EVGA

The bloody RAM support on SR-2 :-)

Author
Bepzinky
iCX Member
  • Total Posts : 281
  • Reward points : 0
  • Joined: 2016/04/30 13:15:31
  • Status: offline
  • Ribbons : 2
2017/04/15 04:12:04 (permalink)
I'm having issues getting the full 24GB of RAM to work on my SR-2 with dual X5670's and it looks to be a common thing of SR-2.
 
So far I've tried
a) 6x4GB ATP DDR3 1333 ECC server kit
b) 6x4GB Kingston HyperX Predator 2133 (have these now)
c) Inserting the memory in either red or black slots.
 
ELEET shows full 24GB installed, AIDA64 sees all 6 modules and when benchmarking it correctly says "Hexa DDR3" but I only get 20GB in BIOS, Windows (10 x64) and in CPU-Z.
I don't think any of the controllers/ DIMM slots are busted since the software does recognize all 6 sticks and doesn't have any crashing / BSOD-ing issues.
 
I've bumped the voltages a bit (now 3.6GHz CPU / 1500MHz RAM) but I still see 20GB. When I disable CPU_0 I get 12GB in CPU-Z, 24GB in ELEET and 8GB in BIOS/Windows 
Also - yes, I tried different dividers/voltages too.
 
 
 
post edited by Bepzinky - 2017/04/15 05:41:39
#1

7 Replies Related Threads

    Bepzinky
    iCX Member
    • Total Posts : 281
    • Reward points : 0
    • Joined: 2016/04/30 13:15:31
    • Status: offline
    • Ribbons : 2
    Re: The bloody RAM support on SR-2 :-) 2017/05/10 06:33:56 (permalink)
    I've noticed that the working stick is a bit cooler to the touch than the others. If I try to boot with the RAM only inserted in the 2 middle slots at the left CPU (other CPU is then disabled, of course) I get some kind of bootloop error at start.
    I've also tried with an X5550 Xeon and got the same result.
     
    Do you think replacing the CPU socket could fix the issue? I know some BGA wizards who can solder a whole new socket.
    post edited by Bepzinky - 2017/05/10 06:47:22
    #2
    loveha
    Omnipotent Enthusiast
    • Total Posts : 10497
    • Reward points : 0
    • Joined: 2009/09/26 01:51:22
    • Location: Beulah Michigan
    • Status: offline
    • Ribbons : 65
    Re: The bloody RAM support on SR-2 :-) 2017/05/11 00:10:25 (permalink)
    How much CPU VTT have you added in the BIOS? Increasing that to +300 to +350 can sometimes fix it.
    I doubt it's the socket unless you have pins bent that can not be straightened.
    Unfortunately, this is one of those issues that was frequent with this chipset that's a pain to solve and is sometimes unsolvable.

    Case - Phanteks Enthoo Evolv X
    Mobo - EVGA X570 FTW
    CPU - AMD Ryzen 9 5900X
    RAM - 32GB Trident Z Royal 3600MHz 14-14-14-34
    GPU - EVGA RTX 3080 Ti FTW3 Ultra Gaming
    PSU - Seasonic Prime TX-1000
    Heatkiller IV CPU Waterblock / Heatkiller V GPU Waterblock
    Heatkiller V EBC Active Backplate / Two Heatkiller Rad 360 L
    Six Noctua NF-A12x25 / Heatkiller D5 Pump
    #3
    Bepzinky
    iCX Member
    • Total Posts : 281
    • Reward points : 0
    • Joined: 2016/04/30 13:15:31
    • Status: offline
    • Ribbons : 2
    Re: The bloody RAM support on SR-2 :-) 2017/05/13 15:17:09 (permalink)
    I'm at +-1.35-1.36 volts. Pushing VTT volts and high BCLK with 1.375+ IOH causes random Alzheimer - it tends to forget even more RAM at boot.
    Strange that I can see every installed module in any diagnostics software but the board makes no use of it. I'd guess it's normal for this board then as both sockets look absolutely mint and I've even cleaned the pins one by one (yes, really ;) ) and tried other CPU's.
    I'm now running 4x4GB G.Skill Ripjaws X but again it's the same story as with the 2 other RAM kits I've already tried.
     
    By the way, which of the VRM's is labelled CPU0/VRM0 in software? My VRM0 runs very hot under load and I might have a thermal pad issue but I don't know which one is it.
    I've seen that EVGA SR2 labelling guide, but man it's still confusing, is it the VRM with the small heatsink above the right hand CPU? That heastsink feels cool but the PCB above is hot to touch and after stopping the stress test the temp drops by 30*C in a second.
    post edited by Bepzinky - 2017/05/13 15:38:22
    #4
    Bepzinky
    iCX Member
    • Total Posts : 281
    • Reward points : 0
    • Joined: 2016/04/30 13:15:31
    • Status: offline
    • Ribbons : 2
    Re: The bloody RAM support on SR-2 :-) 2017/05/27 14:40:13 (permalink)
    Where is the CPU0 PWM temp sensor located? I've changed the thermal pads on the smaller heatsink but I still see 90*C on PWM0 under OC'ed P95 load.
    #5
    d wray
    New Member
    • Total Posts : 85
    • Reward points : 0
    • Joined: 2014/06/03 00:23:23
    • Status: offline
    • Ribbons : 0
    Re: The bloody RAM support on SR-2 :-) 2017/05/29 17:40:52 (permalink)
    I have 12 8GB sticks of Hynix HMT31GR7CFR4A 1333mhz ECC that I got off Craigslist for $5 each that used to boot all the time every time into Windows Server 2016 without any problems. I'm running into issues now trying to get 7 RX 480s to work. 4 cards work without a problem but once I start adding more, it drops cards and starts dropping ram too. Will probably just run 4 and put the other 3 in another rig if I can't get it by the end of the week. But check out those sticks I know they work.
    #6
    Bepzinky
    iCX Member
    • Total Posts : 281
    • Reward points : 0
    • Joined: 2016/04/30 13:15:31
    • Status: offline
    • Ribbons : 2
    Re: The bloody RAM support on SR-2 :-) 2017/06/02 10:30:38 (permalink)
    How could one explain this situation?
    http://i.imgur.com/J1zSQZW.png
    EVGA engineers did some new maths, 2x4=4, Windows also shows 4GB total. The amount of RAM inserted is correct - 2x4GB, 1 module per CPU.
    post edited by Bepzinky - 2017/06/02 10:33:37
    #7
    hzdriver
    New Member
    • Total Posts : 10
    • Reward points : 0
    • Joined: 2011/02/24 22:04:52
    • Status: offline
    • Ribbons : 0
    Re: The bloody RAM support on SR-2 :-) 2017/06/18 01:21:54 (permalink)
    last Tip, check the vcore for Ram, must be 1,65 Volt, Default ist lower!!!

     
      
      
     
    #8
    Jump to:
  • Back to Mobile