Ethernet Products
Determine ramifications of Intel® Ethernet products and technologies
4808 Discussions

Intel 10-Gigabit X540-AT2 (rev 01) Bandwidth Not as spec

TRANG2
Beginner
8,362 Views

Hello Intel,

We have purchased 3pcs of Intel Corporation Ethernet Controller 10-Gigabit X540-AT2 (rev 01).

We have connected this to our HP Desktops in the PCI Express slot(PCI bridge: Intel Corporation Xeon E3-1200 v3/4th Gen Core Processor PCI Express x16 Controller (rev 06)).

Now when we run iperf tool to measure the bandwidth we found that when dual port are used to tx/rx we see total bandwidth as 1st port @~4Gbits/sec and 2nd port @~5Gbits/sec total for 2 ports(port1 + port2) @~9Gbits/sec

When we use only 1 port we see 9Gbps bandwidth.

But as the product spec details from the link (https://ark.intel.com/products/58954/Intel-Ethernet-Converged-Network-Adapter-X540-T2 Intel® Ethernet Converged Network Adapter X540-T2 Product Specifications ) as per port 10Gbps.

Please help us to reach the bandwidth as mentioned in the product spec. Means are we doing something wrong or any configuration in the driver that we need do.

Attachments shows how we are connected and tests performed with the results.

Thanks in advance.

0 Kudos
31 Replies
TRANG2
Beginner
3,933 Views

Adding few more attachments as this forum limits to 5 per post.

0 Kudos
idata
Employee
3,933 Views

Hello Rangam,

 

 

Thank you for posting in Intel Wired Ethernet Communities. We will review the details and provide an update soon. Please let us know if you have any questions.

 

 

Best regards,

 

Daniel D

 

Intel customer support

 

Agent under contract to Intel

 

0 Kudos
TRANG2
Beginner
3,933 Views

Hello Daniel D,

Thanks responding for our query. We would like to know any progress on the issue?

A Quick support will help us a lot in enabling our new hardware based on the same Chipset from Intel.

Thanks,

Lokesh Rangam.

0 Kudos
idata
Employee
3,933 Views

Hello Rangam,

 

 

We are working on a resolution. Could you provide additional information to investigate this issue. What is the Linux distribution and version being used? What is the model and brand of the switch the 10Gb ports are being connected to? Provide an output of ethtool -k. Thank you for your cooperation. If you have any questions please do not hesitate to ask.

 

 

Best regards,

 

Daniel D

 

Intel customer support

 

Agent under contract to Intel
0 Kudos
TRANG2
Beginner
3,933 Views

Hello Daniel,

We are using Ubuntu 16.04 and we are not using any switch in between. The connection is direct between 2 10Gb ports.

The output of ethtool -k as follows.

user@Myhost1:~/Workspace/work/src/lane410G$ ethtool -k enp1s0f0

Features for enp1s0f0:

rx-checksumming: on

tx-checksumming: on

tx-checksum-ipv4: off [fixed]

tx-checksum-ip-generic: on

tx-checksum-ipv6: off [fixed]

tx-checksum-fcoe-crc: on [fixed]

tx-checksum-sctp: on

scatter-gather: on

tx-scatter-gather: on

tx-scatter-gather-fraglist: off [fixed]

tcp-segmentation-offload: on

tx-tcp-segmentation: on

tx-tcp-ecn-segmentation: off [fixed]

tx-tcp-mangleid-segmentation: off

tx-tcp6-segmentation: on

udp-fragmentation-offload: off [fixed]

generic-segmentation-offload: on

generic-receive-offload: on

large-receive-offload: off

rx-vlan-offload: on

tx-vlan-offload: on

ntuple-filters: off

receive-hashing: on

highdma: on [fixed]

rx-vlan-filter: on

vlan-challenged: off [fixed]

tx-lockless: off [fixed]

netns-local: off [fixed]

tx-gso-robust: off [fixed]

tx-fcoe-segmentation: on [fixed]

tx-gre-segmentation: on

tx-gre-csum-segmentation: on

tx-ipxip4-segmentation: on

tx-ipxip6-segmentation: on

tx-udp_tnl-segmentation: on

tx-udp_tnl-csum-segmentation: on

tx-gso-partial: on

tx-sctp-segmentation: off [fixed]

tx-esp-segmentation: off [fixed]

fcoe-mtu: off [fixed]

tx-nocache-copy: off

loopback: off [fixed]

rx-fcs: off [fixed]

rx-all: off

tx-vlan-stag-hw-insert: off [fixed]

rx-vlan-stag-hw-parse: off [fixed]

rx-vlan-stag-filter: off [fixed]

l2-fwd-offload: off [fixed]

hw-tc-offload: off

esp-hw-offload: off [fixed]

esp-tx-csum-hw-offload: off [fixed]

