From: "Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com>
To: benh@kernel.crashing.org, paulus@samba.org, mpe@ellerman.id.au,
Scott Wood <scottwood@freescale.com>,
Denis Kirjanov <kda@linux-powerpc.org>
Cc: linuxppc-dev@lists.ozlabs.org
Subject: Re: [PATCH V2 07/10] powerpc/mm: update PTE frag size
Date: Fri, 27 Nov 2015 12:57:56 +0530 [thread overview]
Message-ID: <87egfbgav7.fsf@linux.vnet.ibm.com> (raw)
In-Reply-To: <1448274825-30289-8-git-send-email-aneesh.kumar@linux.vnet.ibm.com>
"Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com> writes:
> Now that we don't track 4k subpage information we can use 2K PTE
> fragments.
>
> Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
> ---
> arch/powerpc/include/asm/book3s/64/hash-64k.h | 6 +++---
> 1 file changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/arch/powerpc/include/asm/book3s/64/hash-64k.h b/arch/powerpc/include/asm/book3s/64/hash-64k.h
> index 5062c6d423fd..a28dbfe2baed 100644
> --- a/arch/powerpc/include/asm/book3s/64/hash-64k.h
> +++ b/arch/powerpc/include/asm/book3s/64/hash-64k.h
> @@ -39,14 +39,14 @@
> */
> #define PTE_RPN_SHIFT (30)
> /*
> - * we support 8 fragments per PTE page of 64K size.
> + * we support 32 fragments per PTE page of 64K size.
> */
> -#define PTE_FRAG_NR 8
> +#define PTE_FRAG_NR 32
> /*
> * We use a 2K PTE page fragment and another 4K for storing
> * real_pte_t hash index. Rounding the entire thing to 8K
> */
> -#define PTE_FRAG_SIZE_SHIFT 13
> +#define PTE_FRAG_SIZE_SHIFT 11
> #define PTE_FRAG_SIZE (1UL << PTE_FRAG_SIZE_SHIFT)
>
This break THP with 4k hpte support because we need to track 4096
subpage information, and we have only 2048 bytes after this change.
Another thing I noticed is the impact of not tracking subpage
information. We do see some significant impact as shown by the mmtest
results below. The plan now is to go back to 4K pte framgments but
instead of using 16 bits to track 4k subpage valid bit in pte, we use only 4
bits as the last patch in this series does ("[PATCH V2 10/10]
powerpc/mm: Optmize the hashed subpage iteration"). We will track the
secondary and slot information on the second half. This will result in us using
hidx value 0x0, wrongly. This actually indicate primary hash with slot
number zero. But since we are not going to track individual 4k
subpage information we may using slot 0 wrongly. I checked the existing
code and we should be able to handle that case gracefuly.
aim9
guest guest
without-patch with-patch
Min page_test 467386.67 ( 0.00%) 330480.00 (-29.29%)
Min brk_test 4231312.46 ( 0.00%) 4217133.33 ( -0.34%)
Min exec_test 1015.66 ( 0.00%) 610.33 (-39.91%)
Min fork_test 2208.92 ( 0.00%) 1556.89 (-29.52%)
Hmean page_test 475149.52 ( 0.00%) 334539.38 (-29.59%)
