MSI B550M Mortar Motherboard - a few test results

[Buftor]

I finally received this MB a week ago. I wanted to share a few observations that may help others.

My storage is ADATA SX8200 Pro 1TB NVME SSD. Since I plan on adding another M.2 drive in the future, I was curious to see how the secondary M.2 slot performs. The secondary slot is connected to the chipset, and not directly to the CPU. I ran CrystalDiskMark 7 and received best results as follows:

[Read]
Sequential 1MiB (Q= 8, T= 1): 3064.818 MB/s [ 2922.8 IOPS] < 2735.79 us>
Sequential 1MiB (Q= 1, T= 1): 2687.179 MB/s [ 2562.7 IOPS] < 390.03 us>
Random 4KiB (Q= 32, T=16): 1341.326 MB/s [ 327472.2 IOPS] < 1562.22 us>
Random 4KiB (Q= 1, T= 1): 65.486 MB/s [ 15987.8 IOPS] < 62.41 us>

[Write]
Sequential 1MiB (Q= 8, T= 1): 2575.017 MB/s [ 2455.7 IOPS] < 3252.73 us>
Sequential 1MiB (Q= 1, T= 1): 2358.253 MB/s [ 2249.0 IOPS] < 444.40 us>
Random 4KiB (Q= 32, T=16): 1129.526 MB/s [ 275763.2 IOPS] < 1854.98 us>
Random 4KiB (Q= 1, T= 1): 192.614 MB/s [ 47024.9 IOPS] < 21.14 us>

I then moved the ADATA drive to the primary M.2 slot and received the following results:

[Read]
Sequential 1MiB (Q= 8, T= 1): 3441.476 MB/s [ 3282.0 IOPS] < 2435.91 us>
Sequential 1MiB (Q= 1, T= 1): 2674.762 MB/s [ 2550.9 IOPS] < 391.51 us>
Random 4KiB (Q= 32, T=16): 1332.619 MB/s [ 325346.4 IOPS] < 1517.54 us>
Random 4KiB (Q= 1, T= 1): 58.306 MB/s [ 14234.9 IOPS] < 69.95 us>

[Write]
Sequential 1MiB (Q= 8, T= 1): 3313.708 MB/s [ 3160.2 IOPS] < 2525.21 us>
Sequential 1MiB (Q= 1, T= 1): 2788.109 MB/s [ 2658.9 IOPS] < 375.49 us>
Random 4KiB (Q= 32, T=16): 1368.005 MB/s [ 333985.6 IOPS] < 1520.07 us>
Random 4KiB (Q= 1, T= 1): 172.027 MB/s [ 41998.8 IOPS] < 23.62 us>

Obviously, the secondary M.2 slot does not reach the full potential of the drive, even though both drive and slot are PCI-E 3 gen 4 lane. Maybe this is because the B550 chipset is connected to the CPU using PCI-E 3 gen (I am a newbie, I'm just guessing here). Maybe also that's why MSI implements a switch between the secondary M.2 slot and the secondary PCI-E expansion slot, because if both worked together, neither would perform very well. Again, I'm just guessing.

Another quick test that I ran was to monitor the temperature of the SSD drive in the primary slot with and without the MSI provided heatsink. I ran the CrystalDiskMark test twice without a break, and monitored the temp using CrystalDiskInfo. Since this app only updates the temp every minute, I kept refreshing it every second to get more accurate results. The results were as follows:

Without heat sink: 71C degrees max
With heatsink: 53C degrees max

This test may not be stressing the drive long enough, but I think it shows that the heat sink helps.

The rest of my setup is Ryzen 5 3600, MSI GeForce 1650 Super and Ballistix DDR4-3200 32GB. In addition to adding more storage, I also plan to upgrade the CPU at some stage. After assembly, the system boot without a problem, and I just applied the memory XMP profile. So far it's been completely stable and quiet. I will attempt to OC and see how that goes.

Thanks to everyone on this forum that gave good advice and shared a lot of knowledge (especially TheLostSwede).

 

[Hemmingstamp]

Nice, and thanks for the tests. I have my eyes on the MAG B550 TOMAHAWK, MORTAR WI-FI before that.
Still torn between them and an X570 board. I've read so many reviews my eyes have melted,

 

[Vya Domus]

You're going to experience this with any board, the drives that go through the chipset will always be inevitably slower.

 

[Hemmingstamp]

You're going to experience this with any board, the drives that go through the chipset will always be inevitably slower.

I don't even bother benching drives these days, can't be arsed. I mean, how fast do we want these things to go???
Still good of the OP to post the results though.

 

[Buftor]

Nice, and thanks for the tests. I have my eyes on the MAG B550 TOMAHAWK, MORTAR WI-FI before that.
Still torn between them and an X570 board. I've read so many reviews my eyes have melted,

I actually got the Mortar WiFi. It's only $10 more for Wifi 6 and Bluetooth. I already had an older mATX case and 400W power supply, so I was limited by the size of the board and the total amount of power. I actually ordered Asrock B450M Steel Legend ($90) then Gigabyte B550M Aorus Pro ($130) then the Mortar WiFi ($170). At first, when the B550 boards came out, I was shocked by the price hike, compared to B450. But the more I read, the more I understood why they were more expensive.

If you are planning to have more than one NVME drive in your system, I suggest you check again the X570 boards. I think the B550 Tomahawk also has the switch between the secondary M.2 and secondary expansion slot. This may or may not be an issue for you.

