MSI MEG Z490I Unify

[Black Haru]

MSI has stuffed a lot of power into a small package with the MSI MEG Z490I Unify. Featuring a stealthy black aesthetic and a VRM configuration with 90 A power stages, the MSI MEG Z490I Unify could be a top choice for overclockers. Just how much can MSI pack into the Mini-ITX footprint?

Show full review

 

[lemkeant]

Thanks for the write up. For those of us who are older than 25, SFF, ITX shopping, No RGB is a total positive in my mind, not a negative

 

[duynguyenle]

Thanks for the review. Looks like a competent board, however I was interested in this board before, but have some gripes with MSI's implementation of features so I probably will wait for review of other Z490 boards as well.

That said, I found some small errors on the front page and spec sheet of the review:

• CPU Power in page 1 is listed as 9 phases in the specs table, but the main paragraph says 8 phases
• In the same table, Memory section is listed as 2x DIMM but somehow has support for quad channel memory (copy paste error?)

Thanks for including high resolution photos of the board, I appreciate being able to see where things are and how the connectors are arranged (especially relevant for ITX boards where potential buyers may have size/position constrants for board connectors/SATA ports etc...)

Also the inclusion of long duration VRM temperature testing is quite useful. I hope future board reviews on this site continues offering this test.

 

[BSim500]

Excellent review. I hope that Audio Quality, Fan Control, Storage benchmarks, etc, page remain permanent for future motherboard reviews. It really does make TechPowerUp one of the most in-depth sites.

Just one typo though - on the conclusion page you have "No RGB" under "Negative" (thumbs down column). That should be under "Positive" (thumbs up column). Thanks.

 

[RandallFlagg]

What are the specs on the Z390 comparison systems? These reviews show things like Blender getting a 30% improvement on the Z490 10900K boards vs the Z390, but I find no specs on the Z390s in the comparison. I would assume it is an 9900K based on some of the other benchmarks, but 30% is a lot and unexpected here.

 

[EarthDog]

What are the specs on the Z390 comparison systems? These reviews show things like Blender getting a 30% improvement on the Z490 10900K boards vs the Z390, but I find no specs on the Z390s in the comparison. I would assume it is an 9900K based on some of the other benchmarks, but 30% is a lot and unexpected here.

25% more cores and threads + increased clock speeds is my guess.

 

[Black Haru]

Thanks for the review. Looks like a competent board, however I was interested in this board before, but have some gripes with MSI's implementation of features so I probably will wait for review of other Z490 boards as well.

That said, I found some small errors on the front page and spec sheet of the review:

• CPU Power in page 1 is listed as 9 phases in the specs table, but the main paragraph says 8 phases
• In the same table, Memory section is listed as 2x DIMM but somehow has support for quad channel memory (copy paste error?)

Thanks for including high resolution photos of the board, I appreciate being able to see where things are and how the connectors are arranged (especially relevant for ITX boards where potential buyers may have size/position constrants for board connectors/SATA ports etc...)

Also the inclusion of long duration VRM temperature testing is quite useful. I hope future board reviews on this site continues offering this test.

CPU power is accurate, in that it also includes SOC (iGPU) phase. this reflects in the other reviews as well.

What are the specs on the Z390 comparison systems? These reviews show things like Blender getting a 30% improvement on the Z490 10900K boards vs the Z390, but I find no specs on the Z390s in the comparison. I would assume it is an 9900K based on some of the other benchmarks, but 30% is a lot and unexpected here.

Z390 testing used a 9900k, and the latest BIOS for each board (as of 3 or so weeks ago when I did the back testing). Everything else is exactly the same.

 

[duynguyenle]

What are the specs on the Z390 comparison systems? These reviews show things like Blender getting a 30% improvement on the Z490 10900K boards vs the Z390, but I find no specs on the Z390s in the comparison. I would assume it is an 9900K based on some of the other benchmarks, but 30% is a lot and unexpected here.

There's nothing unexpected here, Blender scales basically perfectly with core/threads and very well with frequency (though not quite perfectly).

With the 10900K packing 25% more core/threads and each of those core running at a higher clock, 30% is around where I'd expect the performance advantage to be.

