Re: [vpp-dev] how to load igb_uio to bind interface to dpdk?

Hi, Did you try to use vfio-pci instead? You might need to enable the no-iommu mode [1] Otherwise we do not have enough output to help. Best ben [1] https://dpdk-guide.gitlab.io/dpdk-guide/setup/binding.html#vfio-no-iommu > -Original Message- > From: vpp-dev@lists.fd.io On Behalf Of >

[vpp-dev] how to load igb_uio to bind interface to dpdk?

Hi team, I am trying to bind an interface in centos 8 vm to uio modue. uio_pci genericdriver gave up however. I am trying to enable  igb_uio module  from dpdk by editing *--- a/build/external/packages/dpdk.mk* - $(call set,RTE_EAL_IGB_UIO,n) + $(call set,RTE_EAL_IGB_UIO,y) however make build f

Re: [vpp-dev] AMD Epyc and vpp.

Hi Christian, Everything else being correct (VPP NUMA core placement vs NIC etc) and if you see a bottleneck on IO, you might need to set the NIC as 'Preferred IO' in the BIOS. Best Ben > -Original Message- > From: vpp-dev@lists.fd.io On Behalf Of Christian > Hopps > Sent: mercredi 22

回复：RE: [vpp-dev] Why not break the check flow when the first error occurs of ip4_inout()?

Hi Dave, Thanks for your reply. --  -- ??: "Dave Barach (dbarach)"

[vpp-dev] AMD Epyc and vpp.

Hi vpp-dev, Has anyone done performance analysis with the new AMD epyc processors and VPP? Just naively running my normal build shows a 3GHz Epyc machine under-performing a 2.1GHz intel xeon. Thanks, Chris. signature.asc Description: Message signed with OpenPGP -=-=-=-=-=-=-=-=-=-=-=- Links:

[vpp-dev] Jenkins Sandbox Maintenance Complete

Jenkins sandbox maintenance is complete * Migration to Auth0 o FAQ https://identity.linuxfoundation.org/migration-faq * OS and security updates * Jenkins upgrade * Plugin upgrades Please open a ticket at support.linuxfoundation.org if you experience any issues. Thank you, Anton and

Re: [vpp-dev] VPP vs Kernel proxy performance

Hi, It looks like your linux vm iperf3 client could be saturated, so not sure if there’s more that could be done. You can control fifo size from vcl.conf (rx-fifo-size and tx-fifo-size). Regards, Florin > On Jul 21, 2020, at 10:44 AM, sadhanakesa...@gmail.com wrote: > > hi Team, > i tried t

Re: [vpp-dev] VPP vs Kernel proxy performance

hi Team, i tried to run this for evaluating https://wiki.fd.io/view/VPP/HostStack/LDP/iperf with a linux client (centos 7 vm) and ubuntu kernel 4.15 (vpp server where iperf3 runs) with uio_pci_generic driver i am also getting close to 900 Mbits/sec vs 800 Mbits/sec for a linux server/client iper

Re: [vpp-dev] VPP vs Kernel proxy performance

Hi Sebastiano, The test proxy application is just an example, so it’s far from optimized. Nonetheless, last time I tested it was capable of saturating a 10Gbps nic. So some things to consider while debugging: - fifo size configuration. The wiki page does not set a fifo size and as a result a s

Re: [vpp-dev] app_send_dgram should do ip fragmentation if data more than snd_mss

Hi, By default udp computes its mss starting from a 1500 mtu. You can avoid this by either changing the default, i.e., in startup.conf under the udp stanza add “mtu ”, or alternatively you can force mss on each connect, by explicitly setting it in the remote endpoint’s transport_endpoint_cfg_t

[vpp-dev] VPP vs Kernel proxy performance

Dear all, I was trying to test the performance of the VPP Host Stack, compared to the one of the Linux kernel TCP/IP stack. In particular, I was testing the TestProxy application [1] and compare it with the simpleproxy application available at this URL [2]. My setup is composed of a server1, which

Re: [vpp-dev] Regarding new ipsec interface patch

Removed. I'm glad it works for you. Could I ask for a quid pro quo? Mark the existing ipsec interface APIs as deprecated in the new scheme. /neale tpyed by my fat tumhbs From: Dave Barach (dbarach) Sent: Monday, July 20, 2020 2:50:09 PM To: Christian Hopps ; vp

[vpp-dev] app_send_dgram should do ip fragmentation if data more than snd_mss

hi: When sending 1500 bytes data with session app_send_dgram api, I find session-queue node will split the data to two udp packets, then vpp will sending to two separate IP packets out with no fragmentation. Even I set interface MTU with 9000, It's still sending two separate IP packets. I think

Re: [vpp-dev] Why not break the check flow when the first error occurs of ip4_inout()?

Shouldn't the version check come first and the code avoid further checks if it fails? Probably unlikely to fail though given execution has reached "ip4_input" :) Thanks, Chris. > On Jul 21, 2020, at 7:12 AM, Dave Barach via lists.fd.io > wrote: > > Branches hurt performance more than arithme

Re: [vpp-dev] Why not break the check flow when the first error occurs of ip4_inout()?

Branches hurt performance more than arithmetic and/or conditional moves. From: vpp-dev@lists.fd.io On Behalf Of "?? Sent: Monday, July 20, 2020 10:17 PM To: vpp-dev Subject: [vpp-dev] Why not break the check flow when the first error occurs of ip4_inout()? Hi all, When I'm reading the ip4_inp