Maximus XI Apex and Dimm.2 M.2 SSD speed limit clarification.

[Betroz]

Hi guys 🙂

I recently bought myself two Adata SX8200 PRO SSDs and plugged them in the Dimm.2 module. Ran a CrystalDiskMark bench and noticed that on one of the M.2 SSDs the sequential read and write performance was half of what the other SX8200 PRO had. About ~1750 MB/s. I checked my settings in the BIOS, but could not find anything wrong there. I then confirmed the performance difference with Adata SSD Toolbox, one SSD runs at Gen3 4X mode and the other at Gen3 2X.

My system specs : XI Apex with BIOS 1105, 9900K, 32GB RAM, singel 2080Ti and just one other storage device, a Samsung 860 EVO SATA SSD. Windows 10 1903 fully updated, and using Microsoft NVMe driver.

Is this caused by a BIOS bug or is it a Z390 platform limitation?

Thanks for any reply :cool:

[Ch3vr0n]

Probably a misconfiguration in the bios (rather than a bug). It's definitely not a Z390 platform limitation, at worst it's the dimm.2 implementation of asus, this is not something specified in the Z390 chipset i think.

I'm running 4 standard HDDs, 1 samsung 970 pro, 1 960 pro and 3 optical drives all at full speed.

[Betroz]

Ch3vr0n wrote:
I'm running 4 standard HDDs, 1 samsung 970 pro, 1 960 pro and 3 optical drives all at full speed.

On the XI Apex board?

[Ch3vr0n]

Betroz wrote:
On the XI Apex board?

My entire post only addresses your "is this a platform issue". It's not, as i'm using that same platform (different board though) and i'm running all that hardware. So it's not a chipset limitation. Either as i said bios misconfiguration, or at worst board limitation. It aint exactly the same board as mine 😉

[Carlyle2020]

It must be a wrong Bios setting since i get full speed on both SSDs.

Confirmed via the tool you suggested.

I checked my Bios. I didn't find something to point out except that "Hyper M.2 xxx" should be off since this is a pcie-slot extension card setting as far as i know.

If you can't find the setting please post Bios pics and i will compare.

Rgds
Carlyle

[Betroz]

Carlyle2020 wrote:
I checked my Bios. I didn't find something to point out except that "Hyper M.2 xxx" should be off since this is a pcie-slot extension card setting as far as i know.

Rgds
Carlyle

That setting is off. I tried to set it on one time, and the system would not boot. Had to reset cmos.

[XYchromosone]

Betroz wrote:
That setting is off. I tried to set it on one time, and the system would not boot. Had to reset cmos.

I Believe it would not boot with the setting set on is because it disables some sata ports which the SSD must rely on. I did something similar in my XI Extreme where I enabled 4X PCIe for the SSD M.2_1 in the motherboard. It disables 2 sata ports to get that double bandwidth. But the system would not boot so I had to revert back to 2x.

[Betroz]

XYchromosone wrote:
I Believe it would not boot with the setting set on is because it disables some sata ports which the SSD must rely on. I did something similar in my XI Extreme where I enabled 4X PCIe for the SSD M.2_1 in the motherboard. It disables 2 sata ports to get that double bandwidth. But the system would not boot so I had to revert back to 2x.

I believe I have found the cause of the problem, allthough I don't understand why it got fixed this way. Since I only have one SATA SSD in my system connected to port 1, I have had the other 5 ports disabled in the BIOS. For the sake of troubleshooting I now enabled ALL of them, ran a new CrystalDiskMark bench on the second drive, and now the performance is where it should be.

I would like Asus to explain why this is so. It would be great 🙂

[Carlyle2020]

... Isn't there a mention in the manual about ports 1, 5 and 6 sharing resources ?

[Betroz]

Carlyle2020 wrote:
... Isn't there a mention in the manual about ports 1, 5 and 6 sharing resources ?

I searched the manual now, and could not find anything about that. Anyway all the SATA ports must be enabled in BIOS for both M.2 SSDs to run at full speed in that Dimm.2 module...for some reason. One would think that disabling ports would free up lanes from the PCH, but apparently not.