• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

AMD dual mobile challenge: one full week of uptime

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

xiphmont

Registered
Joined
Jul 22, 2004
OK folks, so, we have a problem on a few other forums (2cpu and EOC).

The short of it is: Lots of people are making dualies from AMD XP mobiles. It always looks more or less like it works. And then the things start crashing inexplicably. Sometimes every few hours. Sometimes only once or twice a week. Sometimes a sudden folding error that locks the machine after four days of flawless operation. A few unlucky people haven't even been able to get them to stay up for 15 minutes without seriously underclocking.

I, myself, have tried to build two, one on a K7D-L, another using the Gigabyte GA-7DPXDW-P. At stock timings and stock voltage, they'd invariably crash a few times a week; last one destroyed my filesystem in the process. This is putting mobiles into machines previously stable (not a single hard or soft crash) for at least a year of continuous operation when using real MPs.

What it comes down to is this: We all know lots of people who have made a dualie with mobiles. But I've not yet found a single person who can back up a claim of stability with proof that their box has been up a whole week at stock timings or better. Screenshot, syslog, both folding logs, whatever, you name it. A full, continuous week of folding with no reboot, crash or folding error. Please, we're on the honor system here, no fabricating logs.


So, who here can show dual folding logs + an uptime report? Just as importantly... who here is having these random crashes and is just too embarrassed to admit it (up until now) because they thought their buddies all had 'stable' mobile dualies?

I've sunk alot of time and alot of money into trying to make a dualie mobile stable, and it's been alot of fun... but I've not had an inch of success thusfar. See the threads over at EOC for the gory details if you wish ("K7D + mobile 2600+ woes" is the start of it). I want to know if, at this point, it's even worth bothering to keep trying. BTW, don't bother claiming you're stable if you're not going to post the proof. Without the goods, you ain't got nothin' :)

Monty
 
Last edited:
If the mobo doesn't like the mobile's then there isn't anything you can do aside from trying a new mobo... a few singal CPU mobo's have the same probs with the mobile's.

So unless someone has a dual mobo that likes mobile's I don't think anyone is going to have true stability.
 
PCGUY112887 said:
If the mobo doesn't like the mobile's then there isn't anything you can do aside from trying a new mobo... a few singal CPU mobo's have the same probs with the mobile's.

So unless someone has a dual mobo that likes mobile's I don't think anyone is going to have true stability.

...and that is *exactly* what I'm trying to find out :) I haven't found anyone who can prove it worked even once anywhere.

BTW, lots of things could be going wrong besides just the mobo... for example, what if the MP cache glue circuitry in the Bartons only actually works right at 1.6v or above? Or a mobile is just an MP where the glue test failed? the real Barton MPs run at less than normal (1.65v) too... So far it looks like all the crashes I've had have been a processor getting bad bits in its cache (in MP mode, two Athlons glue their caches together directly through the Northbridge, and so share the L2 caches. If something goes wrong there.... *shudder*. That circuitry isn't used in non-MP at all). Of course, the cuases are all just speculation. I'm more interested in results, yea or nay. Unless, of course, someone comes up with a plausible theory why it isn't working, and it's something I haven't tried yet.

Monty
 
Last edited:
Wow, this kinda makes me glad I didnt get the mobiles back when I was thinkin about it.
Didnt know so many were having these kind of problems. Kudos to stepping up, dont quite get it if its not stable, I would want to know why and would go apenuts to!

Apologies for cluttering the thread, wanted to subscribe and figured a post wouldnt kill thread either.
If I had the cash to blow Id buy a pair just to give it a shot, but I dont. :shrug:
 
I've got 6 dual Mobile systems running stable.
1 has been unstable in my house
It's been replaced and I haven't tried to get it back up yet
I can furnish you 19 days of uninterupted folding logs.
I'm embarassed to say its way over due for some critical updates.
The system is a Tyan 2469 K7X Pro with XP-M2400 at MP2800+ (16x133)
 
Twin_Turbo said:
I've got 6 dual Mobile systems running stable.
1 has been unstable in my house
It's been replaced and I haven't tried to get it back up yet
I can furnish you 19 days of uninterupted folding logs.
I'm embarassed to say its way over due for some critical updates.
The system is a Tyan 2469 K7X Pro with XP-M2400 at MP2800+ (16x133)

Ah, hi, TT, I should have asked you more over at EOC :)

Yes, I'd like to look over the proof, as being able to post a summary of it will help out other folks who don't know either of us.

First off, questions about the details: What are all of the working combinations (processor, stepping, mobo, ram, timings, mods, etc)? What was the combination that didn't work? I assume these are folding rigs, and as such, they're nominally running multiple F@H instances 24x7? I'd be tickled to turn my long struggle into a successful debugging operation. I'm repeating all my tests myself right now with attention to detail and 'best possible conditions' to eliminate as much possibility as I can that my problems were some other form of instability.

Second: 19 days certainly qualifies for my stated challenge; it's better than I've done by a factor of 5. Despite qualifying, have any of the machines exhibited other worrisome instability? A crash or FS corrupt that you couldn't explain some other way? My own worries have always been cache related; the various lockups never seemed to be FPU/ALU related. The processors have either spontaneously locked (indicating a direct logic fault kiling the kernel outright) or flushed out bad bits to memory until the VFS corrupted (either the kernel getting an esoteric interprocess IO error, or trashing the physical filesystem).

