MSI Z170A GAMING M7 (Intel LGA-1151)

[Ikaruga]

Honestly guys, motherboard reviews here on TPU for like the past 4-5 years have been exactly the same. I've written them all. Seems like a bit of fanboyism in some posts, from where I sit.

You wrote that after my quote, so I assume you meant it for me. I don't fully understand what did you mean, I saw almost 20% speed increase(!) over Haswell refresh on the "3D performance result" page and that made me really curios, that's all. I'm definitely not a fanboy of anyone (and not even a boy anymore for a long time).

 

[cadaveca]

You wrote that after my quote, so I assume you meant it for me. I don't fully understand what did you mean, I saw almost 20% speed increase(!) over Haswell refresh on the "3D performance result" page and that made me really curios, that's all. I'm definitely not a fanboy of anyone (and not even a boy anymore for a long time).

Nah, not for you, just in general. Please don't take it anything but.

Metro Last Light is a weird bench for me for sure. Before I used Shogun2, with pretty much equal settings. Resolution is lowered to make the CPU a bottleneck, which makes small changes in CPU performance stand out. Typically this is due to memory timings, but I have also found how Turbo profiles are set to have an impact as well. ASUS, for example, used to automatically enabled max Turbo to all cores, but today, they ask if that is what you want, or if you'd like the default (good for users with stock cooling).

In the end though, as you say, these differences are really small, if even noticeable, to the end user. With this in mind, focusing on performance figures for CPU and memory when it comes down to a motherboard, seems like a waste of time, unless three is a significant difference. So in the future, you'll find AIDA64, Cinebench, and 3DMark as the only "performance benchmarks", and more drive testing (all interfaces; I have to collect drives for this yet), audio, LAN, fan control, Power consumption, and then a look at automated features, like seen with this review. These are the things that boards, offer, so that's what will be tested and shown.

With so much user customization available across the board from all manufacturers, performance differences are a mere tweak away from each other, and to me, are unimportant. To the majority of users, color and price is more important. Some people want more than that, for sure. But things like CPU overclocking... man, this platform, you set vCore, set multi, test, done.

Either way the most complete set of boards right now are all ASUS, by a large margin in fact. But no doubt the others will get there in due time.

Yeah, because only offering 3 of the first 4 timings is the way to go. Meanwhile, Corsair releases 13-13-14 ram, and you can't set the second two independently. Yeah. That's complete, all right. You're talking Intel IGP and gaming.

I'll admit, nice find though, and since you did, it'll get fixed. But will ASUS give us the fourth timing? I'm asking for it...

 

[ShockG]

Yeah, because only offering 3 of the first 4 timings is the way to go. Meanwhile, Corsair releases 13-13-14 ram, and you can't set the second two independently. Yeah. That's complete, all right. You're talking Intel IGP and gaming.

I'll admit, nice find though, and since you did, it'll get fixed. But will ASUS give us the fourth timing? I'm asking for it...

That's actually on the MSI motherboard and others as well. You can check it in Command Center or CPU-Z even.
This is right now by design from INTEL, it may change in updated ME code in future, but as it stands, if you change tRCD or tRP in the primary settings. The IMC will match the higher of the two value.
That is, if you set 15-16-15-38 for example. What you will end up with is 15-16-16-38. The same goes for setting 15-15-16-38. You will end up with 15-16-16-38.
There currently isn't a way to set them interdependently and that is why ASUS has removed the setting. You can only set 3 out of the 4 timings.

 

[ShockG]

Interesting.... I will see if I can duplicate the provlem with the current bios.

I can't say I tested shutting the audio off or hdmi. But to to say it's not ready for retail with issues that would seemingly affect so few, I believe is being too critical. Perhaps I don't fully understand the issue.

How did you find these issues? Particularly the disabling audio. What prompted you to do that for a review?

EDIT: Disabling Audio does cause a no boot issue. However, it was a sloooooow boot process, hangs on 79 for a bit, I get a signal on the monitor with no picture, and then it ends up on 99 debug (monitor signal stays on). This is BIOS 142.

EDIT2: I have no issues with manually adjusting OC PLL voltage or SFR... but I am not using the onboard GPU either. Do I have to use the iGPU for this issue to show up?

The PLL will likely affect you if you in the following ways.
Set SFR to 1.3 or 1.4V then restart, you'll get 00.
The board auto detects SFR is higher than CPU PLL then hangs. YOu can set CPU PLL first to a high value, then follow it up on the next boot by setting SFR to the required value.
The issue however is that SFR during POST sets, before PLL so you may still get the "00" issue.
142 is alright, but 151 seems a bit more stable, but the two issues are still present though.

