The first ABIT NF7-S motherboard was a fine board, but with a few problems. One obvious issue with enthusiasts was the omission of the four mounting holes around the CPU socket needed for large heatsinks, such as the Swiftech MCX462+. Some water coolers required these holes as well, so we can see this would raise a few concerns. Another problem for overclockers was the fairly limited options for vCore and memory voltage adjustments.
The NF7-S v1.2 addressed some of the shortcomings by adding the four mounting holes, as well as upping the vCore and DIMM voltage options to 2.3v and 2.9v respectively. A couple things that still occurred (though somewhat corrected with a new BIOS update) are issues with Corsair TWINX and BIOS corruption problems. The BIOS corruption issues are well documented with the nforce 2, and something we've been sure to mention in our reviews thus far. Recently, I managed to kill the BIOS on our NF7-M. Basically, after making some memory adjustments, the screen would freeze on "Saving to BIOS", only to never being able to boot up again. This happened with our MSI and Epox boards as well, so this is not an ABIT-only issue.
With the NF7-S v2.0 we'll be looking at today, ABIT has set out to perfect the board… correcting all past problems, as well as making a few changes that should make the board much more attractive to those hoping to take the plunge into the nForce 2. This won't be a long-winded review, since there isn't much of a physical difference between the 1.2 PCB and the 2.0 PCB, so we'll cover the areas that will be of most interest before heading to the overclocking and benchmarks. If you'd like to know more, feel free to read our ABIT NF7-M review and nForce 2 preview.
The ABIT NF7-S
Not a lot has changed on the surface for the NF7-S v2.0 when compared to the v1.2. The board still carries the same colour and basic layout. One change for the better is the chip that is next to one of the mounting holes has been moved slightly to accommodate larger heatsink lugs. We stated in our NF7-M review that some lugs may bump up against this chip (or not fit at all depending on the size). I cannot say for sure with other heatsinks, but our Swiftech MCX462+ has no installation issues with the v2.0 PCB.
The alignment of the AGP and memory slots still presents a problem for those of you using larger video cards. With a Ti4600, you'll almost certainly need to remove the card before removing/changing the ram since the memory clips won't have the necessary clearance needed. This is by no means a huge problem, but it is an inconvenience.
Like the v1.2 PCB, the power connection lie between the I/O back panel and the CPU socket. There is plenty of clearance to get your hands in there to install/remove the power connections, but the complain we have is these very power cables may interfere with the CPU heatsink and fan. Though I've never damaged a power cable with a heatsink fan, the possibility of this happening is present on this board.
One notable change from the v1.2 PCB is the move to a newer Crush18D A1 stepping. The v1.2 PCB used the A3 stepping, and while there are probably a few optimizations done, the notable change is support for 400FSB Athlons.
The MCP-T chipset gives us support for ethernet 10/100, USB 2.0, ATA133, and Soundstorm. It is capable of true 5.1 stereo surround sound, doing all the Dolby Digital encoding in hardware. Like the SSP, the MCP-T has also been upgraded, though I'm unsure of any changes other than probable tweaks here and there.
Overclocking
ABIT boards have always been solid overclockers, but with the NF7-M, we did have some struggles getting our Barton 2500+ to overclock above 200FSB. The board would boot past 200FSB, but it would hardly be anything I'd call stable. Given that the NF7-S v.2.0 is 400FSB (200FSB double pumped), I had some hope we'd have better success this time around.
The first test was to see if we were able to at least hit the same numbers as in our Barton 2500+ review. This proved to be no problem, as the NF7-S didn't flinch at all at 12x200. Keeping the multiplier at 12, I started moving the FSB upwards, adjusting the voltage as needed until we reached the ceiling of 208.
This was an ok result, given the high multiplier, so we dropped down to 10.5. The end result was 240FSB, but it wouldn't be anything I'd call stable. Our vCore was running along at 1.9v, so I lowered the FSB until we reached what I would call acceptable stability (run benchmarks, run Prime95 overnight without lockups). We settled on a 237FSB, which is 2488MHz.
Since we've probably reached the limits of our processor, we then lowered the multiplier to 9.5. Ram timings were kept tight, but we had to adjust our divider to 5/4 (400FSB end result). Moving the FSB upwards again, we landed on a final overclock of 9.5x245. I did manage 9x250, but the stability wasn't the greatest.
Lowering to 8x250 didn't seem to help much either. Unfortunently, it was either 250FSB or 245FSB, as the BIOS has nothing in between.
Our results were much better this time around, and the revision 2.0 PCB further cements ABIT's reputation as an overclocker's friend.
Test Setup
ABIT NF7-S nForce2: Barton 2500+, 2 x 256MB Corsair TWINX PC3200 Ram, ATi AiW Radeon 9700 Pro, 120GB Western Digital SE 8MB Cache, Windows XP SP1, nForce 2 Unified Driver Package 2.03, ATi Catalyst 3.4.
ABIT NF7-M nForce2: Barton 2500+, 2 x 256MB Corsair TWINX PC3200 Ram, ATi AiW Radeon 9700 Pro, 120GB Western Digital SE 8MB Cache, Windows XP SP1, nForce 2 Unified Driver Package 2.03, ATi Catalyst 3.4.
Test software will be:
SiSoft Sandra 2003
PC Mark 2002
PiFast
3D Mark 2003
Unreal Tournament 2003 CPU Tests
Quake 3: Arena
I'm aware that the two ABIT boards are different models, but we factored out the IGP component of the NF7-M by using the AiW 9700 Pro, and we're not using the SATA RAID for the hard drives on the NF7-S. This will hopefully give us as close to apples to apples between the two PCB revisions.
ABIT NF7-S Revison 2.0: The ABIT nForce 2 family gets another tweak to address some issues that have plagued the nForce 2 since the initial release. We take a look at the changes, as well as if there are any performance increases.
