Warning: this post will only be understood by a few people, so I apologize to the rest of you. The content is highly technical, so move along unless you really like computers.
I used to build a PC from parts every two years at Christmas. I did this in 2002, 2004, 2006, and 2008. I skipped 2010 because software still hadn't caught up to my hardware, so the investment seemed wasteful. The same thing occurred in 2012. Other than graphics cards and SSDs, desktop PC technology has remained pretty stagnant. The focus has moved on to laptops, tablets, and other mobile devices. So, I bought those two items last month and figured I can survive on the x58 platform for at least another year. The i7-920 is still a very capable CPU even four years out.
I finally ran across some software that required HPET, which is supported on my motherboard, but not using a 2008-era BIOS. I was hesitant to upgrade because I had a stable overclock in place for over 4 years. Still, I figured it was just a new BIOS version and it enabled some other features I had been living without for years.
I flashed to the latest version and a nightmare ensued. I cleared the CMOS and DMI to prevent conflicts with the new version and rebuild all the tables. After the initial reboot I lost both third party SATA controllers, which controlled my Windows 8 boot volume and my two optical drives. The component drives for my RAID 0 array were detected on the Intel controller, but weren't recognized as an array.
So, it was bad to lose the drives, but then while I was trying to troubleshoot, an alarm went off. Systems temperatures (overclocked) normally look like this.
Mine were approaching 80 degrees. That's when I realized that the PC was silent. No fans. I checked the connections before realizing that the BIOS had them set to 0% until 70 degrees. What sort of bizarre default is that?
After fixing the fans, I went back to the drives. There wasn't much point to fixing the memory timings if I wasn't going to be able to boot anyway. Even though the third party controllers were active, they couldn't see anything connected in IDE, RAID, or AHCI. It turns out that the BIOS flash included ROM updates for all the controllers, which apparently did something bad.
Without the optical drives, I wasn't going to be able reflash, so I started working on the Intel side to try and get an old Windows 7 partition to work. I switched the Intel controller into RAID mode so I could access the configuration utility, which let me use Autodetect to find the array. The utility found it and after a reboot, I could see it in the BIOS. Then I tried to boot.
Apparently the bootloader was stored on the SSD. Ugh. Now I have to create a DOS bootdisk on a USB drive and hope I can write a bootloader to the RAID array. After doing that, I was able to boot into Windows 7. Thankfully I had left everything as MBR. I see GPT becoming a huge issue in the future as drives larger than 2TB become commonplace. There are still major incompatibilities there.
Anyway, once in Windows I started researching the controllers and learned that this is common. Apparently using /CC and /CD parameters in awdflash isn't enough. You actually need to manually clear the CMOS. Why have parameters then?!
After this, the JMicron controllers were able to see the optical drives and the SSD. Unfortunately, the optical drives weren't able to be seen in Windows because of some sort of "Option ROM" error. I guess the old Blu-Ray drives aren't capable of working in IDE+RAID or AHCI+RAID mode. They would only function in pure IDE.
Now the only problem was the SSD. I rewrote the bootloader again to find Windows 8 and it did. When I tried to boot, I was told my system installation was corrupt. It was throwing errors when I set the controller to AHCI mode. When I switched to IDE, it booted fine. For whatever reason, Windows 8 removed the AHCI drivers.
I tried all the common workarounds to fix it, but safe mode was as far as I could get. I finally gave up and wiped the drive. A fresh AHCI install fixed it. What a terrible design! People migrate from IDE to AHCI all the time. Why can't both drive sets be loaded?
After ensuring stability on all 6 drives, I went back to enable my overclock. Unfortunately, I found that the menu settings had changed. There were new options enabled by default that I knew nothing about. This is what happens when you stop following this type of technology for four years.
I added back all the timings and voltages I had set before and disabled everything that appeared to be related to automatic overclocking. The overclock seemed to hold, but now I experience hard freezes every few days. No BSOD, no corruption, no sound looping. It's a hardware failure, but I can't pinpoint it. I can only assume it's the motherboard losing connection to the SSD and causing an immediate freeze. I knew it was going to be risky to use a brand new SATA III device on a motherboard with SATA II and an ancient overclock, but it's frustrating to be close.
