Memory settings for 2 sticks of CMX4GX3M1A1333C9 in a Biostar A780L3B

Hi,

Can you guys please supply me with the ram settings for running two of these sticks: CMX4GX3M1A1333C9

in a Biostar A780L3B motherboard. I recently just bought the board and the 2 sticks of ram as a bundle from an online retailer. Processor is a Phenom II X4 945

 

Thanks :-)

Are they not working properly on Auto/Stock/Default? What does CPUID's utility CPU-Z Memory Tab show? Timings are 9 9 9 24.

You are also running two unmatched sticks which is an unsupported configuration and Corsair can not guarantee rated speeds when mixing/combing memory.

If your machine is unstable at 1333mhz you may try lower ing then to 1066mhz and see if that is any better.

The timings do indeed show up as 9-9-9-24. However in MemTest86+ I am getting exactly one error on Test 8 when running both sticks together.

I realize corsair may not support running both these sticks at 9-9-9-24 if running them together as they are not part of a dual channel pack. But, I was wondering if perhaps this was something specific to my board where I had to turn of channel interleaving or something like that in order to run these specific sticks together.

Can I just loosen the timings and still run at 1333? Would that work? If so, what timings would I need?

If getting an error, recommend to test each RAM stick individually in the primary DIMM slot to try & isolate the stick throwing an error. If each passes, then there is a mismatch problem that may or may not be able to be compensated for. Rather than loosen the timings more than 9 9 9 24, try setting the DRAM V. a bit higher, say to 1.55-1.575v's range & test. Another JEDEC Standard for DDR3-1333 is 10 10 10, but not sure that is going to be much help if the mismatch is just too much.

I tested each stick individually for 5 passes each. And both sticks passed with no errors. Also, the DRAM voltage is already set to 1.60v

I also read around the forums here last night and saw some suggestions about running the ram sticks together at 2T with the values 9-9-9-24 put in manually. So I did that, and ran MemTest86+ last night for 12 hours straight. And I got more errors. 1 on Test 4, 3 on Test 6, and 4 on Test 7. The errors were actually found in the 2nd pass but I didn't check on the computer till much later.

I am going to try running the ram sticks together at 1.65 volts at DDR3-1333 and see what happens. Maybe it needs a little more voltage, as RAM guy posted that they can take up to 1.70..

If that doesn't work then I am going to run another test after lowering the speed to DDR3-1066 (apparently SPD picks up the timings for these sticks at 8-8-8-20 under DDR3-1066). Will update this post then. I figure one of these should work. This was part of a barebones bundle deal so I don't understand why an e-tailer would sell a bundle that will cause issues running at stock settings.

Well, just reporting back. Neither method worked. Memtest86+ still fails at the same address...

Thing is, each stick on its own passes just fine.

But when run together then memtest86+ fails at the exact same address..

0006a8035f8 - 1704.0MB

There are only two slots, so if I swap the sticks around in the slots... then the failing address changes very slightly to:

0006a8037f8 - 1704.0MB

 

Is this supposed to help me narrow down which stick has the issue? Or does this just mean what HU16E suggested - a mismatch problem that may or may not be able to be compensated for?

 

I am going to run 2 x 4gb of dual channel ram from a different manufacturer (these came in a kit and are matched sticks). If they pass memtest86+ then I can rule out the motherboard and its slots, and definitely chaulk it up to the mismatched sticks of corsair ram that came with the barebones bundle.

I am going to try running the ram sticks together at 1.65 volts at DDR3-1333 and see what happens. Maybe it needs a little more voltage, as RAM guy posted that they can take up to 1.70..

 

Whoa there! Yes, the RAM itself can, maybe even a little bit more, but that is risking permanent or fatal damage to the CPU! Having to apply more than 1.50v to modern day RAM stick(s) to achieve 1333 speed is ridiculous to begin with. Don't let this mismatch of sticks discourage you away from Corsair RAM to another brand. I will take an oath & have it notorized, in the past 5 years, with 16 complete rigs currently here in the hobby test room, all (100%) with nothing but Corsair RAM kits (all with Corsair PSU's as well), I have had '0', zero, nada, no RAM failures or RMA's, ever. :):

HU16E,

The Biostar A780L3B board by default sends 1.60volts to the ram. This is even after you clear the CMOS and load up setup defaults in the bios. I can't imagine 1.6volts would damage the cpu, especially since this is ram voltage we are talking about.

I didn't mean any disrespect to corsair as a brand. Actually, I generally always buy corsair ram and am a big fan of their products (such as their HX series power supplies to their XMS and vengeance series ram).

 

But every now and then a really good deal comes across (you know how it is) and I will get a different brand just for throwing into a hobby rig or as a cheap upgrade for a friends computer or what not. I don't have a dual channel DDR3 kit of corsair on hand at the moment, otherwise I would have tested with those. At the moment, memtest86+ has not yet shown any error. It has been 3+ hours since I started the test with the dual channel matched pair sticks in the slots. So, so far it looks like the issue I was facing was most definitely a problem with unmatched sticks not being able to run together in dual channel.

 

Just to clarify:

Every time I make a build for someone I go for only a select few brands for the components - corsair is at the top of that list. Now, at the moment I have several corsair DDR2 kits still running in core2quad and AMD Phenom X3 rigs.

And out of all the corsair ram I have ever bought, I only ever had to RMA one dual channel kit and that was because one stick was completely stock full of errors. The RMA process was very smooth and I had a new kit in my hands within 3 weeks of sending out mine. I understand that sometimes these things happen and you have to RMA a product made by any company.... not just corsair. I am just glad that the RMA experience with corsair is much better than the rest.

I can't imagine 1.6volts would damage the cpu

 

Is the Memory Controller in the CPU or is it not? RAM getting set by the BIOS to 1.6v is 'non' JEDEC Standard for normal DDR3 RAM. You had mentioned maybe trying 1.65v or even more, IMHO, that is just so wrong for a 1333 speed to need that much DRAM V..

 

Good to hear a solution has been found. As suspected, the mismatch was just too much to overcome.

HU16E,

The Biostar A780L3B board by default sends 1.60volts to the ram. This is even after you clear the CMOS and load up setup defaults in the bios. I can't imagine 1.6volts would damage the cpu, especially since this is ram voltage we are talking about.

This is incorrect, AMD's are not like the Intel chips and most do require more voltage than the intel chips. 1.7 isn't out of line for that Phenom CPU and chipset.

 

BTW, the modules are guaranteed up to 2.0v, so no need to worry about dmageing the modules.

 

I am going to run 2 x 4gb of dual channel ram from a different manufacturer (these came in a kit and are matched sticks). If they pass memtest86+ then I can rule out the motherboard and its slots, and definitely chaulk it up to the mismatched sticks of corsair ram that came with the barebones bundle.

How did this turn out?

How did this turn out?

 

At the moment, memtest86+ has not yet shown any error. It has been 3+ hours since I started the test with the dual channel matched pair sticks in the slots.

 

Looks to be resolved by this statement.

 

This is incorrect, AMD's are not like the Intel chips and most do require more voltage than the intel chips. 1.7 isn't out of line for that Phenom CPU and chipset.

 

No wonder I want nothing to do with AMD systems! Yuck!

The dual channel kit with the matched pair of ram worked perfectly. I left memtest86+ to run all night. After 14+ hours, there were still zero errors.

 

By the way, the processor is a Phenom II X4 945 with an integrated memory controller.

 

Thanks everyone for your help. Looks like the matter is resolved now.

I am now left with two unmatched 4gb sticks of DDR3 that work perfectly on their own, but not so well with each other. :-p Oh well. That's life. I can always find a use for them down the road.