Forums before death by AOL, social media and spammers... "We can't have nice things"
|    alt.os.linux    |    Getting to be as bloated as Windows!    |    107,822 messages    |
[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]
|    Message 107,380 of 107,822    |
|    Paul to bad sector    |
|    Re: Asus x870e Proart Creator motherboar    |
|    12 Jul 25 19:14:44    |
   
   From: nospam@needed.invalid   
      
   On Sat, 7/12/2025 11:02 AM, bad sector wrote:   
   > On 7/12/25 3:04 AM, Paul wrote:   
   >> On Sat, 7/12/2025 12:45 AM, bad sector wrote:   
   >>> Minutes after the last posts I went into BIOS to take out the CPU native   
   GPU by setting Integrated Graphics to Disabled and leaving the PCIe GPU as   
   Primary. On reboot (NO beeps on the these new boards) got a steady yellow   
   Q-Led mening RAM problems.    
   Reseated one ddr5 card and went with only that one. It worked, reseated the   
   other one too. The yellow LED now lasts about 1:30 and then the white LED   
   (GPU) takes over. This one doesn't leave, not after reseating the GPU, not   
   after removing it and    
   plugging the HDMI into the onboard slot instead of the GPU one, not after   
   changing HDMI cables and even monitor. CMOS was cleared at every step. The   
   green (boot) LED never lights up, with no disk plugged in BIOS is never   
   entered. Starting to have enough    
   of this outfit (the GPU is also Asus)!   
   >>>   
   >>   
   >> You have to be very careful, with some of the stuff you mentioned in   
   passing.   
   >>   
   >> The CMOS can only be cleared, with mains disconnected.   
   >>   
   >> When reseating materials, it is a good idea to have mains disconnected   
   >> for that too.   
   >>   
   >> You can try taking it back to the single-stick-NON-ECC and see if   
   >> you can bring it up that way.   
   >>   
   >> 1) Power off.   
   >> 2) Remove plugin GPU.   
   >> 3) Put ECC RAM in its antistatic package.   
   >> 4) Put the "teaser RAM" that brought it up one time before,   
   >> into the far slot on one of the memory channels.   
   >> 5) Plug HDMI monitor cable into motherboard.   
   >> 6) Power up, wait patiently for recovery :-)   
   >> Remember the annoying habit of "BIOS, taking longer after a change".   
   >   
   > Stripped it naked..   
   >   
   > Seated single 8gb non-ecc ddr5 in slot 2 [A2].. green   
   >   
   > BIOS-1401 to defaults, put everything back in... green   
   >   
   > Swapped ecc-B into slot 2 [A2].. green   
   >   
   > Swapped ecc-A into slot 2 [A2].. green   
   >   
   > Seated ecc-B into slot 4 [B2].. WHITE led (VGA fail)   
   >   
   > Disabled ecc in BIOS and repeated the above.. green   
   >   
   > Swapped non-ecc 8gb ddr5 into slot 4 [B2].. green   
   >   
   > BIOS-1401, enabled ecc.. green   
   >   
   > Returned ecc-B into slot 4 [B2].. YELLOW led (DRAM fail)   
   >   
   > Removed ecc-B from slot 4 [B2].. green   
   >   
   > BIOS-1401, disabled integrated graphics.. green   
   >   
   > Returned ecc-B into slot 4[B2].. YELLOW (DRAM fail)   
   >   
   > Swapped ecc's in the 2 slots..   
   >   
   > ...... either ecc card works in slot 2 [A2]   
   > ...... slot 4 [B2] fails with either ecc card in it   
   >   
   > taking a break, million other chores awaiting.   
   >   
      
   The CPU is Land Grid Array. (Very sharp metallic spring bites   
   into gold plated land.) Normally, this is reproducible any time   
   the CPU is removed and replaced. And the arm that applies pressure   
   is locked down. The pin sharp point, always goes precisely back   
   into the bite mark.   
      
   Is the CPU arm locked down right now ? You could have some pretty substantial   
   cooler on it, so sometimes inspection is difficult.   
      
   Land Grid Array works up to (so far), around 9000 contacts,   
   but I have no idea how the "pressing-force" is equally distributed   
   so all the lands have good spring contact. The hobby CPUs don't   
   have nearly the same LGA counts.   
      
   There have been CPUs that burned in sockets, because of poor contact.   
   There were two brands of sockets, Lotes (good) and Foxconn.   
   The Foxconn issue was not repeated on the next generation.   
      
   The generation I use here (AM4) is PGA or pin grid array. And the   
   socket is ZIF (Zero Insertion Force). Moving the lever on those,   
   provides side force making the "contacts touch the pins". When you   
   remove a heatsink on those, you can pull the CPU right out of the   
   ("locked") PGA socket :-) That does not hurt anything. There have been   
   cases where PGA pins were crushed. A metallic ball point pen refill, could   
   be used to (carefully) bend a PGA pin upright again. And that kind of   
   surgery is not always successful (but, ya gotta try). When you make a mistake   
   and that happens, you can usually tell the patient is not going to survive   
   the attempt to straighten a pin.   
      
   *******   
      
   DIMM socket pins, almost never foul...   
      
   It would take a determined individual, bent on destruction,   
   to do actual damage to a DIMM socket. The pins are very very stiff.   
   It takes mallet-level attempts to damage it, to mess it up. I   
   highly doubt there is any reason whatsoever, to be "cleaning"   
   the DIMM socket.   
      
   Similarly, the "pink eraser crowd", I'm not with them. Gold on   
   electronics, can be applied 10u or 50u. The latter thickness   
   is used on telecomm contacts. The former number is used   
   on computer industry equipment. Applying abrasives to   
   10u gold is a "mistake", doing more damage than good. There   
   are already pinholes in the finish, due to the thinness of   
   the plating.   
      
   There is so much wiping-force on DIMM insertion, there can't possibly   
   be a contact problem. Only a corrosive liquid (Kings Reagent would do),   
   could do a job on the contacts.   
      
   I just don't see a reason to be fooling around with the socket.   
   I'd rather reseat the CPU, then bash on the RAM (the RAM interface   
   is on the CPU package). There's no question something bad is going on,   
    but it probably isn't the fault of the DIMM contact there. If you actually   
   managed to bend a DIMM contact (!), the DIMM would no longer seat and lock.   
      
   I've visually inspected DIMM sockets with a magnifier, and have   
   not had a reason to be suspicious of what I'm seeing.   
      
    Paul   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)   
|
[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]
(c) 1994, bbs@darkrealms.ca