* Testing results of zswap
@ 2013-06-27 2:03 Bob Liu
2013-07-15 2:56 ` Bob Liu
` (3 more replies)
0 siblings, 4 replies; 6+ messages in thread
From: Bob Liu @ 2013-06-27 2:03 UTC (permalink / raw)
To: Linux-MM
Cc: Andrew Morton, Konrad Rzeszutek Wilk, Seth Jennings,
Dan Magenheimer, Nitin Gupta, bob.liu, Mel Gorman
Hi All,
These days I have been testing zswap.
I found that the total ram size of my testing machine effected the
testing result.
If I limit RAM size to 2G using "mem=", the performance of zswap is
very disappointing,
But if I use larger RAM size such as 8G, the performance is much better.
Even with RAM size 8G, zswap will slow down the speed of parallelio.
I run the testing(mmtest-0.10 with
config-global-dhp__parallelio-memcachetest) after the default
distribution booted every time.
Below are some results:
1) kernel verion 3.10-rc6, mem=2G
parallelio
rc6 rc6
base-2G zswap-2G
Ops memcachetest-0M 14719.00 ( 0.00%) 10617.00 (-27.87%)
Ops memcachetest-200M 14711.00 ( 0.00%) 10152.00 (-30.99%)
Ops memcachetest-433M 14839.00 ( 0.00%) 10245.00 (-30.96%)
Ops memcachetest-666M 14989.00 ( 0.00%) 10134.00 (-32.39%)
Ops memcachetest-900M 15180.00 ( 0.00%) 9821.00 (-35.30%)
Ops memcachetest-1133M 15654.00 ( 0.00%) 9178.00 (-41.37%)
Ops memcachetest-1367M 16100.00 ( 0.00%) 9740.00 (-39.50%)
Ops io-duration-0M 0.00 ( 0.00%) 0.00 ( 0.00%)
Ops io-duration-200M 3.00 ( 0.00%) 5.00 (-66.67%)
Ops io-duration-433M 6.00 ( 0.00%) 10.00 (-66.67%)
Ops io-duration-666M 10.00 ( 0.00%) 14.00 (-40.00%)
Ops io-duration-900M 13.00 ( 0.00%) 21.00 (-61.54%)
Ops io-duration-1133M 17.00 ( 0.00%) 25.00 (-47.06%)
Ops io-duration-1367M 20.00 ( 0.00%) 31.00 (-55.00%)
Ops swaptotal-0M 39.00 ( 0.00%) 81481.00
(-208825.64%)
Ops swaptotal-200M 1400.00 ( 0.00%) 106825.00 (-7530.36%)
Ops swaptotal-433M 16.00 ( 0.00%) 103052.00
(-643975.00%)
Ops swaptotal-666M 8.00 ( 0.00%) 93693.00
(-1171062.50%)
Ops swaptotal-900M 0.00 ( 0.00%) 91009.00 (-99.00%)
Ops swaptotal-1133M 3609.00 ( 0.00%) 103650.00 (-2771.99%)
Ops swaptotal-1367M 95.00 ( 0.00%) 89652.00
(-94270.53%)
Ops swapin-0M 39.00 ( 0.00%) 39283.00
(-100625.64%)
Ops swapin-200M 678.00 ( 0.00%) 50220.00 (-7307.08%)
Ops swapin-433M 16.00 ( 0.00%) 48372.00
(-302225.00%)
Ops swapin-666M 8.00 ( 0.00%) 43669.00
(-545762.50%)
Ops swapin-900M 0.00 ( 0.00%) 42791.00 (-99.00%)
Ops swapin-1133M 716.00 ( 0.00%) 49320.00 (-6788.27%)
Ops swapin-1367M 91.00 ( 0.00%) 42094.00
(-46157.14%)
Ops minorfaults-0M 511226.00 ( 0.00%) 532594.00 ( -4.18%)
Ops minorfaults-200M 507501.00 ( 0.00%) 577895.00 (-13.87%)
Ops minorfaults-433M 507342.00 ( 0.00%) 573211.00 (-12.98%)
Ops minorfaults-666M 506917.00 ( 0.00%) 565424.00 (-11.54%)
Ops minorfaults-900M 513814.00 ( 0.00%) 569420.00 (-10.82%)
Ops minorfaults-1133M 559981.00 ( 0.00%) 592414.00 ( -5.79%)
Ops minorfaults-1367M 511420.00 ( 0.00%) 572809.00 (-12.00%)
Ops majorfaults-0M 6.00 ( 0.00%) 8593.00
(-143116.67%)
Ops majorfaults-200M 200.00 ( 0.00%) 11335.00 (-5567.50%)
Ops majorfaults-433M 55.00 ( 0.00%) 10729.00
(-19407.27%)
Ops majorfaults-666M 68.00 ( 0.00%) 9258.00
(-13514.71%)
Ops majorfaults-900M 94.00 ( 0.00%) 9935.00
(-10469.15%)
Ops majorfaults-1133M 411.00 ( 0.00%) 10902.00 (-2552.55%)
