From: Tonghao Zhang <xiangxia.m.yue@gmail.com>
To: Ferruh Yigit <ferruh.yigit@intel.com>
Cc: "Lu, Wenzhuo" <wenzhuo.lu@intel.com>,
konstantin.ananyev@intel.com, "Xing,
Beilei" <beilei.xing@intel.com>, "Dai, Wei" <wei.dai@intel.com>,
dev@dpdk.org
Subject: Re: [dpdk-dev] [PATCH v3 1/5] net/ixgbevf: set the inter-interrupt interval for EITR.
Date: Wed, 18 Apr 2018 09:01:18 +0800 [thread overview]
Message-ID: <CAMDZJNWOjeDXXBQUXicAKF92sWr3P-Sy5HS0-3UUJx5s7PFwzw@mail.gmail.com> (raw)
In-Reply-To: <3b7d6b8b-270b-18e4-13a0-3676d02ef686@intel.com>
On Tue, Apr 17, 2018 at 7:00 PM, Ferruh Yigit <ferruh.yigit@intel.com> wrote:
> On 3/22/2018 1:01 PM, xiangxia.m.yue@gmail.com wrote:
>> From: Tonghao Zhang <xiangxia.m.yue@gmail.com>
>>
>> Set EITR interval as default. This patch can improve the
>> performance when we enable the rx-intrrupt to process the
>> packets because we hope rx-intrrupt reduce CPU. For example,
>> the 200us value of EITR makes the performance better with
>> the low CPU.
>>
>> Users can configure the value of ITR via DPDK configuration.
>>
>> The default value of ITR is 500us, compatible with RSC of ixgbe PF,
>> and next patch will use the default value.
>>
>> Signed-off-by: Tonghao Zhang <xiangxia.m.yue@gmail.com>
>> ---
>> v1 --> v2:
>> use the configure file, for different user.
>> suggested by Beilei Xing, http://dpdk.org/dev/patchwork/patch/32989
>> ---
>> config/common_base | 2 ++
>> drivers/net/ixgbe/ixgbe_ethdev.c | 7 +++++++
>> drivers/net/ixgbe/ixgbe_ethdev.h | 12 ++++++++++++
>> 3 files changed, 21 insertions(+)
>>
>> diff --git a/config/common_base b/config/common_base
>> index e74febe..2e9fded 100644
>> --- a/config/common_base
>> +++ b/config/common_base
>> @@ -196,6 +196,8 @@ CONFIG_RTE_LIBRTE_IXGBE_DEBUG_DRIVER=n
>> CONFIG_RTE_LIBRTE_IXGBE_PF_DISABLE_STRIP_CRC=n
>> CONFIG_RTE_IXGBE_INC_VECTOR=y
>> CONFIG_RTE_LIBRTE_IXGBE_BYPASS=n
>> +# interval up to 1024 us
>> +CONFIG_RTE_LIBRTE_IXGBE_ITR_INTERVAL=-1
>
> I can understand this is to let user to ability to fine tune this value, the
> down side is when number of these kind of tune parameters increased, it become
> confusing and config options because less useful or perhaps useless.
>
> And overall we are trying to reduce config options we have.
>
> I can see there is already some default values in the header file, what do you
> think removing the config option and go with default values defined in header?
v2 used the default value in header, but for configuring it like as
i40e option (CONFIG_RTE_LIBRTE_IXGBE_ITR_INTERVAL).
so v3 makes some changes. These patches may have been applied. For
reducing config option, should we do it in individual patches
to fix the ixgbe/i40e option ?
>>
>> #
>> # Compile burst-oriented I40E PMD driver
>> diff --git a/drivers/net/ixgbe/ixgbe_ethdev.c b/drivers/net/ixgbe/ixgbe_ethdev.c
>> index e67389f..495e72c 100644
>> --- a/drivers/net/ixgbe/ixgbe_ethdev.c
>> +++ b/drivers/net/ixgbe/ixgbe_ethdev.c
>> @@ -5780,6 +5780,13 @@ static void ixgbevf_set_vfta_all(struct rte_eth_dev *dev, bool on)
>> if (vector_idx < base + intr_handle->nb_efd - 1)
>> vector_idx++;
>> }
>> +
>> + /* As RX queue setting above show, all queues use the vector 0.
