Build-A-PC Chronicles: Intel's QX9770 Rocks, But My Project Hits A SnagBuild-A-PC Chronicles: Intel's QX9770 Rocks, But My Project Hits A Snag

I'm becoming convinced that PCs have distinct personalities. Maybe I've been at this too long, but the idiosyncrasies I encountered in my latest quad-core PC build project have convinced me that gremlins lurk in every system. Fortunately, I was saved by a novel diagnostic indicator, which Asus included with the motherboard.It's been a few weeks since we checked in on my project to build a PC with Intel's not-yet-released Core 2 Extreme QX9770 processor. This will be Intel's top-of-the-line desktop chip when it ships sometime in the second quarter. It's equipped with four cores, each running at 3.2 GHz. This thing is a performance demon -- it's manufactured in a 45-nm process, which means it's got the smallest feature size you can get in chips today. (Smaller features allow more gates and transistors to be crammed onto a given slice of silicon.) It's also equipped with a 1,600-MHz front-side bus. Again, this is Intel's fastest FSB and is great for performance. But it's so new, I had a limited range of motherboards to choose from, since most current mobos top out at a 1,300-MHz FSB.

OK, so here's where we are so far with the build, which has taken far longer than normal (more about exactly why, later). After gathering all the components, I hunted about for an enclosure. I ended up rescuing a dusty old HP case, mainly because I was impressed with its Titanic-like proportions and heft. However, that led to difficulties cutting out a fan hole. You can read and see the whole (hole?) tawdry tale in my earlier post, complete with a video of the first phase of the build (at the bottom of the post).

Which brings us up to date. What happened next was, I crammed all the components into the case. It was surprising how tight everything was, even given that I was using a massive box. This confirms many of the reviews I've seen, at Maximum PC (my favorite enthusiast mag) and others, to the effect that many of the expensive new cases, which purport to be aimed at enthusiast builders, all have some problem or other. So, even given my difficulties with my salvaged HP enclosure -- along with the crudely cut fan hole, the thing has an ugly gun-metal gray finish, which reminds me of an old Hallicrafters shortwave radio.

I'm using an Asus P5E64 WS Professional motherboard. I picked this board because it's one of the few equipped with Intel's x38 chipset. (True, more are hitting the market every day, but I acquired this about two months ago.) The X38 enables it to support the QX9770's fast, 1,600-MHz front-side bus. Like all Asus mobos, this one benefits from its copper heat-pipe design, which inspires confidence that it will stand up to the thermal stresses of performance desktop computing. (That, and the Sunbeam CR-SW-775 Silent Whisper heat sink I attached atop the processor.)

The graphics card is a Gigabyte GeForce 8800 GTS, with 640 MB of video memory. This one is a screamer; in a subsequent post, I detail its performance and report on its 3DMark benchmark results.

For memory, I used 2-GB (2 x 1-GB) of DDR3 SDRAM from Crucial Memory.

Unlike many of my previous projects, this build didn't go smoothly. Possibly this was due to my inattentiveness. I'm not one for reading directions. Heck, if I relied on instructions, I'd buy a PC, rather than build one.

The average project offers so little leeway -- it's more equivalent to a kit, where you're bolting together preassemblies -- that I suspect most people do these things on auto-pilot. I know I do. Which is why I've always wondered what the heck I'd do if I ran into a real problem. It's not like most PC builders have logic analyzers lying around.

Heck, even if I had detailed specs -- and I suspect many components headed to market on 18-month (or less) product life-cycles are cobbled together with precious little documentation -- what would I do if I found a problem? Remove and resolder a multi-hundred-solder-bump IC onto a multilayer printed-circuit board? I don't think so. Which is why it's so gratifying that I was able to get past the big glitch I ran into this time.

My first little stumbling block was that I neglected to read the specs on the Asus P5E64 motherboard. I tried to install some leftover DDR2 memory, but the sticks wouldn't fit, because the middle tab is in a different place on DDR2 and DDR3 SIMMs.

Purchasing the 2 GB of Crucial Memory DDR3 solved the SDRAM installation problem, and I completed the build. However, when I buttoned up the case, after an initial flash of power, the PC wouldn't boot. This was all the more confounding, because the first time I turned on the power supply, the mobo LED lit up. This was a clear indication that power was present, and that there weren't any shorts. (A short between the mobo and the case, where some of the solder on the underside of the board is touching the aluminum of the enclosure, is usually the first culprit when a newly build PC doesn't fire up. That and forgetting to positively plug in all the on-board power connectors.)