user@Myhost1:~/Workspace/work/src/lane410G$ ethtool -k enp1s0f1

Features for enp1s0f1:

rx-checksumming: on

tx-checksumming: on

tx-checksum-ipv4: off [fixed]

tx-checksum-ip-generic: on

tx-checksum-ipv6: off [fixed]

tx-checksum-fcoe-crc: on [fixed]

tx-checksum-sctp: on

scatter-gather: on

tx-scatter-gather: on

tx-scatter-gather-fraglist: off [fixed]

tcp-segmentation-offload: on

tx-tcp-segmentation: on

tx-tcp-ecn-segmentation: off [fixed]

tx-tcp-mangleid-segmentation: off

tx-tcp6-segmentation: on

udp-fragmentation-offload: off [fixed]

generic-segmentation-offload: on

generic-receive-offload: on

large-receive-offload: off

rx-vlan-offload: on

tx-vlan-offload: on

ntuple-filters: off

receive-hashing: on

highdma: on [fixed]

rx-vlan-filter: on

vlan-challenged: off [fixed]

tx-lockless: off [fixed]

netns-local: off [fixed]

tx-gso-robust: off [fixed]

tx-fcoe-segmentation: on [fixed]

tx-gre-segmentation: on

tx-gre-csum-segmentation: on

tx-ipxip4-segmentation: on

tx-ipxip6-segmentation: on

tx-udp_tnl-segmentation: on

tx-udp_tnl-csum-segmentation: on

tx-gso-partial: on

tx-sctp-segmentation: off [fixed]

tx-esp-segmentation: off [fixed]

fcoe-mtu: off [fixed]

tx-nocache-copy: off

loopback: off [fixed]

rx-fcs: off [fixed]

rx-all: off

tx-vlan-stag-hw-insert: off [fixed]

rx-vlan-stag-hw-parse: off [fixed]

rx-vlan-stag-filter: off [fixed]

l2-fwd-offload: off [fixed]

hw-tc-offload: off

esp-hw-offload: off [fixed]

esp-tx-csum-hw-offload: off [fixed]

Thanks,

Trilokesh Rangam.

0 Kudos
idata
Employee
3,933 Views

Hello Rangam,

 

 

Thank you for the quick response. We are working on a solution. We will update you as soon as possible.

 

 

Best regards,

 

Daniel D

 

Intel customer support

 

Agent under contract to Intel
0 Kudos
idata
Employee
3,933 Views

Hello Rangam,

 

 

Please try to run iperf between a different pair of adapters and with different cables. If that does not work try to run the test on another operating system and another system entirely to isolate the issue. If you have any questions please do not hesitate to contact us.

 

 

Best regards,

 

Daniel D

 

Intel customer support

 

Agent under contract to Intel
0 Kudos
TRANG2
Beginner
3,932 Views

Hello Daniel,

You might misunderstood our connectivity, I have attached a image which explains the connection diagram of our environment.

Also we wanted use this on Linux platform only, What do you mean "Try to run the test on another operating system", Will this product doesn't support Ubuntu?

Thanks,

Trilokesh Rangam.

0 Kudos
idata
Employee
3,932 Views

Hello Rangam,

 

 

Thank you for the image. You mentioned you have 3 adapters. Would you be able to use the third adapter in place of one of the two to check if this is being caused by a faulty adapter or port. Although Ubuntu is supported in driver updates; we are recommending another operating system to make sure this is not a compatibility issue with Ubuntu. If you have another system running Windows or another distribution of Linux, please temporarily install the adapter and run iperf to confirm this is not an issue isolated to Ubuntu.

 

 

Best regards,

 

Daniel D

 

Intel customer support

 

Agent under contract to Intel
0 Kudos
TRANG2
Beginner
3,932 Views

Hello Daniel,

Thanks for the advice.

We have tried all possible combinations to isolate the issue as:

With Ubuntu:

PCIe Cards No: 1 -> 2 & 2 -> 3 and 1 -> 3

With Windows 10:

PCIe Cards No: 1 -> 2 & 2 -> 3 and 1 -> 3

With Ubuntu -> Windows 10:

PCIe Cards No:

1 in Ubuntu -> 2 in Windows

2 in Ubuntu > 1 in Windows

2 in Ubuntu -> 3 in Windows

3 in Ubuntu -> 2 in Windows

1 in Ubuntu -> 3 in Windows

3 in Ubuntu -> 1 in Windows

All the combinations has the same behavior. When we use Single port we see 9+Gbps, on Dual ports 1 @4+ Gbps 2nd @4/5+ Gbps Total ~9Gbps. We are expecting Total 18+ Gbps.

We are expecting this should give the rated bandwidth in custom Linux(core-minimal-image yocto).

Thanks,

Trilokesh Rangam.

