What happens when a vendor adds 802.11ac to Killer networking? You get an ACK - specifically, the MSI X299 Gaming M7 ACK. This motherboard includes 3-way SLI and Crossfire support, dual M.2 slots with a unique heatsink design attached to the chipset, both front and rear USB 3.1 (10 Gbps) ports, dual Realtek audio chips, Killer based networking, and plenty of RGB LEDs to light up the case. 

MSI X299 Gaming M7 ACK Overview

With the X299 platform, users tend to drop into one of two categories: the all or the nothing. A user investing in X299 is either trying to get everything, every feature, with multiple graphics cards and displays and lots of memory, or they are trying to get the high-end on the cheap. I remember the days when investing in X58 to get a Core i7-920 and overclocking it was seen as the height of aiming at performance, compared to investing in a Core i7-800 series platform. Back then, X58 was a stepping stone into HEDT. But here, the MSI X299 Gaming M7 ACK, or M7 ACK for short, is clearly aimed at users who want it all: more cores, more memory, more graphics cards, more storage, more network connectivity, and most importantly, more LEDs. If you wanted to light up the night, this is one motherboard that can certainly do that for you.

Buy MSI X299 Gaming M7 ACK on Amazon.com

Aside from features and LEDs, there is one consistent metric in all motherboard reviews. Our performance results placed the M7 ACK right in the middle of the pack, performing well overall. The form of Multi-Core Enhancement it has runs the benchmarks in the 3.6 GHz range, as do the majority of other motherboards tested. The main system performance win was the fact that it was the fastest MSI board to boot that we have tested so far. But users investing in this system should look into the overclocking.

Manually overclocking our Core i9-7900X was met with little fanfare, as this board happily took our i9-7900X CPU to 4.6 GHz before running into our temperature limit. Using the automated OC tool, Game Boost, was a different story. We were only able to reach the 2nd boost level before the CPU was throttling due to the voltage causing temperatures to be too high for the settings. Through it all, the smaller heatsink covering the VRMs did a good job keeping temperatures to an acceptable level during our extended stress testing peaking at 75C while overclocked to 4.5 GHz.

The M7 ACK allows users to have all the options for storage connectivity. There are eight SATA ports along with two M.2 slots. These slots are both covered by the unique hinged heatsink attached to the chipset heatsink. Both M.2 slots support PCIe and SATA modules, and share bandwidth with the U.2 port and SATA ports. 

The board comes with four full-length PCIe slots in positions 1, 3, 4, and 6. Those slots are all connected through the CPU and intended for video cards. PCIe slots 2 and 5, both PCIe 3.0 x1, are connected through the chipset. For connectivity, there is a total of four USB 3.0 ports (three managed by ASMedia 1074, one by the chipset), three USB 2.0 ports (chipset), and two USB 3.1 ports (Type-A and Type-C, ASMedia 3142) on the back panel. Internally there are additional headers for both USB 3.0/USB3.1 as well as two USB 2.0 headers. 

For networking, MSI chose to go with an all Killer-solution here. The board uses Rivet Networks latest Killer E2500 NIC for wired duty and the  Killer Wireless-AC 1535 module for wireless, which comes pre-installed in the M2 (Key E) slot. The 802.11ac module supports Bluetooth 4.1 and speeds up to 867 Mbps. The networking is built into the name of the motherboard - Killer networking and 802.11ac gives 'AC' and 'K', for ACK.

Pricing for the MSI X299 Gaming M7 ACK is $390 at Newegg and $365 (from $420) at Amazon. This price point puts it in some decent company in the the $350-$400 range including the ASRock Fatal1ty X299 Professional Gaming i9 currently listed at $390 (from $399), the ASUS ROG Strix X299-XE Gaming at $370, and the GIGABYTE X299 AORUS Ultra Gaming Pro priced at $350. If U.2 is needed, the MSI board is the hands down choice, otherwise, users will need to filter through faetures they need and want as well as consider looks and price. 

MSI X299 Strategy

MSI brings a current total of eleven X299 boards to choose from: the MSI X299 XPower Gaming AC holds the flagship title and makes its home in the Enthusiast Gaming segment along with the X299 Gaming M7 ACK. There are a total of three boards in the Performance Gaming hierarchy in the X299 Gaming Pro Carbon AC, Gaming Pro Carbon, and X299M Gaming Pro Carbon AC (mATX). The Arsenal line carries the three Tomahawk boards, the X299 Tomahawk, the X299 Tomahawk Arctic and the X299 Tomahawk AC, while the Pro lineup for professionals has three motherboards; X299 SLI Plus, X299 Raider, and X299M-A Pro.