Ops majorfaults-1367M 133.00 ( 0.00%) 9340.00 (-6922.56%)
rc6 rc6
base-2G zswap-2G
User 1068.98 715.36
System 3910.06 2696.03
Elapsed 7871.94 7890.20
rc6 rc6
base-2G zswap-2G
Page Ins 1322220 10206472
Page Outs 24675096 31069608
Swap Ins 100636 1548495
Swap Outs 134568 1734841
Direct pages scanned 14441 16430214
Kswapd pages scanned 7792848 35445764
Kswapd pages reclaimed 6486627 7514907
Direct pages reclaimed 4342 2166004
Kswapd efficiency 83% 21%
Kswapd velocity 989.953 4492.378
Direct efficiency 30% 13%
Direct velocity 1.834 2082.357
Percentage direct scans 0% 31%
Page writes by reclaim 135881 5002899
Page writes file 1313 3268058
Page writes anon 134568 1734841
Page reclaim immediate 111 8048646
Page rescued immediate 0 0
Slabs scanned 1532032 8295040
2) kernel verion 3.10-rc6, mem=8G
parallelio
rc6 rc6
base-8G zswap-8G
Ops memcachetest-0M 16298.00 ( 0.00%) 15496.00 ( -4.92%)
Ops memcachetest-773M 13594.00 ( 0.00%) 12175.00 (-10.44%)
Ops memcachetest-1675M 5937.00 ( 0.00%) 8520.00 ( 43.51%)
Ops memcachetest-2577M 5972.00 ( 0.00%) 8378.00 ( 40.29%)
Ops memcachetest-3479M 2810.00 ( 0.00%) 3422.00 ( 21.78%)
Ops memcachetest-4381M 3579.00 ( 0.00%) 2760.00 (-22.88%)
Ops memcachetest-5284M 2266.00 ( 0.00%) 2166.00 ( -4.41%)
Ops io-duration-0M 0.00 ( 0.00%) 0.00 ( 0.00%)
Ops io-duration-773M 16.00 ( 0.00%) 18.00 (-12.50%)
Ops io-duration-1675M 34.00 ( 0.00%) 51.00 (-50.00%)
Ops io-duration-2577M 46.00 ( 0.00%) 55.00 (-19.57%)
Ops io-duration-3479M 64.00 ( 0.00%) 82.00 (-28.12%)
Ops io-duration-4381M 77.00 ( 0.00%) 102.00 (-32.47%)
Ops io-duration-5284M 91.00 ( 0.00%) 128.00 (-40.66%)
Ops swaptotal-0M 0.00 ( 0.00%) 0.00 ( 0.00%)
Ops swaptotal-773M 86867.00 ( 0.00%) 64298.00 ( 25.98%)
Ops swaptotal-1675M 258889.00 ( 0.00%) 119483.00 ( 53.85%)
Ops swaptotal-2577M 249005.00 ( 0.00%) 110535.00 ( 55.61%)
Ops swaptotal-3479M 317495.00 ( 0.00%) 177709.00 ( 44.03%)
Ops swaptotal-4381M 239539.00 ( 0.00%) 166971.00 ( 30.29%)
Ops swaptotal-5284M 286595.00 ( 0.00%) 154143.00 ( 46.22%)
Ops swapin-0M 0.00 ( 0.00%) 0.00 ( 0.00%)
Ops swapin-773M 39739.00 ( 0.00%) 31843.00 ( 19.87%)
Ops swapin-1675M 128380.00 ( 0.00%) 59499.00 ( 53.65%)
Ops swapin-2577M 115923.00 ( 0.00%) 55016.00 ( 52.54%)
Ops swapin-3479M 143563.00 ( 0.00%) 87498.00 ( 39.05%)
Ops swapin-4381M 115248.00 ( 0.00%) 77465.00 ( 32.78%)
Ops swapin-5284M 120546.00 ( 0.00%) 61451.00 ( 49.02%)
Ops minorfaults-0M 1526324.00 ( 0.00%) 1521203.00 ( 0.34%)
Ops minorfaults-773M 1592495.00 ( 0.00%) 1581814.00 ( 0.67%)
Ops minorfaults-1675M 1652153.00 ( 0.00%) 1603339.00 ( 2.95%)
Ops minorfaults-2577M 1648158.00 ( 0.00%) 1596046.00 ( 3.16%)
Ops minorfaults-3479M 1581514.00 ( 0.00%) 1562471.00 ( 1.20%)
Ops minorfaults-4381M 1587579.00 ( 0.00%) 1526344.00 ( 3.86%)
Ops minorfaults-5284M 1540904.00 ( 0.00%) 1506546.00 ( 2.23%)
Ops majorfaults-0M 0.00 ( 0.00%) 0.00 ( 0.00%)
Ops majorfaults-773M 5371.00 ( 0.00%) 5562.00 ( -3.56%)
Ops majorfaults-1675M 16701.00 ( 0.00%) 10265.00 ( 38.54%)
Ops majorfaults-2577M 14799.00 ( 0.00%) 9205.00 ( 37.80%)
Ops majorfaults-3479M 18279.00 ( 0.00%) 14593.00 ( 20.17%)
Ops majorfaults-4381M 14562.00 ( 0.00%) 12985.00 ( 10.83%)
Ops majorfaults-5284M 15177.00 ( 0.00%) 10454.00 ( 31.12%)
