Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

wjesse

macrumors member
May 14, 2019
31
23
Taiwan
Thx for the photos, could you please take a photo of the Marvell chip? The black one down from the PEX8714.
[doublepost=1557925835][/doublepost] View attachment 837109

the Marvell chip is 88SE9120
88se9120-jpg.837119
 

Attachments

  • 88SE9120.jpg
    88SE9120.jpg
    206.2 KB · Views: 834
Last edited:

tsialex

Contributor
Jun 13, 2016
13,454
13,601
the Marvell chip is 88SE9120
Thx for the photos, really appreciated. Wow, that I really don't expected.

Marvell 88SE9120 it's a PCIe 2.0 1x SATA and PATA controller.:eek:o_O

88SE9120C0-NAA2C000 PCIe 2.0 x1 to two SATA 6 Gb + one PATA ports I/O controller.

PCIe 2.0 endpoint device.
Supports one lane.
Compliant with PCIe 2.0 specifications.
Supports communication speeds of 2.5 Gbps and 5.0 Gbps.
Supports IDE programming interface registers for the SATA controller and the ATA/ATAPI controller.
Supports AHCI programming interface registers for the SATA controller. Supports aggressive power management.
Supports error reporting, recovery, and correction.
Supports Message Signaled Interrupt (MSI).

Checking the traces that connect to the Marvell controller, seems that HPT is using it basically for SPI signals, with the SATA and PATA pins disconnected. Interesting.
 
Last edited:

wjesse

macrumors member
May 14, 2019
31
23
Taiwan

w1z

macrumors 6502a
Aug 20, 2013
692
481
PEX8714 General Features 12-lane, 5-port PCIe Gen3 switch
Seems that 4 lanes go from the PEX8747 to the PEX8714, but from there the lanes change to vias.

Still can't see the Marvell controller inscriptions, this is the one that I want to know what is the function on SSD7102. Manual from SSD7102 makes no mention whatsoever about the hardware itself.

Best guess would be the second switch, PEX8714, is being utilized for background raid and trimming operations.

Edit: just read the amazon reviews for the SSD7102 - manufacturer response to the differences (hardware) between 7101A and 7102 is

The SSD7101A-1 and 7102 uses the same PLX chipset on the board however the SSD7102 has an extra EFI chipset.
 
Last edited:

tsialex

Contributor
Jun 13, 2016
13,454
13,601
Best guess would be the second switch, PEX8714, is being utilized for background raid and trimming operations.
Possible, but what controller is doing it? The Marvell or the Nuvuton? The switch itself don't do management operations.
 

w1z

macrumors 6502a
Aug 20, 2013
692
481
Possible, but what controller is doing it? The Marvell or the Nuvuton? The switch itself don't do management operations.

Best guess again would be that they're using the PEX8714 for handling traffic between the main switch and Marvell/Nuvuton controllers so raid/trimming traffic won't impact the performance of the main switch.
 

tsialex

Contributor
Jun 13, 2016
13,454
13,601
Best guess again would be that they're using the PEX8714 for handling traffic between the main switch and Marvell/Nuvuton controllers so it won't impact the performance of the main switch.
One thing of note, SSD7102 BOM would be at least $60 or more over the BOM for SSD7101-A, PEX8714-A costs $543,46 for 10 units on DigiKey and we still have the Marvell controller.

HighPoint made some interesting choices here, but nothing that would make me buy a SSD7102 over a SSD7101-A.
[doublepost=1557943220][/doublepost]Btw, my statement is valid for MP5,1 use. If you have a PC or plan to upgrade to one in the future, SSD7102 probably is a better fit.
 

w1z

macrumors 6502a
Aug 20, 2013
692
481
One thing of note, SSD7102 BOM would be at least $60 or more over the BOM for SSD7101-A, PEX8714-A costs $543,46 for 10 units on DigiKey and we still have the Marvell controller.

HighPoint made some interesting choices here, but nothing that would make me buy a SSD7102 over a SSD7101-A.

Couldn't agree more. Also, HighPoint has stated that they don't/won't support the SSD7102 on macOS, only the SSD7101-A.

The SSD7102 does not support Mac systems.
Please use the SSD7101A-1 if you wish to use it in a macOS.
 
