BIOS A50 Updates:

• Allows Uncore Adjustability Below 20X multiplier

  ---

  
  CD/DVD Installation - Recommended installation method
• Download the correct .zip file below for your motherboard:

              W555_A50.zip

• First go into your BIOS and load defaults, save and exit
• Go into the folder where you unzipped this package and run the appropriate batch file.
  • For Windows XP, Vista or 7 32Bit: AFUWIN32.BAT
  • For Windows XP, Vista or 7 64Bit: AFUWIN64.BAT
  • If in DOS Mode: AFUDOS.BAT
• Press Y in CAPITAL LETTERS to start the flash process
• DO NOT TURN OFF POWER OR RESET DURING THE FLASH PROCESS
• After the flashing is finished, shut down the computer. Then power on again and go into BIOS, Load Defaults, Save and Exit.

Warning:
**Do not restart or interrupt your PC while the flash is ongoing
**Do not Flash BIOS on overclocked settings
**It is always recommended to power-down after flash and Clear CMOS

juanP

Is that the only fix for this update?

Juan,
 
This was the correction for this update.
 
Is there something else you need addressed?

I'd like to see better memory performance out of a bios then what we currently seem to be able to achieve with #49.

By using MMTool to compare with A49 the following microcode is revised, while 1B module seems not changing a bit.
 
20652 Rev 09 -> Rev 0C
 20656 Rev 05 -> Rev 07
 206C2 Rev 0C -> Rev 13.

Thanks for the new bios - I didn't have trouble running lower uncore on all previous bioses though?
 
Anyway, here's a few things I'd like to see fixed/improved:
 
1. Pushing the Clear CMOS button should not clear CMOS Date/Time, that's annoying as hell
2. Profiles. Can we PLEASE get the ability to name them? I really can't remember which profile was which half of the time.
3. I still get a no post from the board after changing an OC too many times. Even on stable overclocks that I have gained by using the E-Leet tool in windows (if I put those values in bios it won't post and I have to clear CMOS). Happens if I try to change memory divider and BCLK at the same time, or using a BCLK of 170+ in general, and in a bunch of different scenarios as well
 
Thanks :)

Bios 50 has dropped my G19 USb keyboard several times now0haven't had that problem with bios 47 or 49.  Also the memory speed and detection is still a problem, I frequently have to reboot several times to get all of the memory recognized.

The first problem that I think many people are having is that not all the memory is being consistently recognized-I frequently have to warm reboot so that all of the memory is discovered in post, typically two to four times. The other issue would be that I am personally running G Skill T&T 3x4 ddr3-2000.  Their testing photo is 8,8,8,20,4,59,8,6,30,0 at 2T and  I'm at only 1500 not 2000 with timings of 9,9,9,24,74 at 2T--it doesn't seem like we're really getting the memory performance out of the motherboard that I expected based on the 5520 chip supposedly having better memory performance than the x58 chipset.  Just my impressions, if they're correct an improvement in the bios would be nice.

How to find Bios Version??

This is what my splash screen says:
Bios Date: 05/05/10 09:15:13 Ver: 08.00.16

Just seeing if I need to update Bios.
If so, best version?

I don't think using registered DIMMs is the real fix on this.
 
I think the real fix is:
1. Provide full manual control over all memory settings.
2. Provide more visibility into what errors are encountered (and control whether they can be ignored) when the memory controller is setup (extended boot block code).
3. Provide more visibility into all the memory settings configured/derived in the memory controller init path (could show this in the BIOS setup screen).
 
I'm using 6x4GB gskill modules, and my attempts to disable NUMA have still failed.  When the system does boot, the same 4GB memory slot gets "taken out", regardless of what DIMM is installed there as well as using a variety of timing values.  Other reports indicate gskill modules are more suspectible to being "taken out" (*).  I saw this with 2GB gskill modules in my early SR2 testing as well.  For reference, the module being "taken out" on my system is Location bus 254 (0xFE), device 4 (0x04), function 1 (0x01) register 0x48:
(*) See this post for more info: http://forums.evga.com/tm.aspx?m=646413
 
I've started digging into the memory configuration data persisted to NVRAM, and will report on that later.  It's also clear the way memory is configured varies significantly between NUMA and non-NUMA.  For the NUMA case, memory access times appear slower than expected, even when testing memory access on memory allocated on a specific NUMA node and accessed locally to that NUMA node (more on this later).  I also need to check how the gskill SPD settings get read -- cpu-z and eleet read different values for the XMP profile.
 