>> + * Set only the ITR value of IXGBE_MISC_VEC_ID.
>> + */
>> + IXGBE_WRITE_REG(hw, IXGBE_VTEITR(IXGBE_MISC_VEC_ID),
>> + ixgbe_calc_itr_interval(RTE_LIBRTE_IXGBE_ITR_INTERVAL)
>> + | IXGBE_EITR_CNT_WDIS);
>> }
>>
>> /**
>> diff --git a/drivers/net/ixgbe/ixgbe_ethdev.h b/drivers/net/ixgbe/ixgbe_ethdev.h
>> index 1db29bd..c779001 100644
>> --- a/drivers/net/ixgbe/ixgbe_ethdev.h
>> +++ b/drivers/net/ixgbe/ixgbe_ethdev.h
>> @@ -58,6 +58,18 @@
>> IXGBE_EITR_ITR_INT_MASK)
>>
>>
>> +#define IXGBE_QUEUE_ITR_INTERVAL_MAX 1024 /* 1024us */
>> +#define IXGBE_QUEUE_ITR_INTERVAL_DEFAULT 500 /* 500us */
>> +
>> +static inline uint16_t
>> +ixgbe_calc_itr_interval(int16_t interval)
>> +{
>> + if (interval < 0 || interval > IXGBE_QUEUE_ITR_INTERVAL_MAX)
>> + interval = IXGBE_QUEUE_ITR_INTERVAL_DEFAULT;
>> +
>> + return IXGBE_EITR_INTERVAL_US(interval);
>> +}
>> +
>> /* Loopback operation modes */
>> /* 82599 specific loopback operation types */
>> #define IXGBE_LPBK_82599_NONE 0x0 /* Default value. Loopback is disabled. */
>>
>
next prev parent reply other threads:[~2018-04-18 1:01 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-03-22 13:01 [dpdk-dev] [PATCH v3 0/5] ixgbe: fix bugs or just improve xiangxia.m.yue
2018-03-22 13:01 ` [dpdk-dev] [PATCH v3 1/5] net/ixgbevf: set the inter-interrupt interval for EITR xiangxia.m.yue
2018-04-17 11:00 ` Ferruh Yigit
2018-04-18 1:01 ` Tonghao Zhang [this message]
2018-04-18 16:10 ` Ferruh Yigit
2018-04-26 13:14 ` Zhang, Qi Z
2018-04-26 13:54 ` Tonghao Zhang
2018-03-22 13:01 ` [dpdk-dev] [PATCH v3 2/5] net/ixgbe: set the ITR via configuration xiangxia.m.yue
2018-03-22 13:01 ` [dpdk-dev] [PATCH v3 3/5] net/ixgbe: write disable to ITR counter xiangxia.m.yue
2018-03-22 13:01 ` [dpdk-dev] [PATCH v3 4/5] net/ixgbevf: save IXGBE_VTEIMS to intr->mask for performance xiangxia.m.yue
2018-03-22 13:01 ` [dpdk-dev] [PATCH v3 5/5] net/ixgbe: remove the unnecessary call rte_intr_enable xiangxia.m.yue
-- strict thread matches above, loose matches on Subject: below --
2018-02-05 0:51 [dpdk-dev] [PATCH v3 0/5] ixgbe: fix bugs or just improve xiangxia.m.yue
2018-02-05 0:51 ` [dpdk-dev] [PATCH v3 1/5] net/ixgbevf: set the inter-interrupt interval for EITR xiangxia.m.yue
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=CAMDZJNWOjeDXXBQUXicAKF92sWr3P-Sy5HS0-3UUJx5s7PFwzw@mail.gmail.com \
--to=xiangxia.m.yue@gmail.com \
--cc=beilei.xing@intel.com \
--cc=dev@dpdk.org \
--cc=ferruh.yigit@intel.com \
--cc=konstantin.ananyev@intel.com \
--cc=wei.dai@intel.com \
--cc=wenzhuo.lu@intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).