Short Review: Edging ASMedia 1166 PCIe Gen3 x2 M.2 to 6 x SATA HBA Chipset. It doesn't suck 👍

This is why I use ASM1166 PCIe cards rather than M.2s. Still doesn’t look like anybody’s done an armored implementation, though over the past year there’s been a shift to the reinforced NGFF versions discussed upthread in the listings I see.

An NGFF size heatsink’s too small to do much, though aBav’s shown it can be enough with dedicated airflow on unreinforced implementations.

Still not something I’d really want to put a bunch of expensive drives off of if it’s going to get much of a workload, especially as on the ASM1166s I’ve worked with more of the heat dissipation’s still through the board than through the larger heatsink. NGFF reinforcement doesn’t help there as it increases backside thermal resistance.

Doesn’t appear to be any quantification of the NGFF thermal penalty so, personally, I’d prefer to put thermocouples in if doing anything significant.

1 Like

With some soldering it should be possible to access the UART interface on the Radxa board. Pins 17 and 18 on the ASM1166 controller:

From ASM1166 datasheet


1 Like

There is an acrylic back board:

But there is nothing yet to protect the M.2 connector… though in my experience it should be strong enough.

2 Likes

I still like LSI SAS HBA’s and would use them over some no-name chinese m2 SATA adapter in a PCIe to m2 adapter. Daisy chaining adapters is usually not a good idea.

But I (partially) agree.

Broadcoms customer service is miserable. And trying to find the latest firmware on their nightmare of a website is also a miserable experience. But that is - usually - a one time inconvenience, and then you are off to the races.

If all you want to do is use either SATA or SAS hard drives, there is absolutely no reason to buy LSI 9400, 9500 or 9600 series adapters.

The 9300’s and 9305’s are perfect for this. They support 6Gbit/s SATA and 12Gbit/s SAS.

I have a few 9300-8i’s and they are all great. I have two 9305-24i’s and they are fantastic.

I’ve never used a 9305-16i, but they should be good too.

The only one I would caution against would be the 9300-16i, as it is essentially just two 9300-8i’s stuck together on the same board using a PCIe switch, and as a result they run pretty hot and use a fair amount of power.

The only downside with the 9300 and 9305 series is that SATA Trim is not implemented, so if you plan on using SSD’s, this can be an issue.

The best part about the 9300 and 9305 series is that they are very affordable these days, as they are a few years old. You can get them from a reliable Enterprise hardware recycler on eBay fir a fraction of what they once cost.

There are 9300-8i’s on there fro $16 right now. 9305-16i’s for onlt about $100 and 9305-24i’s for only about $175, which is very reasonable for the capacity you get. I’d go with these over some “chinese adapter in a chinese adapter” any day.

As for the 9400, 9500 and 9600 series, I do understand where you are coming from, from an abstraction layer perspective. They seem to interact with NVMe devices and present them to the OS as if they were some sort of SCSI device, with some ugly translation layer in between.

Whole they do this with NVMe devices, they don’t appear to inject themselves in any untoward way between the OS and the drive when it comes to SATA or SAS devices.

Using NVMe devices on LSI 9400, 9500 and 9600 adapters is really more of a convenience thing than it is a performance thing. The 9400 series absolutely kills NVMe performance. The 9500 is a little better, but it really wasn’t until the 9600 series that these “Tri-Mode” HBA’s really became decent performance wise.

And I agree, I’d be a little bit weary of what the adapter is doing there sitting in between my ZFS pools and the drives. They might be validated to Enterprise standards, but it still does make me a little uncomfortable. But again, that is really only an NVMe issue, not an issue when using them for SATA or SAS.

Looks like the first successful flashing of the radxa firmware here (need to scroll to the bottom and hit “Load more comments”, then hit enter with the browser location bar selected) : 7 watts idle on Intel 12th/13th gen: the foundation for building a low power server/NAS | mattgadient.com

I’ll have to add a bit to the ASM1166 error here.
When I calculated the parity on my Snapraid drives, I realized my dmesg put out massive amounts of the before mentioned PCIe errors. While everything seemed to be working and I would have been “ok” with some errors at boot, I couldn’t tolerate this spam.
I ordered another PCIe to 6x SATA ASM1166 from a different company, but same problem. Turned out they both use the same firmware version: 221118-00-48-00

So that company Radxa is apparently selling a M.2 to 6x SATA adapter which also uses the ASM1166. I posted the link to the firmware flashing tool for Linux in my previous comment already. ( Index of /accessories/m2-to-hexa-sata-adapter/tools ) Having a backup ASM1166 now I figured I just try flashing the ROM file ASM1166_10250005.ROM from that link, using the tool on linux.
Now the errors are gone, spindown/-up seem to work fine. And my system is even in C10 most of the time (80-85%). Here is the output of my successfull flashing process, it’s really easy:

ASM116x Firmware Update Tool V1.1.1.0  
Usage:[option]...  
 -U filename   Update firmware.  
 -E            Erase SPI.  
 -S            Display FW Version.  
root@pve:~# ./116xfwdl -S  
  
ASM116x Firmware Update Tool V1.1.1.0  
  
Dev[0]::FWver from PCIe: 22 11 18 00 48 00    
  
root@pve:~# ./116xfwdl -U ASM1166_10250005.ROM    
  
ASM116x Firmware Update Tool V1.1.1.0  
Found 1 ASM deiceFind a SPI flash ROM ID : 5Eh, 60h, 13h is not in Supported List!!!Try to program...ASM116UpdateSpiFlashRom: Chip Erase status = 0  
ASM116UpdateSpiFlashRom: Blank Check status = 0  
ASM116UpdateSpiFlashRom: Write Data status = 0  
Update SPI flash ROM......PASS!!!  
  
root@pve:~# ./116xfwdl -S  
  
ASM116x Firmware Update Tool V1.1.1.0  
  
Dev[0]::FWver from PCIe: 22 11 18 00 48 00    

root@pve:~# reboot  
.
.

root@pve:~# ./116xfwdl -S  
  
ASM116x Firmware Update Tool V1.1.1.0  
  
Dev[0]::FWver from PCIe: 24 10 25 00 00 05
5 Likes

@dmc Sorry for the delay in getting back to you, here are the flash chips identified on my PCIe and NVME-form factor cards by the Windows flashing tool:

P25Q40H
BY25D40

Both claim to be 512kB.

Looks like there’s an even newer firmware - 241224-0000-00 (YYMMDD-****-** format?) available here - Asmedia ASM-2116/116x Sata 6G Controller Firmware Version 12240000

Someone has flashed it successfully here - Latest Firmware for ASM1064/1166 SATA Controllers - #79 by Kommandore - BIOS modules (PCI ROM, EFI and others) - Win-Raid Forum

I updated my ASM-116x Controller from 221118-0000-00 with the newest Firmware from station-drivers ( Asmedia ASM-2116/116x Sata 6G Controller Firmware Version 12240000) and it works fine.

Another person in the thread also notes that (I believe compared to 241025-0000-05) this firmware supports PCIe MSI again - Latest Firmware for ASM1064/1166 SATA Controllers - #80 by boombastik - BIOS modules (PCI ROM, EFI and others) - Win-Raid Forum

Also it enables again message signaling (msi)

3 Likes