rc6 rc6
base-8G zswap-8G
User 538.84 606.68
System 2192.36 2449.63
Elapsed 8569.89 8601.77
rc6 rc6
base-8G zswap-8G
Page Ins 12122756 6361132
Page Outs 107059860 99973488
Swap Ins 2973693 1536266
Swap Outs 3489272 1716774
Direct pages scanned 0 8914107
Kswapd pages scanned 25711001 11516911
Kswapd pages reclaimed 13957707 5221461
Direct pages reclaimed 0 8848421
Kswapd efficiency 54% 45%
Kswapd velocity 3000.155 1338.900
Direct efficiency 100% 99%
Direct velocity 0.000 1036.311
Percentage direct scans 0% 43%
Page writes by reclaim 6587157 3306649
Page writes file 3097885 1589875
Page writes anon 3489272 1716774
Page reclaim immediate 25168 61862
Page rescued immediate 0 0
Slabs scanned 4218112 3148672
Direct inode steals 0 0
Kswapd inode steals 252 0
Kswapd skipped wait 0 0
--
Regards,
--Bob
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Testing results of zswap
2013-06-27 2:03 Testing results of zswap Bob Liu
@ 2013-07-15 2:56 ` Bob Liu
2013-07-15 14:56 ` Seth Jennings
` (2 subsequent siblings)
3 siblings, 0 replies; 6+ messages in thread
From: Bob Liu @ 2013-07-15 2:56 UTC (permalink / raw)
To: Linux-MM
Cc: Andrew Morton, Konrad Rzeszutek Wilk, Seth Jennings,
Dan Magenheimer, Nitin Gupta, bob.liu, Mel Gorman,
Robert Jennings
As my test results showed in this thread.
1. Zswap only useful when total ram size is large else the performance
was worse than disabled it!
2. Zswap occupies some memory but that's unfair to file pages, more
file pages maybe reclaimed during memory pressure.
I think that's why the performance of the background io-duration was
worse than disable zswap.
Seth, any feedback? Do you observe the same issue?
On Thu, Jun 27, 2013 at 10:03 AM, Bob Liu <lliubbo@gmail.com> wrote:
> Hi All,
>
> These days I have been testing zswap.
> I found that the total ram size of my testing machine effected the
> testing result.
>
> If I limit RAM size to 2G using "mem=", the performance of zswap is
> very disappointing,
> But if I use larger RAM size such as 8G, the performance is much better.
> Even with RAM size 8G, zswap will slow down the speed of parallelio.
>
> I run the testing(mmtest-0.10 with
> config-global-dhp__parallelio-memcachetest) after the default
> distribution booted every time.
>
> Below are some results:
>
> 1) kernel verion 3.10-rc6, mem=2G
>
> parallelio
> rc6 rc6
> base-2G zswap-2G
> Ops memcachetest-0M 14719.00 ( 0.00%) 10617.00 (-27.87%)
> Ops memcachetest-200M 14711.00 ( 0.00%) 10152.00 (-30.99%)
> Ops memcachetest-433M 14839.00 ( 0.00%) 10245.00 (-30.96%)
> Ops memcachetest-666M 14989.00 ( 0.00%) 10134.00 (-32.39%)
> Ops memcachetest-900M 15180.00 ( 0.00%) 9821.00 (-35.30%)
> Ops memcachetest-1133M 15654.00 ( 0.00%) 9178.00 (-41.37%)
> Ops memcachetest-1367M 16100.00 ( 0.00%) 9740.00 (-39.50%)
> Ops io-duration-0M 0.00 ( 0.00%) 0.00 ( 0.00%)
> Ops io-duration-200M 3.00 ( 0.00%) 5.00 (-66.67%)
> Ops io-duration-433M 6.00 ( 0.00%) 10.00 (-66.67%)
> Ops io-duration-666M 10.00 ( 0.00%) 14.00 (-40.00%)
> Ops io-duration-900M 13.00 ( 0.00%) 21.00 (-61.54%)
> Ops io-duration-1133M 17.00 ( 0.00%) 25.00 (-47.06%)
> Ops io-duration-1367M 20.00 ( 0.00%) 31.00 (-55.00%)
> Ops swaptotal-0M 39.00 ( 0.00%) 81481.00
