On 22 April 2013 18:07, Martin N <martin.n@???> wrote:
> Hello,
>
> At 17:08 22/04/2013, you wrote:
>
>> On 2013-04-22 13:00, <mailto:hampshire-request@**mailman.lug.org.uk<hampshire-request@???>
>> >hampshire-**request@???<hampshire-request@???>wrote:
>>
>>
>> On 21 April 2013 17:34, Peter B. <<mailto:petyrb@gmail.com>pety**
>> rb@??? <petyrb@???>> wrote:
>>
>> As far as I know usb 2.0 maxes out at 30mbps usb 3.0 100mbps for local
>> data transfer.
>>
>> I think you're confusing things here. USB2 runs at up to 480Mbps raw line
>> speed, and USB3 runs at up to 5Gbps.
>>
>>
>> --
>> Daniel Llewellyn
>>
>> That's as maybe. However USB relies upon a whole heap of CPU Processing
>> to move anything. This is done do keep the costs down. Firewire on the
>> otherhand does DMA and uses a whole lot less CPU.
>> As someone who wrote their first Device Driver before Microsoft existed
>> (for a PDP-11/45 in 1976) Serial I/O using devices such as a UART is
>> really-sllooowwww. DMA is much faster and neater. Sadly in these X86 times
>> there are lots of devices that do I/O character by character or
>> byte-by-byte and not using DMA.
>>
>
>
> Is this the same for e-SATA?
> (high CPU load)
>
> That will probably the way i will go when i can find a 4 bay dock for a
> reasonable price.
I wholeheartedly agree that standard ethernet based disks (gig-e) are the
way to go if you require multiple hosts accessing either simultaneously or
with a minimum of (un)plugging.
However, I would suggest looking into e-sata if you only need to connect
one host.
AFAICT e-sata is just a socket/plug standard and some rules over hot
plugging which expose the internal sata bus directly with no additional
encapsulation (unlike USB disks) - i.e. e-sata should be capable of being
just as performant as internal sata.
--
Daniel Llewellyn
--
Please post to: Hampshire@???
Web Interface:
https://mailman.lug.org.uk/mailman/listinfo/hampshire
LUG URL:
http://www.hantslug.org.uk
--------------------------------------------------------------
