Re: [vpp-dev] aarch64 compilation break

On 02/20 21:26:40, Brian Brooks wrote: > On 02/20 14:43:47, Gabriel Ganne wrote: > > Hi Damjan, Brian, > > > > Damjan, your commit https://gerrit.fd.io/r/c/10670 has an impact in > > vnet_classify that overlaps with what I had written in > > https://gerrit.fd.io/r/c/10476/ > > I see this handles

Re: [vpp-dev] aarch64 compilation break

On 02/20 14:43:47, Gabriel Ganne wrote: > Hi Damjan, Brian, > > Damjan, your commit https://gerrit.fd.io/r/c/10670 has an impact in > vnet_classify that overlaps with what I had written in > https://gerrit.fd.io/r/c/10476/ I see this handles u32x4_zero_byte_mask, but I also see breakage due to

Re: [vpp-dev] Sequence of execution of process nodes

It would help if you’d describe the problem that you’re trying to solve. You can park processes so they wait on specific events, and use events to sequence processes in arbitrary ways. I can’t think of a case where I’ve wanted to do that. Dave From: vpp-dev@lists.fd.io

Re: [vpp-dev] _mm_testz_si128 Woes

The trick I like, when I don't need quite so much detail as the guide provides, is to simply remember to look in 2 places for header files, /usr/include AND /usr/lib/gcc. From /usr/lib/gcc, it is easy to find one's way down to the include directory for intrinsics. The trick seems to cover both

[vpp-dev] [FD.io Helpdesk #52672] Issues with CentOS jenkins job

Marco, The error is due to trying to install a package older than the one already installed. It appears this issue has been resolved as the centos7 builds are no longer failing due to this issue. Thank you, Vanessa On Tue Feb 20 07:09:27 2018, mvarl...@suse.de wrote: > Hi, > > This

[vpp-dev] Sequence of execution of process nodes

Hi All, Is there any way in VPP with which we can control the sequence in which process nodes are executed. If i want say process node 1 to run after process node 2, can i define it? P.S. I am new to VPP, apologies for naive question. Regards Nikhil

[vpp-dev] VPP build FAIL w MLX5 pmd

Hi, I'm following the steps as per https://community.mellanox.com/docs/DOC-2729 To build VPP with MLX5 PMD The following make command fails : make dpdk-install-dev DPDK_MLX5_PMD=y - cd /home/avi/vpp/dpdk/deb/_build/nasm-2.13.01 && sh

[vpp-dev] num-mbufs higher than 420000 is causing VPP daemon to crash #opendaylight-releng

I have tried to adjust the num-mbufs to a value higher than 42 and this resulted in daemon crash about 1.3 minutes after starting it.  This is happening on -vpp v18.04-rc0~218-g07510dd -CentOS 7.4 -   4.15.0-1.el7.elrepo.x86_64

Re: [vpp-dev] maximum number of interfaces ? #vpp

[Edited Message Follows] I managed to get this working with all 8 x 10G interfaces. it seem that the issue was dpdk_mbuf value which was a bit too low.

Re: [vpp-dev] maximum number of interfaces ? #vpp

I manages to get this working with all 8 x 10G interfaces. it seem that the issue was dpdk_mbuf value which was a bit too low.

Re: [vpp-dev] FD.io CSIT rls1801 report published

Thanks Ray, and well spotted re regressions.. We were still collecting data on those, as they have been masked before by another issue related to VPP plugins (jira CSIT-925) that confused CSIT performance trending machinery, and it was unclear if it was CSIT environment or VPP code. We made

[vpp-dev] aarch64 compilation break

Hi Damjan, Brian, Damjan, your commit https://gerrit.fd.io/r/c/10670 has an impact in vnet_classify that overlaps with what I had written in https://gerrit.fd.io/r/c/10476/ However, without some changes in vector_neon.h it has the side effect to break aarch64 compilation. I've updated mine above

[vpp-dev] Damjan Marion will chair today's fd.io vpp project call

I have an unavoidable meeting conflict. Thanks to Damjan for pinch-hitting! Dave

[vpp-dev] CONFIG_RTE_LIBRTE_MLX5_PMD - VPP build

Hi, I set the CONFIG_RTE_LIBRTE_MLX5_PMD in build-root/build-vpp-native/dpdk/dpdk-17.08/config/common_base This is probably not the place to set it , because when I rebuild VPP the CONFIG_RTE_LIBRTE_MLX5_PMD is restored to its default = 'n' Where do I have to set CONFIG_RTE_LIBRTE_MLX5_PMD

[vpp-dev] Issues with CentOS jenkins job

Hi, This morning I see another issue affecting the CentOS jenkins job. All others are fine. The error marking the build as "Verified-1" can be found below. 10:59:17 make[2]: Leaving directory `/w/workspace/vpp-verify-master- centos7/dpdk' 10:59:17 sudo rpm -Uih

Re: [vpp-dev] maximum number of interfaces ? #vpp

I started disabling dpdk on  3 of the working ports and now i got the ones that were not working before in working state. So it look like i might reach some resource limitation when enabling more than 5 interfaces at the same time.  I will need to find out what resource was falling short now.

Re: [vpp-dev] FD.io CSIT rls1801 report published

w00t, much kudos, always hugely valuable. There are some pretty significant regression on both xl710 and x520 for vhost and ipsec use cases at the bottom of the table. The most of significant of these is a 42% drop in the performance of 10ge2p1x520-64b-1t1c-eth-l2xcbase-eth-2vhostvr1024-1vm

Re: [vpp-dev] _mm_testz_si128 Woes

The Intel intrinsics guide are a handy reference for this kind of thing. https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm_testz_si128 You need to trim the additional underscore off the intrinsic to look it up. Ray K On 20/02/2018 02:52, Florin Coras wrote: Okay, I take

Re: [vpp-dev] _mm_testz_si128 Woes

Hopefully this will help: https://gerrit.fd.io/r/10670 compiles fine in my centos7 container. On 20 Feb 2018, at 03:52, Florin Coras > wrote: Okay, I take that back. Google tells me these are SSE4 intrinsics, not SSE2 … Florin On