vhost perf testing with Qemu 5t5pc

Currently we were using 3t3pc for Qemu (2t2pc testpmd). This is
not enough anymore as we are testing VPP with 4t4pc. This patch
is supposed to increase the number of core used by testpmd.

Change-Id: Id58634c050b51943679a1f7cb74c624a98d81a27
Signed-off-by: pmikus <pmikus@cisco.com>

4 files changed, 21 insertions, 23 deletions

diff --git a/resources/libraries/robot/performance.robot b/resources/libraries/robot/performance.robot
index 28f02f9123..e19b8cfbde 100644
--- a/resources/libraries/robot/performance.robot
+++ b/resources/libraries/robot/performance.robot
@@ -1168,8 +1168,8 @@
 | Guest VM with dpdk-testpmd connected via vhost-user is setup
 | | [Documentation]
 | | ... | Start QEMU guest with two vhost-user interfaces and interconnecting
-| | ... | DPDK testpmd. Qemu Guest is using 3 cores pinned to physical cores 5,
-| | ... | 6, 7 and 2048M. Testpmd is using 3 cores (1 main core and 2 cores
+| | ... | DPDK testpmd. Qemu Guest is using 5 cores pinned to physical cores
+| | ... | 5-9, and 2048M. Testpmd is using 5 cores (1 main core and 4 cores
 | | ... | dedicated to io) mem-channel=4, txq/rxq=256, burst=64,
 | | ... | disable-hw-vlan, disable-rss, driver usr/lib/librte_pmd_virtio.so
 | | ... | and fwd mode is io.
@@ -1190,24 +1190,23 @@
 | | Run keyword | ${vm_name}.Qemu Add Vhost User If | ${sock1}
 | | Run keyword | ${vm_name}.Qemu Add Vhost User If | ${sock2}
 | | Run keyword | ${vm_name}.Qemu Set Node | ${dut_node}
-| | Run keyword | ${vm_name}.Qemu Set Smp | 3 | 3 | 1 | 1
+| | Run keyword | ${vm_name}.Qemu Set Smp | 5 | 5 | 1 | 1
 | | Run keyword | ${vm_name}.Qemu Set Mem Size | 2048
 | | Run keyword | ${vm_name}.Qemu Set Disk Image | ${glob_vm_image}
 | | ${vm}= | Run keyword | ${vm_name}.Qemu Start
-| | Run keyword | ${vm_name}.Qemu Set Affinity | 5 | 6 | 7
+| | Run keyword | ${vm_name}.Qemu Set Affinity | 5 | 6 | 7 | 8 | 9
 | | Run keyword | ${vm_name}.Qemu Set Scheduler Policy
-| | Dpdk Testpmd Start | ${vm} | eal_coremask=0x7
+| | Dpdk Testpmd Start | ${vm} | eal_coremask=0x1f
 | | ...                | eal_mem_channels=4
 | | ...                | pmd_fwd_mode=io
 | | ...                | pmd_disable_hw_vlan=${True}
-| | ...                | pmd_disable_rss=${True}
 | | Return From Keyword | ${vm}
 
 | Guest VM with dpdk-testpmd-mac connected via vhost-user is setup
 | | [Documentation]
 | | ... | Start QEMU guest with two vhost-user interfaces and interconnecting
-| | ... | DPDK testpmd. Qemu Guest is using 3 cores pinned to physical cores 5,
-| | ... | 6, 7 and 2048M. Testpmd is using 3 cores (1 main core and 2 cores
+| | ... | DPDK testpmd. Qemu Guest is using 5 cores pinned to physical cores
+| | ... | 5-9 and 2048M. Testpmd is using 5 cores (1 main core and 4 cores
 | | ... | dedicated to io) mem-channel=4, txq/rxq=256, burst=64,
 | | ... | disable-hw-vlan, disable-rss, driver usr/lib/librte_pmd_virtio.so
 | | ... | and fwd mode is mac rewrite.
@@ -1232,19 +1231,18 @@
 | | Run keyword | ${vm_name}.Qemu Add Vhost User If | ${sock1}
 | | Run keyword | ${vm_name}.Qemu Add Vhost User If | ${sock2}
 | | Run keyword | ${vm_name}.Qemu Set Node | ${dut_node}
-| | Run keyword | ${vm_name}.Qemu Set Smp | 3 | 3 | 1 | 1
+| | Run keyword | ${vm_name}.Qemu Set Smp | 5 | 5 | 1 | 1
 | | Run keyword | ${vm_name}.Qemu Set Mem Size | 2048
 | | Run keyword | ${vm_name}.Qemu Set Disk Image | ${glob_vm_image}
 | | ${vm}= | Run keyword | ${vm_name}.Qemu Start
-| | Run keyword | ${vm_name}.Qemu Set Affinity | 5 | 6 | 7
+| | Run keyword | ${vm_name}.Qemu Set Affinity | 5 | 6 | 7 | 8 | 9
 | | Run keyword | ${vm_name}.Qemu Set Scheduler Policy
-| | Dpdk Testpmd Start | ${vm} | eal_coremask=0x7
+| | Dpdk Testpmd Start | ${vm} | eal_coremask=0x1f
 | | ...                | eal_mem_channels=4
 | | ...                | pmd_fwd_mode=mac
 | | ...                | pmd_eth_peer_0=0,${eth0_mac}
 | | ...                | pmd_eth_peer_1=1,${eth1_mac}
 | | ...                | pmd_disable_hw_vlan=${True}
-| | ...                | pmd_disable_rss=${True}
 | | Return From Keyword | ${vm}
 