MSI's X299 Motherboard Lineup (1/29)
Amazon Newegg
X299 XPower Gaming AC [review planned] $450 $450
X299 Gaming M7 ACK [this review] $366 $380
X299 Gaming Pro Carbon AC Review 9/21 $273^ $330
X299 Gaming Pro Carbon   $320 $320
X299M Gaming Pro Carbon AC [in editing] $277 $290
X299 Tomahawk AC   $273 $290
X299 Tomahawk Arctic Review 11/20 $280 $280
X299 Tomahawk   $256 $270
X299 SLI PLUS Review 11/29 $232 $220
X299 Raider   $215 $220
X299M-A Pro   - $237^

^ Third Party Sellers

Information on Intel's X299 and our other Reviews

With Intel's release of the Basin Falls platform, encompassing the new X299 chipset and LGA2066 socket, a new generation of CPUs called Skylake-X and Kaby Lake-X were also released. The Skylake-X CPUs range from the 7800X, a hex-core part, all the way up to an 18-core 7980XE multitasking behemoth. Between the bookend CPUs are five others increasing in core count, as in the table below. The latter HCC models are set to be launched over 2H of 2017.

Skylake-X Processors
  7800X 7820X 7900X   7920X 7940X 7960X 7980XE
Silicon LCC   HCC
Cores / Threads 6/12 8/16 10/20   12/24 14/28 16/32 18/36
Base Clock / GHz 3.5 3.6 3.3   2.9 3.1 2.8 2.6
Turbo Clock / GHz 4.0 4.3 4.3   4.3 4.3 4.3 4.2
Turbo Max Clock N/A 4.5 4.5   4.4 4.4 4.4 4.4
L3 1.375 MB/core   1.375 MB/core
PCIe Lanes 28 44   44
Memory Channels 4   4
Memory Freq DDR4 2400 2666   2666
TDP 140W   140W 165W
Price $389 $599 $999   $1199 $1399 $1699 $1999

Board partners have launched dozens of motherboards on this platform already, several of which we will have an opportunity to look over in the coming weeks and months. 

Other AnandTech X299 Motherboard Reviews:

To read specifically about the X299 chip/platform and the specifications therein, our deep dive into what it is can be found at this link.

X299 Motherboard Review Notice

If you’ve been following the minutiae of the saga of X299 motherboards, you might have heard some issues regarding power delivery, overclocking, and the ability to cool these processors down given the power consumption. In a nutshell, it comes down to this:

  • Skylake-X consumes a lot of power at peak (150W+),
  • The thermal interface inside the CPU doesn’t do much requiring a powerful CPU cooler,
  • Some motherboard vendors apply Multi-Core Turbo which raises the power consumption and voltage, exacerbating the issue
  • The VRMs have to deal with more power, and due to losses, raise in temperature
  • Some motherboards do not have sufficient VRM cooling without an active cooler
  • This causes the CPU to declock or hit thermal power states as to not degrade components
  • This causes a performance drop, and overclocked systems are affected even more than usual

There has been some excellent work done by Igor Wallossek over at Tom’s Hardware, with thermal probes, thermal cameras, and performance analysis. The bottom line is that motherboard vendors need to be careful when it comes to default settings (if MCT is enabled by default) and provide sufficient VRM cooling in all scenarios – either larger and heavier heatsinks or moving back to active cooling. 

Visual Inspection


View All Comments

  • Diji1 - Wednesday, March 7, 2018 - link

    >don't buy that QoS bullshit, Ian, you're not stupid, if some idiot is shittorenting on ADSL no amount of qos

    So you've never used (or properly configured?) a device with QoS then I take it because if you had you wouldn't be dribbling inane stuff from your imagination like this.
  • karakarga - Wednesday, March 7, 2018 - link

    New mainboards may start to use Doby Atmos playback with 9.1.6 output. New UHD films may soon adapt to it. Buying an amplifier is really expensive for hearing it. Computer with the help of CPU may ease it for us. Reply
  • PeachNCream - Monday, March 5, 2018 - link

    Lots of us have a very good understanding of networking so don't discount us as lacking the knowledge necessary to make sense of what Rivet Networks says their hardware is doing with respect to packet analysis and traffic prioritization. Maybe the specific implementation is unique to Killer NICs (I'd even argue that's not the case since they're moving to Intel hardware so if there is packet prioritization happening in hardware, it'll eventually be Intel's implementation with Rivet's interface atop it to allow user configuration). It also seems preconceived (alarmingly so) to assume the readers would just pop in to make negative comments if you go through the trouble of running benchmarks. I'd like to think there's value in what Anandtech publishes which is why most of us, even those with strongly negative opinions about Killer products, are here in the first place reading articles. Some of us don't clearly articulate why we don't like something though. Instead it might come off as general loathing, but I think the feedback you're getting here is rooted in the favorable light Killer is cast in without supporting data that's further combined with negative customer experiences of past Killer NIC products. If you'd like to address that perception and remain as outwardly positive quantified data that isn't marketing material direct from the company would go a long way because claims would have the necessary backing.

