Shuttle SN25P & AMD X2 CPUs

[Doom9]

Does anybody has this combination running stable using stock settings (so no exchanging SATA cables, disabling SATA spread spectrum, disabling USB2 or using only one core)?

[boban10]

look here for more posts about problems with this board:

http://www.xtremesystems.org/forums/showthread.php?t=69416&highlight=Shuttle+SN25P



http://forums.anandtech.com/messageview.aspx?catid=51&threadid=1556194&enterthread=y


and here:
It is reported that with the Athlon X2 CPU and Some NF4 Motherboards, Regardless of BIOS's, and Drivers, USB2.0 will not work when Windows is using Both Cores. It has been solved that Either USB 1.1 will have to do, or 1 core active for USB 2.0. This is something that I find horrible, and Dont know who is at fault here. The Biggest Problem is the Shuttle SFFs (SN25P) Nearly Everyone with the "revision1.0" of this has this problem. Here is a link to another forum where they discuss it in depth, and see how many people accually have the problem. They do tests and everyone who uses USB2.0 with an Athlon X2 (if it works at all) will lock the computer instantly.

http://forums.sudhian.com/messageview.aspx?catid=43&threadid=80305&highlight_key=y&keyword1=SN25P

[Doom9]

I already know all that.. I'm trying to find out if there's anyone not affected because I have the lingering suspicion that the SN25P and X2 is just a no go, period (disabling cores and stuff doesn't count). Plus, I have a certain crasher that even teh USB1.1/Core disabling won't fix: Memory Test in PC Mark 05. It crashes the box reliably even with USB 1.1 and even with one core disabled, but it works out just fine if only you switch out the CPU for a single core model.

[Revgen]

According to Shuttle's site the latest BIOS patched the Cool 'n Quiet function.

http://www.shuttle.com/share/fae/hq/download/bios_rdm/readmefn25.htm

Assuming that your board carries the latest BIOS (FN25S01O), try disabling Cool 'n Quiet and see what happens, or try to revert back to the FN25S01K BIOS.

The BIOS engineers could have made a mistake.

[Doom9]

try disabling Cool 'n QuietIt never is on unless somebody figures out that yuo have to find and download the drivers on the amd site. So by all estimates the majority of people don't get to use that feature.

I haven't had a freeze for hours (the first one since I got the PC back, and the PC was acting weird since the boot with the Firewire HD connected from the getgo).

Note though that I'm looking for people here.. not any help. I know it's hard for us guys.. we always want to help if we hear of a problem.. but that's the way it is. I'm just trying to determine if there's anybody who can run this damned box problem free. I give it a few days, then get in touch with pc news outlets and try to have them write something nasty about Shuttle for this incredible mess they've caused with the X2 chips.

[Revgen]

This might be the real cause of the problems.

The SN25 comes with a 350 watt PSU to drive this system.

http://global.shuttle.com/Product/Barebone/SN25P.asp

I don't know about you, but I don't think running a high-powered dual-core CPU on a 350 Watt PSU is a good idea. It's made even worse when running it in a SFF case where the higher temperatures can tax the efficiency of the PSU.

These conditions are probably made even worse when running CPU intensive applications such as video encoding.

It's common knowledge among computer techs that 30% of all comp problems are caused by a faulty or underpowered PSU.

<Bottom Line>

If Shuttle is indeed marketing their barebones as "Dual-Core ready", they should be scolded publicly for offering such an underpowered PSU to drive their systems.

[Doom9]

the PSU is hardly the problem.. that would be something all the SFF tests ot there would've picked up. AMD CPUs are way less power hungry than their Intel counterparts.. so you'd expect the Intel barebones that use the exact same PSU to have a lot more problems. Yet there's quite a few people using those products with top of the line GFX cards, and no problems reported. Considering that the SilentX PSU has been out for quite a while now (some time last year), if they really weren't up to the task we'd know for sure by now. The AMD dual core chips are only out for about a month, and it's only those creating problems.

I have not found a single shred of evidence that would point to the PSU. Running Prime for 10 hours straight without a glitch, doing the same with the Burn in test should be enough to prove that the PSU has nothing to do with it.. stress isn't the killer. If your ead all the threads linked above you'll see that all those boxes freeze under almost no load, but run stable at full load. It really is an incompatibility between the mainboard and the CPU, nothing else. And I'm not the one to have performed the most tests, there's others that have this box running stable under full load, with more and more powerhungry components that I have.

[puffpio]

have you seen this post yet?

http://hardforum.com/showthread.php?t=930760

I pretty good writeup (ie step by step instructions to minimize problems)

[Doom9]

Yes. I cannot agree with the IDE driver comment though, I have 3 barebones to back up my version. And the processor driver is the same for single and dual core.

BTW, Shuttle is acknowledging the problem: http://forums.sudhian.com/messageview.aspx?catid=43&threadid=81070&enterthread=y

[Doom9]

I found something that really scares me: the X2 has a known bug that could cause to memory corruption/freezes. More info here: http://forums.sudhian.com/messageview.aspx?catid=43&threadid=80219&enterthread=y

[boban10]

every known cpu have its bugs...
p4 too http://www.theinquirer.net/?article=16823

i dont think this bug have anything to do with your problem, as why with other boards all is working right ?
i think this is a Shuttle only problem, and they need to fix it now...and not in 3 months...
thats only my opinion based on forum posts, about Shuttle and X2 problem, i dont see problems with other boards... (maybe there are, but nobody post about that problems for now)

[Mr. Monte]

Maybe a recall in the future?

