Re: [Xen-ia64-devel] [PATCH][RFC] performance tuning TAKE 5

["Yoshinori Katase" <y_katase@xxxxxxxxxxxxxxxx>]

Hi, all

We, Fujitsu team evaluated these patches' effects from the point of view of 
vnif performance improvement.

Conclusion:
These patches are very excellent at the point of  transfer from DomU.
Details of this conclusion is published at Xen summit NOW!

patch list
11334:3f4b18d026de_avoid_long_time_interrupt_masking.patch
11335:05f4b6f69eee_perfc_for_vtlb_flush.patch
11336:2885dcbd70e0_perfc_mm_c.patch
11337:eb2a3b4b22da_perfc_dom0vp_p2m_and_m2p.patch
11338:c8f39b6e1ece_p2m_exposure_xen_side.patch
11339:33d306f87a3d_p2m_exposure_linux_side.patch
11340:fa62cd7eea19_p2m_exposure_test_module.patch
11341:1b9a35a82cf1_pervcpu_vhpt.patch
11342:b158294df78d_fix_pte_flags_conflict.patch
11343:543338390a16_import_linux_hash.h.patch
11344:1aeb66069203_tlb_track.patch
11345:7c8573def928_deferred_page_freeing.patch
11346:a1ae2f9af64d_skbuff_tlb_tracking_xen_side.patch
11347:b3bf1d38f023_skbuff_tlb_tracking_linux_side.patch
11348:9b5a7b16bfcd_tlb_track_netfront_page_xen_side.patch
11349:3a4e3fbd2956_tlb_tracking_on_netfront_page_linux_side.patch
11350:d109c748960c_tlbflush_clock.patch

patched case:     all patches applied.
unpatched case:  only 11335,11336,11337 applied.  These are added only perf 
counter.
patched-2 case:  patches applied except 11346,11347,11348,11349

measurement environment

Tiger4----------GbitEthreHUB-----------Tiger4(Native)
xen cs11333                                             RHEL4.2 memory 8GB
Dom0: RHEL4.2 memory 1GB
DomU: RHEL4.2 memory 1GB

tool : netperf2.4.1
measurement time 100sec
=========================================
Native to Native(top's load of CPU)
 client (0.7%) -> server (1.7%)
 transfer rate 764.81(Mbit/sec)
 
(Mbit/sec)????unpached????patched????patched-2
Dom0->Native??798.18?????798.36?????797.78
Native->Dom0??688.94?????752.22?????740.77

DomU->Native??356.90?????707.10?????354.95 <------LOOK!
Native->DomU??109.27?????129.69?????108.76
==========================================
Attached file is logs of measurement with xentop logs.
Special thanks Isaku!

Best regards.
katase.

----- Original Message ----- 
From: "Isaku Yamahata" <yamahata@xxxxxxxxxxxxx>
To: <xen-ia64-devel@xxxxxxxxxxxxxxxxxxx>
Sent: Thursday, August 31, 2006 12:17 PM
Subject: [Xen-ia64-devel] [PATCH][RFC] performance tuning TAKE 5


> Hi all
>
> These patches are for performance tuning TAKE 5.
> They are for comment, review and evaluation.
> I analyzed the behaviour of these patches based on performance counter.
> I'll send the mail on it separately.
>
> PATCHES:
> - performace counter
> - p2m exposure
> - per vcpu vhpt
> - tlb tracking
>  - grant table transfer 
>  - netback skbuff preregister
>  - netfront page preregister
> - deferred page freeing
> - tlb flush clock
>  Now stabilized
>
> CHANGES:
> - various bug fixes
>  especially tlb flush cloks is stabilized
>
> PATCH DETAIL:
> - per vcpu vhpt
>  It focuses on vcpu migration between physical cpus.
>  With credit scheduler, vcpu is heavily migrated.
>  This patch tries to reduce vTLB flush when vcpu is migrated.
>
> - p2m exposure
>  DMA paravirtualization requires the conversion from pseudo physical address
>  to machine address. Currently it is done by hypercall.
>  This patch tries to reduce the conversion overhead by read-only 
>  mapping the xen p2m table to domain.
> - tlb tracking
>  It forcuses on grant table mapping.
>  When page is unmapped, full vTLB flush is necessary.
>  By tracking tlb insert on grant mapped page, full vTLB flush
>  can be avoided.
>  Especially vbd does only DMA, so dom0 doesn't insert tlb entry
>  on the grant mapped page. In such case any vTLB flush isn't needed.
>  
> - netback skbuff/netfront page tlb tracking
>  This focuses on grant table transfer.
>  When page is transfered, full vTLB flush is necessary on both 
>  sender domain and receiver domain.
>  By preregistering the page, Xen/IA64 begins to track tlb insert on 
>  regestered pages.
> - deferred page freeing
>  When the page in which tlb insert isn't tracked is unmapped/zapped from
>  domain, full vTLB flush is necessary again.
>  Balloon driver and grant table page transfer is the case.
>  This patch focuses on it.
>  It tries to batch freeing/zapping page from domain in order
>  to reduce full vTLB flush.
>
> - tlb flush clock
>  This is intended to be a counter part of Xen/x86 tlb flush clock.
>  But this is used only when vcpu context switch only. not for lazy tlb flush.
>
> thanks.
> --
> yamahata

--------------------------------------------------------------------------------


> _______________________________________________
> Xen-ia64-devel mailing list
> Xen-ia64-devel@xxxxxxxxxxxxxxxxxxx
> http://lists.xensource.com/xen-ia64-devel

Attachment: yamahata.patch.log2.tar.gz
Description: GNU Zip compressed data

_______________________________________________
Xen-ia64-devel mailing list
Xen-ia64-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-ia64-devel