> (-208825.64%)
> Ops swaptotal-200M 1400.00 ( 0.00%) 106825.00 (-7530.36%)
> Ops swaptotal-433M 16.00 ( 0.00%) 103052.00
> (-643975.00%)
> Ops swaptotal-666M 8.00 ( 0.00%) 93693.00
> (-1171062.50%)
> Ops swaptotal-900M 0.00 ( 0.00%) 91009.00 (-99.00%)
> Ops swaptotal-1133M 3609.00 ( 0.00%) 103650.00 (-2771.99%)
> Ops swaptotal-1367M 95.00 ( 0.00%) 89652.00
> (-94270.53%)
> Ops swapin-0M 39.00 ( 0.00%) 39283.00
> (-100625.64%)
> Ops swapin-200M 678.00 ( 0.00%) 50220.00 (-7307.08%)
> Ops swapin-433M 16.00 ( 0.00%) 48372.00
> (-302225.00%)
> Ops swapin-666M 8.00 ( 0.00%) 43669.00
> (-545762.50%)
> Ops swapin-900M 0.00 ( 0.00%) 42791.00 (-99.00%)
> Ops swapin-1133M 716.00 ( 0.00%) 49320.00 (-6788.27%)
> Ops swapin-1367M 91.00 ( 0.00%) 42094.00
> (-46157.14%)
> Ops minorfaults-0M 511226.00 ( 0.00%) 532594.00 ( -4.18%)
> Ops minorfaults-200M 507501.00 ( 0.00%) 577895.00 (-13.87%)
> Ops minorfaults-433M 507342.00 ( 0.00%) 573211.00 (-12.98%)
> Ops minorfaults-666M 506917.00 ( 0.00%) 565424.00 (-11.54%)
> Ops minorfaults-900M 513814.00 ( 0.00%) 569420.00 (-10.82%)
> Ops minorfaults-1133M 559981.00 ( 0.00%) 592414.00 ( -5.79%)
> Ops minorfaults-1367M 511420.00 ( 0.00%) 572809.00 (-12.00%)
> Ops majorfaults-0M 6.00 ( 0.00%) 8593.00
> (-143116.67%)
> Ops majorfaults-200M 200.00 ( 0.00%) 11335.00 (-5567.50%)
> Ops majorfaults-433M 55.00 ( 0.00%) 10729.00
> (-19407.27%)
> Ops majorfaults-666M 68.00 ( 0.00%) 9258.00
> (-13514.71%)
> Ops majorfaults-900M 94.00 ( 0.00%) 9935.00
> (-10469.15%)
> Ops majorfaults-1133M 411.00 ( 0.00%) 10902.00 (-2552.55%)
> Ops majorfaults-1367M 133.00 ( 0.00%) 9340.00 (-6922.56%)
>
> rc6 rc6
> base-2G zswap-2G
> User 1068.98 715.36
> System 3910.06 2696.03
> Elapsed 7871.94 7890.20
>
> rc6 rc6
> base-2G zswap-2G
> Page Ins 1322220 10206472
> Page Outs 24675096 31069608
> Swap Ins 100636 1548495
> Swap Outs 134568 1734841
> Direct pages scanned 14441 16430214
> Kswapd pages scanned 7792848 35445764
> Kswapd pages reclaimed 6486627 7514907
> Direct pages reclaimed 4342 2166004
> Kswapd efficiency 83% 21%
> Kswapd velocity 989.953 4492.378
> Direct efficiency 30% 13%
> Direct velocity 1.834 2082.357
> Percentage direct scans 0% 31%
> Page writes by reclaim 135881 5002899
> Page writes file 1313 3268058
> Page writes anon 134568 1734841
> Page reclaim immediate 111 8048646
> Page rescued immediate 0 0
> Slabs scanned 1532032 8295040
>
> 2) kernel verion 3.10-rc6, mem=8G
> parallelio
> rc6 rc6
> base-8G zswap-8G
> Ops memcachetest-0M 16298.00 ( 0.00%) 15496.00 ( -4.92%)
> Ops memcachetest-773M 13594.00 ( 0.00%) 12175.00 (-10.44%)
> Ops memcachetest-1675M 5937.00 ( 0.00%) 8520.00 ( 43.51%)
> Ops memcachetest-2577M 5972.00 ( 0.00%) 8378.00 ( 40.29%)
> Ops memcachetest-3479M 2810.00 ( 0.00%) 3422.00 ( 21.78%)
> Ops memcachetest-4381M 3579.00 ( 0.00%) 2760.00 (-22.88%)
> Ops memcachetest-5284M 2266.00 ( 0.00%) 2166.00 ( -4.41%)
> Ops io-duration-0M 0.00 ( 0.00%) 0.00 ( 0.00%)
> Ops io-duration-773M 16.00 ( 0.00%) 18.00 (-12.50%)
> Ops io-duration-1675M 34.00 ( 0.00%) 51.00 (-50.00%)
> Ops io-duration-2577M 46.00 ( 0.00%) 55.00 (-19.57%)
> Ops io-duration-3479M 64.00 ( 0.00%) 82.00 (-28.12%)
> Ops io-duration-4381M 77.00 ( 0.00%) 102.00 (-32.47%)
