Re: [RFC] implement QUEUED spinlocks on powerpc

在 2017/2/13 下午5:08, panxinhui 写道:

在 2017/2/7 下午2:46, Eric Dumazet 写道:

quoted

On Mon, Feb 6, 2017 at 10:21 PM, panxinhui [off-list ref] wrote:

quoted

hi all
        I do some netperf tests and get some benchmark results.
I also attach my test script and netperf-result(Excel)

HI, all
I use loopback interface to run netperf tests,
#tc qd add dev lo root pfifo limit 10000
#ip link
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc pfifo state UNKNOWN mode DEFAULT group default qlen 1000
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00

and put the result in netperf.xlsx(excel)

It is a 32 vcpus P8 machine, with 32Gib memory.

This time spinlock is the best one, qspinlock > pvqspinlock. So sad.

This time, I have appiled some optimising patches on pvqspinlock.
When there is a high contention, the performance has a good improvement ans is very similar to spinlock.

Result is attached in netperf.xlsx

thanks
xinhui

thanks
xinhui

quoted

quoted

There are two machine. one runs netserver and the other runs netperf
benchmark. 1000Mbps network is connected with them.

#ip link infomation
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state
UNKNOWN mode DEFAULT group default qlen 1000
     link/ether ba:68:9c:14:32:02 brd ff:ff:ff:ff:ff:ff

According to the results, there is not much performance gap with each other.
And as we are only testing the throughput, the pvqspinlock shows the
overhead of its pv stuff. but qspinlock shows a little improvement than
spinlock. My simple summary in this testcase is
qspinlock > spinlock > pvqspinlock.

when run 200 concurrent netperf, I paste the total throughput here.

        concurrent runners| total throughput | variance
-------------------------------------------
spinlock        | 199 | 66882.8 | 89.93
-------------------------------------------
qspinlock       | 199 | 66350.4 | 72.0239
-------------------------------------------
pvqspinlock     | 199 | 64740.5 | 85.7837

You could see more data in nerperf.xlsx

thanks
xinhui

Hi xinhui

1Gbit NIC is too slow for this use case. I would try a 10Gbit NIC at least...

Alternatively, you could use loopback interface.  (netperf -H 127.0.0.1)

tc qd add dev lo root pfifo limit 10000