  • Like
Reactions: JedNZ

tsialex

Contributor
Jun 13, 2016
13,454
13,601
Couldn't agree more. Also, HighPoint has stated that they don't/won't support the SSD7102 on macOS, only the SSD7101-A.
HighPoint statements should be taken with a truck of salt. People already have SSD7102 working here with MP5,1. They talked the same for SSD7101-A in the earlier days.
 

w1z

macrumors 6502a
Aug 20, 2013
692
481
HighPoint statements should be taken with a truck of salt. People already have SSD7102 working here with MP5,1. They talked the same for SSD7101-A in the earlier days.
People using and having it working on macs is one thing, Highpoint providing support, drivers, web mgmt software for it on macOS or to Mac users is another thing which was the point of my post and not that it can't be used in macOS.

So to all those who are using the 7102 on Mac Pros/macOS, don't be disappointed if Highpoint declines to provide support.
 
  • Like
Reactions: ClaudAA

tsialex

Contributor
Jun 13, 2016
13,454
13,601
People using and having it working on macs is one thing, Highpoint providing support, drivers, web mgmt software for it on macOS or to Mac users is another thing which was the point of my post and not that it can't be used in macOS.

So to all those who are using the 7102 on Mac Pros/macOS, don't be disappointed if Highpoint declines to provide support.
Sure, but it’s already proved that HighPoint support is totally non-existent the moment that you mention Mac.
 
  • Wow
Reactions: ClaudAA

w1z

macrumors 6502a
Aug 20, 2013
692
481
Sure, but it’s already proved that HighPoint support is totally non-existent the moment that you mention Mac.
Non-existent if you are trying to boot a macOS volume on the 7101A, which they stated they don't support despite the card booting macOS on our macs, but for macos data volumes and other hardware related issues they're responsive. That's my experience.
 

tsialex

Contributor
Jun 13, 2016
13,454
13,601
Non-existent if you are trying to boot a macOS volume on the 7101A, which they stated they don't support despite the card booting macOS on our macs, but for macos data volumes and other hardware related issues they're responsive. That's my experience.
Not to be not-picking, but they denied DOA replacement for at least two MP5,1 users here, there’s a thread about that.
 

Mac_User 0101

macrumors regular
Oct 8, 2017
133
43
you mean WebGUI - NVMe Manager?
http://highpoint-tech.com/USA_new/series-ssd7102-download.htm

I think that's for windows, never try it.

But I tried macOS Software Package for SSD7101A, never work.
http://highpoint-tech.com/USA_new/series-ssd7101a-1-download.htm
I stand corrected. The generic info for "NVMe RAID Solutions" states the NVMe Manager is for Windows, Linux and Mac OS and shows the symbol for all 3 operating systems but there is only Windows and Linux currently available for download on the 7102 page. As those guys said earlier, Highpoint isn't supporting Mac with this particular card. I was going to tell you to try the one for SSD7101-A. You say you did that and didn't work? Did the web GUI actually start up and just not detect the card? You probably need to install the Mac driver for 7101-A to enable the NVMe manager GUI to see the card. I would think that it's all the same NVMe manager and driver for the entire 7100 series but maybe not. If you get a minute to test that I'd love to know the results. I have a RocketStor enclosure from them and there is a Mac OS web GUI for that too but you need to install the driver. They are generic for the entire RS61xxV series. The web GUI gives lots of options, statistics and advanced RAID levels.

http://highpoint-tech.com/USA_new/series-ssd7101a-1-download.htm

http://highpoint-tech.com/USA_new/nvme-storage-solution.htm
 
Last edited:

joevt

macrumors 604
Jun 21, 2012
6,966
4,260
Best guess would be the second switch, PEX8714, is being utilized for background raid and trimming operations.

Edit: just read the amazon reviews for the SSD7102 - manufacturer response to the differences (hardware) between 7101A and 7102 is "The SSD7101A-1 and 7102 uses the same PLX chipset on the board however the SSD7102 has an extra EFI chipset."
<guess>The PEX8714 is to connect the Marvell chip, which doesn't do anything except appear as a SATA controller with an EFI PCIe option rom driver. The EFI driver acts as a SATA storage driver, probably ignores the Marvell chip, and talks to the NVMe devices, thus providing boot support. Maybe the EFI driver doesn't work with a Mac's EFI.</guess>

What does this card look like using "lspci", or Windows Device Manager, or ioreg?
 

wjesse

macrumors member
May 14, 2019
31
23
Taiwan
I was going to tell you to try the one for SSD7101-A. You say you did that and didn't work? Did the web GUI actually start up and just not detect the card? You probably need to install the Mac driver for 7101-A to enable the NVMe manager GUI to see the card.

