AM2 X2 - Asus nf570 - Corsair RAM - not stable??

All about them.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
Hifriday
Patron of SPCR
Posts: 237
Joined: Thu Aug 05, 2004 3:32 pm

AM2 X2 - Asus nf570 - Corsair RAM - not stable??

Post by Hifriday » Tue Aug 01, 2006 9:10 am

We needed to add two mid-range database servers at work, and with the recent price drop it was the perfect opportunity to pick up some dual-core s939 Opterons and hope for drawing some 35W samples. Unfortunately Opteron prices haven't dropped as significantly as the regular X2 chips. Even the s939 X2 chips are becoming scarce in the local market, and the vendor we use only had AM2 chips. I was hesitant to go with AM2, motherboards still being relatively new, but the vendor assured me of good stability and actually AM2/DDR2 worked out slightly cheaper compared with the older s939/DDR. So we setup two machines as follows:

X2 4200+ / ASUS M2N-SLI Deluxe
2 x 1GB Corsair VS DDR2-667
WD 74GB Raptor
ASUS 7300LE / ANTEC SU-380

X2 3800+ / ASUS M2N-E
2 x 1GB Corsair VS DDR2-667
2 x WD RE16 2500YD
ASUS 7300LE / SS S12-430

After some initial testing, both machines are having stability problems. I don't know if this is a problem with AM2, the ASUS motherboards, or the Corsair Value RAM. Have others also encountered stability issues or are your AM2 X2s running just fine?

With the first rig, if I enable CnQ and only run one instance of P95, the CPU clock will cycle up and down from 2.2ghz down to 1.8ghz and occassionally to 1.0ghz as one instance of P95 will not fully load both cores. Within 5 minutes or so P95 will error out. However if I run two instances of P95 (one per core) then clock rate will stay maxed out at 2.2ghz and no stability problems yet. Also with CnQ disabled either in PowerMgmt or BIOS, or using CrystalCPUID, I have not experienced any P95 errors. But CnQ is one of AMD's features, certainly it should not cause stability problems?

With the second rig, regardless of CnQ, it won't even run P95 for more than 1 minute!! However after switching the RAM from slots A1/B1 to A2/B2 it seems to be fine, but I will need to test it more. I know motherboards and RAM can have compatibility issues, but this is not performance memory with high clocks and tight timings. There is no overclocking, all RAM settings left on AUTO and timings were properly set per CPUZ. I even tried fixing the RAM to 667mhz in BIOS but it didn't make any difference. If really picking the right RAM is so sensitive, I am wondering how stable is AM2? Any others with AM2 experience?

On a side note, unfortunately TcaseMax is unable to read the sample TDP for these AM2 chips, it seems AMD no longer puts this on each chip. However I did do some AC power measurements for those interested:

RIG-----------(1)----(2)
Idle CnQ------72----70
Idle No CnQ--81----77
1xP95--------115---108
2xP95--------133---122

This is all measured in VA, but from SPCR's reviews, PF should be close enough to 1 that it can be taken as watts. Also both PSU should be operating around 77-82% efficiency.

=assassin=
Posts: 243
Joined: Thu Aug 25, 2005 2:46 am
Location: Blackpool, England, UK
Contact:

Post by =assassin= » Thu Aug 03, 2006 3:10 am

I've heard of a number of cases with RAM being incompatible with this board - you may have to do a BIOS update, or even RMA and get a different motherboard. So many AM2 boards seem to have issues with the DDR2 memory, usually because the voltage for the memory is set too low.

Hifriday
Patron of SPCR
Posts: 237
Joined: Thu Aug 05, 2004 3:32 pm

Post by Hifriday » Thu Aug 03, 2006 11:16 am

=assassin= wrote:So many AM2 boards seem to have issues with the DDR2 memory, usually because the voltage for the memory is set too low.
Indeed this seems to be the case.

Browsing Corsair/Asus forums it does seem quite a few others are having the same problem and even some who can't even get Windows installed!! The suggestions include 1) update BIOS, 2) increase voltage to 2.1v, 3) drop clocks down to 533.

For the M2N-E, Memtest ran fine for 4 passes (around 2 hours), but once in Windows/P95 even with CnQ disabled it would error out within an hour. However setting the voltage manually to 1.95v (the maximum setting for this board's BIOS) has allowed it to run stable for 6 hours so far with CnQ enabled. I will let it run overnight and see the results... fingers crossed.

For the M2N-SLI Deluxe, with CnQ disabled (voltages left at auto) it was P95 stable for 8 hours. I then tried setting the RAM voltage also to 1.95v (this board's BIOS does let you go much higher) and re-enabled CnQ. Again letting it run and hoping for good results tomorrow...

darkdays
Posts: 4
Joined: Mon Mar 20, 2006 11:12 pm

Post by darkdays » Thu Aug 03, 2006 12:18 pm

Wow, thanks for the thread! Was just about to purchase X2 and Asus M2V combo, keep us posted!

vitaminc
Posts: 306
Joined: Mon Mar 27, 2006 8:11 am
Location: Silicon Valley, California

Post by vitaminc » Thu Aug 03, 2006 1:48 pm

You should update the BIOS for M2N-E to 0203, it's much more stable than 0103.

I am running 3200+, M2N-E, Corsair DDR2 800, CnQ enabled. My system has been up for more than a week (full time LogMeIn server and BT machine), and its stable when I play WoW/WC3 at the same time.

But yes, M2N-E isn't very suitable to OC because Vdimm goes up to 1.95V only, and there are some compatibility issues with DDR2 memories. OCZ won't work on this one (RMA a pair).

Hopefully ASUS update the BIOS sometimes soon.

Hifriday
Patron of SPCR
Posts: 237
Joined: Thu Aug 05, 2004 3:32 pm

Post by Hifriday » Sat Aug 05, 2006 7:35 am

vitaminc wrote:You should update the BIOS for M2N-E to 0203, it's much more stable than 0103.
I considered updating the BIOS as well, but from forum posts it seems this didn't solve the stability problems and some users even reported it being worse.

In anycase, both systems managed to run Prime95 stable for 21 hours WITH C'n'Q enabled. I would have liked to test them further, but unfortunately these machines needed to be shipped to their install site, where they'll do more testing. During this time I concurrently ran some Sisoft Burn-in runs, as well as briefly launching two additional instances of P95, and nicely all remained stable. So it seems that increasing the DDR2 voltage to 1.95v did the trick in this case.

Post Reply