Top line is 40-CMB000-G210 and bottom line V58CV6E0070C5G210X. It might be tomorrow before I can try the bios but will of course let you know how it goes! UPDATE: I've tried with ImgBurn, Winimage and MagicISO to add this bios to the floppy image and make a bootable CD but can't seem to get it right! Any pointers on how to get this into an iso would be great as I don't have access to a floppy drive.
Thanks for the tips, bios successfully loaded once I set the emulation correctly. It's booted fine and scaling OK, will update following some more uptime.
Hi all any huge thanks to the posters in this thread, esp tqhoang for the bios updates. Have been battling with a hdama rev g with dual opteron 250's and 4GB ram since I bought it. Regardless of operating susyem ie linux distros, windows xp,sbs 2003 etc. Thanks to your bios updates and info here I now have sbs2008, dell slic in bios and all working, with 2 sata raid drives. However my original problem remains " it freezes after running well for hours or even a couple days". I have followed all I've read in the forums ref memtest, I even replaced the ram and hard drives, all drivers are correct from what I've read. I was going down the cpu or chipset temp route, but having logged and benchmarked the system, it doesn't fail at higher temps. It just seems to freeze when it wants to. Last test was to run it with only cpu 0 only in use, and I now have a stable system! I have used both cpu's in cpu 0 socket and the system remains stable for days. So i know both cpu's are good. But the system will not run well when using both cpu's. The supplier of the server suggested it may be due to a difference in cpu temp is the cause, on testing the difference is about 5 dec C. Any ideas guys ? LJ
tghoang Thanks for your quick reply. 1.. Yes stepping revision is E4. 2.. I am using your bios "HDAMAG_BIOS_2.18B_SI5500_DSDT_OPTERON250E4" 3.. When using 2 cpu's YES I have the ram shared 2GB in each set of memory banks. At the mo with 1 cpu they are all in 1 bank. 4.. I have read the chipset heatsink temp at 51 dec C max, with a IR temp probe. Have downloaded your rebuilt bios as suggested, will try later today and reply. Many thanks LJ
Tqhoang, Sadly, still getting freezes after a couple of hours with the PLL lock modded bios (didn't adjust the fans at all so not an overheating issue). Have just reloaded the original bios and will try rmclock. Thanks again for your efforts!
Have to report so far my server is still stable since 9:45 this AM Maybe early days as my system has run a day or so in the past, but it certainly still running now. The other thing I have noticed is that on some start ups the server runs for about 1 min, then shuts down, on first power up. But subsequent power ups are normal. Not sure if that occurs with notginger's ? I only mentioned that because it never happens when using only 1 cpu ! Will report status over a few days, see how it goes. LJ
Hi and also from my side thanks especially to tqhoang. I am also seeing the same problems on my system, 2 DC 270 and 8GB ram. I am starting to suspect that the system does not support frequency scaling. It runs fine under Opensolaris that only do Cstate transitions. Under Linux or Windows 7 with frequency scaling it hangs after a while. I was not even able to install Linux, setting acpi=off solved that problem and with that it is running fine but at full speed. While investigating frequency scaling for Opensolaris I found several comments that the syncronization of clocks between the processore were not stable with the 200 series processors and was the reason they had not implemented it. Anyway I managed to modify the SIL module in the BIOS to work as a non raid card. This was needed for Opensolaris to install. If anybody is interested I will provide it. I also removed 2 unneeded modules (Promise Sata and LAN for 5704) and exchanged the LAN boot room for the gpxe. Works fine and will allow booting from iscsi Best Regards Peter
Since the bios change I must say both cpu's are running about 10 deg C cooler than before. Both cpu's run at about 22 deg C at idle. Confident with this improvement I have run prime95 torture tests running both cpu's at 100% whilst watching the temps. Noticed that cpu0 raised to 51 deg C and cpu1 got to 33 deg C. This is strange because cpu1 normally runs a little hotter due to location in the case, cpu0 has more headroom, and isn't below hard drives. That said I have been running the cpu's 1 at a time as mentioned, so the cpu now in slot 0 was the cpu from slot 1 previously. Maybe that cpu runs hotter. After stopping the tests and letting the cpu's cool down to 22 deg C ish I the uninstalled groupshield, just after the completed uninstall the server froze. I should have left it alone after the stress test and see if it ran well, now I don't know if this has confussed my trial. LJ
Peter - Could you upload the SIL module or describe what you needed to change? Also are you able to boot off the SATA drives when in non-RAID mode? I think the freezing problem is more isolated to the motherboard hardware. I've had Opteron CG 248's and E6 280's doing the PowerNow/OPM frequency & voltage scaling on my HDAMB's just fine. Also there are HDAMA-I owners with E6 280's working fine too. I can build you a BIOS with PowerNow for the Opteron 270's but I want to get some feedback from the BIOS I made with the default VST change from 100usec to 200usec.
LJ - Which BIOS are you running? The PLL3 or the VST200? BTW don't worry about which CPU is hotter...that's all dependent upon the OS and how well/evenly it schedules the tasks. My Opteron 280's kind of sit around 39C using the stock AMD heatsinks.
Tqhoang, I will try to upload the SIL module. It boots fine no problems at all. If you compare the SIL rom bios they are identical except some bytes in the end. These bytes determine the behaviour of the bios, they are loaded during boot. The builtin controller does not have these bytes. Reading the SIL manual I found the positions to store the non raid values during inisialization. I simply added some code to do this. Peter
With WHS Vail, lockups are a bit random but tend to be after hours rather than minutes. It was horribly unstable with Fedora though when I was trying to get the PowerNow dmesg info and would freeze after a couple of minutes. I've just loaded the bios with VST 200usec - it boots ok but freezes when I try to run CPU-Z. Will see how it goes overnight!
tghoang. I'm still running the PLL3 bios at the mo. After a restart of the server about 20:00 last night it is still running this AM, following an overnight AV scan and full system backup. So it has been busy for a few hours overnight, not just idle. Happy to try the VST200 bios if you want. LJ
ljcomp - Maybe just stick with that one for now. notginger - Since you've already gotten a lockup, do you want to test a BIOS with both PLL=3usec and VST=200usec?
Peter - Thanks for the SIL ROM. You said that "I simply added some code to do this". Did you disassemble & rebuild the SIL ROM or did you just hexedit something? Could you explain this a little more?
tghoang I agree with that, I have been working the server quite hard so far today. Uninstalling and re installing applications to see if that contributed to my crash yesterday and so far no problems at all. Still early days perhaps but fingers crossed. Not done any more prime95 torture tests though. Will not run them untill the server has run a few days. After that I will torture it see what happens. Many thanks mate LJ