I wish I could say that I attempted a systematic diagnosis of my potential troubles, but that's not how these things work. After checking the mobo for shorts (i.e., loosening its nine screws and trying the power again), and verifying that all the power connectors were attached, I set about looking for jumpers. Clearing the real-time clock, which entailed moving the RTC jumper from pins one and two over to pins two and three, waiting 10 seconds and reseating the jumper in the original position, did the trick.

I was finally able to fire up the PC. Because I was using a 500 GB Maxtor hard drive onto which I had previously installed Ubuntu Linux -- that install was for a project done on another system -- I was able to completely boot up the PC. Success! Unfortunately, I couldn't remember my Ubuntu user password, so I wasn't able to begin doing "real work."

No matter -- my intention wasn't to use Ubuntu, it was to install Windows Vista. I'll discuss how the Vista install went in a subsequent post. Before I got to loading the Microsoft operating system, though, I was plagued with the problem of the computer locking up after shutting down. That is, I could turn it off, but many times it wouldn't reboot. I soon realized that the machine needed frequent resets of the RTC clock -- that pesky jumper again.

Here's where I turned to the nifty Asus diagnostic device (picture below) included with the P5E64 motherboard. Called the G.P. Diagnosis Card, this 1-inch wide by 2-inch high card attaches to a connector on the edge of the motherboard. It's got two LEDs, which display a status code. The codes are outlined in Asus's manual for the mobo, which makes this the first time I had a reason to RTFM.

Turns out my system wasn't making it past the memory test, during its power-on system test (POST) sequence. This indicated that there was a problem involving the DDR3 SDRAM. Of course, I was afraid I'd fried the memory. If this had been an older machine, I'd have swapped out the RAM for a spare. However, since it used the expensive, new DDR3, and I only had the two 1-GB sticks, all I could do was remove one bank and then the other. I also tried reconfiguring the locations, since the mobo had two A-channel and two B-channel DRAM slots.

None of this had any effect. This PC would boot after RTC reset, and then get stuck at the POST memory test on a subsequent restart. At one point, I thought the problem might have been metal on the underside of the GeForce 8800 GTS graphics card, which was resting right atop the CMOS battery. (It's that battery which maintains the RTC.) I applied a little electrical tape (sure to be a gooey discovery several years hence) between the two, which is good practice anyway. But it did nothing to solve the problem.

Realizing finally that I had to do some investigation, I hit the online forums, including Asus's own. It quickly became apparent that many of the newer systems -- specifically, cutting-edge motherboards running ultra-fast DDR3 memory -- have issues with getting full compatibility in place. Namely, much of this stuff is so new, it takes time before, say, motherboard X, is compatible with memory from each and every vendor. Perhaps that's why the Asus manual contained a list of DDR3 SDRAM specifically approved for use with the mobo. I had wondered about that, because I rarely recall seeing such lists before. It did concern me that my Crucial Memory modules were not on that list -- though they are listed as being compatible on Crucial's Web site -- but not too much, because, again, who really pays attention to the stuff?

Are you guessing where I'm headed here? Correct; I found an updated BIOS on Asus's support Web site, which did the trick. The new BIOS said that the update was to address "memory compatibility issues." Bingo. Since I've loaded the BIOS onto the motherboard, the system has been rock solid.

I still crane my neck to look at the G.P. Diagnosis Card as the system goes through POST, to make sure it gets past the memory test, but I'm sure I'll get over that shortly.

Next time, and in subsequent posts, I'll talk about loading Vista onto the Core 2 Extreme QX9770 quad-core PC, benchmarking, and converting it into a dual-boot Vista and Windows XP system.

For now, here again is the video from the first half of my quad-core PC build project, in which I focus on cooling issues (i.e., I drill a nasty hole for a fan in the side of the case). I'll have a video of the rest of the project -- installing all the cards, buttoning up the hardware, and loading Vista -- soon.

To read the previous two posts in this series, see "Build-A-PC Chronicles: Reviving A Dusty Old Case", and "Video: How To Build An Intel QX9770 Quad-Core PC".

Like this blog? Subscribe to its RSS feed, here.

For a mobile experience, follow my daily observations on Twitter.

Check out my tech videos on this YouTube channel.