0
I own a server and the motherboard is a supermicro H8QM3-2+.
Board specs located here: http://www.supermicro.com/Aplus/motherboard/Opteron8000/MCP55/H8QM3-2.cfm
Anyways, I gotten this board in 2009 and built my own server. Over the years business picked up and I kept adding onto it. I went from one cpu with 2 sticks of ram. This year I added about 3 cpu's and bought 16 sticks of ram each 4 gigs of ram again using 667 mhz and dual channel ecc. The rams are supported. I then bought more to max out all the ram slots. However, I noticed all 3 cpus with all full 8 slots of ram works fine. I then go to populate the 4th cpu ram slots. The first 2 slots works but what's odd is I would have ram in slots 1a, 1b and then 2b it will work fine. If I Do 1a,1b, 2a, 2b the server on boot won't even boot bios. I see a black screen. From what I can tell when i tried moving the ram around. All ram work since the first 2 slots work. I tried all the ram in those 2 slots to swap in and out to at least check if the ram is ok. All the ram works.
So, I am at the point where it's either the 4th cpu that has an issue or it's the ram slots that don't work.
I would like advice on what to do next? I am right now thinking to swap out the 4 cpu with another one. I have about 10 cpus of the same kind and time where I can swap them in. I personally don't think it's the CPU and I am guessing it's either the actual board or I might need to set something into bios.
There's a setting that says 8 channel something. I read up on that setting an found that it's normally needed to be enabled to support 8 gig sticks. I have this on and have tried it both while it's on and off and there's no difference in operations.
When I installed the CPU i was careful and I do know that no pins were bent or broken. I was careful when installing it. I would like to know if it's worth a try to replace the cpu? Or if there's something I need to set in bios to get this to work. In the manual it does say that when all ram slots are populated the ram speeds drops to 533 when using 667 or 833 mhz ram sticks. Again doesn't say anything about setting anything just says that assuming that it does this automatically.
I tried to manually set the rams speed to 533 mhz but after setting this the server board wouldn't boot at all. I had to take the battery out to clear bios and reset the settings. I got back in but the bios shows all ram sticks in cpu 1, 2,3 to be running at 200 mhz while cpu4 runs at 333 mhz and there's nothing to change the ram speed for just one set or for one cpu. I am now thinking to set all ram to 200 mhz and give that a try. i have really no clue what to do and would rather try something other than replacing the cpu since I have limited thermal paste. Right now I got 3 ram in that cpu 4 slot bank. They're in 1a,1b,2b that is it. If I put ram in 2a,3a,3b,4a,4b etc.. it just won't boot the system.
I would like to fill all the slots up to get me a total of 128 gigs of ram. Right now I think it's registered to 111 gigs of ram. That's what bios shows as of now.
Any tips or suggestions what I should do or try?
Ram sticks used is Nemix Ram 4gb ddr2-667 MHz pc2-5300 240-pin 1.8v 2Rx4 ECC registered server memory module
As requested here's a pic of the server in working condition:
2 pictures above shows the back where there's the IO connectors on the right side is cpu 4 where the issue is at. This config works. The server is currently running and working. If you look at the back there's 2 ram slots with ram in it for the cpu4. If you look left from the capacitors there's one ram slot next to the capacitors on the left side that one ram slot is part of cpu4 and is DIMM1b. On the left is cpu3 and that's it's ram bank. Each cpu has 8 slots. The cpu3 you will count 9 because the one closes to cpu4 is actual DIMM 1b for CPU4 it's part of CPU4's ram bank. If you look on the right you would count 7 ram slots. There should be 8 which means the one on the other side of the capacitors in the cpu3 ram bank the last slot next tot he capacitors closer to CPU 4 is part of CPU4 ram bank.
The problem is that when I add any ram to any other slot in that cpu4 ram bank it won't boot the system. I checked the ram and have put all the free ram I have in those 3 ram slots and it works. So, I know the ram I have remaining are working and good so it's not the ram. Now, it could be the CPU or the ram slots itself on the motherboard. Or I might need to set something up. In Bios right now with the pictures provide with that config it shows cpu1,2,3 running ram at 200 mhz and cpu4 running ram at 333 mhz. I have no clue if I have to manually set cpu4 ram to run at 200 mhz or it will do this automatically? All I know this config works and once I add more working ram into the other slots the server cannot boot anymore. Not even bios boots and there's no beeps or post. The fans run at it's highest RPM rate and just stays at that state until I power off and restart the server manually by pressing the power off button.
user520108
Posted 2018-02-05T20:48:55.463
Reputation: 1
Is there anyway you can take an overhead picture of the current configuration that wont boot? – Narzard – 2018-02-05T21:33:39.080
Narzard, I just edited the OP and put the pics of it. – user520108 – 2018-02-06T21:21:21.860
Everything seems correct. It is weird that they are underclocking so much though. When every ram slot is full it should downclock to 533mhz at the lowest. Maybe try setting all banks to clock at 533mhz (limitation of hardware) and then populating all the slots. The mobo supports non-interleaved memory so, ultimately where the sticks go doesnt matter too much in your case. Give that a shot and see what happens – Narzard – 2018-02-06T22:09:27.670
I actually did try that before. It was one of my first things to try when trying to figure out what might be wrong. I did this for all memory banks. When I set them to 533 mhz then on startup it would not boot at all even if I took out ram or set it back to same config as in the pictures provided. I had to take out the battery to clear bios. I then had to setup bios again and leave the clocks for ram to auto. That's where I am at right now. The option in the bios allows me to set clock from anywhere from 200 mhz to 533 mhz manually. – user520108 – 2018-02-06T22:53:49.030
I want to know if I should try and replace the CPU? I was told by others that the pins in the chip might be bent or the cpu might be defective. However, I don't want to do this first since i have a limited supply of artic thermal paste. I would think if the cpu had an issue detecting the ram dimms I would think it still would boot the system just won't recognize the ram. – user520108 – 2018-02-06T23:11:38.193