michaelk, what kind of LE-1640 do you have? The old 2.6ghz or the new 2.7ghz(LE-1640B)?

_________________
HP EX470, 4 GIG, 4200+ EE SFF, WHS V1, Production
HP EX485, 4 GIG, E5800, WHS 2011, Stablebit Drive Pool, Transfer Server, Backup Server
HP EX490, 4 GIG, Q9400S, WHS 2011, Stablebit Drive Pool, Mediaserver
HP EX485, 4 GIG, Q8200S, WHS 2011, ITUNES Server

Yeup, I am. And the failure with the original RAM happened with the Northbridge fan mod done as well.

I don't (umm, "no longer" ) think my failures are heat related.

_________________
:: Mark

Mark, in looking at your posts, you seem to be running ok with the exception of the orthos error which I don't think is an error, could very well be an bug in orthos. I think I read somewhere where people were able to get by that error by running mutiple instances of orthos and when they did that it ran 7-8 hours with no errors. But, single instance caused the same error at the same time every run. I looked at the voltages and all I tell you is the 5 volt measurements were exactly like yours using the stock power supply, when I went to the Shuttle power supply it lowered them to slighly under 5 volts. all the rest look ok. Forget the -12 volt, etc because stock or not they vary all over the place because there is no -12 volt supply coming into the motherboard.

_________________
HP EX470, 4 GIG, 4200+ EE SFF, WHS V1, Production
HP EX485, 4 GIG, E5800, WHS 2011, Stablebit Drive Pool, Transfer Server, Backup Server
HP EX490, 4 GIG, Q9400S, WHS 2011, Stablebit Drive Pool, Mediaserver
HP EX485, 4 GIG, Q8200S, WHS 2011, ITUNES Server

Cool, thanks sir!

I think I said this somewhere up above in the muck, but at this point I'm kinda' done chasing this and am willing to just leave it until I see a failure while doing "normal duty".

Though that SNM issue's got me tweaked too...

_________________
:: Mark

honestly no I'm not running at 100% (first it never got far enough booting to adjust fan speeds with the memory stick but without the stick and the 1640 I assumed it was fine and the reported temps were ok so i didn't bother)- I just want a nice and solid machine so if I need to fiddle to get it to work then it's not safe enough for my use.

I plan to run home automation on the thing with mcontrol so it just needs to work for my partiular needs.

I thought it was the memory and even got an rma (although havent sent it back)- but after the 1640 crapped alone I'm not so sure. I assume most sticks are dead or good and it's rare to get one that's iffy enough that it can get booted all the way up such that I can ping the pc and then it craps. So this weekend I plan to try again with just the memory upgrade when i get a few minutes. I'll try to remember to report back.

it reports 2.6 anyplace on the pc that says so.

I got it from tigerdirect and it says part number ADH1640DHBOX on their website.

I need to double check the packaging at home but I'm fairly certain it didn't say 2.7 anywhere.

I never tried running fans at 100%. I assumed the fan speed add-in whould have ramped the fan speeds up as temps increased (which it did). Whenever the failures occurred in Orthos or OCCT, it forced a shutdown. All I was left with was the frozen screen of the RDP session indicating the failure.

Wow... quite a few posts to catch up on...
I also agree... Despite the fact that erail has nearly eliminated the PSU as being the culprit, I still think the shutdowns are power demand related... only now I'm leaning more to power components on the motherboard (which we can't do anything about).Consider for a moment what Orthos is actually doing: It's performing a complex calculation which it already knows the answer to... so when the computer doesn't generate the result it's expecting it means the computer has performed a computational error.. I don't know about you guys but it would bug the crap out of me if i knew my computer (or server) was able to to make mistakes whenever it starts to break a sweat.

But to play a bit of devils advocate: Aside from some RAM (depending on the test you select), Orthos really only uses the FPU part of the Processor (this is the part that deals with fractional numbers vs whole numbers)... and most day-to-day tasks (aside from perhaps encoding/decoding and other similar numbercrunching) don't really use that part of the CPU...
I passed my tests with the fans at the stock settings... that said I've been running the 'optimized defaults' (-20 threshold temps) fan control setting now for weeks (without any shutdowns).

I can't express how disappointing it is that even the BE-2300 & 3800 SFF may not work for everyone... based on the early results from myself & others I thought it was nearly a sure thing!

Is there some critical difference/variable we've missed that seperates those who are having complete sucess from those with (partial) stability and/or shutdown issues?

At a face value, the problem I experience (not able to connect a Hauppauge HD-PVR to the MSS) doesn't seem related to the processor's problem many are experiencing. But underneeth it may.

The project I'm on now is to replace my BE-2350 with the stock processor and try to install HD-PVR, then I wish to flash the original BIOS and do the same. If it works, and it should since I read somewhere that someone has got it to work on its MSS, I will try to mod the BIOS alternately with the new AGESACPU then with the ACPITBL. If nothing works, then I'm off. If something works, than maybe we will have more info on the two modded parts of the BIOS we are actually using.

