Date: Thu, 5 Jun 2008 23:49:11 +0900
To: andi@firstfloor.org
Cc: fujita.tomonori@lab.ntt.co.jp, linux-kernel@vger.kernel.org,
       mgross@linux.intel.com, linux-scsi@vger.kernel.org
Subject: Re: Intel IOMMU (and IOMMU for Virtualization) performances
From: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
In-Reply-To: <873antuojg.fsf@basil.nowhere.org>
References: <20080604235053K.fujita.tomonori@lab.ntt.co.jp>
	<873antuojg.fsf@basil.nowhere.org>
Mime-Version: 1.0
Content-Type: Text/Plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Message-Id: <20080605231914Z.fujita.tomonori@lab.ntt.co.jp>
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1043
Lines: 20

On Wed, 04 Jun 2008 18:56:35 +0200
Andi Kleen <andi@firstfloor.org> wrote:

> FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp> writes:
> >
> > I'm just interested in other people's opinions on IOMMU
> > implementations, performances, possible future changes for performance
> > improvement, etc.
> 
> I think using the bitmap is an excellent idea and your numbers look good. 
> Do you have numbers on the memory consumption too?
> Trading some memory for performance is ok for something as performance critical 
> as the IOMMU. 

If we use 4GB virtual DMA address space (as the patch does), we need
128 KB for the bitmap for one domain.

With the RB tree, the memory consumption depends on how many addresses
are mapped (it needs one entry for one address though we could merge
multiple addresses).
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/