 

[EarthDog]

Thanks shock for the follow up.

Odd that I have a different debug code/end result on the HD Audio bug, ehh?

Again I'm curious, how the heck did you run into these things? I mean, did you try a discrete sound card so you shut off the audio? Why did you raise the OC PLL/SFR voltages? Even using LN2 that isn't something I touch. The HDMI thing I can see why you hit that...

 

[buggalugs]

Anandtech has a full review of this platform, it looks like skylake is at best 5% faster than a 4790K haswell system, and that's with the haswell system running 1600Mhz memory vs 2133Mhz memory on skylake. Some games show Skylake as slower than haswell, but skylake is faster in a few rare synthetic benchmarks. With the same memory speed, that 5% difference would most likely disappear, making it about the same. Of course, you can get much faster DDR4 memory, but memory doesn't seem to make much difference to any tasks, anyway. I just thought there would be an improvement clock for clock, but anyway.

My 4790K does 4.7 @ around 1.25v, Skylake seems to need heaps of volts to reach 4.5Ghz like 1.3-1.4v , and they don't even have the voltage regulator on the CPU anymore.

I'm a bit disappointed. I think I'll hang on to my 4790K and spend some money on a new monitor instead, like a 4K or maybe Asus 144Hz IPS. Or I can upgrade my 290X to a fury with the money, theres some performance gains there.

 

[EarthDog]

With the same memory speed, that 5% difference would most likely disappear, making it about the same.

Ehh, I dont think so fried. Look around for reviews that compared apples to apples bugg. You then go on to say memory doesn't make much of a difference?? In my review, I compared the CPUs at 4.9GHz and 2666 memory. The only difference was the timings CL11 vs CL14. I still showed over 6% and that was with using SuperPi 1M/32M which responds well to memory changes (tight timings and speed). If I removed those, I was looking at 8% in the items tested.

My 4790K does 4.7 @ around 1.25v, Skylake seems to need heaps of volts to reach 4.5Ghz like 1.3-1.4v , and they don't even have the voltage regulator on the CPU anymore.

Remember, they start off around 1.3v (give or take)... so you are really only raising it .15 to get there... about the same as your 4790K which generally starts off around 1.1v (give or take). I would imagine that without the FIVR, that is part of the reason why the starting voltage is much higher even on a smaller process (just a guess).

 

[Sasqui]

Dave, very entertaining read (and good review). Too bad "Spinal Tap" is copyrighted, it'd make for a great name for a motherboard series, lol.

 

[Ikaruga]

Nah, not for you, just in general. Please don't take it anything but.

Thanks, no worries

Metro Last Light is a weird bench for me for sure. Before I used Shogun2, with pretty much equal settings. Resolution is lowered to make the CPU a bottleneck, which makes small changes in CPU performance stand out. Typically this is due to memory timings, but I have also found how Turbo profiles are set to have an impact as well. ASUS, for example, used to automatically enabled max Turbo to all cores, but today, they ask if that is what you want, or if you'd like the default (good for users with stock cooling).

In the end though, as you say, these differences are really small, if even noticeable, to the end user. With this in mind, focusing on performance figures for CPU and memory when it comes down to a motherboard, seems like a waste of time, unless three is a significant difference. So in the future, you'll find AIDA64, Cinebench, and 3DMark as the only "performance benchmarks", and more drive testing (all interfaces; I have to collect drives for this yet), audio, LAN, fan control, Power consumption, and then a look at automated features, like seen with this review. These are the things that boards, offer, so that's what will be tested and shown.

I'm sad to hear that you are about to drop real world tests from your reviews, I like low-res tests, I was just curious about the resolution because high ones do not show much difference.

With so much user customization available across the board from all manufacturers, performance differences are a mere tweak away from each other, and to me, are unimportant. To the majority of users, color and price is more important. Some people want more than that, for sure. But things like CPU overclocking... man, this platform, you set vCore, set multi, test, done.

Life got really busy this summer, I did not have time to do a Skylake build yet, but you just made the ich even more serious

 

[cadaveca]

That's actually on the MSI motherboard and others as well. You can check it in Command Center or CPU-Z even.
This is right now by design from INTEL, it may change in updated ME code in future, but as it stands, if you change tRCD or tRP in the primary settings. The IMC will match the higher of the two value.
That is, if you set 15-16-15-38 for example. What you will end up with is 15-16-16-38. The same goes for setting 15-15-16-38. You will end up with 15-16-16-38.
There currently isn't a way to set them interdependently and that is why ASUS has removed the setting. You can only set 3 out of the 4 timings.