Hmean brk_test 4277644.85 ( 0.00%) 4294441.01 ( 0.39%)
Hmean exec_test 1042.32 ( 0.00%) 625.09 (-40.03%)
Hmean fork_test 2391.28 ( 0.00%) 1663.53 (-30.43%)
Stddev page_test 3315.58 ( 0.00%) 1978.01 ( 40.34%)
Stddev brk_test 42167.57 ( 0.00%) 76647.45 (-81.77%)
Stddev exec_test 18.02 ( 0.00%) 8.16 ( 54.73%)
Stddev fork_test 77.69 ( 0.00%) 49.06 ( 36.84%)
CoeffVar page_test 0.70 ( 0.00%) 0.59 ( 15.27%)
CoeffVar brk_test 0.99 ( 0.00%) 1.78 (-81.02%)
CoeffVar exec_test 1.73 ( 0.00%) 1.30 ( 24.50%)
CoeffVar fork_test 3.25 ( 0.00%) 2.95 ( 9.20%)
Max page_test 479513.33 ( 0.00%) 338526.67 (-29.40%)
Max brk_test 4412066.67 ( 0.00%) 4454430.38 ( 0.96%)
Max exec_test 1071.62 ( 0.00%) 637.00 (-40.56%)
Max fork_test 2500.00 ( 0.00%) 1732.18 (-30.71%)
guest guest
without-patch with-patch
User 0.23 4.49
System 0.44 2.43
Elapsed 723.30 726.21
guest guest
without-patch with-patch
Minor Faults 38893956 25714580
Major Faults 25 228
Swap Ins 0 0
Swap Outs 0 0
Allocation stalls 0 0
DMA allocs 12760829 8717992
DMA32 allocs 0 0
Normal allocs 0 0
Movable allocs 0 0
Direct pages scanned 0 0
Kswapd pages scanned 0 0
Kswapd pages reclaimed 0 0
Direct pages reclaimed 0 0
Kswapd efficiency 100% 100%
Kswapd velocity 0.000 0.000
Direct efficiency 100% 100%
Direct velocity 0.000 0.000
Percentage direct scans 0% 0%
Zone normal velocity 0.000 0.000
Zone dma32 velocity 0.000 0.000
Zone dma velocity 0.000 0.000
Page writes by reclaim 0.000 0.000
Page writes file 0 0
Page writes anon 0 0
Page reclaim immediate 0 0
Sector Reads 4252 27096
Sector Writes 69488 69948
Page rescued immediate 0 0
Slabs scanned 0 0
Direct inode steals 0 0
Kswapd inode steals 0 0
Kswapd skipped wait 0 0
THP fault alloc 0 0
THP collapse alloc 0 0
THP splits 0 0
THP fault fallback 0 0
THP collapse fail 0 0
Compaction stalls 0 0
Compaction success 0 0
Compaction failures 0 0
Page migrate success 0 0
Page migrate failure 0 0
Compaction pages isolated 0 0
Compaction migrate scanned 0 0
Compaction free scanned 0 0
Compaction cost 0 0
NUMA alloc hit 12757558 8716737
NUMA alloc miss 0 0
NUMA interleave hit 0 0
NUMA alloc local 12757558 8716737
NUMA base PTE updates 646 648
NUMA huge PMD updates 0 0
NUMA page range updates 646 648
NUMA hint faults 878759 610567
NUMA hint local faults 878759 610567
NUMA hint local percent 100 100
NUMA pages migrated 0 0
AutoNUMA cost 4393% 3052%
pft timings
guest guest
without-patch with-patch
Min system-1 0.0900 ( 0.00%) 0.0600 ( 33.33%)
Min system-3 0.1500 ( 0.00%) 0.1300 ( 13.33%)
Min system-5 0.2400 ( 0.00%) 0.2000 ( 16.67%)
Min system-7 0.2700 ( 0.00%) 0.2600 ( 3.70%)
Min system-8 0.2400 ( 0.00%) 0.2400 ( 0.00%)
Min elapsed-1 0.1000 ( 0.00%) 0.0800 ( 20.00%)
Min elapsed-3 0.0700 ( 0.00%) 0.0600 ( 14.29%)
