F3-17000CL9-4GBXM and TZ68A+

[Tradesman]

Asus is pretty good, my top three are ASRock, Asus and GigaByte and those are about the only ones I build on - unless a client specifies a particular brand/model....I haven't been real pleased w/ support from Asus or GB, and while they can be slow about things, once prodded, I've been happy w/ ASRock....Been using them more and more lately. I've got a 2500K on a P67 Fatal1ty and a 3570L on a Z77 EX IV

[rthaney]

Important Update

OK. So did some more testing and tried the old "One stick at a time" test.

Result: worked at specified timings and better. But with a serious hitch.

Here is the story:
Was testing one stick at a time and decided to push all the timings very high and lower them gradually to see where my thresholds were. Got down to 9-11-10-28 @2133MHz but still had the other timing very high. Eventually got them down to where they were before and even better.

Could not figure out what I did wrong before. Swapped each module to test and both passed memtest and prime 95. So, what did I do wrong? decided to check each of the memory slots. Put the memory on the second bank of DDR slots and WHAMO! Did not read the rimming correctly. I set the timings to 9-11-10-28 but it read them as 6-6-6-10 @ 2133MHz. This did not happen in the first Memory Bank. No wonder there were errors galore.

So put them back in as dual channel and again, errors all over the place. In fact the errors were worse in this configuration than in the second set of banks alone.

So now the question is, is this Biostar's issue or will they pawn it off on Intel, since the memory controller is on the chip.

So the memory is not the problem and the memory will work on the computer in the right configuration.

Wish me luck.

[Tradesman]

Sounds like a bad slot, the MC should see each stick individually unless the slot is messed up

[rthaney]

Biostar had me replace CPU to test. Still getting errors on two slots. So looks like a Motherboard issue, as you said.

Hope they are fast at turn around, I work from home and need to use my computer for work. Got a back up but I do software testing which, as i am sure you know, needs specific configurations for testing.

So, wish me luck and thanks for all the help.

[qiqi1021]

VCC IO is the voltage setting for the memory controller, not VCC SA!

VCC SA is the system agent voltage and doesn't need to be touched even at extreme overclocks. It runs at 0.925V by default, and Intel specs state max is 0.971V.

[Tradesman]

rthaney
Let us know how it goes

qiqi1021

Ideally it shouldn't have to be raised........if you run everything at stock and 2 sticks at CPU specified DRAM (generally 1333 or 1600) or if you have a Z77 (with a good BIOS)...but since most run faster sticks on Intel and OC the CPU it often needs to raised when OCing the CPU, having high capacity DRAM, multiple sticks and most importantly at times (which is quite often) the the mobo makers don't have the BIOS right to run all components corrects. Further the VCCSA (System Agent Voltage) in many cases is beginning to incorporate the MC (Memory Controller) voltage as part of the VCCSA instead of using a separate (depending on chipset) VVCIO, CPUVTT, QPI/VTT, etc (numerous different names based on socket and chipset). As far as Ivy Bridge, my 2 Z77 systems both default to .0925 and on the 3570K to run 32GB I had to run it at 1.1 until the BIOS was corrected to handle the sticks, now it runs at stock....if it was true that there's not a need for it to be adjusted (either up or down) it would be a locked setting, or wouldn't even appear. If you actually takes what Intel prints to heart, one would wonder why they even make CPUs that can OC (since they say OCing is a no-no......or why they ever came out with the XMP standard since their CPUs are basically all rated for 1333/1600, yet XMP runs right on up through 2800.

[qiqi1021]

I'm just saying in this case we should be looking to raise VCCIO voltage as that is the voltage of the memory controller for Sandy Bridge. With my 2 Sandy Bridge systems increasing VCCSA had no effect on stability but VCCIO did, something that appears to be consistent with most reviews and user experiences I've seen.

A few quotes from Sin0822:

VCCIO: more commonly known at QPI/VTT voltage, this is the VTT voltage. Formally known as Processor Power for I/O it is the voltage for the integrated memory controller as well as the PCI-E controller. While Intel?s Maximum is 1.05 +/- 3% = 1.08v, you can go higher, much higher. I would recommend staying below 1.2v for 24/7 use, but depending on the quality of the IMC on your chip, I have seen 2133 MHz done on as little at 1.1v.

VCCSA: More commonly known as System Agent Voltage. Intel?s maximum System Agent Voltage (Vccsa) is 0.971v and minimum is 0.879v. Stock is 0.925v. System Agent Voltage should NOT be touched, it is supposed to be a fixed voltage, and it powers many things that the VCC does not power. One of the most important is the Power Control Unit (PCU) which controls internal power allocation of the processor. This voltage is to be generated by a separate VRM than used for SVID.

[Tradesman]

To a degree what is being said is rather contradictory, with VCCIO (MC voltage) which is common known as as QPI/VTT (for P55/X58 chipsets), or VTTCPU, CPUVTT to others, etc, as touched upon above in my earlier post Sinn0822 states it's max (per Intel is 1.05 +/- 3% = 1.08v), he then says you can go higher, much higher than the max yet with VCCSA he says you DO HVE TO ADHERE to Intel's max - why on one, but not the other, isn't a max a max?

To go a bit further, with some sets of DRAM particularly high freq sticks and large amounts of DRAM you are are in effect pushing the rest of the system to higher levels which may necessitate the use of a higher VCCSA to keep up with the rest of the system - same can apply to high OCs of the CPU is you want the entire system to be stable, simply raising the DRAM or MC voltage can help stabilize DRAM but the entire system needs to be sable else you still will get BSOD and other crashes

[rthaney]

Finaly Got the Motherboard Back from RMA

Finally got the Motherboard back from Biostar.

The Good News: They replaced my motherboard and the new motherboard can read the XMP profiles now.

The Bad News: Did not help. Same problem. Channel A is great and both sticks work on channel A. When any stick is put in channel B, either by itself or in Dual Channel Mode...Errors!

So what have we learned? Motherboard is working. Unless their configuration is such that all their motherboards have this same issue. The memory is working...both stick work at the specified speed in XMP.

Now I guess it is on to Intel. Don't know where else to go. Any thoughts?

[Tradesman]

It's possible you have a bad CPU? Can you try it in a different mobo, friend, co-worker etc? Other than that, I'm still thinking a problem with their BIOS, which would affect all the line of mobos running that BIOS or a flaw in the mobo itself

[rthaney]

It's the CPU

Luckily I have a backup computer with an i3 in it. Plugged it in and passed with flying colors. This is strange since the last time when I switched CPU's I got errors with the i3 as well. This time, no errors.

What else is strange is that, in MEM Test, it would still read the memory in channel B incorrectly. The speed was good but it would read the timings as 6-6-6-28, which could never pass without errors. But there were no errors so who knows?

Anyway looks like it is now a call to Intel. Hope they honor their warranty. Do you happen to have a link to their RMA? At first glance it looks like they hide their contact info to make it difficult to return stuff.

[Tradesman]

They have a live chat area where you can get started:

http://www.intel.com/support/feedbac...ssor#processor

[rthaney]

Awesome! Thanks.

[Tradesman]

Keep us updated

[rthaney]

Intel: since I have been running memory out of spec (1333MHz / 1.5v) They will not warranty the CPU. They will however do a one time RMA IF...IF I get some 1333MHz 1.5v DDR3 RAM and the system fails or has errors on that.

So now I have to go and find such RAM. I know it is cheap but I am still not convinced it is the CPU. Need to do some more tests.

Maybe just start upgrading to 3rd Gen Intel. Little depressed at this point but what are ya gonna do?