Single page Print

DFI's LANParty 875 Pro motherboard

Party on, Wayne
— 12:00 AM on July 28, 2003

ModelLANParty Pro875
Price (Street)$186

SO LET ME START OFF with a blunt statement: DFI has never been known as an enthusiast motherboard maker. Sure, their stuff was cheap, and very reliable, and would probably do just great in grandma's PC, but the typical overclocker wanted things that DFI boards just didn't have, like, say, the ability to overclock. Hence, DFI was like the nerdy girl in the John Hughes film: You were aware of its existence, but didn't seriously consider it an option.

Of course, by the end of the John Hughes film, the nerdy girl had been magically transformed into the Really Hot Girl. Recently, DFI's products have undergone a similar metamorphosis. We've already seen the AMD side of this with the LANParty KT400A and NFII Ultra, but today we're going to look at one of DFI's Intel offerings, the LANParty Pro875, based on the Canterwood chipset. Does this 875P board wind up looking like Ally Sheedy at the end of The Breakfast Club? Let's take a look.

The specs
We'll start off with the obligatory specs chart.

CPU supportSocket 478-based Intel Pentium 4 processors
Form factorATX
ChipsetIntel "Canterwood" 875P
North bridgeIntel 875P
South bridgeIntel ICH5R
InterconnectIntel Accelerated Hub (266MB/sec)
PCI slots5 32-bit/33MHz
AGP slots1, 4X/8X AGP Pro 50 (0.8/1.5V)
AMR/CNR slotsNone
Memory4 184-pin DIMM sockets
Maximum of 4GB of DDR400/333/266 SDRAM
Storage I/OFloppy disk
2 channels ATA/100 via ICH5R south bridge
2 channels ATA/133 via Highpoint 372N
Serial ATA2 channels Serial ATA 150 via ICH5R south bridge
RAIDSerial ATA RAID 0, 1 support via ICH5R south bridge
ATA/133 RAID 0, 0+1, 1.5 support via Highpoint 372N
Legacy ports1 PS/2 keyboard, 1 PS/2 mouse, Serial and Parallel ports
USB4 USB 2.0/1.1 ports
2 additional USB 2.0/1.1 ports via FrontX port replicator
Audio6-channel audio via ICH5R/CMI9739A codec
line out, line in, mic in connectors (motherboard)
headphone out, mic in connectors (FrontX)
coaxial digital S/PDIF input and output (backplate)
MIDI/game connector (backplate)
Ethernet10/100/1000 Gigabit Ethernet via Intel CSA GbE controller
Bus speedsFSB: 200-400MHz in 1MHz increments (w. "C" CPU)
AGP: fixed @ 66MHz
PCI: fixed @ 33MHz
Bus dividersDRAM: 3:2, 5:4, 1:1
VoltagesCPU: +0mv to +350mv in 50mv increments
DRAM: 2.6V, 2.7V
AGP: 1.5V, 1.6V
MonitoringVoltage, fan status, and temperature monitoring

The LANParty Pro875 has a lot of the features you've come to expect on a Canterwood board, such as four DIMM sockets supporting DDR400 memory, AGP 8X, USB 2.0 and two channels of Serial ATA. However, there are also some notable additions, such as the FrontX connectivity and the two channels of ATA/133 RAID via the Highpoint 372N controller. The specs chart doesn't tell the whole story, however, and you'll soon see that DFI has thrown in quite a bit more than a FrontX and a Highpoint chip.

A brief word about the Highpoint controller: Those of you familiar with RAID levels may have scratched your heads at the mention of RAID 1.5 in the chart above. DFI is pushing this as a big feature on the LANParty Pro875, claiming that it "integrates the advantages of RAID 0 and RAID 1 with only 2 HDDs instead of 4." Neat trick, that, but unless I missed something none of the documentation actually explains how they do it.

I finally found a Flash animation on the DFI web site that tells you what's going on here. Essentially, RAID 1.5 is a RAID 1 (mirror) array which processes read requests by grabbing alternating chunks of data off each drive, similar to the striping of RAID 0. Of course, in a RAID 0, writes are striped as well, so you're not exactly getting all the advantages of a RAID 0. Still, in theory, RAID 1.5 should provide read performance close to that of a two drive RAID 0 array, with write performance (and data redundancy) equal to that of a RAID 1.

I'm not up on the specific internal workings of typical RAID hardware, but I was always under the impression that SCSI RAID cards, for example, striped reads off RAID 1 arrays as a matter of course. After all, it only makes sense to do it this way when you have two drives sitting there with identical data on them. Perhaps I'm mistaken, or perhaps up until now, IDE solutions haven't striped reads like this. On the other hand, if other IDE RAID solutions do in fact stripe their reads on RAID 1 arrays, then RAID 1.5 is just a new name for an old practice.