When I got SSD7102, I tried the SSD7101-A web GUI/driver for MacOS, didn't work, but maybe just like you said, probably I missed something.
I will give it one more try tomorrow.

I used to boot to Windows 10 via Boot Camp for test Windows driver and RAID 0 array, but there's an issue stopped me, it can't find my Windows boot drive when SSD7102 was plug in PCIE slot. I do have to test this again and report it here.
 

Mac_User 0101

macrumors regular
Oct 8, 2017
133
43
*Deleted.... Please see post by tsialex below
 

Attachments

  • Linux_Driver_Failed_to_connect_to_WebGUI(SSD7100)_18_12_12.pdf
    822.3 KB · Views: 544
Last edited:

tsialex

Contributor
Jun 13, 2016
13,454
13,601
#lspci | grep 8747

02:00.0 PCI bridge: PLX Technology, Inc. PEX 8747 48-Lane, 5-Port PCI Express Gen 3 (8.0 GT/s) Switch (rev ca)
03:08.0 PCI bridge: PLX Technology, Inc. PEX 8747 48-Lane, 5-Port PCI Express Gen 3 (8.0 GT/s) Switch (rev ca)
03:10.0 PCI bridge: PLX Technology, Inc. PEX 8747 48-Lane, 5-Port PCI Express Gen 3 (8.0 GT/s) Switch (rev ca)
06:00.0 PCI bridge: PLX Technology, Inc. PEX 8747 48-Lane, 5-Port PCI Express Gen 3 (8.0 GT/s) Switch (rev ca)
07:08.0 PCI bridge: PLX Technology, Inc. PEX 8747 48-Lane, 5-Port PCI Express Gen 3 (8.0 GT/s) Switch (rev ca)
07:10.0 PCI bridge: PLX Technology, Inc. PEX 8747 48-Lane, 5-Port PCI Express Gen 3 (8.0 GT/s) Switch (rev ca)
This is not useful, you have to show the full lspci, since the controllers will have different PCIe IDs and you are doing grep for just the PEX-8747, SSD7102 has another PCIe switch, PEX-8714.

Use something like this, change the 5:* and 6:* to the correct PCIe device domains:

Code:
sudo lspci -vvx -s 5:* >SSD7102.vvx.txt; sudo lspci -vvx -s 6:* >>SSD7102.vvx.txt
 
Last edited:

Mac_User 0101

macrumors regular
Oct 8, 2017
133
43
This is not useful, you have to show the full lspci, since the controllers will have different PCIe IDs and you are doing grep for just the PEX-8747, SSD7102 has another PCIe switch, PEX-8714.

Use something like this, change the 5:* and 6:* to the correct PCIe IDs:

Code:
sudo lspci -vvx -s 5:* >SSD7102.vvx.txt; sudo lspci -vvx -s 6:* >>SSD7102.vvx.txt
Thank you. I just posted from the Highpoint support info via a web search. Hopefully wjesse will run your command and post the results for the thread to view.
 

handheldgames

macrumors 68000
Apr 4, 2009
1,943
1,170
Pacific NW, USA
<guess>The PEX8714 is to connect the Marvell chip, which doesn't do anything except appear as a SATA controller with an EFI PCIe option rom driver. The EFI driver acts as a SATA storage driver, probably ignores the Marvell chip, and talks to the NVMe devices, thus providing boot support. Maybe the EFI driver doesn't work with a Mac's EFI.</guess>

What does this card look like using "lspci", or Windows Device Manager, or ioreg?

Good guess. Perhaps This is how they are achieving boot support for NVME raid 0? Plugging the 7102 into a modern PC would help clarify this mystery.
 

wjesse

macrumors member
May 14, 2019
31
23
Taiwan
This is not useful, you have to show the full lspci, since the controllers will have different PCIe IDs and you are doing grep for just the PEX-8747, SSD7102 has another PCIe switch, PEX-8714.

