Rear I/O

The rear I/O contains all the necessary connectors - like PS2 mouse and keyboard, six USB 2.0, 6-pin Firewire and RJ45 for Gigabit Ethernet. A single eSATA is included which should suffice, even if it is powered by JMicron, and audio is supplied from either RCA or optical Toslink S/PDIF, or six 3.5mm analogue stereo jacks. The stereo jacks provide the eight channel audio as well as separate microphone and line inputs.

All in all, it's a good array features - six USB sockets is a nice median, although there is space for a couple more given the PCI brackets aren't provided. S/PDIF inputs are missing from the rear I/O and also unfortunately from the board itself, but both types of outputs should make more people happy.

The large gap in the middle looks like it could be filled with more ports, but it could just be space for hot air to exhaust from the heatpipes... or not: as the solid rear I/O plate doesn't seem to suggest this. So it's not deliberate space and there are no extra ports to fill the gap, which speaks of a missed opportunity there for Foxconn.

BIOS

Included again is the FoxOne chip as well as the new "Gladiator" branded BIOS. The POST screen maybe Foxconn Mars red, but once you get inside we've still got a basic blue screen and red selection box. Foxconn has a red (physical) box and black board - so is it so hard to have some coordination like others go to the effort of making?

Abit, for example, don't only colour their BIOS red or black (or even pink with the AB9 Pro) to reflect its board, but the uGuru is also a whole different sub section that makes it feel different, unique and special. A standard Pheonix BIOS will however appeal to those use 3rd party applications and rely on on a regular interface between BIOS and clock generator - the uGuru upsets this but the Gladiator BIOS should not.

It is all in the details of what the BIOS can do, true, but the special "Gladiator" feel is lost in simple marketing hype, however it's a sacrifice that others might be happy with - you can't please everyone.


The voltage selection is just insane - being able to push the CPU over 2V is simply nuts for a Core 2. Memory and northbridge voltages of 3.3V+ and 2.4V+ respectively are also just as black and crisp inducing. The best method of upping the CPU voltage is a combination of the multiplier and the "CPU voltage" option to give an optimum overall voltage, of which there are literally hundreds of options - click the graph to the right and aim for the blue/grey areas.

Unfortunately the CPU PLL voltage is tied to the southbridge voltage - so while you'll want to up the PLL voltage as you overclock and increase the CPU voltage, there should be little reason for you to abuse the southbridge in the same way. Given that it just has a simple low profile heatpipe cooler on it, it does worry us a little. However, the native core voltage for ICH9 is 1.05V so we're already overvolting to start with. We did churn it up to 1.725V as we overclocked and found it was certainly hotter but not so hot we couldn't touch it, however I'm naturally curious as to what a 64 percent voltage increment does to the life of the chip.

The voltage options might be extremely liberal and offer enough options for many people, but it's variety is still a little lacking - there's no GTL Reference voltage, no memory controller or channel specific voltage options, no clock overcharging, and as I said above - no separation between CPU PLL and Southbridge voltage, let alone separation between Southbridge I/O and Core.

The memory timings are pretty comprehensive, and even though with early BIOSes we had some trouble with them, since then the compatibility and performance has improved greatly to a point where the most recent BIOS that will ship on retail boards (featured here), allowed us to run 3-3-3-9, tWR-4, tRFC-18 and 1T command and "Fast Chip Select" enabled at 800MHz.

Tuning the extra tWR-4, tRFC-18 and 1T gave us an extra 200-300MB/s in Everest read scores and dropped the latency by several nanoseconds, but "Fast Chip Select" seemed to offer an additional 200MB/s but dropped the latency by around 10ns! Unfortunately even though "Fast Chip Select" was simple test stable, it wasn't extensively stable for us so wasn't included during testing, however the other performance optimisations were.

Obviously this depends on what memory you use, but if you're running some quality MicronD9s then you should hopefully get a similar or better result.

Other features include the "Overclocks Gear" (I hope that's not your bad grammar again, Richard? - Ed.) which stores the BIOS profiles and is extremely useful if you don't enjoy reinserting all your options every time the Overclock protection kicks in and resets everything. The clock adjustment options for PCI, PCI-Express and CPU/Memory are just like you'd find elsewhere and include increments as low as single MHz.

What does worry us is the "Instant OC" feature which seems to offer overclocking up to a massive (yet impossible) 75 percent and jumps in large five percent increments. It tries to compensate this by running the CPU at a massive 1.55V up to 20 percent OC, then an even more insane 1.6V above this speed. 25 percent is only 416MHz FSB which should be do-able for most CPUs at far lower voltages. We raised this concern with Foxconn and were told it was being looked at, but the fact it's remained in a shipping BIOS gives us cause for concern considering it's a newbie feature which means such people are less aware of the cooling requirements of trying to run a Core 2 CPU at over 1.5V vCore.

The Foxconn engineers have worked extremely hard on this and it shows in many aspects. However there are still a few issues we've raised here that need addressing to not only make it equivalent to the other performance boards on the market, but to also make it safe for those who don't know any better.