 | Guest VM with Linux Bridge connected via vhost-user is setup
diff --git a/tests/perf/Long_Bridge_Domain_Vhost_Intel-X520-DA2.robot b/tests/perf/Long_Bridge_Domain_Vhost_Intel-X520-DA2.robot
index 3764b7af11..cef94c1b44 100644
--- a/tests/perf/Long_Bridge_Domain_Vhost_Intel-X520-DA2.robot
+++ b/tests/perf/Long_Bridge_Domain_Vhost_Intel-X520-DA2.robot
@@ -39,10 +39,10 @@
 | ... | *[Cfg] DUT configuration:* DUT1 and DUT2 are configured with L2 bridge-
 | ... | domain and MAC learning enabled. Qemu Guest is connected to VPP via
 | ... | vhost-user interfaces. Guest is running DPDK testpmd interconnecting
-| ... | vhost-user interfaces using 3 cores pinned to cpus 5,6,7 and 2048M
-| ... | memory. Testpmd is using socket-mem=1024M (512x2M hugepages), 3 cores
-| ... | (1 main core and 2 cores dedicated for io), forwarding mode is set to
-| ... | io, rxq/txq=2048, burst=64. DUT1, DUT2 are tested with 2p10GE NIC X520
+| ... | vhost-user interfaces using 5 cores pinned to cpus 5-9 and 2048M
+| ... | memory. Testpmd is using socket-mem=1024M (512x2M hugepages), 5 cores
+| ... | (1 main core and 4 cores dedicated for io), forwarding mode is set to
+| ... | io, rxd/txd=256, burst=64. DUT1, DUT2 are tested with 2p10GE NIC X520
 | ... | Niantic by Intel.
 | ... | *[Ver] TG verification:* TG finds and reports throughput NDR (Non Drop
 | ... | Rate) with zero packet loss tolerance or throughput PDR (Partial Drop
diff --git a/tests/perf/Long_IPv4_Vhost_Intel-X520-DA2.robot b/tests/perf/Long_IPv4_Vhost_Intel-X520-DA2.robot
index 2d29d48c48..4ee03b30ae 100644
--- a/tests/perf/Long_IPv4_Vhost_Intel-X520-DA2.robot
+++ b/tests/perf/Long_IPv4_Vhost_Intel-X520-DA2.robot
@@ -39,10 +39,10 @@
 | ... | *[Cfg] DUT configuration:* DUT1 and DUT2 are configured with IPv4
 | ... | routing and two static IPv4 /24 route entries. Qemu Guest is connected
 | ... | to VPP via vhost-user interfaces. Guest is running DPDK testpmd
-| ... | interconnecting vhost-user interfaces using 3 cores pinned to cpus 5,6,7
-| ... |  and 2048M memory. Testpmd is using socket-mem=1024M (512x2M hugepages),
-| ... | 3 cores (1 main core and 2 cores dedicated for io), forwarding mode is
-| ... | set to mac, rxq/txq=2048, burst=64. DUT1, DUT2 are tested with 2p10GE
+| ... | interconnecting vhost-user interfaces using 5 cores pinned to cpus 5-9
+| ... | and 2048M memory. Testpmd is using socket-mem=1024M (512x2M hugepages),
+| ... | 5 cores (1 main core and 4 cores dedicated for io), forwarding mode is
+| ... | set to mac, rxd/txd=256, burst=64. DUT1, DUT2 are tested with 2p10GE
 | ... | NIC X520 Niantic by Intel.
 | ... | *[Ver] TG verification:* TG finds and reports throughput NDR (Non Drop
 | ... | Rate) with zero packet loss tolerance or throughput PDR (Partial Drop
diff --git a/tests/perf/Long_Xconnect_Vhost_Intel-X520-DA2.robot b/tests/perf/Long_Xconnect_Vhost_Intel-X520-DA2.robot
index 6eb8547fea..d64b325068 100644
--- a/tests/perf/Long_Xconnect_Vhost_Intel-X520-DA2.robot
+++ b/tests/perf/Long_Xconnect_Vhost_Intel-X520-DA2.robot
@@ -39,9 +39,9 @@
 | ... | *[Cfg] DUT configuration:* DUT1 and DUT2 are configured with L2 cross-
 | ... | connect. Qemu Guest is connected to VPP via vhost-user interfaces.
 | ... | Guest is running DPDK testpmd interconnecting vhost-user interfaces
-| ... | using 3 cores pinned to cpus 5,6,7 and 2048M memory. Testpmd is using
-| ... | socket-mem=1024M (512x2M hugepages), 3 cores (1 main core and 2 cores
-| ... | 2 cores dedicated for io), forwarding mode is set to io, rxq/txq=2048,
+| ... | using 5 cores pinned to cpus 5-9 and 2048M memory. Testpmd is using
+| ... | socket-mem=1024M (512x2M hugepages), 5 cores (1 main core and 4 cores
+| ... | dedicated for io), forwarding mode is set to io, rxd/txd=256,
 | ... | burst=64. DUT1, DUT2 are tested with 2p10GE NIC X520 Niantic by Intel.
 | ... | *[Ver] TG verification:* TG finds and reports throughput NDR (Non Drop
 | ... | Rate) with zero packet loss tolerance or throughput PDR (Partial Drop