Hey man can you upload the CPU microcode file please?

You got me interested - just what exactly are we talking about?
 
How do we get it, how risky is it to install, and what benefit is this likely to bring me?
 
Assume I am am stupid, ignorant, and catch on real slow. 

Hmm, I guess he means the BIOS-internal CPU Microcode table.
 
This means, it's bound/integrated in the BIOS (not Linux, Windows or anything like that, its like a selfmade BIOS file).
 
You cannot "just install" it, if its like I think. You have to remove/replace the old CPU Microcode out of an actual BIOS.bin and add the new one with the right tools.
 
Its not that hard if you know what you do and the BIOS is "customizablefriendly" but if you don't - then you might trash your BIOS on the flashingprocess or make it unusable.
 
I'm not sure if the SR-2 has 3 really independent BIOS chips onboard or just one flashed BIOS Version with 2 Emergency-memoryareas, if its the first thing then you can flash all the way (until the eeprom breaks) and if its faulty, boot back with second/third BIOS and flash a origin BIOS again.. If its like the second option.. then you might have a problem. ;)
 
So the best way would be if EVGA releases a new BIOS with just the new integrated microcode.. that's the official, tested way.
 
Or someone does the work and uploads/releases/gives a hint/somewhat, somewhere... a selfbuild version - which might be not wanted there in the official EVGA forums (not sure). ;)

Or someone tells you what to do (thats your problem then, if it doesnt work)..
 
But I'm sure, if enough people ask nicely, then why shouldnt they release a new Binary with a working uncore multi unlock?

Thanks.. do we have an answer on the 3 BIOS chips?
 
If this only effects one of the 3 bios banks I am happy to give it a go - but if it could muck them all then that is a different story.
 
And this should work on x5650 and x5660?

This is the specific microcode version information (note the size includes padding):
 
MicroCode    Platform               Update        Date         Size
   ID          Type      CPU ID    Revision    YYYY/MM/DD     (Hex.)
---------    --------    ------    --------    ----------    --------
0036C213        03        06C2        13       2010/09/07    00002000
 
 
The three switches indeed toggle between independent BIOS images.
 
If you boot into Windows with the switch set to say selection (1),  you can change the switch to selection (3), and then run the update procedure within Windows (assuming you are using afuwin).  If you leave the switch in position (3), the update will be applicable next time you reboot.  With the update -- at least after changing the uncore -- the machine may reset several times (including a power cycle).
 
At 185x23, 19x uncore, I can lower the VTT from 1.35v to 1.325v fully stable.  20x uncore required 1.35v VTT.
 
p.s. If you are running Radeon 5970s, you may want to update BIOS slots (2) and (3), since at least my board still had the older firmware flashed to those slots which fails to POST.
  
 

may i be worthy

Thanks.. do we have an answer on the 3 BIOS chips?

If this only effects one of the 3 bios banks I am happy to give it a go - but if it could muck them all then that is a different story.

And this should work on x5650 and x5660?

Flashed and worked successfully on my x5650 rig, successfully booted at 16x uncore (mem is 2:8) just as a test, will see what we can do from here. 
 
Thanks guys, much appreciated.
 
Is this something that we could push our CPU clocks higher with, or just relieve stress on memory/vtt?

Update - working fine still. Uncore does work - the worst is that when changing uncore you will often get a FF failure to post, but as said above, it is no drama - just hit reset and things boot up fine, no need to reset cmos.
 
I have done one successful overclock validation, where I reduce my vtt from 1.375 to 1.35, but as I tweaked a few other things at the same time (going for a lower power summer overclock) I can't say for sure that this uncore was the sole benefit, but it is likely to have helped. But one thing that is proved to my satisfaction is that 19x uncore can be completely stable.
 

@may i be worthy
 
Glad to hear it works. Have fun!
 
I've still struggling @ 200 bclk, 3D Mark run slower than all default. IOH @1.425v

@Christefan:

Just reset your BIOS, then it should recognize all again. ;)

One thing to check -- did you also update the boot code when flashing?  afuwin defaults to flashing the main BIOS without touching the boot code.  The boot code was revised at some point, I think to (at least) address compatibility with new 24/48GB kits -- the extended boot block code appears to be involved in programming the IMC.  I believe this is what ends up mapping memory modules out when timing values are not correct.  This would be an area where more visibility into what happens at this stage of starting the system would be useful.
 
You shouldn't see any boot problems once the system has "settled" after the initial power cycle/reboots after flashing or picking a new uncore multiplier.
 

may i be worthy

My x5650 has some weird behavior - despite testing fine in linpack and memtest etc, it could not fold - second or third frame in any of my bigadv test work unit would hang or run for an hour.

At any of my tested stable overclocks, even at stock optimised defaults.

Flicked the switch back to official A49 and all was well, so I reflashed the modified bios again and now things are working again, with uncore unlocked.

Loving the multiple bios banks - although saved presets are forever broken on this machine - A49 fixed them on the x5660 rig, but this one they have always been flaky.

Just FYI.

If you ran afuwin64.bat, then that cover the boot block and other regions (checkout the switches it ultimately passes to afuwinx64.exe).
 
I've seen cases where folding exposes a problem where nothing else does.  I've only seen this hapen on 6701 work units though, and typically it indicates another notch of vcore or VTT is needed.  If you think this is VTT and are playing with the uncore multiplier, this may indicate that dropping the multiplier one setting should cover it.  It's unfortunate there isn't finer grained voltage control on the VTT.
 
I was running x5650 (not retail) @ 205x20, 1.34375v vcore, 1.35v VTT, 17x uncore (3485), DDR3-2050 8-9-8-24-2T rock solid with all work-loads until I switched to x5680s.
 

may i be worthy

I do not know - I inserted the rom file posted here in to the unzipped official A49 bios files, changed the name to match the official 1w555a49.rom, and flashed using the afuwin64.bat provided.

The problems started on the x5650 box after a day or two of perfect operation. And it was bizarre to have a problem that was only detectable by folding - memtest and linpack were passing fine.

Side note, would the perhaps work with the E56xx procs as well? E5620 in particular? :D

Keep an eye out today,
 
We are finalizing a Bios update to include more refined CPU microcode for better uncore control below 20X multi.
 

Sorry to dig this thread out of the ground but I'm trying to get my uncore lower than 20x as well on an E758 x58 sli board with a Xeon X5670 but it seems that even the latest bios that claims to support westmeres will not allow me to lower the uncore multiplier. I've tried 12-16x with 200blck but my uncore is stuck at 4000 (20x) vs. the 2400-3600 that it should be.
Can someone post the CPU microcode update bin and tool so I can do the bios mod as well or point me in the right direction?

Thanks!