Min elapsed-5 0.0600 ( 0.00%) 0.0600 ( 0.00%)
Min elapsed-7 0.0600 ( 0.00%) 0.0600 ( 0.00%)
Min elapsed-8 0.0600 ( 0.00%) 0.0600 ( 0.00%)
Amean system-1 0.1896 ( 0.00%) 0.1071 ( 43.51%)
Amean system-3 0.1805 ( 0.00%) 0.1640 ( 9.14%)
Amean system-5 0.3372 ( 0.00%) 0.3261 ( 3.30%)
Amean system-7 0.3609 ( 0.00%) 0.4526 (-25.42%)
Amean system-8 0.4686 ( 0.00%) 0.4309 ( 8.06%)
Amean elapsed-1 0.2313 ( 0.00%) 0.1281 ( 44.59%)
Amean elapsed-3 0.0779 ( 0.00%) 0.0714 ( 8.35%)
Amean elapsed-5 0.0916 ( 0.00%) 0.0898 ( 2.05%)
Amean elapsed-7 0.0711 ( 0.00%) 0.0877 (-23.37%)
Amean elapsed-8 0.0786 ( 0.00%) 0.0749 ( 4.77%)
Stddev system-1 0.3466 ( 0.00%) 0.2874 ( 17.07%)
Stddev system-3 0.0228 ( 0.00%) 0.0278 (-21.83%)
Stddev system-5 0.1883 ( 0.00%) 0.1865 ( 1.01%)
Stddev system-7 0.0437 ( 0.00%) 0.3484 (-696.61%)
Stddev system-8 0.2732 ( 0.00%) 0.2877 ( -5.33%)
Stddev elapsed-1 0.4832 ( 0.00%) 0.3672 ( 24.01%)
Stddev elapsed-3 0.0085 ( 0.00%) 0.0095 (-11.61%)
Stddev elapsed-5 0.0464 ( 0.00%) 0.0515 (-11.10%)
Stddev elapsed-7 0.0063 ( 0.00%) 0.0645 (-920.30%)
Stddev elapsed-8 0.0379 ( 0.00%) 0.0430 (-13.49%)
CoeffVar system-1 182.7586 ( 0.00%) 268.2875 (-46.80%)
CoeffVar system-3 12.6305 ( 0.00%) 16.9365 (-34.09%)
CoeffVar system-5 55.8482 ( 0.00%) 57.1714 ( -2.37%)
CoeffVar system-7 12.1199 ( 0.00%) 76.9777 (-535.14%)
CoeffVar system-8 58.2889 ( 0.00%) 66.7748 (-14.56%)
CoeffVar elapsed-1 208.9340 ( 0.00%) 286.5759 (-37.16%)
CoeffVar elapsed-3 10.8759 ( 0.00%) 13.2441 (-21.78%)
CoeffVar elapsed-5 50.6194 ( 0.00%) 57.4107 (-13.42%)
CoeffVar elapsed-7 8.8904 ( 0.00%) 73.5236 (-727.00%)
CoeffVar elapsed-8 48.1691 ( 0.00%) 57.4055 (-19.18%)
Max system-1 2.9900 ( 0.00%) 2.6600 ( 11.04%)
Max system-3 0.2900 ( 0.00%) 0.2700 ( 6.90%)
Max system-5 1.5200 ( 0.00%) 1.4000 ( 7.89%)
Max system-7 0.5700 ( 0.00%) 2.6200 (-359.65%)
Max system-8 1.8700 ( 0.00%) 1.9300 ( -3.21%)
Max elapsed-1 4.1200 ( 0.00%) 3.3900 ( 17.72%)
Max elapsed-3 0.1100 ( 0.00%) 0.1000 ( 9.09%)
Max elapsed-5 0.3800 ( 0.00%) 0.3700 ( 2.63%)
Max elapsed-7 0.1000 ( 0.00%) 0.4900 (-390.00%)
Max elapsed-8 0.2700 ( 0.00%) 0.2900 ( -7.41%)
pft faults
guest guest
without-patch with-patch
Min faults/cpu-1 4339.4990 ( 0.00%) 4883.1470 ( 12.53%)
Min faults/cpu-3 42093.3690 ( 0.00%) 45489.7530 ( 8.07%)
Min faults/cpu-5 8458.7940 ( 0.00%) 9107.9580 ( 7.67%)
Min faults/cpu-7 21747.4000 ( 0.00%) 4824.7950 (-77.81%)
Min faults/cpu-8 6778.2410 ( 0.00%) 6542.3320 ( -3.48%)
Min faults/sec-1 3168.3390 ( 0.00%) 3844.8860 ( 21.35%)
Min faults/sec-3 122912.2040 ( 0.00%) 124417.2460 ( 1.22%)
Min faults/sec-5 33891.4780 ( 0.00%) 35242.9170 ( 3.99%)
Min faults/sec-7 126138.4850 ( 0.00%) 26380.8510 (-79.09%)
Min faults/sec-8 47493.9370 ( 0.00%) 44882.3140 ( -5.50%)
