Advice for overclocking XMS3200XLPT PCB rev 1.2

[In-Fluence]

After some success with overclocking this pair of 512 MB sticks, I'm wondering if if I could call upon your professional opinion on what I could do to push for better results... Call me greedy for having a 50% overclock already :biggrin:

 

I'm currently using a DFI-LP NF4 SLI-DR Expert with an Opteron 170 (stable at 280x10), and after some experimenting with a good guide from xtremesystems (written specifically for TCCD), I have managed to get them memtest stable at 300MHz, whilst managing to retain reasonable 2.5-4-3-6-1T timings (amongst the plethora of other settings that can be adjusted). All this on only 2.66v (measured with a multimeter).

 

So, when backing down the memory speed to 275MHz, I pretty much assumed it should quite easily manage 2.5-3-3-6-1T timings. Unfortunately, I just cannot seem to get it stable when altering just that one timing (trcd), and it just pops out the odd error in ramdom locations during tests 5 & 8 after 2-3 hours. 2.66v still seems to give the least errors and increasing produces more, whilst less just causes a hang (not suprising on such low vDIMM).

 

So, my questions are:

 

1) My TCCD are revision '537' chips, which I undestand are a reasonably new revision. In the aforementioned guide, it says that newer TCCD revisions seem to like more voltage. Another thing I thought might be a contributing factor was the PCB revision. I understand v1.2 are labelled 'Brainpower', and I also read that these PCBs are designed to be more efficient at lower voltages. How true is this likely to be considering my results, and what voltages could safely I push to for optimal results? If I did, should other settings be altered as well?

 

2) What settings could you suggest tweaking to help me reach 2.5-3-3-6-1T at 275-280MHz? As suggested in the guide, I have tried varying (but not all) combinations of:

 

Skew values +/-

TRC

TRFC

TREF : which I understand should be set according to memory speed??

Drive strength

Drive Data Strength

Async latecncy & Read preamble

 

Which one / combinations of settings would you consider to be most important?

 

3) My recent attempts at reaching 275MHz 2.5-4-3-6-1T were 8 hours memtest stable. However, It seems that when running superpi I cannot bench for over a minute without rounding errors (on CPU affinity 1,2 or both). Even after going to 2T, I still get rounding errors and I'm wondering if there might be any info you would be able to offer to help me a little more here. I kept it superpi stable on an ASRock board at the same speeds (but on 2.81v), all other components the same as before. The only difference being the number of memory settings available :-)

 

Any advice would be very gratefully received. I would be happy to post detailed results if I can overcome these queries & issues :biggrin:

[CMC_SAVAGE]

Your Dimm Voltage is way too low.

2.8 at least.

[In-Fluence]

Well, my board has been on RMA for nearly a month now, so testing will have to wait. But like I said earlier, higher voltges seem to consistently throw out more errors. I could certainly try something a bit higher when I get the board back, though it seems I may just be better off sticking with the looser timings until I go DDR2 anyway :D:

 

Thnks for taking the time to read :)

[In-Fluence]

Well, after a long wait, I managed to get the board up & running with memory stable (8hrs) at 275-2.5-3-3-6-1T. This was eventually managed with 2.95v which I know is rather high, but memory didn't seem too hot to the touch. I have some spare sensors, and I was thinking about putting these on each of my sticks. What sort of temperatures does anyone think is considered safe for TCCD?

 

I am still getting superPI errors, but I'm dual prime stable for 8 hours - bit baffled by this :confused:

 

p.s. does anyone have TCCD revision 537? I'm keen to find out what others might have acheived!