 

[Hemmingstamp]

I was looking at the Mortar WiFi but then the Tomahawk was on show for only 30 sheets more.
Not many reviews where these boards have been hammered so I'm torn between them.
I'm not keen on the lower end MSI X570 boards, some look poor and not all reviews are good.
Decison - decisions......

 

[Buftor]

I was looking at the Mortar WiFi but then the Tomahawk was on show for only 30 sheets more.
Not many reviews where these boards have been hammered so I'm torn between them.
I'm not keen on the lower end MSI X570 boards, some look poor and not all reviews are good.
Decison - decisions......

Hardware Unboxed channel on YouTube reviewed all of these boards (you may have seen it already). Both the Mortar and Tomahawk will OC the top Ryzen easily, but Tomahawk runs cooler. Also, if it's within your budget, maybe you should look at the X570 Tomahawk.

 

[tabascosauz]

I was looking at the Mortar WiFi but then the Tomahawk was on show for only 30 sheets more.
Not many reviews where these boards have been hammered so I'm torn between them.
I'm not keen on the lower end MSI X570 boards, some look poor and not all reviews are good.
Decison - decisions......

If you can fit the Tomahawk for a little bit more $ there's no reason to limit yourself to the Mortar. Especially if you want to upgrade to Vermeer and make use of its inevitably better IF capabilities (by extension, higher 1:1 DRAM speeds), you'll want the 6-layer Tomahawk for the memory alone. MSI only certifies 2 stick dual-rank RAM up to 3733 on the 4-layer Mortars. All the mATX boards - Steel Legend, Aorus Pro, TUF, and Mortar are all 4-layer.

Mortar also has the single most questionable Clear CMOS header positioning I've ever seen in any desktop computer. It would be fine if the Mortar had a Clear button on the rear I/O, but it doesn't, so better hope you never need to touch the inside of the PC again after you build it. The Tomahawk isn't much better, but as long as you have a 2-slot GPU, you can still hit it with a screwdriver easily without having to remove the GPU like you would on the Mortar.

Tomahawk and Aorus Pro are both pretty solid ATX boards.

 

[Hemmingstamp]

Hardware Unboxed channel on YouTube reviewed all of these boards (you may have seen it already). Both the Mortar and Tomahawk will OC the top Ryzen easily, but Tomahawk runs cooler. Also, if it's within your budget, maybe you should look at the X570 Tomahawk.

I usually look at Der Baur's reviews since he carries out extensive testing, but he seems quiet at present, on his website at least and hasn't tested any B550 boards to date.
He tested a batch of X570 boards a while back, a few of the MSI X570 boards failed miserably on the VRM tests and weren't recommended.

 

[tabascosauz]

[Read]
Sequential 1MiB (Q= 8, T= 1): 3064.818 MB/s [ 2922.8 IOPS] < 2735.79 us>
Sequential 1MiB (Q= 1, T= 1): 2687.179 MB/s [ 2562.7 IOPS] < 390.03 us>
Random 4KiB (Q= 32, T=16): 1341.326 MB/s [ 327472.2 IOPS] < 1562.22 us>
Random 4KiB (Q= 1, T= 1): 65.486 MB/s [ 15987.8 IOPS] < 62.41 us>

[Write]
Sequential 1MiB (Q= 8, T= 1): 2575.017 MB/s [ 2455.7 IOPS] < 3252.73 us>
Sequential 1MiB (Q= 1, T= 1): 2358.253 MB/s [ 2249.0 IOPS] < 444.40 us>
Random 4KiB (Q= 32, T=16): 1129.526 MB/s [ 275763.2 IOPS] < 1854.98 us>
Random 4KiB (Q= 1, T= 1): 192.614 MB/s [ 47024.9 IOPS] < 21.14 us>

That seems like a win, if anything, for the chipset slot. The low queue depth Random is what you'll feel in daily performance. The sequential loss you'll hardly feel in the rare instance you're actually leveraging sequential performance.

 

[Max(IT)]

Another quick test that I ran was to monitor the temperature of the SSD drive in the primary slot with and without the MSI provided heatsink. I ran the CrystalDiskMark test twice without a break, and monitored the temp using CrystalDiskInfo. Since this app only updates the temp every minute, I kept refreshing it every second to get more accurate results. The results were as follows:

Without heat sink: 71C degrees max
With heatsink: 53C degrees max

This test may not be stressing the drive long enough, but I think it shows that the heat sink helps.

wow 71ºC even without the heat sink seem to be quite high

 

[kn00tcn]

wow 71ºC even without the heat sink seem to be quite high

106C? https://www.techpowerup.com/review/crucial-p5-1tb-m-2-nvme-ssd/7.html

 

[Max(IT)]

106C? https://www.techpowerup.com/review/crucial-p5-1tb-m-2-nvme-ssd/7.html

and do you think that's normal ???

Never seen my SSDs (I have 3 of them in two PCs, 2 NVMe and 1 SATA) above 50°

 

[kn00tcn]

and do you think that's normal ???

Never seen my SSDs (I have 3 of them in two PCs, 2 NVMe and 1 SATA) above 50°

i dont think 71 without heatsink at GB/s speeds during a benchmark is quite high, a lot of computer chips reach 80s or 90s by design

106 without heatsink is what i would say 'quite high'

one problem, OP didnt state case or ambient temp (or if the drive has direct airflow)

 

[Buftor]

It was tested with the case open, no direct airflow. Ambient temp was 21C.