Re: [PATCH] vhost: adaptively batch small guest memory copies

From: Yuanhan Liu <yliu@fridaylinux.org>
To: Tiwei Bie <tiwei.bie@intel.com>
Cc: dev@dpdk.org, maxime.coquelin@redhat.com,
	Zhihong Wang <zhihong.wang@intel.com>,
	Zhiyong Yang <zhiyong.yang@intel.com>
Subject: Re: [PATCH] vhost: adaptively batch small guest memory copies
Date: Fri, 8 Sep 2017 15:41:08 +0800	[thread overview]
Message-ID: <20170908074108.GE9736@yliu-home> (raw)
In-Reply-To: <20170824021939.21306-1-tiwei.bie@intel.com>

On Thu, Aug 24, 2017 at 10:19:39AM +0800, Tiwei Bie wrote:
> This patch adaptively batches the small guest memory copies.
> By batching the small copies, the efficiency of executing the
> memory LOAD instructions can be improved greatly, because the
> memory LOAD latency can be effectively hidden by the pipeline.
> We saw great performance boosts for small packets PVP test.
> 
> This patch improves the performance for small packets, and has
> distinguished the packets by size. So although the performance
> for big packets doesn't change, it makes it relatively easy to
> do some special optimizations for the big packets too.

The number showed in other replies looks really impressive. Great work!
This patch also looks good to me. I have one minor comment though.

[...]
> +/*
> + * Structure contains the info for each batched memory copy.
> + */
> +struct burst_copy_elem {
> +	void *dst;
> +	void *src;
> +	uint32_t len;
> +	uint64_t log_addr;
> +};

Like the title says, it's more about batch (but not burst). Also, it's
not a good idea to mix burst and batch. I'd suggest you to use the term
"batch" consistently.

	--yliu