This thread from shamino indicated some early testing of a gskill 12x4GB gskill kit (variation of the same 4GB modules I'm using):
http://forums.evga.com/tm.aspx?m=432275
 
And then a BIOS update (A47) followed.  It would be nice to know more detail on the changes in that BIOS update -- what exactly was changed?  Was NUMA disabled tested with these changes?   Was testing done with only 24GB (3 modules per CPU) as well?  If I roll-back to A46, I've never been able to boot successfully with NUMA disabled -- later BIOSes booting w/out NUMA works sometimes, albeit with the one module "taken out".
 
EVGA Classified SR-2 - BIOS A47 Released Friday, June 25, 2010 3:34 PM (permalink)
BIOS A47 Updates:
•Improves Memory OC
•Adds support for XMP on select 48GB kit
 
 

thefreeaccount

The first problem that I think many people are having is that not all the memory is being consistently recognized

This is a common issue with large amounts of memory in any system. Evga has already provided a fix - you can use registered memory. If you went unbufffered to save $40 after you bought a $600 M/B, it's not really Evga's fault IMO.

Why nobody mentions the fact that with s1 setting in acpi in windows it shows that the mobo can go sleep mode but it is just a fake sleep mode which turns off only drives and screen?
SO if you want to hybernate you have to set s1,if you want to sleep you have to manually set s3 in the acpi setting in the bios.On my netbook i can choose whether to hibernate or sleep without messing with naything. And then the ability to sleep is simply not there beyond certain bclock. I think this is quite urgent.

Mosie, I think you can work-around the missing/hidden C1E Bios option, and possibly the thermal management option as well -- as well as other obscure processor features that are ultimately controlled by manipulating model specific registers (MSRs).
 
It seems like most tools of this nature, like cpu-tweaker, were only designed for single socket systems.
 
There's a pretty simple work-around, which is to run the tool twice -- once while pinned to CPU0, and again while pinned to CPU1.  This will ensure the MSR edits happen on the proper CPU.   I verified this works disabling C1E on each processor separately.
 
Basically, follow the same steps as outlined here, though run cpu-tweaker or other alternate tool instead of IntelBurnTest:
http://forums.evga.com/tm.aspx?m=651731
 
Note this approach doesn't overcome all multi-socket limitations of these tools, but it will work for some.
 
Personally, I've seen no need for messing with C1E.  I have all the power management features and C states enabled on my system (turbo mode=off) with no stability issues.
 

mosie 

PPLLEEEAASSSEEE 

ADD  C1E OPTION  and termal temp shutdown OPTION

  ITS IMPORTANT AND URGENT

and strange litle problem:   with bios 50 I can t modifie the real IOH PLL its fixed at 1.76V and whene aplic 1.85V or more its not work.
IOH PLL always at 1.76V

Hello Mosie,
 
let me understand the graph you posted:

You mean, if C1E is enabled, then the CPU switches very fast between different multipliers (so the frequency changes) and this causes instabilities in your case (at bootup)?
 
If I remember correctly (didnt flash Bios 50 yet, sorry) then to see the C1E Function in the BIOS you have to enable Intel Speed Step or some other switch (I type this out of my memory, could be wrong). Then the option appears.

Did you check this? In the advanced CPU functions setting menu the C1E appeared only after activating some function and I guess it was ISS..?

Any possibility of an updated BIOS that allows C1E to be disabled?

erokhutch

I flashed my BIOS and now my computer will not boot into Win 7. What can I do? Should I remove the battery?

Did you remember to change your SATA settings back to AHCI or RAID depending on what your settings were before?

just updated bios to version 50 on a new build (nothing installed except windows 7 and kaspersky internet security) and having trouble getting into windows 7 after the update too.  Safe mode works, starting windows 7 normally crashes at the login screen.
 
Note that I didn't get into bios after the flash to load optimal defaults, maybe that's why (didn't know the key to hit while the Classified SR2 image was displayed during the bios load - arrgh rtfm casper).  Gonna use the switch to load up.  Would like to know the anser to erokhutch's question - after you've loaded into windows while the bios switch was at a different position, can you switch it back to the first position and then downgrade the bios for that position?