Monty
 
The motherboard is a Tyan 2469 K7X Pro.
I think the CPUs are AQYHA, but I'm not 100%. XP-M2400
I don't want to turn it off to see, but I would if you want.
Memory Crucial ECC Registered PC2100
The only mod I've done is setting the multiplier to 16x. It POSTs as MP2800+, 2 cpus
It runs 24/7 folding the whole time.
The only other thing that I do with it is rip, shrink, and backup my DVDs
 
i've got 2 XP-M 2200's at 140x17 running stable on a K7D Master

once i finish putting my case back together, i'll get my folding on.
 
mumrah said:
i've got 2 XP-M 2200's at 140x17 running stable on a K7D Master

once i finish putting my case back together, i'll get my folding on.

Proof, Sir, we need verification... uptime screenshot? Folding logs? Under a week of two cores *continuous* or a reboot/restart doesn't count...

Monty
 
Here's a 19 day uptime log:
CPU1
CPU2
The system is a Tyan 2469 K7X Pro with XP-M2400 @ MP2800 (16x133)

I'm really glad you requested logs, after looking at mine, CPU2 has been folding as Anonymous since January
 
I've left my computer on the last 2 nights folding and when I have come back to it, there's just a unrecoverable black screen. Maybe I'm seeing some of these problems??
 
Dang, and i was gonna get those mobiles from you xip. Once I read that post over @ eoc when i was in the market for a dualie, i backed out. After what you went thru, I wasn't gonna suffer the same fate, sry ;)

Looks like i won't be able to get in this challenge. I've got 2x XP1700 Tbred B's L-5 modded and pin modded running 15x140 on default vcore on a K7D-L rock steady stable and Folding away.

Good luck to those that accept :beer:
 
IIRC, mumrah is using my patched BIOS that allows the use of mobile CPUs in SMP mode without the cutting of the L5. This could be a factor in the stability of the system.

Somewhat mobile related, but not directly ... I'm (currently) running two mobilised + SMPised Applebreds on my K7D with my patched BIOS (same one as mumrah IIRC). At the stock multiplier (12.0x) I've had it prime95'ing for 48 hours continuous, and probably close to two weeks total (set to start when Windows starts). It hasn't come up with any errors as of yet. I'll do an endurance run next week though.

Finally, this seems to be a good thread to do a cheap plug for my K7DMSR program :) My primary suspicion is that mobiles have different boot settings for various MSRs, or perhaps the BIOS configures them differently. If all of you who replied to this thread could run the program from
http://www.emboss.co.nz/k7d/downloads/msrdump/k7msrdump.0.01.win32.zip
and PM or email me the logs (the account "michael" at the domain "emboss.co.nz") it would be much appreciated. Also, give details such as your motherboard and CPUs (ideally all three lines but these are not critical), and whether your system is stable or not.
 
emboss, mind explaining what the program is, the requirements for it and what it does? Id imagine quite a few of us would like to know details before we install a prog and then send out info inside it to someone we barely know. No offense meant here, but a bit more details might get you a few more volunteers.
 
Glad I found this thread!

Thanks for posting about this Xiphmont. I was this close (*gestures*) to buying a K7D-MasterL or a GA-7DPXDW-P and a pair of 35W mobiles to replace the single OC'd Tbird-1.4ghz in my Linux box. But from this thread and the one over at EOC, ht seems that at minimum both the MSI and Gigabyte boards have shown instability with mobiles even at stock speeds (any word on the Asus boards?).

It would have been nice to have been able to set up an SMP box that's 3x-4x faster while dissipating less heat but since this is my development box, I need stability.

Thanks for the heads-up. Looks like you saved me some headaches. :)

-- JM
 
I haven't read the other threads (meaning anything but this one), so I don't know if this has already been covered, but does this happen if you de-mobilize your mobiles? If it has to do with settings alone, then making it appear as a desktop chip might fix it? Seems to me that if modded regular XPs work just fine, mobiles ought to work just fine too . . .

Z
 
I have gone 4 weeks 1 day without a reboot on my xp-m 2400's in a ASUS a7m266-d running 2244mhz. I have had a freeze or two but im not sure if its the mobiles or the fact im using a software hack to adjust multipliers or peak heat problems while im away. The only 2 times it has frozen was around 1 pm and around 3 pm while i was at work and my rigs are outdoors in a shed so the environment isn't the best. I do have a couple of weeks of folding logs to prove and also sysinfo on IRC as well. My folding log file is 300+kb so it can't be posted here. If you have suggestions on how to get uptime and record uptime from xp let me know and i can post that if you wish. I think the bridge and wire mods may be the real issue here as most folks i see having stablilty issues are running K7D's while folks running the ASUS and the tyan's (not needing these types of mods for the most part). unfortunatly i dont have the time or extra $$ to replace my xps in the K7D. but i have run the mobiles in teh rev 1.03 asus and 1.04 and other than the crappy voltage control of the 1.03 have had no issues with the mobiles. Note also im not pushing these things to the max either as they will easily run 2400+ mhz by themselves. that may also have something to do with it. SO if your intending on info gathering,, we really need all the beans to sort this thing out right. We need board/rev/BIOS, procs used, mods used and so on.
 
i have an asus a7m266-d revision 1.03 with two unmodified 2600 mobiles at 17x142 (2.415ghz) and 2gb of pc2100 registered RAM...

and i have gone as much as 2 weeks without a crash or reboot...but this is only because some program install, or program update causes me to manually reboot before then...

i don't fold...but i do encode DVDs, etc and frequently hit 80-90% CPU usage
 
I can say "Hand on Heart" that my Iwill MPX2 with a pair of M 2500 is 100% stable, its been running Seti@Home Boinc, since day one of the project 24/7 100% on load.

I promise I would admit and moan if it was not stable :)

The only thing I can show, is This where at Number 37 RAC (atm) is my Iwill, in Total Credit it would be higher :)
 
Back