[Revgen]

I found something that really scares me: the X2 has a known bug that could cause to memory corruption/freezes. More info here: http://forums.sudhian.com/messageview.aspx?catid=43&threadid=80219&enterthread=y

This thread (http://www.nuendo.com/phpbb2/viewtopic.php?t=5406&postdays=0&postorder=asc&start=0) may confirm your fears. Although the guy is using a beta BIOS to use his Athlon X2, the problems could possibly be related to the article you just mentioned.

[Doom9]

speaking of beta bios, I tried to flash my box with the beta that should fix the freezes, but I ended up with a dead box instead :( That's certainly a first in my career.

[Revgen]

I had nearly a similar experience. I flashed a new BIOS for my old Athlon XP ABIT board and it wouldn't boot up at all.

I then got an idea in my head to remove the lithium battery that powers the CMOS chip and then put it back in just to see if it would do anything.

Lo and behold the thing ended up working okay again.

[Mr. Monte]

I had nearly a similar experience. I flashed a new BIOS for my old Athlon XP ABIT board and it wouldn't boot up at all.

I then got an idea in my head to remove the lithium battery that powers the CMOS chip and then put it back in just to see if it would do anything.

Lo and behold the thing ended up working okay again.

Same here on my Shuttle SG62.

Flashed, no worky, removed battery and it worked

[Doom9]

Naturally I thought of all that;) Reset CMOS button at the back of the case, reset jumper inside the case and the battery. All checked, and all lead nowhere. I wouldn't give away my useless but very expensive piece of machinery if I could salvage it somehow. If only Shuttle were to use dual BIOS...

[Joe Fenton]

This thread (http://www.nuendo.com/phpbb2/viewtopic.php?t=5406&postdays=0&postorder=asc&start=0) may confirm your fears. Although the guy is using a beta BIOS to use his Athlon X2, the problems could possibly be related to the article you just mentioned.

The bug in question:
Product Errata 75
Revision Guide for AMD Athlon™ 64 and AMD Opteron25759 Rev. 3.51 May 2005 ™ Processors
123 Bypassed Reads May Cause Data Corruption or System Hang
in Dual Core Processors
Description
An internal data path allows some cache line fill requests to bypass the L2 cache and initiate an early DRAM read before receiving hit/miss status from the cache. Under certain timing conditions at low core frequencies, read data from such a request may be returned to the processor core before the core is ready, resulting in an internal bus hang or a corrupted victim data buffer.
Potential Effect on System
Data corruption or system hang.
Suggested Workaround
Contact your AMD representative for information on a BIOS update.
Fix Planned
Yes

Notice the key phrase - at low core frequencies. The "fix" may just be to restrict how much throttling is allowed.

[Doom9]

I guess another thing to mention to Shuttle. But, the issue in question has nothing to do with voltage reduction, it happens even without Cool & Quiet (I haven't figured out how to use that until very recently).

[lexor]

also doom, if you motherboard is using nf3 chipset, then that thing you posted on the main page about problems with 1T timings is to be expected.

I have a full desktop nf3 board and I can't run my memory at 1T timings at all, it's basically very picky about what memory it will allow to run at 1T, some higher grade sticks will work, were-as a lot of "value" sticks (by same manufacturers) won't. It's just the way it is with nf3 chipsets (from what I understand motherboard makers didn't even want to allow 1T option in bios when they first appeared, but they left it in just for the lucky ones for whom it works) so I have always run my mem at 2T.

[Revgen]

First of all, I believe that Doom9's system is an Nforce 4 chipset.

Also, the memory controller on AMD 64 based systems is integrated into the AMD processor, not the chipset. So any 1t/2t timing problems are probably going to happen on all AMD 64 systems regardless of chipset.

[Doom9]

actually, I've been searching for 1T and found that there's reviews where they don't find any problems with those memory timings and that chipset.

Also, my memory has the following stock timing settings: 2,2,2,5,1, so it's not exactly your average RAM module (actually it's a kit of two) but high end stuff (Kingston HyperX Ultra CL2 3200 to be precise)

[Doom9]

So any 1t/2t timing problems are probably going to happen on all AMD 64 systems regardless of chipset.This is definitely not correct.. the proof is currently visible on my screen where PC Mark just successfully completed the memory test on my trusty old Newcastle A64 3500+ at 1T (using the exact same RAM configuration (different modules of course but same model, and the same memory settings in the BIOS). Clearly, Shuttle has some more fixing to do, and this time it's not an X2 thing as it happens with single core chips as well.

[Revgen]

This is definitely not correct.. the proof is currently visible on my screen where PC Mark just successfully completed the memory test on my trusty old Newcastle A64 3500+ at 1T (using the exact same RAM configuration (different modules of course but same model, and the same memory settings in the BIOS). Clearly, Shuttle has some more fixing to do, and this time it's not an X2 thing as it happens with single core chips as well.

You're correct, I just installed my new Athlon X2 4600+ yesterday and it runs at 1T settings fine.