Review on Syba SD PEX40054 Port SATA Hyper by Chase Hosea

Terrible. I can't believe this item is for sale. There's so much wrong with him that I don't even know where to start, but I'll try. First, let me tell you where I'm from. I've already read the reviews and wasn't expecting much at first, but I've been looking for a long time for an AHCI chipset (especially AHCI) that is FBS-capable (FIS-based switching) and these guys advertised it. Unfortunately, this chipset does not implement the AHCI specification sufficiently enough for our AHCI driver to be able to check drives. He only mentions three possibilities. It reports AHCI v1.2 and S64A, NCQ and PMD and that's it. It reports no SSNTF or SCLO and definitely no FBS (FIS Based Switching Support). Without SSNTF, and especially without SCLO, this chipset cannot be even remotely reliable. It also doesn't report low-power or other mode features, and many other standard features are missing. For example, the AHCI motherboard chipset reports: S64A, NCQ, SSNTF, SMPS, SALP, SAL, SCLO, SPM, PMD, SSC, and PSC. (Obviously I'm testing on a Linux or BSD system, not Windows, so I have full control over the driver and know exactly what the chipset is doing at the PCIe host interface level). If this chipset supports the correct port multiplier definition, it would report SPM (port multiplier support per port), and if it supports FIS-based switching via the AHCI port multiplier specification, it would report FBS. He reports neither the one nor the other. This chipset basically collects real ports behind the scenes and performs port multiplier detection and scanning behind the scenes and also reports them as individual AHCI ports. In other words, it hides the port multiplier behind some firmware and chip magic instead of implementing the port multiplier specification on host-facing ports. Honestly, I'm not sure how it's supposed to work when the chip reports only 8 ports (one containing a MARVELL controller, also out of spec), aggregated PM ports behind the scenes, but two external and two internal SATA has ports (which should theoretically support at least 20 drives if each port has a port multiplier attached). This might be fine except for three things: (1) Without SCLO you can't have a reliable AHCI chip, period. (2) If it implements FIS based internal switching, it definitely doesn't show up when I read the tests on the ports. With two disks in the port multiplier I get 70MB/s reading from both, but reading from both at the same time I get 35MB/s from both. In other words, this is clearly NOT FIS based internal switching if you have an actual port multiplier attached to the E-SATA port. And (3) it doesn't handle hot-swapping properly. When I first booted up the machine, it found two of the three drives in my RM. When I disconnected and reconnected two of them, it found that two ports were disconnected and connected, but then one of the two reconnected drives could not be recognized. Also, the console reported DMA kernel errors about 30 seconds before detecting either of the two reconnected drives. When I rebooted it got stuck in the BIOS (couldn't even boot the machine). twice. After a few reboots it went through the BIOS and reported all three drives. This is a real mess. Not even remotely reliable. This chip is basically trying to do the PM detection behind the scenes and present individual drives to the host as individual AHCI ports instead of implementing the port multiplier AHCI chipset feature and making it accessible to the host. And it does a terrible job. I understand why others have reported Windows BSODs with this family of chips. The firmware is insanely bad, even worse than SiS cards. I was hoping that it would implement the AHCI port multiplier specification. It's not like this. I was hoping he implemented FIS based switching. actually it isn't. I recommend that everything based on this family of chips is a very broad place. Avoid this at all costs. This is complete rubbish. - Matt