MSI FM2-A85XA-G65 Review: Know Your Platform
by Ian Cutress on December 11, 2012 1:21 AM EST- Posted in
- Motherboards
- MSI
- FM2
- A85X
When looking at a motherboard for the first time, my initial thoughts tend towards two areas. Firstly, does it look like it would be a good performer – there is a certain level of confidence that comes from a positive aesthetic perspective, and if it looks like a dump it would probably act like one. The second thought is usually geared towards the PCIe layout, and I imagine trying to place two GPUs and a sound card (either PCI or PCIe x1). Some manufacturers have an odd way of designing this layout, optimizing it for build time rather than usability. Thankfully though, MSI does sort of reasonable on both accounts.
Visually the board is sleek and seems uncluttered compared to most motherboards, despite the features on board. We get access to power/reset buttons, easy access voltage read points, a full eight SATA ports, five fan headers, the full array of video outputs and that all important PCIe layout for two GPUs + one or two extra devices. In contrast, we do not get a two-digit debug LED, the memory slots are not single latched, the USB 3.0 port is in an odd place in the middle of the board, two of the SATA ports could be blocked by a second GPU, and there are no extra controllers for functionality.
On the hardware side we get a Realtek audio/NIC combo with the ALC892 and 8111E respectively. Previous A85X motherboards tested at AnandTech use a total of three full-length PCIe slots whereas the MSI uses two to optimize usage scenarios with PCI and PCIe x1 slots.
The BIOS design is good in giving the important information, albeit the wall of options in the OC menu could take some getting used to. The fan controls are basic at best, and the automatic OC control leaves a lot to be desired when pushing the performance of a Trinity system. Manual overclocking performance with automatic LLC was not spectacular either.
One of the best bits of the MSI motherboard is the software, with the Live Update software making sure the system is up to date, and the Control Center for all things information, detection and control. The Click BIOS system is great to introduce users to the BIOS while still being in the comfort of a full operating system.
Due to the ‘bug’ in the BIOS relating to top turbo speed and actually getting it to initiate, there are aspects to the performance where the MSI may have spent more time fine tuning or developing a different aspect of implementation. Despite this, multi-threaded performance seems to be fine. The system could be better in terms of DPC Latency (329 microseconds, ideally should be under 200) and time to POST (16.75 seconds), relating back to fine tuning. I should point out that improvements could come with a later BIOS than what was tested here.
At $100, the MSI FM2-A85XA-G65 actually falls on the cheaper end of the range of FM2/A85X motherboards available. The MSI has some tough competition and does not really pull itself away from the rest of the pack unless you want easy-to-use physical read points or a nice system of additional software to deal with. When the results are placed side by side with some of the cheaper FM2 boards we have tested, the performance/price scales are often in favor of the competition, leaving the MSI as a choice if you are really gunning for an MSI build.
Addendum 11/12/2012: Based on the findings in this review, MSI have released a BIOS update to v1.4 which should fix the turbo core issue. The issue was related to the default choice of the C6 state in BIOS 1.3 which was disabled. It should be enabled in BIOS v1.4, or users of the 1.3 BIOS can manually change the BIOS option to get the stock performance back on track. We will hopefully release a news update with the latest benchmark results in due course.
Facebook
Twitter
RSS
11 Comments
View All Comments
torp - Tuesday, December 11, 2012 - link
It makes no sense to build a system like that, you'd go FM2 if you want to use the integrated video... and then a 3-400W power supply, or even a PicoPSU would suffice.The power consumption test basically gives no useful information.
axelthor - Tuesday, December 11, 2012 - link
If you'd read the article instead of just skimming through it an looking at the graphs you would know why.ssj3gohan - Tuesday, December 11, 2012 - link
Yeah, but just keep in mind that for almost all reviews (especially on anandtech) the power consumption figures are just there as a checkbox feature of the review, they rarely if ever mean something useful. Judging from their more polished reviews and the podcast, it's not a case of lack of intelligence or craftmanship that they bodge up every power measurement, it just seems that they don't have enough time to properly execute this part of a review.So, if I start slagging off these otherwise awesome people I should give them some advice as to how they can properly execute this, right?