> Ops io-duration-5284M 91.00 ( 0.00%) 128.00 (-40.66%)
> Ops swaptotal-0M 0.00 ( 0.00%) 0.00 ( 0.00%)
> Ops swaptotal-773M 86867.00 ( 0.00%) 64298.00 ( 25.98%)
> Ops swaptotal-1675M 258889.00 ( 0.00%) 119483.00 ( 53.85%)
> Ops swaptotal-2577M 249005.00 ( 0.00%) 110535.00 ( 55.61%)
> Ops swaptotal-3479M 317495.00 ( 0.00%) 177709.00 ( 44.03%)
> Ops swaptotal-4381M 239539.00 ( 0.00%) 166971.00 ( 30.29%)
> Ops swaptotal-5284M 286595.00 ( 0.00%) 154143.00 ( 46.22%)
> Ops swapin-0M 0.00 ( 0.00%) 0.00 ( 0.00%)
> Ops swapin-773M 39739.00 ( 0.00%) 31843.00 ( 19.87%)
> Ops swapin-1675M 128380.00 ( 0.00%) 59499.00 ( 53.65%)
> Ops swapin-2577M 115923.00 ( 0.00%) 55016.00 ( 52.54%)
> Ops swapin-3479M 143563.00 ( 0.00%) 87498.00 ( 39.05%)
> Ops swapin-4381M 115248.00 ( 0.00%) 77465.00 ( 32.78%)
> Ops swapin-5284M 120546.00 ( 0.00%) 61451.00 ( 49.02%)
> Ops minorfaults-0M 1526324.00 ( 0.00%) 1521203.00 ( 0.34%)
> Ops minorfaults-773M 1592495.00 ( 0.00%) 1581814.00 ( 0.67%)
> Ops minorfaults-1675M 1652153.00 ( 0.00%) 1603339.00 ( 2.95%)
> Ops minorfaults-2577M 1648158.00 ( 0.00%) 1596046.00 ( 3.16%)
> Ops minorfaults-3479M 1581514.00 ( 0.00%) 1562471.00 ( 1.20%)
> Ops minorfaults-4381M 1587579.00 ( 0.00%) 1526344.00 ( 3.86%)
> Ops minorfaults-5284M 1540904.00 ( 0.00%) 1506546.00 ( 2.23%)
> Ops majorfaults-0M 0.00 ( 0.00%) 0.00 ( 0.00%)
> Ops majorfaults-773M 5371.00 ( 0.00%) 5562.00 ( -3.56%)
> Ops majorfaults-1675M 16701.00 ( 0.00%) 10265.00 ( 38.54%)
> Ops majorfaults-2577M 14799.00 ( 0.00%) 9205.00 ( 37.80%)
> Ops majorfaults-3479M 18279.00 ( 0.00%) 14593.00 ( 20.17%)
> Ops majorfaults-4381M 14562.00 ( 0.00%) 12985.00 ( 10.83%)
> Ops majorfaults-5284M 15177.00 ( 0.00%) 10454.00 ( 31.12%)
>
> rc6 rc6
> base-8G zswap-8G
> User 538.84 606.68
> System 2192.36 2449.63
> Elapsed 8569.89 8601.77
>
> rc6 rc6
> base-8G zswap-8G
> Page Ins 12122756 6361132
> Page Outs 107059860 99973488
> Swap Ins 2973693 1536266
> Swap Outs 3489272 1716774
> Direct pages scanned 0 8914107
> Kswapd pages scanned 25711001 11516911
> Kswapd pages reclaimed 13957707 5221461
> Direct pages reclaimed 0 8848421
> Kswapd efficiency 54% 45%
> Kswapd velocity 3000.155 1338.900
> Direct efficiency 100% 99%
> Direct velocity 0.000 1036.311
> Percentage direct scans 0% 43%
> Page writes by reclaim 6587157 3306649
> Page writes file 3097885 1589875
> Page writes anon 3489272 1716774
> Page reclaim immediate 25168 61862
> Page rescued immediate 0 0
> Slabs scanned 4218112 3148672
> Direct inode steals 0 0
> Kswapd inode steals 252 0
> Kswapd skipped wait 0 0
>
--
Regards,
--Bob
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Testing results of zswap
2013-06-27 2:03 Testing results of zswap Bob Liu
2013-07-15 2:56 ` Bob Liu
@ 2013-07-15 14:56 ` Seth Jennings
2013-07-16 12:16 ` Bob Liu
2013-08-06 11:37 ` Wanpeng Li
2013-08-06 11:37 ` Wanpeng Li
3 siblings, 1 reply; 6+ messages in thread
From: Seth Jennings @ 2013-07-15 14:56 UTC (permalink / raw)
To: Bob Liu
Cc: Linux-MM, Andrew Morton, Konrad Rzeszutek Wilk, Dan Magenheimer,
Nitin Gupta, bob.liu, Mel Gorman
On Mon, Jul 15, 2013 at 10:56:17AM +0800, Bob Liu wrote:
> As my test results showed in this thread.