0 Kudos
idata
Employee
3,932 Views

Hello Rangam,

 

 

Thank you for the thorough troubleshooting. This will be of great assistance, as we now know the issue is not isolated to a single adapter or operating system. We will update you as soon as possible. Please let us know if you have any questions.

 

 

Best regards,

 

Daniel D

 

Intel customer support

 

Agent under contract to Intel
0 Kudos
idata
Employee
3,932 Views

Hello Rangam,

 

 

We will try to replicate this in our lab to further diagnose the issue. Have you checked with HPE if this issue is known on their end? They might be able to recommend a BIOS update or setting to resolve this issue. Please let us know if you have any other questions. We will update you as soon as possible.

 

 

Best regards,

 

Daniel D

 

Intel customer support

 

Agent under contract to Intel
0 Kudos
TRANG2
Beginner
3,932 Views

Hello Daniel,

Is there any update on this issue?

We have tried other PCI peripheral which give us the expected results on the same PCI slot where we connected the Intel's 10G x540 PCI card.

This even says that is only problem with the Intel's Card. Can you please speed up the investigation, as we are at the end of our design of our new platform, where we are integrating x540 chipset on our board itself.

If this is the problem with chipset we may need to change it.

Thanks,

Lokesh Rangam.

0 Kudos
idata
Employee
3,932 Views

Hello Rangam,

 

 

We are still trying to replicate the issue. We will update you as soon as possible. Please let us know if you have any other questions.

 

 

Best regards,

 

Daniel D

 

Intel customer support

 

Agent under contract to Intel
0 Kudos
idata
Employee
3,932 Views

Hello Rangam,

 

 

We were unable to reproduce the issue here in the lab. Please provide the brand and model of the 10Gb adapter that demonstrated the expected results on the same PCI-E slot. If you contacted HPE, please inform us if they were able to provide you any advice regarding this concern. Thank you for your patience and cooperation. Please let us know if you have any other questions.

 

 

Best regards,

 

Daniel D

 

Intel customer support

 

Agent under contract to Intel
0 Kudos
TRANG2
Beginner
3,932 Views

Hello Daniel,

Thanks for the update. Currently we don't have any information from HPE on this issue.

Can you please detail the steps that you followed to reproduce issue including the setup and commands used on a linux system.

we will try to replicate the setup and verify from our side as well.

Thanks,

Trilokesh Rangam.

0 Kudos
idata
Employee
3,932 Views

Hello Rangam,

 

 

We would be happy to share our configuration information with you. We used the X540-T2 (PBA G45270-002) on Windows 10 x64 1803 with PROSet Software 23.2.

 

 

X540-T2 - port 1 -> port 1 of another Intel 10Gbps 192.168.1.x

 

X540-T2 - port 2 -> port 2 of another Intel 10Gbps 192.168.2.x

 

 

Configurations:

 

Interrupt moderation - off

 

flow control - off

 

offloading - enabled

 

jumbo frame - off

 

RSS - default

 

 

Iperf2:

 

iperf -s command on system with other Intel 10Gbps

 

iperf -c 192.168.x.x command from two command prompts simultaneously on the system with the X540

 

 

Result:

 

Average iperf connection from X540 to the other Intel 10Gbps using port 1 and 2 simultaneously = 7-8Gbps per port

 

 

We recommend contacting HPE for their recommendations if you have not already. Let us know the results if you test the connection using this configuration. Please do not hesitate to contact us with any questions.

 

 

Best regards,

 

Daniel D

 

 

0 Kudos
idata
Employee
3,932 Views

Hello Rangam,

 

 

Is this configuration working for you? Please let us know if you identified the cause. Please feel free to ask any questions.

 

 

Best regards,

 

Daniel D
0 Kudos
idata
Employee
3,932 Views

Hello Rangam,

 

 

If you have any other questions please let us know.

 

 

Best regards,

 

Daniel D
0 Kudos
idata
Employee
3,712 Views

Hello Rangam,

 

 

We tried another setup, and achieved similar results.

 

 

Ubuntu 16.04 with X540-T2 connecting to CentOS 7.4 with another Intel 10GbE.

 

 

X540-T2 port 1 192.168.1.x connecting to other 10GbE port 1 19.168.1.x

 

X540-T2 port 2 192.168.2.x connecting to other 10GbE port 2 19.168.2.x

 

 

X540-T2 configured with default settings

 

 

From 10GbE CentOS system iperf2 server

 

 

iperf -s from Intel 10GbE adapter

 

iperf -c 192.168.x.x from two command prompts simultaneously on each port of the X540-T2.

 

 

Results to the other 10GbE port 1 and 2 = 8-9Gb/s per port simultaneously.

 

 

Let us know if you still have any questions.

 

 

Best regards,

 

Daniel D

 

 

 

0 Kudos
Reply