Review: MSI P35 Neo2-FR
01.03.08 - 10:54pm
I feel awfully bad coming out with this review considering how this board has been available for quite some time, but here it goes. Today I’ll be reviewing MSI’s P35 Neo2-FR motherboard with a slew of benchmarks stressing memory performance, harddrive performance, and 3D performance. The Neo2-FR is by far one of the cheapest P35 motherboards available on the market but it shares the same PCB and components with it’s higher priced sibling, MSI’s P35 Platinum. Today I’ll be looking solely at what I consider to be “normal” overclocking in that all the tests will be performed under my current air-based heatsink, so nothing too extreme today, sorry to disappoint.
The Package
For as long as I can remember, computer component boxes have typically featured ridiculous images of robots, unrealistically curvy women, and monsters. The Neo2-FR continues this trend with a most interesting box cover, I sure hope it rips through the benchmarks like the claws ripping through steel do on the box art. Opening the box revealed a jumbled mix of manuals, disks, and a motherboard sleeved in an electro-static bag. For $90 I guess you can’t ask for much in the way of organization, but I was a bit thrown off compared to a few other boards I had unboxed recently.


The Board
The board itself is quite impressive featuring all-solid capacitors, 4-phase PWM, 16x and 4x PCIe slots, ICH9R Southbridge, and a modest heatpipe cooling system for the PWM, Northbridge, and Southbridge. The color selection for the plastic slots were all neutral colors, nothing too garish, but the memory slots weren’t color coded for dual-channel operation which has become a sort of standard.





BIOS
Here lies the command center from which all sorts of fun things can be tweaked. The Neo2-FR’s BIOS was surprisingly well equipped compared to other budget boards, from what I have been told it actually mirrors the P35 Platinum’s BIOS, no surprise there considering how they share the same PCB. Within the BIOS you have your typical options such as setting the time, setting the AHCI/RAID modes, and monitoring the system temperatures, but stuffed under the “CELL” menu are all the settings we really want to tweak. Below is a list of the various settings along with their minimum and maximum values along with their available increments.
- FSB: Stock to 700MHz in 1MHz increments
- Multiplier: 6x to stock
- FSB/Memory Divider: 1:1, 1:1.2, 1:1.25, 1:1.5, 1:1.6, 1:1.66, 1:2
- PCIE Frequency: 100MHz to 200MHz in 1MHz increments
- VCORE: stock to stock+0.7875v in 0.0125v increments
- VDIMM: 1.80v to 2.60v in 0.05v increments
- VTT FSB: 1.175v to 1.550v in 0.025v increments
- NB: 1.250v to 1.625v in 0.025v increments
- SB I/O: 1.5v to 1.8v in 0.1v increments
- SB: 1.05v to 1.15v in 0.1v increments
Benchmarks
I’ll be using a slew of benchmarks to thoroughly test the board along with a series of various BIOS settings that’ll show both the tame and wild sides of this board. Considering how easy it is to overclock with this board, I think it’s safe to say that the settings at 3600MHz are easily attainable by most individuals while the maximum clockspeed settings may be out of reach for the majority without aftermarket cooling.
- SuperPi 1M and 32M: SuperPi is a high-performance single-treaded application that stresses the processor and memory.
- wPrime 32M and 1024M: WPrime is a multithreaded benchmark similar to SuperPi but permitting simultaneous testing of all cores.
- 3DMark 01, 03, 05, 06: These benchmarks by Futuremark typically stress the GPU while 3D06 features a CPU test that favors multi-core processors.
- HD Tach and HD Tune: These two benchmarks will measure SATA throughput with RAID0 enabled.
- Everest Ultimate Edition: This benchmark will test memory latency between the modules and the Northbridge
- Sandra 2008: This series of benchmarks will test the memory throughput.
Results
I’ve got to say the Neo2-FR is quite a remarkable board for the price-point or even for the P35 chipset. Below you can see the raw results for the three sets of tests. I performed tests at 2400MHz, 3600MHz, and maximum speed per benchmark. You may notice how the maximum speed varies with benchmark, this is due to the fact that certain benchmarks stress the processor more and less and stability becomes an issue. For the most part the single-treaded applications can be run at the edge of stability while the multi-threaded applications are much more intense and therefore require lower clockspeeds.
Maximum Clockspeed and Front-side Bus
Overclocking quadcore processors can be quite the struggle as the heat can be a bit excessive and compared to dualcore processors they are very finicky on various settings within the BIOS. On this particular board I was able to hit a maximum FSB on aircooling of 490MHz from a stock of 266MHz. I was also able to hit a Windows-stable frequency of 3950MHz which had to be dropped to 3923MHz to be SuperPi 1M stable. The clockspeed itself isn’t all that impressive as I routinely hit 4000Mhz however the 490MHz FSB on aircooling is very promising. Edit: About a week or two later I tested this board under my cascade and hit a 523FSB.
Processor Intensive Benchmarks
There were no optimizations used during these tests so you should be able to match these results with a fresh installation of Windows. SuperPi was stable at up to 3923MHz which was impressive and set a new personal record for air-cooled SuperPi and most remarkably it shaved a second off my previous overall record which was set at 4200MHz on my single-stage cooled E6400. Very impressive to say the least which also hints to things to come. wPrime scaled very well with clockspeed bumps but after looking over my results with my E6400 there seems to be something wrong with the results. Searching around on the net, it sounds like the version I did the tests with has a bug such that it doesn’t evenly compute and a single-thread lags behind severely, impacting the score. I’m still reporting the scores but understand they aren’t proper.