 

[RandallFlagg]

There's nothing unexpected here, Blender scales basically perfectly with core/threads and very well with frequency (though not quite perfectly).

With the 10900K packing 25% more core/threads and each of those core running at a higher clock, 30% is around where I'd expect the performance advantage to be.

Right. So no thermal throttling on 10th gen i9 despite having 25% more cores and higher frequency on same process tech.

 

[Assimilator]

@Black Haru not sure how long it's been like this, but the AIDA64 memory tests are living on the "CPU performance" page - maybe you should rename that page to "General performance" or similar?

Right. So no thermal throttling on 10th gen i9 despite having 25% more cores and higher frequency on same process tech.

There isn't going to be any thermal throttling at stock clocks with the test cooling setup that Haru is using (custom 360mm loop).

 

[EarthDog]

Right. So no thermal throttling on 10th gen i9 despite having 25% more cores and higher frequency on same process tech.

What are you rambling on about?

 

[duynguyenle]

Right. So no thermal throttling on 10th gen i9 despite having 25% more cores and higher frequency on same process tech.

I know it's popular to bash on Intel at the moment, but in this particular case it's very simple. Thermal throttling happens when the cooling solution at hand is incapable of dissipating the thermal output of the CPU. For Intel CPUs, this is typically done on the basis of a temperature threshold (modern AMD CPUs also take into account power draw amongs other things).

If the cooling solution being used can adequately dissipate the thermal load, then you're not going to see thermal throttling, it's as simple as that. Based on the Intel documentation so far, the die packaging itself also saw some updates to make sure the heat load is dissipated away from the chip more efficiently by die itself being slimmer compared to previous gen CPUs.

Thermal throttling is not some sort of black magic (i.e. PACKING MORE CORES USING THE SAME PROCESS NODE MEANS IT MUST BE THROTTLING!!!!). It's all down to physics. Having more cores in the same process node means that the physical die is larger so you've got a bit more surface area to interface with the heatspreader, offers a little bit of offset. They've made the die thinner, and could also be binning the higher end SKUs more aggressively for low leakage. We don't know how true this is yet until there are more samples.

It's understandable to be skeptical, and I encourage you to read more reviews and compare the results from other publications in order to ascertain whether 30% performance bump in this specific CPU workload is to be expected or is an outlier (based on the reviews I've read so far, a 30% bump in well scalable workloads like Blender is well within the realms of possibility

 

[RandallFlagg]

I know it's popular to bash on Intel at the moment, but in this particular case it's very simple. Thermal throttling happens when the cooling solution at hand is incapable of dissipating the thermal output of the CPU. For Intel CPUs, this is typically done on the basis of a temperature threshold (modern AMD CPUs also take into account power draw amongs other things).

If the cooling solution being used can adequately dissipate the thermal load, then you're not going to see thermal throttling, it's as simple as that. Based on the Intel documentation so far, the die packaging itself also saw some updates to make sure the heat load is dissipated away from the chip more efficiently by die itself being slimmer compared to previous gen CPUs.

Thermal throttling is not some sort of black magic (i.e. PACKING MORE CORES USING THE SAME PROCESS NODE MEANS IT MUST BE THROTTLING!!!!). It's all down to physics. Having more cores in the same process node means that the physical die is larger so you've got a bit more surface area to interface with the heatspreader, offers a little bit of offset. They've made the die thinner, and could also be binning the higher end SKUs more aggressively for low leakage. We don't know how true this is yet until there are more samples.

It's understandable to be skeptical, and I encourage you to read more reviews and compare the results from other publications in order to ascertain whether 30% performance bump in this specific CPU workload is to be expected or is an outlier (based on the reviews I've read so far, a 30% bump in well scalable workloads like Blender is well within the realms of possibility

You got me wrong. My comment was sarcasm about no thermal throttling in response to all the negativity to intel and its purported heat generation simply due to something most (including me) don't fully understand (7nm vs 14nm). It turns out that this is not really the case, gen 10 10900K transfers heat more efficiently and thus gets full use of its extra 2 cores and 4 threads.