Hmean faults/cpu-1 65728.7295 ( 0.00%) 112975.9206 ( 71.88%)
Hmean faults/cpu-3 67715.8785 ( 0.00%) 74340.0153 ( 9.78%)
Hmean faults/cpu-5 36339.2283 ( 0.00%) 37396.4486 ( 2.91%)
Hmean faults/cpu-7 33959.8828 ( 0.00%) 27199.2785 (-19.91%)
Hmean faults/cpu-8 26145.9644 ( 0.00%) 28495.1995 ( 8.99%)
Hmean faults/sec-1 56465.3726 ( 0.00%) 100362.9738 ( 77.74%)
Hmean faults/sec-3 166282.3786 ( 0.00%) 183015.5259 ( 10.06%)
Hmean faults/sec-5 142135.4993 ( 0.00%) 144915.9134 ( 1.96%)
Hmean faults/sec-7 182191.8573 ( 0.00%) 147016.2008 (-19.31%)
Hmean faults/sec-8 166712.4745 ( 0.00%) 175603.5422 ( 5.33%)
Stddev faults/cpu-1 21957.2462 ( 0.00%) 23554.5092 ( -7.27%)
Stddev faults/cpu-3 7091.4813 ( 0.00%) 9985.8867 (-40.82%)
Stddev faults/cpu-5 7541.9566 ( 0.00%) 8947.1204 (-18.63%)
Stddev faults/cpu-7 3980.6287 ( 0.00%) 8638.2507 (-117.01%)
Stddev faults/cpu-8 6429.2045 ( 0.00%) 7858.0479 (-22.22%)
Stddev faults/sec-1 21569.0114 ( 0.00%) 21640.5914 ( -0.33%)
Stddev faults/sec-3 14715.2173 ( 0.00%) 21064.4060 (-43.15%)
Stddev faults/sec-5 30084.3765 ( 0.00%) 34040.2952 (-13.15%)
Stddev faults/sec-7 13592.3120 ( 0.00%) 45590.2568 (-235.41%)
Stddev faults/sec-8 31470.9091 ( 0.00%) 40231.8893 (-27.84%)
CoeffVar faults/cpu-1 23.2163 ( 0.00%) 14.9965 ( 35.41%)
CoeffVar faults/cpu-3 10.3387 ( 0.00%) 13.1506 (-27.20%)
CoeffVar faults/cpu-5 18.9342 ( 0.00%) 21.4360 (-13.21%)
CoeffVar faults/cpu-7 11.5648 ( 0.00%) 26.3385 (-127.75%)
CoeffVar faults/cpu-8 22.0578 ( 0.00%) 23.8259 ( -8.02%)
CoeffVar faults/sec-1 23.8979 ( 0.00%) 14.7121 ( 38.44%)
CoeffVar faults/sec-3 8.7752 ( 0.00%) 11.3428 (-29.26%)
CoeffVar faults/sec-5 19.3505 ( 0.00%) 21.0165 ( -8.61%)
CoeffVar faults/sec-7 7.4160 ( 0.00%) 25.8994 (-249.24%)
CoeffVar faults/sec-8 17.3858 ( 0.00%) 20.3632 (-17.13%)
Max faults/cpu-1 135058.1560 ( 0.00%) 179611.8510 ( 32.99%)
Max faults/cpu-3 81575.7040 ( 0.00%) 91510.1490 ( 12.18%)
Max faults/cpu-5 50969.3560 ( 0.00%) 62628.1740 ( 22.87%)
Max faults/cpu-7 45106.2770 ( 0.00%) 47090.8720 ( 4.40%)
Max faults/cpu-8 50540.8730 ( 0.00%) 51169.9640 ( 1.24%)
Max faults/sec-1 130267.4950 ( 0.00%) 159347.5220 ( 22.32%)
Max faults/sec-3 193635.1080 ( 0.00%) 223454.3190 ( 15.40%)
Max faults/sec-5 204045.3880 ( 0.00%) 219350.7390 ( 7.50%)
Max faults/sec-7 217209.2480 ( 0.00%) 225028.8150 ( 3.60%)
Max faults/sec-8 215116.6780 ( 0.00%) 231008.5060 ( 7.39%)
guest guest
without-patch with-patch
User 15.01 16.37
System 146.03 164.43
Elapsed 52.21 57.82
guest guest
without-patch with-patch
Minor Faults 5481327 5493919
Major Faults 0 90
Swap Ins 0 0
Swap Outs 0 0
Allocation stalls 0 0
DMA allocs 5271143 5277377
DMA32 allocs 0 0
Normal allocs 0 0
Movable allocs 0 0
Direct pages scanned 0 0
Kswapd pages scanned 0 0
Kswapd pages reclaimed 0 0
