On Sunday 06 March 2005 07:01, Jeff Dike wrote:
> [EMAIL PROTECTED] said:
> > Is this like it is implemented now?
>
> Yup. In both the AIO and non-AIO case, there is a separate thread to which
> you send IO requests, and at some later point, UML gets interrupted with
> the results.
To answer anoth
[EMAIL PROTECTED] said:
> Is this like it is implemented now?
Yup. In both the AIO and non-AIO case, there is a separate thread to which
you send IO requests, and at some later point, UML gets interrupted with
the results.
Jeff
has anybody compared the performance of UML's built-in blockdevice,
NetworkBlockDevices, ATAoverEthernet, yet?
I don't know how UML's blockdevice works, but i know how NBD works. The
kernel sends multiple requests to the NBD server, and receives the
responses asap. The worse case for UML's block-de
On Friday 04 March 2005 04:01, Sven Köhler wrote:
> Hi,
>
> has anybody compared the performance of UML's built-in blockdevice,
> NetworkBlockDevices, ATAoverEthernet, yet?
>
> I don't know how UML's blockdevice works, but i know how NBD works. The
> kernel sends multiple requests to the NBD server
Hi,
has anybody compared the performance of UML's built-in blockdevice,
NetworkBlockDevices, ATAoverEthernet, yet?
I don't know how UML's blockdevice works, but i know how NBD works. The
kernel sends multiple requests to the NBD server, and receives the
responses asap. The worse case for UML's