> 1. Zswap only useful when total ram size is large else the performance
> was worse than disabled it!
I have not observed this. In my kernbench runs, I was using VMs with ~512MB
of RAM and saw significant improvement from zswap.
>
> 2. Zswap occupies some memory but that's unfair to file pages, more
> file pages maybe reclaimed during memory pressure.
This is true. It remains to be explored how the policies that balance anon
reclaim and page cache reclaim can be respected by zswap. Until then though,
the growth of the zswap pool does add memory pressure which causes more
reclaim in general, both anon and page cache.
> I think that's why the performance of the background io-duration was
> worse than disable zswap.
The I/O load during the parallelio-memcached test shouldn't be effected by
page cache reclaim since it is not re-reading anything. Again, I say that
that test is not a good and repeatable (across different systems and kernel
versions) to test zswap. parallelio-memcached is designed to test
suboptimal page reclaim decisions, not swap performance.
Have you tried running kernbench in a memory environment restricted enough
to cause swapping with zswap enabled? I think that would be a better test.
Seth
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Testing results of zswap
2013-07-15 14:56 ` Seth Jennings
@ 2013-07-16 12:16 ` Bob Liu
0 siblings, 0 replies; 6+ messages in thread
From: Bob Liu @ 2013-07-16 12:16 UTC (permalink / raw)
To: Seth Jennings
Cc: Linux-MM, Andrew Morton, Konrad Rzeszutek Wilk, Dan Magenheimer,
Nitin Gupta, bob.liu, Mel Gorman
Hi Seth,
On Mon, Jul 15, 2013 at 10:56 PM, Seth Jennings
<sjenning@linux.vnet.ibm.com> wrote:
> On Mon, Jul 15, 2013 at 10:56:17AM +0800, Bob Liu wrote:
>> As my test results showed in this thread.
>> 1. Zswap only useful when total ram size is large else the performance
>> was worse than disabled it!
>
> I have not observed this. In my kernbench runs, I was using VMs with ~512MB
> of RAM and saw significant improvement from zswap.
>
Could you confirm the results? Since zswap changed a lot from the beginning.
I tried with 1G of RAM based on kernel v3.10 with you zswap patches,
but there isn't performance improvement.
I have no idea what's the problem might be.
Using make -j4:
kernbench
base1 frontswa
base1G frontswap
User min 1025.27 ( 0.00%) 1024.79 ( 0.05%)
User mean 1025.27 ( 0.00%) 1024.79 ( 0.05%)
User stddev 0.00 ( 0.00%) 0.00 ( 0.00%)
User max 1025.27 ( 0.00%) 1024.79 ( 0.05%)
System min 52.07 ( 0.00%) 52.56 ( -0.94%)
System mean 52.07 ( 0.00%) 52.56 ( -0.94%)
System stddev 0.00 ( 0.00%) 0.00 ( 0.00%)
System max 52.07 ( 0.00%) 52.56 ( -0.94%)
Elapsed min 374.21 ( 0.00%) 370.52 ( 0.99%)
Elapsed mean 374.21 ( 0.00%) 370.52 ( 0.99%)
Elapsed stddev 0.00 ( 0.00%) 0.00 ( 0.00%)
Elapsed max 374.21 ( 0.00%) 370.52 ( 0.99%)
CPU min 287.00 ( 0.00%) 290.00 ( -1.05%)
CPU mean 287.00 ( 0.00%) 290.00 ( -1.05%)
CPU stddev 0.00 ( 0.00%) 0.00 ( 0.00%)
CPU max 287.00 ( 0.00%) 290.00 ( -1.05%)
base1 frontswa
base1G frontswap
User 1027.02 1026.44
System 52.90 53.49
Elapsed 401.51 404.19
base1 frontswa
base1G frontswap
Page Ins 1526804 1531812
Page Outs 2230280 2229688
Swap Ins 440 0
Swap Outs 2743 2
---------------------------------------------
You can see that the swapins/swapouts reduced significantly. But the
run time didn't reduced accordingly.
