Why I have loss during capture data?

[Miladdona]

Hi everyone,
I'm working with Intel(R) Ethernet CNA x710 DA4 FH that has 4 ports and I want to capture high rate data (more than 3 Gb per second per port). First, I tested to capture 3.3 Gb rate data from one port. It was good without any loss. Fig.1 shows the CPU cores load. But when I want to capture data from two or more ports simultaneously, I have a lot of losses, also CPU load is just on one core. Figs 2-4 show CPU load for 2, 3, and 4 port data capturing respectively.
anyone can help me to distribute load on different cores? I think it can remove losses.
Or how can I capture data without loss?
I work by Windows server 2016 OS, WinPcap 4.1.3 to capture data, Visual studio 2013 and C program.
Fig.5 shows my Ethernet card feature.
Thank you in advance.

[Caguicla_Intel]

Hello Miladdona,

Good day!

This is just a follow up to check if you have read our previous message. In case you need clarifications on this one, please let us know and we will be glad to communicate this to our engineers. 

Looking forward to your reply.

We will follow up after 3 business days if we don't hear from you.

Best regards,

Crisselle C.

Intel Customer Support

[Miladdona]

Hi,

Thank you for response.

I read your message and I did your suggestions but I didn't get good new results!

Please find attached!

Miladdona

[Caguicla_Intel]

Hello Miladdona,

Appreciate your effort in trying out the recommended steps to improve multi port performance on Windows.

We are sorry to hear that it didn't help with the issue. Please allow us to re-escalate this issue with our engineers and ask for their insight. We will get back to you as soon as possible but no later than 2-4 business days.

Hoping for your kind patience.

Best regards,

Crisselle C.

Intel Customer Support

[Caguicla_Intel]

Hello Miladdona,

Thank you for the patience on this matter.

Please see below details for the recommendations from our engineering team. 

The screenshot provided shows that you are using 32 RSS queues. The platform used is a client board with an 8 core i7. With this, they would recommend to reduce the RSS queue count to 4 for each of the 10G ports. Set the first adapter RSS base processor number to 0 and the second to 4. This will ensure each queue gets pinned to unused cores. 

Please dump the RSS config for each port using "Get-NetAdapterRss"

Additionally, they suggested that you attempt to sanity check the test environment using a traditional network benchmark tool such as netperf or iperf. 

Also, we hope you don't mind providing the additional information below as per their request.

1.) Is the adapter installed in the correct speed PCIe slot?

2.) What is the link partner in this case, is it also a client platform? 

3.) What is the memory configuration of the system, if not all memory channels are populated it could reduce the available bandwidth.

Looking forward to hear from you.

Should there be no response, I’ll make sure to reach out after 3 business days. 

Best regards,

Crisselle C.

Intel Customer Support

[Caguicla_Intel]

Hello Miladdona,

Good day!

We'd like to check if you are able to read our previous message. If in case you have questions or clarifications regarding our update, feel free to let us know and we will be glad to check it with our engineers.

Awaiting to hear from you.

We will follow up after 3 business days in case we don't receive a reply.

Best regards,

Crisselle C.

Intel Customer Support

[Caguicla_Intel]

Hello Miladdona,

How are you doing?

Please be informed that we will now close this request since we haven't received any response from our previous follow ups. Just feel free to post a new question if you may have any other inquiry in the future as this thread will no longer be monitored.

May you have an amazing day!

Best regards,

Crisselle C.

Intel Customer Support