|
|
Autumn 2003 |
Highlighted Components
|
|
|
|
"The Beast", Sapphire's balls-out Radeon 9800 Pro All-In-Wonder multimedia card. Built around the fastest consumer-level graphics processor on the market, unusually for a multimedia card the AIW variant has the same core and memory clock speeds as regular 9800 Pro cards. It has full support for DirectX 9 (not that any DX9 software actually exists as yet!) and all the TV input, S-Video output, and DVD acceleration facilities of the previous AIW cards. It's definitely extremely fast, and although I haven't actually bothered benchmarking it as yet, it has enough oomph for me to enable all the weird and wonderful graphical effects in StarDock's marvellous little WindowFX applet without slowing anything noticeably. None of the games I own at present are demanding enough to really stretch its abilities, though, and I'm hoping that, like its predecessor, it will last the two or three years it takes for the games to catch up. The Sapphire card is very close to ATI's reference design, but there's an odd quirk in the shape of the complete absence of the auxiliary audio/video input socket. Traditionally, the audio output from a DVD-ROM drive is connected to the AIW card, then a short flying lead connects between the card's CD output and the CD input of the sound card - this allowed the soundtrack from the TV tuner and DVDs to be passed through to the soundcard, but presumably in the more modern design the data is carried by some other route. The CD output is present on the Sapphire AIW, but although I've connected it I'm currently unsure whether it's actually any use or not! Installation with the recent Catalyst 3.8 driver suite was as painless as these things ever are, and fortunately I was aware of the DirectX 9.0b bug before I started, so the complete lack of a TV tuner didn't surprise me. The ATI software is extremely fond of forcibly associating itself with every multimedia file type in existence, unfortunately, and this time is was even harder to make it give them back to WinAmp and Windows Media Player - but persistence and an upgrade to the latest version of WMP eventually paid off, and right now everything seems to be working the way it should. I've been extremely fond of ATI's graphics cards ever since the Mach-64 workhorse, and with the 9800 they're still going strong.
|
|
|
|
Although I ordered the Supermicro DAC-ZCRINT Zero-Channel S-ATA RAID card from a Supermicro dealer recommended by Supermicro, for some unknown reason I actually ended up with the ICP Vortex GTD 8500RZ reference design instead of Supermicro's own bespoke variant. There are no significant differences, it seems, except for the addition of the pretty blue heatsink on the ICP card... This didn't appear to present a problem until I came to install the drive cabling, and then I noticed that the heatsink fouls one of the motherboard's S-ATA connectors, and I've had to shave down the moulded plastic housing and strain-relief on one end of a cable so that it would slip between the heatsink's fins. Not a problem, as it turned out, but it gave me a nasty moment when I first came across it... The card ships with write-caching enabled in the BIOS, and although I'm somewhat dubious of this feature when there's no battery-backup on the card, without it enabled one loses all the best go-faster features of elevator seeks, adaptive delayed writes, etc... I'm leaving it in place at present, but the first time it loses a bunch of data and fries my primary partition I'll probably turn it off. :-) In the meantime, though, I've seen repeated peaks of 35Mb/sec or more during routine read and write operations between the RAID-1 arrays, and I'm very pleased with that. As an aside, buried on the Supermicro web site is an FAQ document for the JSLED jumper block for connecting drive activity LEDs. The search engine for the FAQ documents is almost completely useless, unfortunately, and having spotted the document once in passing I had to mail their helpdesk before I could find it again. It's a great shame, as there's a tremendous amount of information in the database, but with only one word in ten apparently indexed it's not much bloody use!