First and foremost, and this is not meant as flaming or anything: if you don't know what you're doing, don't publish it. By 'not knowing' I don't mean you're an idiot but I mean that you do not fully understand all aspects of power consumption. Power consumption is a ridiculously complicated matter, it depends not just on recognizable BIOS features, the high-level OS, drivers and hardware composition, but also to a fairly high level on temperature, implementation quirks/bugs and simply offset errors on voltage regulators. The only way to make a truly level playing field is for the reviewer to orthogonalize *all* these factors. Always use the same OS on the same drive with the same BIOS features enabled and verified as working. Measure, directly, the DC power consumption of all non-motherboard components if you just want to isolate the effect of the motherboard. Orthogonalize power distribution by not measuring AC but going for DC power measurements. NEVER extrapolate or guess (e.g. silentpcreview uses a lookup table to go from AC to DC watts, which is categorically wrong), always measure directly. And last but not least: always do a sanity check, look at the manufacturer's data sheets (they are basically always right) and check your findings with other reviewers. It doesn't matter if you differ somewhere, as long as you have a complete explanation of why it differs and why your numbers are correct.
For instance, Anandtech publishes a lot of SSD reviews and always does a power consumption test. These are basically all wrong, as are nigh-on all other SSD power consumption figures on other review sites. They test idle power consumption under a fairly old linux kernel that apparently doesn't understand the DIPM (device initiated power management) power management feature. This is a feature that is supported on every major OS release and chipset since Vista and greatly reduces idle power consumption. The consequence is that measured idle power consumption is galaxies away from actual real-world idle power consumption. Not only that, but they apparently forget that on the first page of their review is a table with manufacturer specs that clearly states *much* lower power consumption. Why doesn't that make them scratch their head and think 'hey... maybe we did something wrong? Shouldn't we go check up on this issue?'. They check every other possible performance metric on these drives and aren't satisfied until everything is explained into great detail, so why not also put some effort into power consumption metrics?
As a result, even though intel specs almost all its consumer drives at 75-150mW idle and 150-400mW load power consumption, all reviews of intel SSDs state their idle power consumption as 0.6W and load at 1.2 or something. So if you're looking for a drive with the lowest possible power consumption for use in e.g. a windows tablet, ultrabook or laptop, you won't find any useful information on the web.
/rant
Wwhat - Wednesday, December 12, 2012 - link
That might scientifically make sense, but in real life the only reason to know the power is to see how much AC you pull and have to pay for (or generate), so in fact you only need to measure the AC pull really.cfaalm - Wednesday, December 12, 2012 - link
I think you're both right. Measuring AC pull is only valid in comparison if you use the same PSU on the same voltage. If you go nitpicking over a couple of mV's then even the efficiency curve of the PSU comes into play. I think it's really hard to get it right.IanCutress - Tuesday, December 11, 2012 - link
In order to keep our testing consistent with ALL other motherboard reviews, we test with multiple GPUs and have to have a power supply capable of 3/4 GPUs at full whack with the CPU as well. We use a 1250W Gold power supply, which as noted in the review has a relatively high efficiency - even more so given that I am on the 240 V input.Two important points:
1) We do not have access to every hardware ever released. In my own testing I use the 1250W gold for desktop environments, and a 500W Platinum for mini-ITX environments, because these are the power supplies I own and allow me to complete testing without spending hours changing everything back and forth. We don't have unlimited space to have 18 test beds set up for individual component XYZ either.
2) Even if you feel the individual values mean nothing to you, then as a comparison against other components put in exactly the same position, conclusions can be drawn as to which is better than the others. You examine the gradient of change rather than the absolute value - a technique used often in scientific circles when the absolutes cannot be obtained.
My testing methodology and scientific background that I do have allow for reasoned interpretation and my testing is equivalent to the rigor I placed in my scientific peer-reviewed papers I published. I never aim to mislead or pre-suppose bias on any bit of kit, and only aim to give the readers the best possible explanation with the tools at my disposal.
woogitboogity - Wednesday, January 2, 2013 - link
It is quite simple. They need to keep things consistent when comparing this motherboard and components to others, including all the crazy SLI/CF rigs. So long as you have a decent power supply in terms of quality the supply will not draw more power than it actually needs.SolMiester - Tuesday, December 11, 2012 - link
What is the point of of building APU, then CF with dGPUS?...ridiculous!..frozentundra123456 - Tuesday, December 11, 2012 - link
I agree with what you are saying, but the tested games were surprisingly playable at demanding settings with the right video card(s). However, I cant see buying such a cheap processor, especially an APU, and pairing it high end cards either.It would have been interesting to test something like a 3570k and FX 6300/8350 under the same conditions (with their appropriate MB of course) to see how much faster they were. Surprisingly, it looks like the tests are GPU bound even with a lowly A10 cpu.
Origin64 - Wednesday, December 12, 2012 - link
I still don't get why people are getting so worked up about this. 2.0 x8 offers 95% of the performance, x16 something like 99. We don't need PCIe 3.0 yet. Probably won't for another year or two, at least.