Asus probe?
6. Do you regularly check the temps?
7. Does the error occur ever time you try to run a backup?

Sorry for litany of questions, but it is very hard to summarize all the data I got from 127 pages of back and forth; much of it dealing w/ benchmarks. You would be amazed just how in depth these guys went in dealing with the problem. After reading that I would really like to help you fix this, if for no other reason than to put the effort to good use. ttyl

Charles Burnett Aug 03, 2003, 12:31am EDT Report Abuse
1) fan on chipset
2) 32ZGOJ3117
3) April 3, 2003
4) no canned stress tests, but I run heavy-duty chip design software that crashes for no reason
5) yes. I have tried not running ASUS Probe and still get problems.
6) I watch the temps all the time, especially when doing chip design. Are OK. The Intel spec. is 70 C on my CPU and it's never been over 61. Most of the time the CPU is in the low 40's. I get failures when it's in the low 40's as Backup barely tasks the CPU.
7) Yes, error every time on Backup. Not always the same files, but they are always large (10's to 100's of MB's) files.

Thanks for the pointer to that thread. I haven't read it yet, but I will definitely do so.

I ran Motherboard Monitor on my P4T machine for a couple of years. It seemed to match temperatures from other probes I have in the machine. I switched to Asus Probe, though, recently. I didn't run ASUS Probe at first on the P4T machine because ASUS got two of the fan connectors reversed (on the circuit board I am guessing), so the display is wrong. I ended up switching the cables in the box just so ASUS Probe would display the correct information under the corresponding description.

I've done rounded cables, too. I think it only makes 1 degree of difference at the most. I like them because it's easier to route all the cables/wires in the box since they don't take as much room as ribbon cables. When you get 6 disk drives in one box, it gets messy without rounded cables.

Gray Aug 03, 2003, 01:55am EDT Report Abuse
Alright I was in the midst of writing this a few minutes ago and then everything crashed, so I'll just cover the highlights. First, you have one of the later revisions of this board. Memory usage seems to be the relevant issue here; however your ability to mitigate the problem by disabling the L1 and L2 caches definately throws me for a loop. What effect does this have on the way that the system (OS and HW) deals with memory? How does the chip design application deal w/ 512MD RAM? I know that is hard to test since the crashes are random, but it would be interesting to know. Does disabling the caches have any other significant effects on the OS and hardware?

On a random point, my crash when I was writing this post before, what kind of video card do you use and why? Just curious.

Charles Burnett Aug 03, 2003, 11:34am EDT Report Abuse
Disabling the caches has definite effects on memory access. It undoubtedly affects the sequence of hardware events. Say a memory write needs to happen and the block is in the cache. A read-modify-write cycle must occur in external memory before the next read of this memory or when the block is removed from cache. The point in the future when this block is removed from cache or a read occurs of this memory is unknown and depends solely on what the applications are doing. If the caches are disabled, this update is done immediately and thus nothing is left to be cleaned up later.

Hyperthreading complicates the possibilities as both logical CPU's share the same caches, but not the same registers. That's why hyperthreading isn't exactly the same as a dual-processor system, but there are huge benefits.

Not many programs take advantage of hyperthreading, yet. WinDVD does if you get the high-end version (I don't have it.). Outlook 2002 (Office XP) does. I leave Outlook 2002 running constantly, so I wondered if this was the problem. I disabled hyperthreading in the BIOS a while back and still get problems.

The chip design applications (simulation, especially) can use 100's of MB's easily. I have seen a high-water mark of about 850 MB's before on this machine, so it's clearly using most of it at times. I don't think it has gone over 1GB with the designs I have done on this machine because when that happens and the disk swapping kicks in for more virtual memory, it becomes painfully slow. The disk drive LED would stay on constantly.

Yes, disabling the caches kills system performance. Instead of taking two minutes to boot (and load start-up applications), for example, it takes 10-20. This CPU was never designed to be used without cache. It's just not practical.