Right, but if ASUS can't fix it yet... I expect them to. They have "OC socket"... and it does nothing.

I am a very sarcastic person, so please, don't take my words for what they mean directly. You called ASUS the most complete. I don't agree. Here they have a "feature" that does...nothing? I'm talking about feelings that are created by the ASUS experience, not what's actually happening. Pretend you don't know squat and have this board and then discover these things.

I am in the process of testing this ASUS board for review now. Started yesterday. I'll leave the rest for the review.

Thanks shock for the follow up.

Odd that I have a different debug code/end result on the HD Audio bug, ehh?

Again I'm curious, how the heck did you run into these things? I mean, did you try a discrete sound card so you shut off the audio? Why did you raise the OC PLL/SFR voltages? Even using LN2 that isn't something I touch. The HDMI thing I can see why you hit that...

Wait, you didn't recognize the name? See, this is why I use same UID all over for the last decade+.

 

[Folterknecht]

...
Stock is 1.24V on my CPU, and I did mention that the board warned me that the voltage was too high.
...

I m of the opinion motherboard companies shouldnt offer these "insane" voltages as auto OC option.
The whole auto-OC topic is problematic, because it attracts those users who have no idea what they are doing. And offering voltages that high is asking for trouble - I know we dont have long time experience with these chips yet, but I ve serious doubts about 1.44V being good in the long run 24/7 under air. If you offer auto-OC it should be limited to safe settings for air cooling.

... But things like CPU overclocking... man, this platform, you set vCore, set multi, test, done.
...

Dont know what testing you did beside the things you covered in your review, but what I gathered from ~10 reviews I read/watched now, it might be benefical for performance to use also you BCLK/Cache OC with these chips.

I d be interested in a comparison between normal multi-OC (44x 100 MHz) SandyBridge Style and a combination of Multi, BCKL/Cache OC (22x 200).

Do you know if BCLK-OC will be possible on Z170 with non-K CPUs and Xeons?

 

[cadaveca]

Yeah, I think high blck might add efficiency, but time will tell. Not sure about how it will work with non-K chips.

But, the point I was trying to make was about how easy it could be.

Point taken about the auto-OC, but given the media presented, and how silly using that 11 setting is, I give MSI a pass here. The whole point was pushing to a ridiculous level, and the board immediately tells you the voltages are too high. Had it done that and not notified you...then we'd be having a different discussion.

 

[Deleted member 138597]

looks like motherboards are the things that'll push 6th Gen more than CPUs . For a person like me that uses Wolfdale, Skylake looks really lucrative than Haswell O_O

But anyway, some mistakes in the review. First in the "Test system" page. It says DDR3 RAMs & not DDR4. Second is in the "CPU performance" test. It says "mounted my Corsair H110 cooler" and not H90 (or was it H110 afterall ?)

And don't you think it's too early and sort of "making less sense" to give it 9.9? I mean CPU haven't been reviewed yet by TPU, nor there are other Z170 boards to compare. Comparing M7 against X99 or Z97 doesn't actually tell the real deal you know. Maybe 9.7 or 9.6 should have been a better and a safer score. What do you say @cadaveca?

EDIT: Oh and were you stable at 5GHz?

 

[AsRock]

I missed this review, i really think you should of noted in the title that it was a review .

Nice review btw.

 

[haswrong]

is it possible to run x16 + x4 pci-e config? (graphics + ssd)

edit: ok, skylake has 16 pcie lanes, not 20 as i previously thought. im taking my question back.

 

[cadaveca]

looks like motherboards are the things that'll push 6th Gen more than CPUs . For a person like me that uses Wolfdale, Skylake looks really lucrative than Haswell O_O

But anyway, some mistakes in the review. First in the "Test system" page. It says DDR3 RAMs & not DDR4. Second is in the "CPU performance" test. It says "mounted my Corsair H110 cooler" and not H90 (or was it H110 afterall ?)

And don't you think it's too early and sort of "making less sense" to give it 9.9? I mean CPU haven't been reviewed yet by TPU, nor there are other Z170 boards to compare. Comparing M7 against X99 or Z97 doesn't actually tell the real deal you know. Maybe 9.7 or 9.6 should have been a better and a safer score. What do you say @cadaveca?