The same result by using make -j16:
kernbench
base1 frontsw1
base16 frontsw16
User min 1071.42 ( 0.00%) 1067.70 ( 0.35%)
User mean 1071.42 ( 0.00%) 1067.70 ( 0.35%)
User stddev 0.00 ( 0.00%) 0.00 ( 0.00%)
User max 1071.42 ( 0.00%) 1067.70 ( 0.35%)
System min 56.29 ( 0.00%) 57.06 ( -1.37%)
System mean 56.29 ( 0.00%) 57.06 ( -1.37%)
System stddev 0.00 ( 0.00%) 0.00 ( 0.00%)
System max 56.29 ( 0.00%) 57.06 ( -1.37%)
Elapsed min 360.41 ( 0.00%) 357.24 ( 0.88%)
Elapsed mean 360.41 ( 0.00%) 357.24 ( 0.88%)
Elapsed stddev 0.00 ( 0.00%) 0.00 ( 0.00%)
Elapsed max 360.41 ( 0.00%) 357.24 ( 0.88%)
CPU min 312.00 ( 0.00%) 314.00 ( -0.64%)
CPU mean 312.00 ( 0.00%) 314.00 ( -0.64%)
CPU stddev 0.00 ( 0.00%) 0.00 ( 0.00%)
CPU max 312.00 ( 0.00%) 314.00 ( -0.64%)
base1 frontsw1
base16 frontsw16
User 1073.24 1069.44
System 57.14 57.91
Elapsed 387.61 389.91
base1 frontsw1
base16 frontsw16
Page Ins 1783848 1774536
Page Outs 2241616 2238868
Swap Ins 612 22
Swap Outs 2569 40
--
Regards,
--Bob
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Testing results of zswap
2013-06-27 2:03 Testing results of zswap Bob Liu
2013-07-15 2:56 ` Bob Liu
2013-07-15 14:56 ` Seth Jennings
@ 2013-08-06 11:37 ` Wanpeng Li
2013-08-06 11:37 ` Wanpeng Li
3 siblings, 0 replies; 6+ messages in thread
From: Wanpeng Li @ 2013-08-06 11:37 UTC (permalink / raw)
To: Bob Liu
Cc: Linux-MM, Andrew Morton, Konrad Rzeszutek Wilk, Seth Jennings,
Nitin Gupta, bob.liu, Mel Gorman
[-- Attachment #1: Type: text/plain, Size: 626 bytes --]
On Thu, Jun 27, 2013 at 10:03:52AM +0800, Bob Liu wrote:
>Hi All,
>
>These days I have been testing zswap.
>I found that the total ram size of my testing machine effected the
>testing result.
>
>If I limit RAM size to 2G using "mem=", the performance of zswap is
>very disappointing,
>But if I use larger RAM size such as 8G, the performance is much better.
>Even with RAM size 8G, zswap will slow down the speed of parallelio.
>
>I run the testing(mmtest-0.10 with
>config-global-dhp__parallelio-memcachetest) after the default
>distribution booted every time.
>
Hi Bob,
I see improvement against v3.11-rc1 w/ 2G memory.
[-- Attachment #2: test --]
[-- Type: text/plain, Size: 2061 bytes --]
nozswap2 zswap2
nozswap2G zswap2G
Ops memcachetest-0M 12731.00 ( 0.00%) 11561.00 ( -9.19%)
Ops memcachetest-201M 11373.00 ( 0.00%) 11084.00 ( -2.54%)
Ops memcachetest-672M 11350.00 ( 0.00%) 10910.00 ( -3.88%)
Ops memcachetest-1142M 11057.00 ( 0.00%) 11060.00 ( 0.03%)
Ops io-duration-0M 0.00 ( 0.00%) 0.00 ( 0.00%)
Ops io-duration-201M 4.00 ( 0.00%) 5.00 (-25.00%)
Ops io-duration-672M 9.00 ( 0.00%) 7.00 ( 22.22%)
Ops io-duration-1142M 11.00 ( 0.00%) 13.00 (-18.18%)
Ops swaptotal-0M 451.00 ( 0.00%) 26208.00 (-5711.09%)
Ops swaptotal-201M 158775.00 ( 0.00%) 37548.00 ( 76.35%)
Ops swaptotal-672M 139599.00 ( 0.00%) 42514.00 ( 69.55%)
Ops swaptotal-1142M 137789.00 ( 0.00%) 34580.00 ( 74.90%)
Ops swapin-0M 451.00 ( 0.00%) 11830.00 (-2523.06%)
Ops swapin-201M 29082.00 ( 0.00%) 16169.00 ( 44.40%)
Ops swapin-672M 26611.00 ( 0.00%) 19574.00 ( 26.44%)
Ops swapin-1142M 27238.00 ( 0.00%) 15625.00 ( 42.64%)
Ops minorfaults-0M 557891.00 ( 0.00%) 575357.00 ( -3.13%)
Ops minorfaults-201M 743922.00 ( 0.00%) 595238.00 ( 19.99%)
Ops minorfaults-672M 727870.00 ( 0.00%) 653777.00 ( 10.18%)
Ops minorfaults-1142M 722946.00 ( 0.00%) 595093.00 ( 17.68%)
Ops majorfaults-0M 116.00 ( 0.00%) 4053.00 (-3393.97%)
Ops majorfaults-201M 4251.00 ( 0.00%) 4412.00 ( -3.79%)
Ops majorfaults-672M 3854.00 ( 0.00%) 4971.00 (-28.98%)
Ops majorfaults-1142M 3910.00 ( 0.00%) 4033.00 ( -3.15%)
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Testing results of zswap
2013-06-27 2:03 Testing results of zswap Bob Liu
` (2 preceding siblings ...)
