[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <alpine.DEB.2.02.1110031615490.20248@p34.internal.lan>
Date: Mon, 3 Oct 2011 16:16:52 -0400 (EDT)
From: Justin Piszcz <jpiszcz@...idpixels.com>
To: Pasi Kärkkäinen <pasik@....fi>
cc: linux-kernel@...r.kernel.org, linux-scsi@...r.kernel.org
Subject: Re: HighPoint RocketRaid 620 vs. RocketRaid 640 -- supported?
On Mon, 3 Oct 2011, Pasi Kärkkäinen wrote:
> On Wed, Sep 21, 2011 at 07:28:26AM -0400, Justin Piszcz wrote:
>> Hi,
>>
>> Incase anyone is wondering the same thing, I contacted Highpoint and they
>> said (confirmed) the 620 support (which I am using now) is built-in to
>> the kernel and the 640 is not, so unless you like compiling kernel
>> modules/out of tree drivers for a _boot_ volume and want a regular SATA
>> 6.0gbps controller that supports TRIM, I'd go for the 620 even though it
>> maxes out around 390-420MiB/s read where a real SAS controller supports
>> 490-500MiB/s for an Intel 510 SSD, but remember, SAS controllers (at
>> least the one I tested with does not support TRIM and there is no Intel
>> SSD optimizer for Linux.
>>
>
> Hmm.. does SAS "SCSI UNMAP" automatically get translated to
> ATA TRIM if you hook up SATA disk to SAS HBA ?
When attached to a SAS HBA, it simply says:
Jul 14 16:12:03 p34 kernel: [ 22.258855] EXT4-fs warning (device sda2): release_blocks_on_commit:2679: discard not supported, disabling
>
> Is the ATA<->SCSI translation being done by the HBA firmware,
> or is it handled in the driver?
I'm not sure, it would appear the driver handles it, when its on the Sata 6gbps
card (not SAS), it works fine:
Sep 20 13:53:34 p34 kernel: [ 6.786898] EXT4-fs (sdb2): re-mounted. Opts: discard
>
> -- Pasi (who is too lazy to grep in the driver right now ;)
Justin.
Powered by blists - more mailing lists