- Page 1 eVGA 131-K8-NF44
- Page 2 eVGA 131-K8-NF44
- Page 3 Single Task 2D Performance Results
- Page 4 Multiple Task 2D Performance Results
- Page 5 Call of Duty 2 Performance Results
- Page 6 Quake 4 Performance Results
- Page 7 Counter-Strike:Source Performance Results
- Page 8 Battlefield 2 Performance Results
- Page 9 3DMark06 Performance Results
My experience with MicroATX motherboards is the BIOS always tends to be a little lacking. I’m not sure why this is – perhaps they aren’t considered profit making enough to justify development time. It’s catch 22 though really, as if companies did develop the BIOS’ a little more, then they would probably sell more.
Unfortunately, this eVGA gets put under the same label. There is a limited number of memory timing adjustments, although you do have control over the Data Command Rate (1T/2T). HT frequency is adjustable and the VID/FID settings are adjustable too to control the CPU multiplier and voltage. You can even boost the memory voltage a little. Not too bad really.
However, all these adjustments were let down by the board’s inability to boot at anything above 220MHz. Not only this, but it has a distinct habit of hanging whenever you hit reset.
I can only assume that as there was no option to do with PCI Express frequency in the BIOS, that there was no PCI Express lock in place. From experience, this is often what limits a board to the 220-230MHz area. I can only hope that both this issue and reset issue are solved in a future BIOS update, but who knows.
I did try downloading a new BIOS from the eVGA website, but there was only one available for the nForce 4 platform. I assumed this was correct and that perhaps it used the same BIOS for its ATX and microATX motherboards, which is quite plausible. If the BIOS was wrong, the flash update software should have kicked up a fuss, but it flashed fine. Unfortunately, the board wasn’t happy at all after the update and stalled on the eVGA logo.
So how did the board perform? We tested it using an FX60, 2GB of Corsair CMX1024-3500LLPRO, a Seagate Barracuda ST340083A8 and a pair of standard clocked 7900 GTs running in SLI.
For 2D testing, we used the same suite of software as our Core 2 Duo review. We haven’t tested this in any other Socket 939 platforms, so instead we compared to an AM2 5000+ and an AM2 FX62.
For 3D testing, we used the same suite of software we use in our graphics card testing (see here for details). For comparison, we have an Asus A8N32-SLI motherboard with an identical setup. For a graphics testing we usually have a 2T data command rate setting (for no other reason that that is how it defaults in the BIOS). But as the eVGA BIOS defaulted to 1T, I re-tested the Asus with 1T for a better comparison.
In the audio compression/decompression tests, this machine performed almost identically to the AM2 5000+. The extra 512K cache per core on the FX60 perhaps offset the DDR2 memory on the 5000+.
Virtualdub rendering shows the 5000+ clipping off 32 seconds, but the eVGA still gives it a pretty good run for its money.
Photoshop shows a 33 second difference, which is more noticeable as the whole test only takes around 500 seconds to run. That’s a seven per cent difference! This is even more noticeable in the Photoshop and Virtualdub multitasking test where there was a 12 per cent difference.
File compression and Encryption was strangely faster on the eVGA setup than the 5000+. This also affected the audio/file compression multitasking test, where once again the eVGA was faster.
In gaming, the results were surprising. Generally speaking, because the memory controller is on the Athlon 64 itself, performance from board to board is almost the same. However, in this case there was around a ten per cent performance difference and as much as 50 per cent in Counter-Strike: Source. Naturally, this difference was less noticeable in the higher resolutions where GPU limitation kicked in a little.
Each card has only eight lanes instead of 16, but I find it hard to believe that this was the cause for so much deviation. If I hadn’t killed the board trying to update the BIOS, I’d be re-testing things to make doubly sure.
This board certainly proved my original point, you can have power while maintaining a small footprint. As a board, as long as you don’t want to overclock (which is often difficult to do in small confined areas anyway), it’s fine. If you are building an entirely new machine right now, investing in Socket 939 may be an unwise move. But for those of you looking to upgrade an existing Socket 939 machine, or if you don’t want to get rid of that expensive DDR memory in your Socket A machine – then this board is definitely worth considering. Just keep an eye out for an updated BIOS that fixes the issue with resetting.