Network Performance
We used to test the networking speed of the Realtek PHY, and Windows Task Manager for CPU usage. We copied a variety of install files, totaling 758 MB, varying in sizes of 300kb to as much as 60MB per file to and from the MSI K8T Neo2-FIR machine, to our ASUS P4C800-E box, which uses an Intel Gigabit CSA controller. We also performed the same test with an ISO image, totaling 761MB.
Both systems were connected via a CAT-5E crossover cable, which should prevent any bottlenecks that would arise with our standard 10/100 router.
Small Files Test - 758MB Total
|
Time to Copy
|
Ave Transfer mB/sec
|
CPU %
|
| Upload |
0:21.2
|
35.87
|
14
|
| Download |
0:27.3
|
27.76
|
19
|
Compared with some past results, MSI's Realtek PHY did very well in our small file transfer tests. What is amazing isn't the copy speed, but rather the CPU usage. CPU usage was relatively low, but still a bit higher than we've seen with the K8N Neo2 which uses the same controller.
Large File Test - 761MB Total
|
Time to Copy
|
Ave Transfer mB/sec
|
CPU %
|
| Upload |
0:20.1
|
39.07
|
12
|
| Download |
0:23.4
|
33.18
|
18
|
With the large file, our results were similar to the small files, but slightly improved.
Overclocking
Using the latest public BIOS, and armed with a Koolance EXOS-Al water cooling solution, we capped off at 253 FSB. We did get as high as 262 FSB, but the system failed to load Windows after POST. We maxed the vCore and tried again, but this time the system would not POST at all. Between 254 to 261 FSB, we had various levels of success, but the setup was still not what I considered stable.
Although 253 FSB was relatively stable, allowing us to run a number of benchmarks, I did notice the system would exhibit strange behaviour in certain tasks. For example, Internet Explorer would continually crash on any site loading Flash content. We reinstalled Flash, but the problem did not go away until we moved back to 245 FSB. I considered the failed OCs may have corrupted some Windows based files, so we reinstalled and went straight for 253 FSB. Everything was fine, except this time around, Java based sites failed to load. Again, moving down to 245 FSB made the problems go away.
Nonetheless, it may be a quirk with our setup, but 253 FSB is entirely possible with this board. I've seen some higher overclocks reported with the K8T Neo2, so it looks like the K8T800 Pro's AGP/PCI lock is in full swing here.
Final Words
Based on what we've observed today, MSI put together a very nice package for those of you looking into building a loaded Athlon 64 system. Despite not being on par with with their NVIDIA solution, there are a number of advantages the K8T Neo2-FIR has over its Athlon 64 cousin.
The review board we received, which is available in retail, is about as feature rich as they come. From 8-channel sound, to Gigabit Ethernet, to multiple RAID controllers, MSI has you covered. We did not get a chance to test the wireless connection due to some router issues, but that is one networking option that is available if you choose to use it. The motherboard supports both the FX and non-FX versions of the Athlon 64, supporting Dual Channel, as well as dropping the need to use registered ram (thanks to the new memory controller built into the CPU).
While we were happy with the number of storage options, having no native SATA support is in our opinion, unacceptable. While we can pin blame on Microsoft and their insistence on users keeping floppy drives handy for RAID installations, Intel, ATI and NVIDIA managed to put native SATA support in their controllers, so why can't VIA? Speaking of controllers, we were also disappointed that despite MSI classifying the K8T Neo2-FIR as an enthusiast level board, they chose passive cooling for the North Bridge, and none for the South Bridge.
Performance was good, though a bit mixed when compared to the nForce 3 250Gb. The K8T-Neo2-FIR trailed in all the application tests, both synthetic and real-world, but the results were very close. Gaming is where the K8T Neo2 gained the upperhand, but again, not by much.
The BIOS options are quite good, allowing a number of user level tweaking to squeeze a bit more performance if desired. I did find some of the voltage options a little on the "light" side of things, which may have contributed to some of our stability issues at 250 FSB and up, but we know others have had a bit more success than we did in this department. I do wish MSI would implement a "Watchdog" feature, such as that found in Albatron's boards, as it would make overclocking more convenient by automatically resetting to stable speeds if OCing has gone awry.
Other than some of the overclocking issues past 250 FSB, the K8T Neo2-FIR was a very reliable board throughout testing. We never blue screened once, and the motherboard accepted ram from Corsair, Crucial, Kingston, TwinMOS, and Mushkin without issue. We were even able to populate all four memory slots with Kingston HyperX without any problems, though overclocking capped at 232 FSB. One thing to point out, which MSI has documented, that in order to run the ram at 400MHz, all the modules must be single sided DDR400.
On the topic of memory, we ran into an odd problem with our Transcend USB 2.0 flash drive and the K8T Neo2, where if we left the drive plugged into the system after a reboot (which I often do since I always forget to unmount and remove it) the system would not load Windows at all. MSI is aware of this issue, and are working on a fix.
Ultimately, none of our issues kept us from using the board. They are what I would define as workable issues, and the only people who may run into problems are SATA owners with no floppy drives, or those of you with a lot of double sided ram DIMMs and plan to fill in all the slots. The performance was very good, albeit a bit slower than the K8N Neo2 in application benchmarks, and very reliable. The onboard peripherals will cover most bases, leaving very little to be desired. At , the board is pretty well priced and worth the money when looking at the package as a whole.
Editor's Note: We received an email from one of the co-writers of . Basically, there may be some confusion of interpretation in the terms "native SATA". From my impression of the article, the VT8237 supports SATA natively, meaning, there is no PCI bus chipset or bridge to get in the way.
While this is a valid argument, our stance here at VL is native SATA means you do not need a floppy disk to install Windows on to a SATA drive (pressing F6). It is possible there may be a newer revision of the VT8237 that does away with this, but this was certainly not the case with our review board.