After all this, I may have to upgrade the motherboard, which makes the CPU and RAM worthless. The other alternative would be to switch back to a very old RAID array that could fail at any moment or boot off of a regular 2TB drive and lose all the speed benefits of the SSD. No good options.
The moral of the story: technology is fickle. Components don't always play nice with each other, especially when they come from multiple generations. Four years is an eternity in this world.
 |
| Instability, damage, and voided warranties. |
I finally ran across some software that required HPET, which is supported on my motherboard, but not using a 2008-era BIOS. I was hesitant to upgrade because I had a stable overclock in place for over 4 years. Still, I figured it was just a new BIOS version and it enabled some other features I had been living without for years.
I flashed to the latest version and a nightmare ensued. I cleared the CMOS and DMI to prevent conflicts with the new version and rebuild all the tables. After the initial reboot I lost both third party SATA controllers, which controlled my Windows 8 boot volume and my two optical drives. The component drives for my RAID 0 array were detected on the Intel controller, but weren't recognized as an array.
So, it was bad to lose the drives, but then while I was trying to troubleshoot, an alarm went off. Systems temperatures (overclocked) normally look like this.
 |
| 31 degrees, nice and cool. |
After fixing the fans, I went back to the drives. There wasn't much point to fixing the memory timings if I wasn't going to be able to boot anyway. Even though the third party controllers were active, they couldn't see anything connected in IDE, RAID, or AHCI. It turns out that the BIOS flash included ROM updates for all the controllers, which apparently did something bad.
Without the optical drives, I wasn't going to be able reflash, so I started working on the Intel side to try and get an old Windows 7 partition to work. I switched the Intel controller into RAID mode so I could access the configuration utility, which let me use Autodetect to find the array. The utility found it and after a reboot, I could see it in the BIOS. Then I tried to boot.
Apparently the bootloader was stored on the SSD. Ugh. Now I have to create a DOS bootdisk on a USB drive and hope I can write a bootloader to the RAID array. After doing that, I was able to boot into Windows 7. Thankfully I had left everything as MBR. I see GPT becoming a huge issue in the future as drives larger than 2TB become commonplace. There are still major incompatibilities there.
Anyway, once in Windows I started researching the controllers and learned that this is common. Apparently using /CC and /CD parameters in awdflash isn't enough. You actually need to manually clear the CMOS. Why have parameters then?!
After this, the JMicron controllers were able to see the optical drives and the SSD. Unfortunately, the optical drives weren't able to be seen in Windows because of some sort of "Option ROM" error. I guess the old Blu-Ray drives aren't capable of working in IDE+RAID or AHCI+RAID mode. They would only function in pure IDE.
Now the only problem was the SSD. I rewrote the bootloader again to find Windows 8 and it did. When I tried to boot, I was told my system installation was corrupt. It was throwing errors when I set the controller to AHCI mode. When I switched to IDE, it booted fine. For whatever reason, Windows 8 removed the AHCI drivers.
I tried all the common workarounds to fix it, but safe mode was as far as I could get. I finally gave up and wiped the drive. A fresh AHCI install fixed it. What a terrible design! People migrate from IDE to AHCI all the time. Why can't both drive sets be loaded?
After ensuring stability on all 6 drives, I went back to enable my overclock. Unfortunately, I found that the menu settings had changed. There were new options enabled by default that I knew nothing about. This is what happens when you stop following this type of technology for four years.
I added back all the timings and voltages I had set before and disabled everything that appeared to be related to automatic overclocking. The overclock seemed to hold, but now I experience hard freezes every few days. No BSOD, no corruption, no sound looping. It's a hardware failure, but I can't pinpoint it. I can only assume it's the motherboard losing connection to the SSD and causing an immediate freeze. I knew it was going to be risky to use a brand new SATA III device on a motherboard with SATA II and an ancient overclock, but it's frustrating to be close.
After all this, I may have to upgrade the motherboard, which makes the CPU and RAM worthless. The other alternative would be to switch back to a very old RAID array that could fail at any moment or boot off of a regular 2TB drive and lose all the speed benefits of the SSD. No good options.
The moral of the story: technology is fickle. Components don't always play nice with each other, especially when they come from multiple generations. Four years is an eternity in this world.
No comments:
Post a Comment