EDIT: Oh and were you stable at 5GHz?

stable enough to run 3DMark. 100% stable, no.

You know we have an editor, right? LOL. BUt yeah, typos...I'll fix them. And no, no change to score. I do have other boards. I've have the GAMING M7 for nearly a month now, and have put it through the paces.

 

[EarthDog]

Oh......I forgot to mention, and not sure if Dave did in the review (apologies if you did), but If you want to use W7 on here, be sure to enable an option in the bios under advanced (I will post a screenshot at home) that talks about W7 and USB. Otherwise, you will not have a mouse in windows, nor upon installation of windows. However, if you use W10, it works just fine.

ASUS and EVGA really screwed the pooch and do not have that option. You are essentially forced to slipstream drivers into an install... or use W10.

 

[cadaveca]

Oh......I forgot to mention, and not sure if Dave did in the review (apologies if you did), but If you want to use W7 on here, be sure to enable an option in the bios under advanced (I will post a screenshot at home) that talks about W7 and USB. Otherwise, you will not have a mouse in windows, nor upon installation of windows. However, if you use W10, it works just fine.

ASUS and EVGA really screwed the pooch and do not have that option. You are essentially forced to slipstream drivers into an install... or use W10.

Yep, and the same applies for Win8, too. And yes, I did the review. LuLz.

It's been some time that these problems have been an issue.. the same happens with X99 at times too, depending on how the USB is wired. No driver for the controller on the install disk = no USB devices.

 

[EarthDog]

Heh, must have been lost in translation from Canadian English (ehh?) to USA English... hahaha!

I forgot to mention, and not sure if Dave did in the review (apologies if you did),

I did not ask if you did the review, but mentioned what I was saying IN the reivew.

 

[cadaveca]

Heh, must have been lost in translation from Canadian English (ehh?) to USA English... hahaha!



I did not ask if you did the review, but mentioned what I was saying IN the reivew.

Sheesh, my bad, and no, I didn't mention it. I used WIn10 for the review, but I did test out Win7, Win8, and Win 8.1 installs as well, since our cooler reviewer wants this board for his test bench, but needed certain features.

 

[Delta6326]

@cadaveca do you have a Gigabyte GA-Z170X-Gaming 7? I'm debating this MSI and the Gigabyte.

Going to have 2x m.2 Samsung EVO's and need atleast 4+ usb's

 

[EarthDog]

Sheesh, my bad, and no, I didn't mention it. I used WIn10 for the review, but I did test out Win7, Win8, and Win 8.1 installs as well, since our cooler reviewer wants this board for his test bench, but needed certain features.

Yeah, that bios option in this board really makes me feel like ASUS and EVGA screwed the pooch big time. Having to slipstream USB drives into an install disk is a PITA. BIG plus there for this board and W7 users...

 

[cadaveca]

@cadaveca do you have a Gigabyte GA-Z170X-Gaming 7? I'm debating this MSI and the Gigabyte.

Going to have 2x m.2 Samsung EVO's and need atleast 4+ usb's

Gigabyte is sending me a board, probably GAMING G1. I have a tracking number, but not a board yet.

 

[Delta6326]

@cadaveca I was looking and couldn't tell in the fan page if the fans are NOT PWN but 3pin can you still adjust them via the software with voltage control or something? Looking to get these would have them connected in groups of 3 front 2 front, 2 CPU, 1 back, do you think they can support enough juice from the fan headers or only 2 per header?

I also like the MSI M5 Looks like a M7 killer if you don't need a few "gimmicky features" and I/O cover.
And the G45 is the same as the M5 with only no Covers on the Audio chips and slightly different heatsink
M5

G45

 

[beccahoo]

Hi!
Thanks for the review and for everyone's comment.

Could you guys help me choose between and specify why?
A) Asus Z170 Pro Gaming Socket LGA1151 ATX Motherboard -
http://www.centrecom.com.au/asus-z170-pro-gaming-socket-lga1151-atx-motherboard-ddr4
B) MSI Z170A Gaming Pro LGA1151 ATX Motherboard - DDR4
http://www.centrecom.com.au/msi-z170a-gaming-pro-lga1151-atx-motherboard-ddr4

I have a friend who says MSI makes excellent motherboards and asus is known for graphics cards. Price point these two boards are the same and I don't want to spend more than the $279.

I plan to get back to gaming and have decided to buy either the i5 skylark or the i7..skipping gen 4 since it's more pricey than 6th gen skylark!

Thanks for your help! I am going to the shops in 12 hours!