Hi Stephen,
> -----Original Message-----
> From: Stephen Hemminger <step...@networkplumber.org>
> Sent: Friday, February 3, 2023 8:31 AM
> To: Su, Simei <simei...@intel.com>
> Cc: Zhang, Qi Z <qi.z.zh...@intel.com>; Guo, Junfeng
> <junfeng....@intel.com>; dev@dpdk.org; Wu, Wenjun1
> <wenjun1...@intel.com>
> Subject: Re: [PATCH v3 2/2] net/igc: enable launch time offloading
>
> On Thu, 2 Feb 2023 15:18:01 +0800
> Simei Su <simei...@intel.com> wrote:
>
> >
> > +static uint32_t igc_tx_launchtime(uint64_t txtime, uint16_t port_id)
> > +{
> > + struct rte_eth_dev *dev = &rte_eth_devices[port_id];
> > + struct igc_adapter *adapter = IGC_DEV_PRIVATE(dev);
> > + uint64_t base_time = adapter->base_time;
> > + uint64_t cycle_time = adapter->cycle_time;
> > + uint32_t launchtime;
> > +
> > + launchtime = (txtime - base_time) % cycle_time;
> > +
> > + return rte_cpu_to_le_32(launchtime); }
>
>
> Divide in transmit path will slow things down.
> Better to use something like rte_reciprocal_divide_64() to avoid slow 64 bit
> divide.
Thanks for your comments.
The performance won't be affected if the launchtime function is turned off.
If the function is turned on, it indeed doesn't need high performance based on
the function.
Thanks,
Simei
