[2/2] ring: empty optimization

Message ID 20200519152725.63486-3-mb@smartsharesystems.com (mailing list archive)
State Accepted, archived
Delegated to: David Marchand
Headers
Series ring: empty optimization |

Checks

Context Check Description
ci/checkpatch success coding style OK
ci/Intel-compilation fail Compilation issues

Commit Message

Morten Brørup May 19, 2020, 3:27 p.m. UTC
Testing if the ring is empty is as simple as comparing the producer and
consumer pointers.

In theory, this optimization reduces the number of potential cache misses
from 3 to 2 by not having to read r->mask in rte_ring_count().

The modification of this function were also discussed in the RFC here:
https://mails.dpdk.org/archives/dev/2020-April/165752.html

Signed-off-by: Morten Brørup <mb@smartsharesystems.com>
---
 lib/librte_ring/rte_ring.h | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
  

Comments

Stephen Hemminger May 19, 2020, 3:52 p.m. UTC | #1
On Tue, 19 May 2020 15:27:25 +0000
Morten Brørup <mb@smartsharesystems.com> wrote:

> diff --git a/lib/librte_ring/rte_ring.h b/lib/librte_ring/rte_ring.h
> index 9078e7c24..f67141482 100644
> --- a/lib/librte_ring/rte_ring.h
> +++ b/lib/librte_ring/rte_ring.h
> @@ -733,7 +733,9 @@ rte_ring_full(const struct rte_ring *r)
>  static inline int
>  rte_ring_empty(const struct rte_ring *r)
>  {
> -	return rte_ring_count(r) == 0;
> +	uint32_t prod_tail = r->prod.tail;
> +	uint32_t cons_tail = r->cons.tail;
> +	return cons_tail == prod_tail;
>  }

Blank line after declarations?

Are the temporary variable even needed?
  
Morten Brørup May 19, 2020, 4:02 p.m. UTC | #2
> From: dev [mailto:dev-bounces@dpdk.org] On Behalf Of Stephen Hemminger
> Sent: Tuesday, May 19, 2020 5:52 PM
> 
> On Tue, 19 May 2020 15:27:25 +0000
> Morten Brørup <mb@smartsharesystems.com> wrote:
> 
> > diff --git a/lib/librte_ring/rte_ring.h b/lib/librte_ring/rte_ring.h
> > index 9078e7c24..f67141482 100644
> > --- a/lib/librte_ring/rte_ring.h
> > +++ b/lib/librte_ring/rte_ring.h
> > @@ -733,7 +733,9 @@ rte_ring_full(const struct rte_ring *r)
> >  static inline int
> >  rte_ring_empty(const struct rte_ring *r)
> >  {
> > -	return rte_ring_count(r) == 0;
> > +	uint32_t prod_tail = r->prod.tail;
> > +	uint32_t cons_tail = r->cons.tail;
> > +	return cons_tail == prod_tail;
> >  }
> 
> Blank line after declarations?
> 
> Are the temporary variable even needed?

Personally, I agree with you, but I was trying to match the existing coding style of the closely related rte_ring_count() function - only to avoid this kind of feedback.

Damn if you do, damn if you don't. :-)
  
Ananyev, Konstantin May 22, 2020, 12:32 p.m. UTC | #3
> 
> Testing if the ring is empty is as simple as comparing the producer and
> consumer pointers.
> 
> In theory, this optimization reduces the number of potential cache misses
> from 3 to 2 by not having to read r->mask in rte_ring_count().
> 
> The modification of this function were also discussed in the RFC here:
> https://mails.dpdk.org/archives/dev/2020-April/165752.html
> 
> Signed-off-by: Morten Brørup <mb@smartsharesystems.com>
> ---
>  lib/librte_ring/rte_ring.h | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/lib/librte_ring/rte_ring.h b/lib/librte_ring/rte_ring.h
> index 9078e7c24..f67141482 100644
> --- a/lib/librte_ring/rte_ring.h
> +++ b/lib/librte_ring/rte_ring.h
> @@ -733,7 +733,9 @@ rte_ring_full(const struct rte_ring *r)
>  static inline int
>  rte_ring_empty(const struct rte_ring *r)
>  {
> -	return rte_ring_count(r) == 0;
> +	uint32_t prod_tail = r->prod.tail;
> +	uint32_t cons_tail = r->cons.tail;
> +	return cons_tail == prod_tail;
>  }
> 
>  /**
> --

Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>

> 2.17.1
  
David Marchand July 1, 2020, 9:19 a.m. UTC | #4
On Tue, May 19, 2020 at 6:02 PM Morten Brørup <mb@smartsharesystems.com> wrote:
> > Blank line after declarations?
> >
> > Are the temporary variable even needed?
>
> Personally, I agree with you, but I was trying to match the existing coding style of the closely related rte_ring_count() function - only to avoid this kind of feedback.
>
> Damn if you do, damn if you don't. :-)

Yes, looking at the code, it seems fair taking this patch as is.
  

Patch

diff --git a/lib/librte_ring/rte_ring.h b/lib/librte_ring/rte_ring.h
index 9078e7c24..f67141482 100644
--- a/lib/librte_ring/rte_ring.h
+++ b/lib/librte_ring/rte_ring.h
@@ -733,7 +733,9 @@  rte_ring_full(const struct rte_ring *r)
 static inline int
 rte_ring_empty(const struct rte_ring *r)
 {
-	return rte_ring_count(r) == 0;
+	uint32_t prod_tail = r->prod.tail;
+	uint32_t cons_tail = r->cons.tail;
+	return cons_tail == prod_tail;
 }
 
 /**