Direct pages reclaimed 0 0
Kswapd efficiency 100% 100%
Kswapd velocity 0.000 0.000
Direct efficiency 100% 100%
Direct velocity 0.000 0.000
Percentage direct scans 0% 0%
Zone normal velocity 0.000 0.000
Zone dma32 velocity 0.000 0.000
Zone dma velocity 0.000 0.000
Page writes by reclaim 0.000 0.000
Page writes file 0 0
Page writes anon 0 0
Page reclaim immediate 0 0
Sector Reads 72 14632
Sector Writes 508 3760
Page rescued immediate 0 0
Slabs scanned 0 0
Direct inode steals 0 0
Kswapd inode steals 0 0
Kswapd skipped wait 0 0
THP fault alloc 0 0
THP collapse alloc 0 0
THP splits 0 0
THP fault fallback 0 0
THP collapse fail 0 0
Compaction stalls 0 0
Compaction success 0 0
Compaction failures 0 0
Page migrate success 0 0
Page migrate failure 0 0
Compaction pages isolated 0 0
Compaction migrate scanned 0 0
Compaction free scanned 0 0
Compaction cost 0 0
NUMA alloc hit 5271051 5277260
NUMA alloc miss 0 0
NUMA interleave hit 0 0
NUMA alloc local 5271051 5277260
NUMA base PTE updates 19661 11194
NUMA huge PMD updates 0 0
NUMA page range updates 19661 11194
NUMA hint faults 19 14
NUMA hint local faults 19 14
NUMA hint local percent 100 100
NUMA pages migrated 0 0
AutoNUMA cost 0% 0%
ebizzy Overall Throughput
guest guest
without-patch with-patch
Min Rsec-1 6614.00 ( 0.00%) 4366.00 (-33.99%)
Min Rsec-3 10621.00 ( 0.00%) 7221.00 (-32.01%)
Min Rsec-5 10792.00 ( 0.00%) 7634.00 (-29.26%)
Min Rsec-7 10994.00 ( 0.00%) 7649.00 (-30.43%)
Min Rsec-12 13824.00 ( 0.00%) 7520.00 (-45.60%)
Min Rsec-18 12508.00 ( 0.00%) 7465.00 (-40.32%)
Min Rsec-24 14684.00 ( 0.00%) 9897.00 (-32.60%)
Min Rsec-30 14917.00 ( 0.00%) 10430.00 (-30.08%)
Min Rsec-32 14566.00 ( 0.00%) 10135.00 (-30.42%)
Hmean Rsec-1 6693.18 ( 0.00%) 4393.27 (-34.36%)
Hmean Rsec-3 10900.07 ( 0.00%) 7536.06 (-30.86%)
Hmean Rsec-5 11428.67 ( 0.00%) 7776.80 (-31.95%)
Hmean Rsec-7 11632.77 ( 0.00%) 7862.60 (-32.41%)
Hmean Rsec-12 14513.82 ( 0.00%) 7719.10 (-46.82%)
Hmean Rsec-18 13291.53 ( 0.00%) 7785.99 (-41.42%)
Hmean Rsec-24 14897.25 ( 0.00%) 10232.64 (-31.31%)
Hmean Rsec-30 15259.81 ( 0.00%) 10658.01 (-30.16%)
Hmean Rsec-32 14884.59 ( 0.00%) 10651.59 (-28.44%)
Stddev Rsec-1 64.40 ( 0.00%) 24.20 ( 62.41%)
Stddev Rsec-3 178.18 ( 0.00%) 170.17 ( 4.49%)
Stddev Rsec-5 374.77 ( 0.00%) 78.30 ( 79.11%)
Stddev Rsec-7 495.33 ( 0.00%) 124.71 ( 74.82%)
Stddev Rsec-12 825.33 ( 0.00%) 113.93 ( 86.20%)
Stddev Rsec-18 733.83 ( 0.00%) 393.07 ( 46.44%)
Stddev Rsec-24 118.49 ( 0.00%) 279.08 (-135.54%)
Stddev Rsec-30 270.38 ( 0.00%) 145.14 ( 46.32%)
Stddev Rsec-32 342.51 ( 0.00%) 441.92 (-29.03%)
CoeffVar Rsec-1 0.96 ( 0.00%) 0.55 ( 42.73%)
CoeffVar Rsec-3 1.63 ( 0.00%) 2.26 (-38.10%)
CoeffVar Rsec-5 3.28 ( 0.00%) 1.01 ( 69.27%)
CoeffVar Rsec-7 4.25 ( 0.00%) 1.59 ( 62.69%)