Use something like this, change the 5:* and 6:* to the correct PCIe device domains:

Code:
sudo lspci -vvx -s 5:* >SSD7102.vvx.txt; sudo lspci -vvx -s 6:* >>SSD7102.vvx.txt
Thank you. I just posted from the Highpoint support info via a web search. Hopefully wjesse will run your command and post the results for the thread to view.

Thank you, I'll try later.

Let's focus on SSD7102's raid array function first, here shows System Report of my cMP.
Now I have installed SSD7101A driver for MacOS in another boot drive, but as you can see 'IDE controller' and 'Raid controller' shows no driver installed, is it normal? IS your real SSD7101A shows like that?
7102driver-jpeg.837434
 

Attachments

  • 7102Driver.jpeg
    7102Driver.jpeg
    1.8 MB · Views: 619
Last edited:

tsialex

Contributor
Jun 13, 2016
13,454
13,601
Thank you, I'll try later.

Let's focus on SSD7102's raid array function first, here shows System Report of my cMP.
Now I have installed SSD7101A driver for MacOS in another boot drive, but as you can see 'IDE controller' and 'Raid controller' shows no driver installed, is it normal? IS your real SSD7101A shows like that?

40898770413_0c77bc8902_o.jpg
No, this two PCIe devices are the controllers that SSD7101-A don’t have. I’m not sure if the Marvell has anything with 0x7102 device ID, but 0x91a4 I’m sure.

0x7102 could be a virtual device, created by HPT to provide the RAID.
 

joevt

macrumors 604
Jun 21, 2012
6,966
4,260
No, this two PCIe devices are the controllers that SSD7101-A don’t have. I’m not sure if the Marvell has anything with 0x7102 device ID, but 0x91a4 I’m sure.

0x7102 could be a virtual device, created by HPT to provide the RAID.
I think it's two functions (0 and 1) of the same device (16,0). Only three M.2 devices are connected. One is AHCI and the other are NVMe. The bus numbers of the Slot-2 devices seems strange - I would expect that the M.2 devices would use consecutive bus numbers. The device tree may be interesting. I think the following command should give all the information about Slot-2 devices (PXS2 is the name for slot 2 on my MacPro3,1):
Code:
ioreg -fiw0 -n PXS2 -rl > ioregslot2.txt
 

tsialex

Contributor
Jun 13, 2016
13,454
13,601
I think it's two functions (0 and 1) of the same device (16,0). Only three M.2 devices are connected. One is AHCI and the other are NVMe. The bus numbers of the Slot-2 devices seems strange - I would expect that the M.2 devices would use consecutive bus numbers. The device tree may be interesting. I think the following command should give all the information about Slot-2 devices (PXS2 is the name for slot 2 on my MacPro3,1):
Code:
ioreg -fiw0 -n PXS2 -rl > ioregslot2.txt

Device (16,0) could be the ID of the PEX-8314, where 0x7102 and 0x91a4 are connected, no?
 

wjesse

macrumors member
May 14, 2019
31
23
Taiwan
Big issue!
SSD7101-A's MacOS driver works on SSD7102, I use "HighPoint RAID Management" to create array, guess what?
Awesome!
Create Raid0 Array.jpg


In Disk Utility, you can find a external drive model name listed as HPT DISK 3_0 Media, then I erase/format it and named as SamsungRaid0, no issue.
Disk Utility.jpg


I test R/W speed of the SamsungRaid0 drive, the performance seems great, maybe the same as SSD7101-A.
HardwareRaid0 Speed test.jpg

HardwareRaid0 Speed test2.jpg



Then I use CCC to clone my MacOS High Sierra 10.13.6 boot drive to this SamsungRaid0 drive and reboot from it, succeed!
Raid0 Boot.jpg


I try to boot from other drive without install SSD7101-A driver, the SamsungRaid0 drive is not work and disappear in Disk Utility.
And more, I clone a boot drive with no SSD7101-A driver to the SamsungRaid0 drive, it is not bootable.
So I think maybe clean installation of MacOS on SSD7102 array drive and boot from it is not useful, because of no SSD7101-A driver on it.

That's my test report.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.