[Pc_Support] [Revisited] PCI-Express SATA RAID cards -- HighPoint RocketRAID 2320

Bryan J. Smith b.j.smith at ieee.org
Tue Dec 13 15:05:44 EST 2005


I was just doing some further research on the HighPoint
RocketRAID 2320 (PCIe) and 2220 (PCI-X) versions.

Apparently, in addition to the Marvell 88SX6041/81 (4/8 SATA
to PCI-X -- not sure about PCIe), there is a HighPoint HPT601
controller.  It seems to be an XOR ASIC and capable of other
operations.

I'm not sure how it overall affects RAID-5, let along RAID-0,
1 or 10 -- and if the ASIC is completely controlling.  I have
no idea how much SRAM (if any) is in the IC either (or on the
Marvell chip).

But it's clear from the limited I/O benchmark in Tom's
review(s) (no DTR) that the HPT601 approach isn't as bad as
the Intel ICH7-R or nVidia MCP-04 southbridges at ATA RAID. 
It actually is fairly competitive with the Broadcom (and the
3Ware 9550SX, although I'm sure that's more about the
9550SX's firmware maturity at this point).

So, since the 4-channel PCIe x4 card is sub-$200, it might
not be a bad option to look into.  It claims Linux support,
but I'm not sure how "open" that support is.  Does it hide
the SATA channels?  I don't think so, because the PCI-X/PCIe
arbitrator is on the Marvell chip.  But they claim the HPT601
allows hot-swap and off-loading from host -- so I don't know.

Supposedly the 3Ware 9550SX's 2 PowerPC 400 series
controllers have PCI-X to PCIe bridges on-IC, so hopefully
we'll see a 9550SX for PCIe soon.  But how much it will
compete is anyone's guess, I think AMCC needs to "shakedown
time" in its PPC400 architecture before we'll see the
9550SX's full potential (right now, even on PCI-X, it's very
good, but not good enough to push me towards it).


-- 
Bryan J. Smith                | Sent from Yahoo Mail
mailto:b.j.smith at ieee.org     |  (please excuse any
http://thebs413.blogspot.com/ |   missing headers)



More information about the Pc_support mailing list