CoeffVar Rsec-12 5.67 ( 0.00%) 1.48 ( 73.97%)
CoeffVar Rsec-18 5.50 ( 0.00%) 5.04 ( 8.51%)
CoeffVar Rsec-24 0.80 ( 0.00%) 2.73 (-242.68%)
CoeffVar Rsec-30 1.77 ( 0.00%) 1.36 ( 23.13%)
CoeffVar Rsec-32 2.30 ( 0.00%) 4.14 (-80.09%)
Max Rsec-1 6792.00 ( 0.00%) 4429.00 (-34.79%)
Max Rsec-3 11131.00 ( 0.00%) 7700.00 (-30.82%)
Max Rsec-5 11933.00 ( 0.00%) 7842.00 (-34.28%)
Max Rsec-7 12508.00 ( 0.00%) 8027.00 (-35.83%)
Max Rsec-12 16002.00 ( 0.00%) 7847.00 (-50.96%)
Max Rsec-18 14516.00 ( 0.00%) 8567.00 (-40.98%)
Max Rsec-24 15013.00 ( 0.00%) 10599.00 (-29.40%)
Max Rsec-30 15656.00 ( 0.00%) 10853.00 (-30.68%)
Max Rsec-32 15385.00 ( 0.00%) 11381.00 (-26.03%)
ebizzy Per-thread
guest guest
without-patch with-patch
Min Rsec-1 6614.00 ( 0.00%) 4366.00 (-33.99%)
Min Rsec-3 3464.00 ( 0.00%) 2387.00 (-31.09%)
Min Rsec-5 2100.00 ( 0.00%) 1486.00 (-29.24%)
Min Rsec-7 1494.00 ( 0.00%) 1016.00 (-31.99%)
Min Rsec-12 1010.00 ( 0.00%) 586.00 (-41.98%)
Min Rsec-18 571.00 ( 0.00%) 373.00 (-34.68%)
Min Rsec-24 473.00 ( 0.00%) 330.00 (-30.23%)
Min Rsec-30 398.00 ( 0.00%) 283.00 (-28.89%)
Min Rsec-32 364.00 ( 0.00%) 250.00 (-31.32%)
Hmean Rsec-1 6693.18 ( 0.00%) 4393.27 (-34.36%)
Hmean Rsec-3 3627.65 ( 0.00%) 2504.84 (-30.95%)
Hmean Rsec-5 2283.84 ( 0.00%) 1554.53 (-31.93%)
Hmean Rsec-7 1637.70 ( 0.00%) 1121.58 (-31.51%)
Hmean Rsec-12 1192.25 ( 0.00%) 642.41 (-46.12%)
Hmean Rsec-18 687.06 ( 0.00%) 428.11 (-37.69%)
Hmean Rsec-24 606.35 ( 0.00%) 415.99 (-31.39%)
Hmean Rsec-30 497.96 ( 0.00%) 349.99 (-29.72%)
Hmean Rsec-32 455.21 ( 0.00%) 323.05 (-29.03%)
Stddev Rsec-1 64.40 ( 0.00%) 24.20 (-62.41%)
Stddev Rsec-3 156.31 ( 0.00%) 147.43 ( -5.68%)
Stddev Rsec-5 95.61 ( 0.00%) 30.04 (-68.58%)
Stddev Rsec-7 286.41 ( 0.00%) 40.64 (-85.81%)
Stddev Rsec-12 178.07 ( 0.00%) 19.11 (-89.27%)
Stddev Rsec-18 250.94 ( 0.00%) 69.96 (-72.12%)
Stddev Rsec-24 97.93 ( 0.00%) 69.89 (-28.63%)
Stddev Rsec-30 81.68 ( 0.00%) 42.31 (-48.20%)
Stddev Rsec-32 70.76 ( 0.00%) 62.86 (-11.16%)
CoeffVar Rsec-1 0.96 ( 0.00%) 0.55 ( 42.73%)
CoeffVar Rsec-3 4.30 ( 0.00%) 5.87 (-36.40%)
CoeffVar Rsec-5 4.18 ( 0.00%) 1.93 ( 53.78%)
CoeffVar Rsec-7 17.21 ( 0.00%) 3.62 ( 78.97%)
CoeffVar Rsec-12 14.68 ( 0.00%) 2.97 ( 79.76%)
CoeffVar Rsec-18 33.90 ( 0.00%) 16.15 ( 52.35%)
CoeffVar Rsec-24 15.79 ( 0.00%) 16.40 ( -3.86%)
CoeffVar Rsec-30 16.07 ( 0.00%) 11.92 ( 25.79%)
CoeffVar Rsec-32 15.22 ( 0.00%) 18.88 (-24.05%)
Max Rsec-1 6792.00 ( 0.00%) 4429.00 (-34.79%)
Max Rsec-3 3968.00 ( 0.00%) 2818.00 (-28.98%)
Max Rsec-5 2480.00 ( 0.00%) 1607.00 (-35.20%)
Max Rsec-7 3290.00 ( 0.00%) 1193.00 (-63.74%)
Max Rsec-12 1842.00 ( 0.00%) 678.00 (-63.19%)
Max Rsec-18 1495.00 ( 0.00%) 1074.00 (-28.16%)
Max Rsec-24 987.00 ( 0.00%) 738.00 (-25.23%)
Max Rsec-30 824.00 ( 0.00%) 456.00 (-44.66%)