These days I reread this whole thread and understood some issues I could not grab at the beginning. ymboc had explained we could use other MB BIOSes for modding. But after someone failed with a newer version of the AGESACPU, he rightly recommended us to stick with what was safe. I am still wondering if there are other BIOS donors we could use.

In earlier posts I pointed out that the MSS motherboard was an Acer according to its code. But CPUID reports it is a Wistron. Well Wistron was a division of Acer until 2002, now they are MB providers for Acer and HP and also others. I tried to find a Wistron MB on Internet and couldnot, but there are a lot of Acer MBs (even if Acer declares they don't manufacture MBs, in fact they get many of them through Wistron). So gessing that Wistron MB are still called Acer, I tried to find an Acer MB with the SiS chipset, but nothing found until now. If ACPITBL deals with power management, system description and affinity tables, etc. maybe there is something there to examine.

If this has nothing to do with failures we encounter, just tell me and I'll get off this route. But I would like to give it a try.

If on the other hand it is related to power components on the motherboard, as ymboc puts it, we can't do much about it.

_________________
HP-ProLiant N40L 8GB WHS2011
HP-EX470 4GB Athlon-X2-BE-2350-(Stepping G1) Windows 8 (server build)

I don't remember reading that anyone having the computational errors with Orthos has RMA'd their processor. It would be quite telling if that was done and a different processor did not encounter the errors. If it did it should eliminate the processor as the source (wouldn't eliminate it 100% but pretty close). Any volunteers?

TXDot:
I think you'd have to first test your AM2 processor in a different platform -- one with a bit more headroom and where our processors are on the motherboard's respective supported CPU list..

On the flip side...And I realise hindsight is 20-20, but those having trouble should (have) test(ed) their system with the stock sempron (before voiding their warranty) to see if the stock system passes stability testing.

Motherboard limitations must be eliminated before you try to RMA a CPU

VieuxJules:
I did look for potential acer donor bioses when I started (having made the wistron connection early on)... I know they have a few very small form factor amd-based systems but while I found ones with the same northbridge I didn't find any that used the same southbridge (966). If you compare the specs of the shuttle bios to that of the MSS you'll realise its a close match and that's why we got lucky.

From what I figure the MSS is closely(kinda-sorta?) based off of AMD/Gigabyte's Churchill Reference Design (761GX/966 combo)... and that either AMD or SIS may have been heavily involved in the development of the BIOS and may typically be involved in the generation of many bios's -- judging from SIS's contributions to the linuxbios / coreboot project... and because the menu structure & layout of the Shuttle Bios is very similar to that of the MSS (with similar limitations and general lack of options typical of oem hardware).

The agesacpu component won't/can't be responsible (it's provided by AMD) for any of the quirks you're seeing but the acpitbl component might be... aside: what exactly isn't working with your HD-PVR?

You may remember (it was sometime ago) that my stock processor had not passed the Orthos test, but the BE-2350 did at several trials at least two of them for more than 8 hours.

By this logic, my Sempron is being clearly declared defective. Anyway, since I'm putting it back in for my tests, I will give Orthos another try. My system is ready, I'm just waiting for the vga cable to be delivered. What I don't remember is: did I test the Sempron before or after the BIOS mod. Since I will flash back the original BIOS, I will orthos-it under the two conditions... I'm writing that down to be sure I don't forget.

_________________
HP-ProLiant N40L 8GB WHS2011
HP-EX470 4GB Athlon-X2-BE-2350-(Stepping G1) Windows 8 (server build)

My system passed stress testing with flying colors using the stock processor and has never shut down.

The LE-1640 was never stress tested, but did shutdown about once a week.

The 3800+ failed stress testing both on OCCT and Orthos and experienced shutdowns.

I'd say the chances of two bad replacement CPUs belonging to two different families of processors are highly unlikely. I for one am convinced this is a MB issue (most likely related to power regulation) and barring a miracle there is probably no fix.

Aside: for the HD-PVR, the ArcSoft TME (TotalMedia Extreme) will report a: "cannot find HD-PVR, or it is used by another program" kind of message. I was in touch with Hauppauge for this issue and after some basic recommendations, they throwed the towel and just said it could be not supported for my machine. But I read somewhere that someone got it right... I didn't kept a trace of that message.

So I'm looking to test the BIOS components, but if I want to stay on the safe side, my tests will be over a little number of BIOSes. I don't want to brick my machine and don't intend to solder a BIOS socket. Unless I find this BIOS switch ...

It seems to me there is still a small chance that another ACPI table could do the trick before we throw the towel and resort to the no-fixable MB issue. I understand clearly that you (we) got lucky with the Shuttle...

_________________
HP-ProLiant N40L 8GB WHS2011
HP-EX470 4GB Athlon-X2-BE-2350-(Stepping G1) Windows 8 (server build)

2.6 for sure.