Video Intensive Benchmarks
3DMark has always been a test where a balance of CPU and GPU performance was necessary to score a high score. This becomes very obvious in 3D01 which is greatly CPU limited and a 50% increase in clockspeed results in a huge boost in score. 3D03 and onward however are more GPU limited for my setup but they scores still scale rather nicely with the clockspeed bumps. 3D06 really scales well considering how there is a CPU specific test within it’s suite, the quadcore processor was the only reason this setup was scoring around 6.7k 3DMarks.

I/O Benchmarks
The ICH9R southbridge used on the Neo2 is a chip primarily used for enthuasiast boards, so the fact that there is one attached to this board was a shock. Unlike the ICH9 used on budget and mainstream boards, the ICH9R supports RAID which I decided to implement for these tests. I setup a RAID0 setup with a 128kb stripe size for these tests. If you don’t understand how RAID0 works, I am using two hard drives and splitting the data between the drives such that each drive can read and write the data, in theory increasing performance drastically. To compare the throughput I tested with the RAID0 setup and with a single-drive, the performance increase is quite substantial. For testing the memory with the different clockspeeds and FSBs I simply set the timings to 5-5-5-15 and using dividers attempted to keep the memory as close to it’s stock speed of 1000MHz. For a more indepth view of memory performance on this platform I suggest my review of the PC2-8000 Firestix using a nearly identical setup.


Conclusion
It was a pleasure to review the MSI P35 Neo2-FR and it will continue to give me hours of joy as I continue to tweak the system under more intense cooling methods. The Neo2 overclocked my quadcore processor perfectly and frankly I was shocked with how well it overclocked the memory also. The inclusion of the ICH9R southbridge, function CrossFire, and fantastic performance puts this board into a much higher league than it’s price conveys. I have dealt with boards in the past that it felt like I was fighting tooth and nail for stability but the Neo2 was very easy to overclock and very forgiving when there was a failed overclocking attempt. All in all, if you are looking for a budget motherboard with enthusiast heritage then look no further, the P35 Neo2-FR is the board for you.
Final Score: 10/10
Pros
- Incredible Price
- ICH9R, 4-phase PWM, Stable Quad Clocking
- Ease of Use
Cons
- Unorganized Package

“I feel awfully bad coming out with this review considering how this board has been available for quite some time”
Why? You are actually doing many people a favour, what you have done is reviewed a potentially finished product with final revisions.
One thing is to review a product, it is quite another to review the finished article. Especially in component terms.
There are many choices in the P35 selection, and the extra cost for a top quality mobo may justify a wait for the X38/X48 product array.
If you are able to hit a stable 3ghz+ on differing processors and work in good RAM and gfx then your price outlay is extremely cost effective.
Anyway, cheers for the read, glad of the review before I build for a friend.
If it was for me though, and it sort of is, I would wait for the X38 which is what I am doing, but that is because I need a new case and PSU, my friend however is going to spend £250 and will end up fairly safeguarded for the next couple of years.
Also, recently I did a mini review on a A7N8X deluxe with a 7900GS agp etc, and I have to say many people found it useful. You would be surprised how many people are still living in the dark ages!!!
Cheers for the review. Just picked up this board for £49, seems very good build with all the attention to the components, rather than “buy ours it has all this free stuff”. Everything good here Q6600 @ 4.2 on water, tops 65C, max it has been is 68C.