    I do worry though, that the Killer NIC's benefits can't be realized in benchmarks. If its the case that the key selling points aren't measurable, repeatable, or demonstrable then do they really exist? Are the use cases where a Killer NIC's benefits most likely to be realized commonplace enough to begin with and if they are, what sorts of difficulties make measurement so elusive? It's hard to accept the sales pitch if there's not a trustworthy third party out there that can show they actually exist.

    I understand why you'd like to consult with Rivet's personnel in the process of finding a way to measure and demonstrate what their hardware can do, but surely they already went through this trouble to get the numbers for their marketing materials that would put their product performance into bar graphs alongside other common network adapters. They've got to have invented that wheel already if they're claiming those benefits since false claims would expose the company to legal liability. It'd be odd that they're mum about it since their hardware appeals to a crowd of technically-inclined consumers that get very excited over a 2ms improvement in ping times to a game server.
  • npz - Monday, March 5, 2018 - link

    I'd say that given their well deserved bad reputation in the past, they have a higher bar to climb to really convince the more knowledgeable consumer. And that is just merely for performance claims, not reliability, which is where the majority of their bad rep comes from. To be honest, I do not expect any tech journalist outfit to become a third party QA dev team either. Reply
  • PeachNCream - Tuesday, March 6, 2018 - link

    I hope I'm not misunderstanding what you're saying, but please correct me if I'm reading your intention wrong. I'm not expecting Anandtech to perform quality assurance at all. What I think is reasonable to ask is that if AT identifies Killer NICs in an outwardly positive manner over and over again despite the nearly universally negative response it generates among readers, then the onus is on Anandtech to provide some kind of evidence that proves there's a good reason for it. Good reason beyond not getting future free samples from a hardware company like say MSI if the site doesn't write a positive article.

    I've been watching and reading AT for years and aside from a 2006 review of very, very old Killer NIC hardware, there's been no published benchmark data. In the meantime, Rivet Networks' personnel were fast to respond to comments about their products in other articles AND Ian has been a strong supporter, often defending what is a questionable location to place QoS on a network (individual device NIC versus the home router, for example). Taking a dim view of the readers' capability to comprehend how TCP/IP networks handle data transmission and assuming there's no point in presenting benchmarks because the feedback in comments will be negative shows contempt for readers and is worrisome. That naturally leads to a certain amount of suspicion and I think combined altogether paints a picture that leads to a cycle of negative feedback plus a loss of reader engagement. I'm hoping that benchmarks done independently of Rivet Networks' involvement or influence with a genuine, honest assessment of company claims can break the cycle, because I don't think there really is an underlying ethical issue even though there are some worrying indicators.
  • jabber - Monday, March 12, 2018 - link

    Are you the guy that used to put up glowing posts about Killer NICs going back to when they first came out? Every review that came out showing they offered nothing for the huge mark up would be a post from that guy stating the opposite. If only I could remember his name. But yeah every review you would find this guy bigging up the Killers. It was quite hilarious. Reply
  • PeachNCream - Monday, March 12, 2018 - link

    I'm absolutely not that guy in the slightest bit. My opinion about Killer NICs has swung between apathetic to mildly suspicious over time. I hoped that my posts conveyed that while I'm open to the idea of giving Rivet Networks the benefit of doubt, it bothers me quite a bit that Anandtech paints Killer NICs in a positive light without providing readers with supporting data. I just want to see some sort of evidence OR for the push to boost Killer NICs dialed back to a dull roar. Reply
  • npz - Monday, March 5, 2018 - link

    But for non-whitelisted programs, won't the packet analytics only work if the data is not encrypted? And everyone's is encrypted now.

    I can see a very realistic scenario of doing these concurrently: torrent downloading/uploading + gaming + live streaming. I can also envision another more niche but professional use case of video conferencing + DAW collaboration program + maybe live streaming your session
  • PeachNCream - Tuesday, March 6, 2018 - link

    It depends on where the encryption is happening, but if its say a software package that wraps up communication inside a VPN tunnel, then the NIC will be unable to distinguish one stream from another and won't be able to exercise any prioritization. At least, that's what I believe will be the case anyway unless the ethernet adapter itself is performing encryption which is the case in most modern hardware when you're transmitting data over WiFi using modern WPA.

    Additionally, performing packet analysis takes compute time somewhere. Whether that happens inside the NIC or relies on the CPU, it will add latency which is another reason why I'm not so sure about Rivet's claims of improved performance. Finally, there's the problem of other devices on the network outside of the NIC's control and the path data takes after leaving the router through the ISP's network and then other provider wires. A Killer NIC can't address any of those problems which, besides persistent problems with their firmware and driver quality, is why the performance claims are suspect.
  • nevcairiel - Tuesday, March 6, 2018 - link

    >If you are gaming, plus downloading, plus streaming, plus watching youtube on another monitor etc,
    > the priority thing does its job.

    I have my router doing proper QoS, I can download at full speed and not notice a dip in gaming. IMHO the "bottleneck" in the network needs to perform this task for good results, and thats often the gateway from ethernet to the internet.

Log in

Don't have an account? Sign up now