I'm using the video card with the world's longest name:

Gainward GeForce4 PowerPack! Ultra/750XP Ti4600 128

With a name that long and an exclamation point, it must be good, ha. I'm using this one because it's the only one I have found with dual DVI connectors and S-video in/out. I'm not into gaming and I don't need a cable TV tuner, so it's good for me. On one system I'm driving a DVI LCD and a TV with the S-video out. Someday I'd like to get a "TV"/monitor with a DVI input and this card should drive it nicely. If you use the s-video output, you must use the bottom DVI connector (this is not in the manual). Unfortunately, the bottom DVI connector is not driven with a signal until Windows loads, so you can't see the boot-up info, etc (but you can on the TV). This is about the only short-coming I have found.

I used some ATI cards for years, but finally got aggravated with their software. The hardware seems to be fairly good (except TV tuner). They never fix the drivers, though - they just move on to their next product. There were quite a few features I was never able to use due to bugs. So now I'm trying my luck with nVidia. These companies are under massive pressure to produce the fastest card for all the gamers out there, so stability means nothing.

Gray Aug 03, 2003, 03:35pm EDT Report Abuse
Well, so must for the video card. Back to the real problem. Your answer confirms my fear, which is that it is a basic design problem with the way that the chipset handles that amount of memory. Frankly, this is a pretty far reach for me, but it seems to fit the current facts. First we have the fact that it works with 512MB and it also works with 1GB, but only when the caches are disabled. It just sounds like a memory addressing problem. I am assuming that data written to the cache starts to be cleared after it has filled the entire space and has more use for a different set of bit than those it currently holds. Thus even when using 512MB the entire cache is being used. If this is true, then we can eliminate the possibility that the cache itself is the problem. I am also making the assumption that this is not an OS issue, primarily because I have a 1GB of RAM and have used all of it without crashing. However, I do not use either the backup program or any graphics rendering apps. Speaking of which, have you tried using a different backup application? And I also assume that the OS has been updated regularly? and which OS are you using?

You know I don't usually get involved in forums like this, but at some level I am kind of enjoying this. I am sure you aren't since you are the one experiencing the problem, but there is something about interacting with another technical person on the same level. I guess I just don't run into it very often; certainly not at work.

Charles Burnett Aug 03, 2003, 07:00pm EDT Report Abuse
I'm using Windows XP SP1 with all the latest updates from Windows update. I have not tried another backup application.

If I run the same version of Backup from another Windows XP SP1 machine, I can back up these same files with no problem (source and destination are the same). The problem only happens when I'm running Backup on the machine that holds the files being backed up.

I have another new P4T533 motherboard. Today I swapped it and I get the exact same problem. I just finished a backup and not all of the files verified.

I think this is a design problem with the P4T533 board. As I don't want to run with 512MB of memory, I may start looking around for another manufacturer that makes boards that takes the Pentium 4 3.06 GHz CPU and 32-bit RDRAM. Last I looked this was the only board.

I guess I could try ECC memory as that might mask the problem, but then I would be stuck with this non-ECC memory I can't use.

I think I will call ASUS and see if I can get anything out of them.

Charles Burnett Aug 03, 2003, 07:12pm EDT Report Abuse
By the way, Windows XP Pro supports 2GB of memory 'as is' out of the box. The Backup program should work up to 2 GB in other words. My chip design software will use up to 3GB if Windows XP is configured correctly. This is done by adding the /3GB switch to boot.ini. Without that switch, every application is limited to 2GB. Even with the switch, each application must be specially coded to use the extra GB.

So 1GB should be a piece of cake...

Continue Reading on Page: 1, 2, Next >>



  Topic Tools 
RSS UpdatesRSS Updates

  Related Articles 

A weekly newsletter featuring an editorial and a roundup of the latest articles, news and other interesting topics.

Please enter your email address below and click Subscribe.