|
|
|
|
[Homer Simpson Mode = ON] Mmmmmm, Fans... YS-Tech's wonderful TMD fans to be exact, which have the motor's magnetic drive around the circumference of the fan rotor rather than at the centre, giving a far smaller turbulence zone in the middle of the airflow and so extremely appropriate for CPU heatsinks. The heatsinks themselves are rather low-tech compared to most of the current offerings for regular Pentium 4 chips, but Xeon coolers are still few and far between and the Akasa AK-680Cu units make up for their lack of panache with a massive copper base and a solid aluminium fin assembly. I'd envisioned something rather more exotic, but in the event my CPUs are holding steady at around 35°C under load and I certainly can't complain about that. I have a pair of orb-style units en route from JMC in Japan, and I'm eyeing Swiftech's exotic new MCX-603V units, but given how hard it was prising the retention clips on and off the Akasa units I think I may well just leave them as they are... For once, function might triumph over form!
|
|
|
|
Four Maxtor 160Gb Serial-ATA disk drives, 7200RPM with the now standard 8Mb cache, nestled safely in the SuperFlower case's dismountable drive bay. What I didn't know, when this photograph was taken, was that the one on the right had a subtle fault - and a lot of head-scratching ensued while I repeatedly tried and failed to build it into a RAID array with the others. I was a little worried about using four hard disks in this bay in case they overheated, and my change from the Western Digital Caviar drives I'd originally specified to the Maxtor units was partly prompted by their lower operating temperature. Each pair of drives is right in the blast of one of the 80mm front case fans, though, and now they're in place they all run perfectly cool to the touch even in heavy use. The Serial-ATA cabling is going to take some getting used to though, I'd say. The data connectors don't make a very positive, snappy sort of connection onto the drives, and given how stiff and springy the cables are it seems quite easy to slightly dislodge or displace them while working in the vicinity. I had a few problems with drives disappearing from the arrays when rebooting after some hardware tweaking, and until the promised retention improvements are introduced with the next generation of S-ATA devices I think any work inside the case is going to have to be followed by a quick check of all the drive cabling. The ICP RAID controller is very sensitive to having it's devices suddenly jerked out from underneath it, and although the hardware RAID engine is happy to rebuild the mirror in the background while Windows is up and running, it's not something I'm keen to make a habit of.
|
|
|
|
A selection of pin removers for various types of Molex and other power connectors, obtained from FrozenCPU and Action 2K. These were worth their weight in gold while I was adding the braided sheaths to the cables of the Antec power supply, and although they're not quite as easy to use as the product descriptions suggest, they're a great improvement over the old way of poking and prying with a fine screwdriver. Highly recommended, with the caveat that over-enthusiastic use can actually bend the pin inside the Molex housing, which causes all sorts of problems!
|
|
|
|
I have a love/hate relationship with the wires inside my PCs... I probably spent around fifteen hours wrapping all the internal cables in braided sheathing, and although I love the neat and professional effect it brings to the inside of a fully populated PC, it certainly is a right, royal pain in the ass! Even with the Molex pin removers to save the chore of wiggling braiding and heatshrink over the sharp, angular Molex connectors, it's still a fiddly, time consuming job. Last time I wimped out and covered most of the PSU cables in corrugated plastic split loom, but this time I wanted the best cosmetic effect I could achieve and was firm with myself... braid was the only way, but the Antec Truepower PSU had an absolute plethora of wires (one of the reasons I chose it, of course!) and after a while I was thoroughly sick of the whole thing! FrozenCPU are selling pre-sleeved power supplies these days, and had they stocked the EPS12V variant I needed for a dual Xeon motherboard, I would have bought it in a flash!
|
|
|
|
I also spent some time redoing the wiring for the red Neon case lights and for a new addition, a set of pulsing red "Knight Rider" LEDs that I'm hoping to bury deep in the bowels of the case to provide a dancing, flickering sort of effect on the more reflective components. Unfortunately the module turned out to be faulty, but as soon as the replacement arrives I'll be able to drop it into the newly built wiring loom, with switches for both lighting systems neatly mounted in two convenient apertures on the back panel of the case. Watch the main weblog page for details and, if I feel particularly adventurous, some video clips...
|
|
And now, onto the saga of the rebuild itself... |
|
|
