SM H8DGU-F (2P) compatibility with IL 6272 ES ZS212045TGG44

B

bowlinra

Limp Gawd
Joined
Apr 30, 2012
Messages
195
I have a Supermicro H8DGU-F 2P motherboard that has run sets of 6172 and 6176SE in the past with no problems. So I decide to make things more interesting by getting a pair for 6272 ES (ZS212045TGG44) Stepping B0.

I get it all put together and it won't post.. I figured it probably requires a BIOS update and I put the old chip back in, flash it from R 2.0b to R 3.0b. Reinserted the 6272 ES and it still won't post.. Fans are spinning and the Green LED is blinking, but not sounds, no video.

What ideas do you have for me?
 
Try baby steps. Put a single CPU and ram in. If that does not post, swap out CPUs. Next cycle through different ram.
 
2.0b should've sufficed to run them... something strange is going on...
 
When I went to flash the BIOS, I pulled the 6272 ES out of CPU1 and inserted the 6172 (left the 6272 ES in CPU2). It would post, until I removed the CPU2 6272 ES.. It seems like I notice some pencil lead marks on the bottom.. Maybe the CPUs pad need to be cleaned? If so, do we have a "How To" some where?
 
Alcohol + cotton buds / coffee filters

The pads are easy to clean as long as you don't use something too abrasive
 
The gold coating on chip pads/pins in measured in nanometers. Pencil eraser or cotton swap with alcohol is about as aggressive as you want.

If you are screwed, put mineral oil on wet/dry 1000 grit sandpaper and lightly lap it. It might take off the gold, but it might break a short circuit.
 
Spend two hours trying the memory and cpus in several different configurations.. Cleaned the cpus with Qtips and alcohol. I was able to get it One time to post with an ES in CPU1 and 2 DIMMS in (A1 and B1). I then mirrored the config in CPU2 an no post.. Reverted by to the "one time" posted setup and it wouldn't post. Try a few more thing with no luck. So, I decided it would be good to validate the setup and put in the 6172 (known good cpu) and I couldn't get it to post either. So at that point throw up my hands and asking for the lords help in several languages, I decided to call it a night.

Have I killed the motherboard (I'm still getting the blinking green light and the fans spinning) ? This is my first experience with the extra spicy stuff, and got to say I'm not too excited about pulling apart one of my perfectly working 4P boxes to see if these chips work. What kind of PPD should be able to get on the 6272 ES?
 
A dual G34 16-core at 2.7ghz makes about 250k PPD average. I'm running six ES 16-cores, and all are 6200 44's. The 43's are not predictable, but the 44's seem to run solid on ASUS and Tyan. Haven't run them in Supermicro yet.

One of my boxen with removable BIOS had two pins touching. This created a situation where booting was unreliable, sometimes it would be fine, other times not.

I've actually unplugged a boxen while folding, pulled out the CPU's and stuffed other ones in without resetting CMOS. About a dozen times with no glitches. When a boxen was going to miss deadline, I'd swap CPU's with a faster machine to finish the jobs.
 
Clearing the CMOS worked. I was able to get one of the ES chips to post regularly. The other ES chip will not post at all.. Suggestion for the seller was to downgrade the BIOS from 3.0b to 2.0a.. Hopefully can find the older BIOS.
 
Last edited:
Change BIOS version and got the same results. Apparently one of the CPUs is dead. Unless anyone has another idea. Thanks for the help.
 
The popular South Korea seller is telling me the CPUs were tested good before shipping and believes it the mobo.. Anyone had this experience, where two seemingly the same ES CPUs, one would work and the other not? Seems odd to me.
 
Never seen this kind of behavior... I'd honestly vote "bad" CPU. Hopefully popular south korean seller can
replace the CPU.

Dismantling a 4P to troubleshoot this system seems a bit excessive to me, tbh...
 
Shouldn't have to vote on the CPUs at all: If one worked in CPU slot #1 and the other CPU didn't when you swapped them out, it's a bad CPU. Fact...end of story.
 
It's not always that clean-cut given complexity of CPU<->rest-of-the-world interface, esp.
when dealing with prerelease silicon.
 
bowlinra said:
The popular South Korea seller is telling me the CPUs were tested good before shipping and believes it the mobo.. Anyone had this experience, where two seemingly the same ES CPUs, one would work and the other not? Seems odd to me.
Click to expand...

I switch plans from upgrade my 2P box to upgrade the 4P box and got 2 more ZS212045TGG44 cpus.
As it turns out the "Bad" cpus did work on the SM H8QGi board.. I had to reinstall SM BIOS 3.0b and clear BIOS. I did notice a slight different in the cpus BIOS -> Advance -> Processor & Clock Options shows:
AMD Eng Sample, ZS21245TGG44_31/21/2_2/16 (CPU 1,2,3 )
AMD Eng Sample, ZS212045TGG44_31/21/20_2/16 (CPU 4 "Orginally thought bad")

Note the slight differences in RED.

I've also notice it taking longer to boot with the "Bad" CPU 4 in the system. Anyone have ideas to the differences in the cpus? and/or is this a big deal?
 
Last edited:
bowlinra said:
I did notice a slight different in the cpus BIOS -> Advance -> Processor & Clock Options shows:
AMD Eng Sample, ZS212045TGG44_31/21/20_2/16 (CPU 1,2,3 )
AMD Eng Sample, ZS21245TGG44_31/21/2_2/16 (CPU 4 "Orginally thought bad")

Note the slight differences in RED.
Click to expand...
*Correction* I transposed the models it should be:
AMD Eng Sample, ZS21245TGG44_31/21/2_2/16 (CPU 1,2,3 )
AMD Eng Sample, ZS212045TGG44_31/21/20_2/16 (CPU 4 "Orginally thought bad")

Note the slight differences in RED.

For clarity. All ES cpus appear to be working on the 4P (H8QGi+-F) board (completed 8105 WU), but I could NOT get the thought "Bad" chip (ZS212045TGG44_31/21/20_2/16) to boot with 2P (H8DGU-F) board.

Should I be concerned all 4 ES cpus don't perfectly match? I have a possible option to exchange it with shipping expense.
 
You must log in or register to reply here.
Back
Top