2013-08-06 11:37 ` Wanpeng Li
@ 2013-08-06 11:37 ` Wanpeng Li
3 siblings, 0 replies; 6+ messages in thread
From: Wanpeng Li @ 2013-08-06 11:37 UTC (permalink / raw)
To: Bob Liu
Cc: Linux-MM, Andrew Morton, Konrad Rzeszutek Wilk, Seth Jennings,
Nitin Gupta, bob.liu, Mel Gorman
[-- Attachment #1: Type: text/plain, Size: 626 bytes --]
On Thu, Jun 27, 2013 at 10:03:52AM +0800, Bob Liu wrote:
>Hi All,
>
>These days I have been testing zswap.
>I found that the total ram size of my testing machine effected the
>testing result.
>
>If I limit RAM size to 2G using "mem=", the performance of zswap is
>very disappointing,
>But if I use larger RAM size such as 8G, the performance is much better.
>Even with RAM size 8G, zswap will slow down the speed of parallelio.
>
>I run the testing(mmtest-0.10 with
>config-global-dhp__parallelio-memcachetest) after the default
>distribution booted every time.
>
Hi Bob,
I see improvement against v3.11-rc1 w/ 2G memory.
[-- Attachment #2: test --]
[-- Type: text/plain, Size: 2061 bytes --]
nozswap2 zswap2
nozswap2G zswap2G
Ops memcachetest-0M 12731.00 ( 0.00%) 11561.00 ( -9.19%)
Ops memcachetest-201M 11373.00 ( 0.00%) 11084.00 ( -2.54%)
Ops memcachetest-672M 11350.00 ( 0.00%) 10910.00 ( -3.88%)
Ops memcachetest-1142M 11057.00 ( 0.00%) 11060.00 ( 0.03%)
Ops io-duration-0M 0.00 ( 0.00%) 0.00 ( 0.00%)
Ops io-duration-201M 4.00 ( 0.00%) 5.00 (-25.00%)
Ops io-duration-672M 9.00 ( 0.00%) 7.00 ( 22.22%)
Ops io-duration-1142M 11.00 ( 0.00%) 13.00 (-18.18%)
Ops swaptotal-0M 451.00 ( 0.00%) 26208.00 (-5711.09%)
Ops swaptotal-201M 158775.00 ( 0.00%) 37548.00 ( 76.35%)
Ops swaptotal-672M 139599.00 ( 0.00%) 42514.00 ( 69.55%)
Ops swaptotal-1142M 137789.00 ( 0.00%) 34580.00 ( 74.90%)
Ops swapin-0M 451.00 ( 0.00%) 11830.00 (-2523.06%)
Ops swapin-201M 29082.00 ( 0.00%) 16169.00 ( 44.40%)
Ops swapin-672M 26611.00 ( 0.00%) 19574.00 ( 26.44%)
Ops swapin-1142M 27238.00 ( 0.00%) 15625.00 ( 42.64%)
Ops minorfaults-0M 557891.00 ( 0.00%) 575357.00 ( -3.13%)
Ops minorfaults-201M 743922.00 ( 0.00%) 595238.00 ( 19.99%)
Ops minorfaults-672M 727870.00 ( 0.00%) 653777.00 ( 10.18%)
Ops minorfaults-1142M 722946.00 ( 0.00%) 595093.00 ( 17.68%)
Ops majorfaults-0M 116.00 ( 0.00%) 4053.00 (-3393.97%)
Ops majorfaults-201M 4251.00 ( 0.00%) 4412.00 ( -3.79%)
Ops majorfaults-672M 3854.00 ( 0.00%) 4971.00 (-28.98%)
Ops majorfaults-1142M 3910.00 ( 0.00%) 4033.00 ( -3.15%)
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2013-08-06 11:38 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2013-06-27 2:03 Testing results of zswap Bob Liu
2013-07-15 2:56 ` Bob Liu
2013-07-15 14:56 ` Seth Jennings
2013-07-16 12:16 ` Bob Liu
2013-08-06 11:37 ` Wanpeng Li
2013-08-06 11:37 ` Wanpeng Li
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).