Having said that, I'm actually more interested in the 600/600K, and 700/700K.

 

[Turmania]

I could just buy this board, for the looks alone. Put it inside a glass shelf to exhibit tge inner geek in me even if I won't use an Intel cpu. Well done MSI.

 

[Hinnerk]

The MSI MEG Z490I Unify does include the ever-important post code display, as well as BIOS flashback.

Where?

 

[Black Haru]

Where?

Fixed! The Z490I does not have either.

 

[Upgrayedd]

Could someone please explain why the ASRock boards are getting 10-20fps more in BFV than the MSI boards? Thanks

 

[John Naylor]

Could someone please explain why the ASRock boards are getting 10-20fps more in BFV than the MSI boards? Thanks

Have seen that in the past whereby vendor sneals in a BIOS tweak where BCLK Frequency gets a teeny boost or something along those lines usually resulting in charges of 'cheating" by not dosclosing it or the tester, not checking it.

CPU Power in page 1 is listed as 9 phases in the specs table, but the main paragraph says 8 phases

8 for the VRM + 1 for SOC

 

[Upgrayedd]

Have seen that in the past whereby vendor sneals in a BIOS tweak where BCLK Frequency gets a teeny boost or something along those lines usually resulting in charges of 'cheating" by not dosclosing it or the tester, not checking it.



8 for the VRM + 1 for SOC

Yeah I've noticed it before. Would like to see someone with the resources to confirm it.

 

[arjun90]

Does this motherboard support 2x NVME and 4x SATA Ports to run concurrently without compromise to performance on one of the SATA ports? Some of the other brands either disable one of the SATA ports or lowers the read/write speed if both NVME slots are occupied.

 

[EarthDog]

Yeah I've noticed it before. Would like to see someone with the resources to confirm it.

Confirmed.

The back and forth of how it is counted is confusing....

Does this motherboard support 2x NVME and 4x SATA Ports to run concurrently without compromise to performance on one of the SATA ports? Some of the other brands either disable one of the SATA ports or lowers the read/write speed if both NVME slots are occupied.

Read the manual or check the specifications page.

Though I'm surprised the review doesn't mention this......

 

[The Mad Monk]

The choice in Z490 ITX boards is quite limited and if you want Thunderbolt it is the MSI board or the ASRock Z490 Phantom Gaming ITX/TB3 - I went for this one.

Some points:
- if buying as an upgrade be aware the v1.0 firmware does not support "legacy" setups (i.e DOS partitions or MBR boot). it is GPT/UEFI only
- There is a firmware update on the MSI site which adds the CSM for legacy boot but it might do weird stuff (like not letting you into the BIOS settings after enabling CSM)

- If you run Linux be aware that the current kernel (as of 5.8 RC2) only supports the RTL8125A, this board has the RTL8215B
- as long as you don't mind compiling it yourself RTL has a driver on their web site which works just fine.

Otherwise no complaints so far, it could have done with more USB ports but the space on the back panel for another connector stack is taken by the teeny tiny fan for cooling the VRMs

 

[Drachenherz]

Great review, thank you.
A quick question, though: is it possible to install a CPU tower cooler like the Dark Rock 4 pro with dual fans? How do they connect to the mainboard so that fanspeed of both CPU fans can be controlled?

edit: I feel so stupid now... there‘s a Y-Splitter cable in the package of the Cooler... so, question answered.

 

[wanderer]

I just started setting up my system using this motherboard. I did not find any m.2 screws. If using the plate that covers the nvme drive, is a screw not required? Did anyone find m.2 screws included with this motherboard? Thanks

 

[duynguyenle]

I just started setting up my system using this motherboard. I did not find any m.2 screws. If using the plate that covers the nvme drive, is a screw not required? Did anyone find m.2 screws included with this motherboard? Thanks

There should be a small baggie in the motherboard box with a standoff and a M2 machine screw (not to be confused with M.2). It's easy to miss, they're tiny)

If you can't find it, probably time to contact MSI customer service and they'll sort you out (or grab a generic M.2 mounting kit off Amazon or ebay or something, if you're impatient)