Max Rsec-32 690.00 ( 0.00%) 581.00 (-15.80%)
ebizzy Thread spread
guest guest
without-patch with-patch
Min spread-1 0.00 ( 0.00%) 0.00 ( 0.00%)
Min spread-3 69.00 ( 0.00%) 38.00 ( 44.93%)
Min spread-5 126.00 ( 0.00%) 54.00 ( 57.14%)
Min spread-7 79.00 ( 0.00%) 78.00 ( 1.27%)
Min spread-12 262.00 ( 0.00%) 36.00 ( 86.26%)
Min spread-18 667.00 ( 0.00%) 39.00 ( 94.15%)
Min spread-24 300.00 ( 0.00%) 156.00 ( 48.00%)
Min spread-30 330.00 ( 0.00%) 100.00 ( 69.70%)
Min spread-32 220.00 ( 0.00%) 158.00 ( 28.18%)
Hmean spread-1 0.00 ( 0.00%) 0.00 ( 0.00%)
Hmean spread-3 162.44 ( 0.00%) 125.30 ( 22.86%)
Hmean spread-5 161.03 ( 0.00%) 65.41 ( 59.38%)
Hmean spread-7 143.51 ( 0.00%) 102.37 ( 28.67%)
Hmean spread-12 376.64 ( 0.00%) 53.48 ( 85.80%)
Hmean spread-18 758.21 ( 0.00%) 58.48 ( 92.29%)
Hmean spread-24 348.60 ( 0.00%) 215.23 ( 38.26%)
Hmean spread-30 358.66 ( 0.00%) 126.49 ( 64.73%)
Hmean spread-32 264.70 ( 0.00%) 199.06 ( 24.80%)
Stddev spread-1 0.00 ( 0.00%) 0.00 ( 0.00%)
Stddev spread-3 160.65 ( 0.00%) 134.27 ( 16.42%)
Stddev spread-5 20.56 ( 0.00%) 14.92 ( 27.43%)
Stddev spread-7 667.59 ( 0.00%) 23.11 ( 96.54%)
Stddev spread-12 190.86 ( 0.00%) 11.03 ( 94.22%)
Stddev spread-18 64.39 ( 0.00%) 244.20 (-279.25%)
Stddev spread-24 63.41 ( 0.00%) 78.27 (-23.44%)
Stddev spread-30 34.19 ( 0.00%) 25.58 ( 25.20%)
Stddev spread-32 27.74 ( 0.00%) 63.89 (-130.27%)
CoeffVar spread-1 0.00 ( 0.00%) 0.00 ( 0.00%)
CoeffVar spread-3 58.00 ( 0.00%) 50.71 (-12.57%)
CoeffVar spread-5 12.54 ( 0.00%) 21.81 ( 73.99%)
CoeffVar spread-7 144.19 ( 0.00%) 21.60 (-85.02%)
CoeffVar spread-12 42.68 ( 0.00%) 19.63 (-54.00%)
CoeffVar spread-18 8.43 ( 0.00%) 139.55 (1554.83%)
CoeffVar spread-24 17.71 ( 0.00%) 33.28 ( 87.89%)
CoeffVar spread-30 9.46 ( 0.00%) 19.49 (106.17%)
CoeffVar spread-32 10.36 ( 0.00%) 29.61 (185.76%)
Max spread-1 0.00 ( 0.00%) 0.00 ( 0.00%)
Max spread-3 492.00 ( 0.00%) 386.00 ( 21.54%)
Max spread-5 185.00 ( 0.00%) 89.00 ( 51.89%)
Max spread-7 1796.00 ( 0.00%) 146.00 ( 91.87%)
Max spread-12 686.00 ( 0.00%) 69.00 ( 89.94%)
Max spread-18 869.00 ( 0.00%) 662.00 ( 23.82%)
Max spread-24 479.00 ( 0.00%) 384.00 ( 19.83%)
Max spread-30 426.00 ( 0.00%) 173.00 ( 59.39%)
Max spread-32 302.00 ( 0.00%) 312.00 ( -3.31%)
guest guest
without-patch with-patch
User 1585.46 996.79
System 6802.57 7388.01
Elapsed 1352.70 1355.62
guest guest
without-patch with-patch
Minor Faults 102375790 67314425
Major Faults 0 1
Swap Ins 0 0
Swap Outs 0 0
Allocation stalls 0 0
DMA allocs 102297177 67229359
DMA32 allocs 0 0
Normal allocs 0 0
Movable allocs 0 0
Direct pages scanned 0 0
Kswapd pages scanned 0 0
Kswapd pages reclaimed 0 0
Direct pages reclaimed 0 0
Kswapd efficiency 100% 100%
Kswapd velocity 0.000 0.000
Direct efficiency 100% 100%
