how often do you succeed in fully populating DIMMs on the 1st shot?

How often have you successfully booted a system on the 1st try after fully populating all DIMMs?

  • Never. Always requires adding one DIMM at a time

    Votes: 3 13.6%
  • Almost never, but it has happened.

    Votes: 2 9.1%
  • Almost all the time. Only occasionally do I need to re-seat the DIMM and retry

    Votes: 8 36.4%
  • Always, I am master of DIMMs.

    Votes: 9 40.9%

  • Total voters
    22
Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

BLinux

cat lover server enthusiast
Jul 7, 2016
2,669
1,081
113
artofserver.com
I spent a little time adding RAM to a X9DRD-7LN4F board today with 16x16GB DIMMs. Well, that "little" time took much longer than I hoped. I should know better by now, but I still try. For many years now, I have *never* been able to successfully populate all DIMM slots on a motherboard in one shot and successfully boot the machine. I can't recall the last time I was able to successfully throw a bunch of DIMMs into a system and hit the power button and have everything just work. Yet, I always still try; only to be disappointed. In the end, it always turns into a tedious process of adding 1 DIMM at a time, power on, does BIOS see the DIMM? if yes, power off and add next DIMM. If not, remove and re-seat or try another DIMM. For the 16 DIMMs tonight, I had to power on and off the system over 20+ times until I finally got it to recognize all 256GB. Additionally, there are times I get systems that stubbornly get stuck with the dreaded "System Initializing .... (B6/B7/BA/BF)" codes. This usually results in taking all DIMMS out, doing a CMOS reset, and starting over 1 DIMM at a time. In the end, it eventually all works most of the time; it's not often that I conclude i have bad DIMMs.

So, am I the only one who keeps experiencing this? How often have you successfully booted a new build on the 1st try after filling up all the DIMMs in one shot?

I remember working on old PCs and servers back in the days when IBM made hard drives, Compaq and Gateway computers were still around, and my beloved DEC Alpha was still a "thing". I don't ever recall having to perform such a tedious ritual for RAM upgrades. Are DDR3/DDR4 DIMM slots just more delicate?
 

EffrafaxOfWug

Radioactive Member
Feb 12, 2015
1,394
511
113
Whenever I build a server, first boot will only ever have just the motherboard, one CPU and one stick of RAM plugged in, simply so that if there's a short of a broken PSU or something like that the magic smoke only escapes from the bare minimum of components.

Can't say I've ever experienced a palaver such as you describe though. The closest I've had was setting up a hugely expensive IBM xseries server with an exotic RAM arrangement in the form of DIMMs on memory riser cards each with its own memory controller (this was back when the xeons were still using FSB from the northbridge, so I think each riser card had its own memory controller); it was a quad CPU with a mind-blowingly huge 16GB of RAM destined to have vmare on it. That took about three or four pulls of the memory cards before it would POST correctly, gave us no end of problems, and the performance on virtualisation workloads was piss-poor compared to the opterons we already had as our testbed (we weren't allowed to use AMD for production kit but bought some for our testbed). Think it lasted less than a year in service.
 

Evan

Well-Known Member
Jan 6, 2016
3,346
598
113
Did over 50 servers recently with 16 or 24 dimms each, 100%, I really never ever have an issue... fingers get sore though !!
I didn’t know anybody actually had issues with this.
 

BLinux

cat lover server enthusiast
Jul 7, 2016
2,669
1,081
113
artofserver.com
Did over 50 servers recently with 16 or 24 dimms each, 100%, I really never ever have an issue... fingers get sore though !!
I didn’t know anybody actually had issues with this.
OMG, I am humbled and must bow to your greatness!
 

T_Minus

Build. Break. Fix. Repeat
Feb 15, 2015
7,625
2,043
113
I used to have issues it wasn't about being full or not... it was my installation technique and adding more RAM increased my likelihood of bad install.


  1. Clean RAM contacts with Alcohol Swab if RAM is used. If new I may just glance it over for wear/marks or clean to be safe.

  2. I now make sure to push in enough to align both ends, then push firmly on both ends, then once it clicks I put my finger in the middle and push down firmly and slightly move it back and forth to properly seat.

Since going to this technique I've yet to have a RAM installation issues and I used to have to deal with it like you too.

I've gotten 40+ DIMMS in a USPS box with 0 padding, none were failed.. RAM is rather physically 'sturdy'... so don't be afraid to push it in there and most importantly move it back and forth to seat 100%.


I used to start with 1 stick, test, add more, but that's a huge hassle and since doing the above install method I"ve reversed my 'testing'.
I'll start by loading the board 100% full of RAM then if that works awesome keep going, if not I'll go back to 1 stick. This has saved countless hours.
 
