Tweet
I was used to employ e-SATA HDD-connection with my ASRock Alive e-Dual SATA2 MB before I changed my Motherboard to 890FX Deluxe 5 (BIOS/UEFI 1.60 version). Chipset used on my old MB was JMICRON SATA 3G (II) and one of two SATAII connectors was linked to my backplate. Performance was equal to the internal SATA II and identification of the e-SATA by hotplug, or switch-on of HDD-power was almost immediate. Notes: SATA II on JMICRON was used in the AHCI-mode. My boot disk was first connected on IDE-basis, later on one of the SATAI ports of NVDIA (NFORCE 3 250, also set to AHCI-mode.
On each boot of XP (full screen logo disabled) or recovering out of the sleep state, one could see whether the e-SATA HDD was present before in the list of HDD's being identified at boot time.
Quite different is my Deluxe 5: no identification of the e-SATA on the Marvell for many minutes if XP is already started up(sometimes failing altogether). And even when booting from fresh with external HDD active, the external HDD need not come up automatically in XP afterwards. On a number of Fora I found people withe the same experience. One link led to a later Marvell SATA3 driver version (WHQL) than ASRock provides, but this did NOT change the e-SATA Hotplug behaviour.
So I changed my HDD to an USB 3.0 enclosure and connected to one of the backplate NEC ports: immediate recognition of the HDD and disk performance comparable to (e-) SATA standards. I use a very old disk testfacility inside NERO Reloaded for a basic performance test (Sisoft SANDRA just takes to long). However, when a second hard disk was connected to the other USB3 NEC backplate connection, I found interaction between the two ports: one HDD became slower when the other one was connected and up.
Parttime solution found now: reinstall one HDD back to the e-SATA basis and the MArvell port, but with the option of same removal disabled (see device manager for the MArvell 91xx SATA 6G controller and tab policy settings lowest checkbox now open). Now XP no longer sees the HDD as USB-equivalent and hotplug is disabled. I also installed the Hotswap facility advised in other fora, just in case. e-SATA port now works as SATA, meaning one should in principle not switch it off with XP running, but wait until standby or sleep state is reached. Switching the HDD on before reactivating XP is sufficient for keeping it on. Waking from sleep works sligthly better, because the MB tests more on awakening.
Change in procedure tried for: When setting the Marvell to bootable, its is forced to recognise all drives (HDD or DVD/BD) connected to ports 7 (SATA) and 8 (eSATA) in the bootperiod and it shows this in a special bootscreen. Then the e-SATA disk is just as visible as any fixed drive. But boot takes so long using this option, that I won't use it. I went back againg to the safe removal option in the driver tab and regained fast bootup, but lost fast e-SATA drive recognition.
Whenever I do not use the e-SATA disk for a number of Windows-sessions, one has to wait for XP to find it again after many minutes, sometimes the Hot Swap tool helps, sometimes not. An altenative is activate "search for changed PnP devices" in device manager. And very seldomely user sign-off and sign-on before this activation is needed for finding it..
Conclusion: this Marvell 91xx SATA 6G controller is not capable of proper e-SATA behaviour and using the USB 3.0 alternative on more than one NEC-port is also not advisable. My guess is that 990FX-boards will suffer the same problems with this Marvell chip as the 890FX ones (but USB 3 seems a different chipset) I invite others finding similar behaviour on e-SATA to comment or advise on similar or other solutions.
Other solution found now: use Sharkoon's removable drive bay solution on Marvell ports SATA 7 and 8 (AHCI mode retained). Hot swapping works fine! Of course manipulating bare drive assemblies is not to everyone's taste, but it works!
:-D on thuis
PS UEFI/BIOS update did not help my e-SATA. On the contrary: it made me lose my PS2 ports instead
On each boot of XP (full screen logo disabled) or recovering out of the sleep state, one could see whether the e-SATA HDD was present before in the list of HDD's being identified at boot time.
Quite different is my Deluxe 5: no identification of the e-SATA on the Marvell for many minutes if XP is already started up(sometimes failing altogether). And even when booting from fresh with external HDD active, the external HDD need not come up automatically in XP afterwards. On a number of Fora I found people withe the same experience. One link led to a later Marvell SATA3 driver version (WHQL) than ASRock provides, but this did NOT change the e-SATA Hotplug behaviour.
So I changed my HDD to an USB 3.0 enclosure and connected to one of the backplate NEC ports: immediate recognition of the HDD and disk performance comparable to (e-) SATA standards. I use a very old disk testfacility inside NERO Reloaded for a basic performance test (Sisoft SANDRA just takes to long). However, when a second hard disk was connected to the other USB3 NEC backplate connection, I found interaction between the two ports: one HDD became slower when the other one was connected and up.
Parttime solution found now: reinstall one HDD back to the e-SATA basis and the MArvell port, but with the option of same removal disabled (see device manager for the MArvell 91xx SATA 6G controller and tab policy settings lowest checkbox now open). Now XP no longer sees the HDD as USB-equivalent and hotplug is disabled. I also installed the Hotswap facility advised in other fora, just in case. e-SATA port now works as SATA, meaning one should in principle not switch it off with XP running, but wait until standby or sleep state is reached. Switching the HDD on before reactivating XP is sufficient for keeping it on. Waking from sleep works sligthly better, because the MB tests more on awakening.
Change in procedure tried for: When setting the Marvell to bootable, its is forced to recognise all drives (HDD or DVD/BD) connected to ports 7 (SATA) and 8 (eSATA) in the bootperiod and it shows this in a special bootscreen. Then the e-SATA disk is just as visible as any fixed drive. But boot takes so long using this option, that I won't use it. I went back againg to the safe removal option in the driver tab and regained fast bootup, but lost fast e-SATA drive recognition.
Whenever I do not use the e-SATA disk for a number of Windows-sessions, one has to wait for XP to find it again after many minutes, sometimes the Hot Swap tool helps, sometimes not. An altenative is activate "search for changed PnP devices" in device manager. And very seldomely user sign-off and sign-on before this activation is needed for finding it..
Conclusion: this Marvell 91xx SATA 6G controller is not capable of proper e-SATA behaviour and using the USB 3.0 alternative on more than one NEC-port is also not advisable. My guess is that 990FX-boards will suffer the same problems with this Marvell chip as the 890FX ones (but USB 3 seems a different chipset) I invite others finding similar behaviour on e-SATA to comment or advise on similar or other solutions.
Other solution found now: use Sharkoon's removable drive bay solution on Marvell ports SATA 7 and 8 (AHCI mode retained). Hot swapping works fine! Of course manipulating bare drive assemblies is not to everyone's taste, but it works!
:-D on thuis
PS UEFI/BIOS update did not help my e-SATA. On the contrary: it made me lose my PS2 ports instead