Direct velocity 0.000 0.000
Percentage direct scans 0% 0%
Zone normal velocity 0.000 0.000
Zone dma32 velocity 0.000 0.000
Zone dma velocity 0.000 0.000
Page writes by reclaim 0.000 0.000
Page writes file 0 0
Page writes anon 0 0
Page reclaim immediate 0 0
Sector Reads 428 620
Sector Writes 4164 4620
Page rescued immediate 0 0
Slabs scanned 0 0
Direct inode steals 0 0
Kswapd inode steals 0 0
Kswapd skipped wait 0 0
THP fault alloc 0 0
THP collapse alloc 0 0
THP splits 0 0
THP fault fallback 0 0
THP collapse fail 0 0
Compaction stalls 0 0
Compaction success 0 0
Compaction failures 0 0
Page migrate success 0 0
Page migrate failure 0 0
Compaction pages isolated 0 0
Compaction migrate scanned 0 0
Compaction free scanned 0 0
Compaction cost 0 0
NUMA alloc hit 102297213 67229443
NUMA alloc miss 0 0
NUMA interleave hit 0 0
NUMA alloc local 102297213 67229443
NUMA base PTE updates 43814 45596
NUMA huge PMD updates 0 0
NUMA page range updates 43814 45596
NUMA hint faults 35181 36243
NUMA hint local faults 35181 36243
NUMA hint local percent 100 100
NUMA pages migrated 0 0
AutoNUMA cost 176% 181%
next prev parent reply other threads:[~2015-11-27 7:28 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-11-23 10:33 [PATCH V2 00/10] Reduce the pte framgment size Aneesh Kumar K.V
2015-11-23 10:33 ` [PATCH V2 01/10] powerpc/mm: Don't hardcode page table size Aneesh Kumar K.V
2015-11-23 10:33 ` [PATCH V2 02/10] powerpc/mm: Don't hardcode the hash pte slot shift Aneesh Kumar K.V
2015-11-23 10:33 ` [PATCH V2 03/10] powerpc/nohash: Update 64K nohash config to have 32 pte fragement Aneesh Kumar K.V
2015-11-23 10:33 ` [PATCH V2 04/10] powerpc/nohash: we don't use real_pte_t for nohash Aneesh Kumar K.V
2015-11-23 10:33 ` [PATCH V2 05/10] powerpc/mm: Use H_READ with H_READ_4 Aneesh Kumar K.V
2015-11-23 10:33 ` [PATCH V2 06/10] powerpc/mm: Don't track 4k subpage information with 64k linux page size Aneesh Kumar K.V
2015-11-23 10:33 ` [PATCH V2 07/10] powerpc/mm: update PTE frag size Aneesh Kumar K.V
2015-11-27 7:27 ` Aneesh Kumar K.V [this message]
2015-11-27 11:56 ` Aneesh Kumar K.V
2015-11-23 10:33 ` [PATCH V2 08/10] powerpc/mm: Update pte_iterate_hashed_subpages args Aneesh Kumar K.V
2015-11-23 10:33 ` [PATCH V2 09/10] powerpc/mm: Drop real_pte_t usage Aneesh Kumar K.V
2015-11-23 10:33 ` [PATCH V2 10/10] powerpc/mm: Optmize the hashed subpage iteration Aneesh Kumar K.V
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=87egfbgav7.fsf@linux.vnet.ibm.com \
--to=aneesh.kumar@linux.vnet.ibm.com \
--cc=benh@kernel.crashing.org \
--cc=kda@linux-powerpc.org \
--cc=linuxppc-dev@lists.ozlabs.org \
--cc=mpe@ellerman.id.au \
--cc=paulus@samba.org \
--cc=scottwood@freescale.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).