  • Like
Reactions: BLinux

Patrick

Administrator
Staff member
Dec 21, 2010
12,511
5,792
113
I build a lot of systems so I cannot say "always" but I have been 100% on the last 300+ DIMMs installed.
 
  • Like
Reactions: William

BLinux

cat lover server enthusiast
Jul 7, 2016
2,669
1,081
113
artofserver.com
I used to have issues it wasn't about being full or not... it was my installation technique and adding more RAM increased my likelihood of bad install.
  1. Clean RAM contacts with Alcohol Swab if RAM is used. If new I may just glance it over for wear/marks or clean to be safe.

  2. I now make sure to push in enough to align both ends, then push firmly on both ends, then once it clicks I put my finger in the middle and push down firmly and slightly move it back and forth to properly seat.
Since going to this technique I've yet to have a RAM installation issues and I used to have to deal with it like you too.

I've gotten 40+ DIMMS in a USPS box with 0 padding, none were failed.. RAM is rather physically 'sturdy'... so don't be afraid to push it in there and most importantly move it back and forth to seat 100%.
Thanks for sharing your technique. I've got another box of 16 DIMMs on the way and I will give your technique a try. My insertion technique (LOL) is similar to yours, but what do you mean by "slightly move it back and forth to properly seat?" Say, if the DIMM is horizontally oriented in your view, is "back and forth" up and down? or left and right?

I haven't bothered to clean the contacts on used DIMMs, so that's a new concept to me.

I used to start with 1 stick, test, add more, but that's a huge hassle and since doing the above install method I"ve reversed my 'testing'.
I'll start by loading the board 100% full of RAM then if that works awesome keep going, if not I'll go back to 1 stick. This has saved countless hours.
Yes, it's a huge time hassle to go 1 by 1 and I've had to resort to that every time in recent times. But some how, I still try to see if everything would just work on the 1st go, so I insert all the DIMMs and hit the power button, and sure enough, I always get the stuck at B7/B9/BA code during "System Initializing..." Sometimes, the system will get stuck in that state too... even when I reduce it down to just a single DIMM still gets stuck at B7; at which point I do a CMOS clear and it starts to recognize the DIMMs again.

I build a lot of systems so I cannot say "always" but I have been 100% on the last 300+ DIMMs installed.
Hmm... i guess I must have bad technique or something?

I'm glad my experience doesn't seem to be the norm; it means there's hope that maybe there's something I can change that will improve the experience.
 

dswartz

Active Member
Jul 14, 2011
610
79
28
I have a 2-cpu xeon SM motherboard (don't remember the model offhand). 8 slots, 4 blue, 4 black. You're supposed to populate the blue in a channel before the black. Mobo AND ram were brand-new. Could NOT get the 4th blue slot to work. At all. POST would fail with some cheesy error message. Looking into the slot with flashlight and magnifier, nothing obviously wrong. I assume that slot was DOA. It bothered my OCD, but I finally settled for populating blue+black and blue+black and everything worked fine. In theory, I'm losing some theoretical memory bandwidth, but meh. I really wasn't interested in RMA'ing the board... Built a new Ryzen workstation with 4 8GB DDR4 sticks. It seemed to POST okay, but windows 10 kept insisting that 16GB was 'reserved for hardware' or some such. I did notice in the BIOS that it reported 32GB but in single channel mode. Pulled all 4 sticks and reinserted them (rocking them back and forth as recommended.) POSTed 32GB in dual channel mode. Boot windows 10. Voila, 32GB of usable RAM :)
 

Evan

Well-Known Member
Jan 6, 2016
3,346
598
113
I wonder if it’s a SM thing or rather a non HPE,Dell,Cisco,Lenovo etc.
Certainly on the big vendors systems they are strong, I position the dimm in slot, push down one end keeping the pressure on then push down the other end. Really rather firm push. It’s a solid click as it’s seats and the retainers come to position themselves. You don’t need to touch the retainers, they should be already hard in place due to how hard you pushed down to insert the dimm.

But I know those big vendors there is no board flex or anything since they are very solid systems, good design, maybe better than SM,Tyan,Asrock-rack, etc ?
 

BLinux

cat lover server enthusiast
Jul 7, 2016
2,669
1,081
113
artofserver.com
I wonder if it’s a SM thing or rather a non HPE,Dell,Cisco,Lenovo etc.
Certainly on the big vendors systems they are strong, I position the dimm in slot, push down one end keeping the pressure on then push down the other end. Really rather firm push. It’s a solid click as it’s seats and the retainers come to position themselves. You don’t need to touch the retainers, they should be already hard in place due to how hard you pushed down to insert the dimm.

But I know those big vendors there is no board flex or anything since they are very solid systems, good design, maybe better than SM,Tyan,Asrock-rack, etc ?
there might be something to that point. i never did seem to have these kind of issues back when i mostly dealt with HP/Compaq/Dell/Sun/SGI/DEC servers. recently, it's mostly been used Supermicro and some ASRock (workstation boards).
 

Terry Kennedy

Well-Known Member
Jun 25, 2015
1,140
594
113
New York City
www.glaver.org
there might be something to that point. i never did seem to have these kind of issues back when i mostly dealt with HP/Compaq/Dell/Sun/SGI/DEC servers. recently, it's mostly been used Supermicro and some ASRock (workstation boards).
The Optiplex 755/960/9010/9020 have been problematic since day 1. I've never put all 4 (wow, a whole 4 slots) modules in and had things work. Plus (being a desktop) the diagnostics are minimal at best - usually the failure mode is "nothing happens". On a half dozen of these systems ordered new from Dell, 50% of them needed memory replacements during the warranty period.

Dell servers (Rx00/x10, mostly) have been well-behaved, even with memory from the junk drawer, except for an R410 that didn't see any memory, no matter what I put in it. I wrote it off as a dud motherboard.

As far as home-built stuff, the Supermicro X8DTH-iF, 56x0 CPUs and Hynix memory have been solid when all were new. Used ones have been more of a pain, mostly due to bad RAM. All parts were "supposed" to be HMT31GR7AFR4C-H9 but there were some other ones in the box - a few 16GB but slower, a few 4GB, etc. And a few out-and-out duds, and a few that threw ECC errors during testing.

Speaking of testing, Memtest86+/MemTest86 aren't as useful as they could be. The non-UEFI versions don't know about modern chipsets, and SPD reporting is incomplete (reporting SPD data for half of the installed modules or reporting all 0's, etc.) and I've never gotten the UEFI versions to work, even on UEFI-only motherboards.

But the absolutely worst bloodbath I ever had was trying to assemble a system where a friend selected and purchased all of the parts new and asked me to help assemble it (he picked the parts himself). It was an AMD B350 system with an A10-9700 CPU. I complained repeatedly and loudly to him about "toy parts" due to things like the memory "spec" sheet going on for 2 pages about its fancy chaser LEDs but not saying anything about being single, double, or quad rank. He replaced the motherboard twice (once with an identical one, then with a different brand), the CPU once, and the memory 3 times (the 3rd time I made him purchase memory that was definitely single-rank). The 2nd motherboard vendor's "compatibility list" had many obvious errors in its supported memory table. Anyway, after getting all of the parts swapped and after well over 2 months, I got the thing to post - but it takes over a minute for anything to happen on the video output - until then it is just noise. The motherboard manufacturer claimed "oh, all B350 boards do that". Certainly all of their boards do - their forum is overrun with complaints, and despite releasing "beta" BIOS updates for a year or so that claim to fix it, they haven't. Of course, the icing on this cake was him telling me "PC Partpicker says that all of these are compatible" when they obviously weren't. I don't know if they're just optimistic or their Magic 8-ball is broken, but they definitely didn't know what they were talking about in this case.
 

BLinux

cat lover server enthusiast
Jul 7, 2016
2,669
1,081
113
artofserver.com
I used to have issues it wasn't about being full or not... it was my installation technique and adding more RAM increased my likelihood of bad install.
Wow! Today I tried your advice; cleaned the DIMM contacts, pressed down on the middle of the DIMMs and wiggled them a bit. Boom! all 16x DIMMs worked on the 1st shot! Haven't had that experience in a long time... I've been assembling computers for over 20 yrs (not continuously), and I still had something to learn! Thanks for your advice man...
 
  • Like
Reactions: T_Minus

acquacow

Well-Known Member
Feb 15, 2017
784
439
63
42
I just added 8 new DIMMS to another supermicro X9 board and it definitely booted, but had errors after a few hours once my VMs all got spun up.

Memtest86 showed errors, so I began the process of isolating the buggy dimm.

Turns out it just had to be re-seated a few times and now it works fine.

I have this issue occasionally with brand new DIMMS and I have no idea why. I always have to seat them a few times before they work reliably.
 

T_Minus

Build. Break. Fix. Repeat
Feb 15, 2015
7,625
2,043
113
Wow! Today I tried your advice; cleaned the DIMM contacts, pressed down on the middle of the DIMMs and wiggled them a bit. Boom! all 16x DIMMs worked on the 1st shot! Haven't had that experience in a long time... I've been assembling computers for over 20 yrs (not continuously), and I still had something to learn! Thanks for your advice man...
Wooohoo!

That's awesome :)