TPLINK XDR5430v1 and v2 CPU performance real test

TPLINK’s XDR5430 has two versions, one is the V1 version of the Broadcom solution, and the other is the recently released V2 version of the Qualcomm solution. v1 and v2 have different CPUs and wireless chips, so this article tests the CPUs of these two models to see what the difference is or is not.

IPQ5018 is dual-core 1GHz, A53 architecture, 14nm.

BCM6750 is triple-core 1.5GHz, A7 architecture, 28nm.

On the surface it seems to be BCM6750 strong, but the architecture is not the same.

I hope there is a quantifiable test, in the end, who is stronger, how much stronger, there is direct data to compare would be best, for example, the packet forwarding rate.

The CPU of the router, as the core, needs to handle the forwarding of the external network and the internal network IP, this forwarding capacity is called the packet forwarding rate, there are large packets and small packets of bean curd and barbecued pork, usually take the forwarding rate of small packets to measure the ability of high and low. What does it mean?

Translated with www.DeepL.com/Translator (free version)

The number of packets (minimum packets) of 64byte sent per unit time is used as the basis for calculation. For Gigabit Ethernet, the calculation is as follows: 1,000,000,000bps/8bit/(64 + 8 + 12) byte = 1,488,095pps Note: When the Ethernet frame is 64byte, the fixed overhead of 8byte frame header and 12byte frame gap needs to be considered. Therefore, the packet forwarding rate for a wire-speed Gigabit Ethernet port forwarding 64byte packets is 1.488Mpps

 

What about the smallest packet that can forward 1.488 million packets per second, and the largest packet? The maximum 1518byte.

Conversion: 1,000,000,000bps/8bit/(1518+8+12)byte=81,274pps.

“FC2544 standard requires the frame length: 64byte, 128byte, 256byte, 512byte, 768byte, 1024byte, 1280byte, 1518byte, these data with the above formula to calculate the corresponding pps, the results are as follows: (gigabit)

Network transmission in the packet has a large and small, are in the range of 64 and 1518, if the large packet forwarding can reach the line speed, small packets may not also be able to reach.

To test the forwarding rate of small packets, use a professional network tester on the line, to get a network tester is very simple as long as the money, the problem is that I do not have money. Individual users to ask for quotes will not necessarily care about you.

IxCahrio and iperf can not measure the small packet forwarding rate, the operating system and CPU overhead is too large, measured only double-digit speed.

Then I looked up DPDK and test-related keywords and found TRex and minismb network tester. I didn’t try TRex because I don’t know how to knock the code and I can’t get it. minismb is suitable for me because there is a UI interface console and I can use it by connecting to another computer with minismb system installed. Website: http://www.minismb.com/

I took minismb test out of the value compared to the above table in the understanding of the value is a little bit lower, but I am very satisfied. The feeling of saving hundreds of thousands of dollars. (Professional tester measured the same as the theoretical value)

 

Let’s get started.

The following are the results of XDR5430v1.

The following are the results for XDR5430 v2.

Port 01 in the interface is connected to the router’s LAN port, 02 is connected to the WAN port, from Port2 (wan port) to Port1 (lan port) to send packets, see the second line of “RX packet Rate” is the result of packet forwarding rate.

The two ports send packets to each other is two-way. The result is as follows.

For easy browsing, I put all the results for 64byte, 128byte, 256byte, 512byte, 768byte, 1024byte, 1280byte, 1518byte in a table for clarity.

XDR5430v1 and v2 versions of the wireless router, packet forwarding rate test results in the following table.

The result shows that the packet forwarding rate of V2 version is significantly higher than V1 version.

Next test WAN port when there are 1000 streams, 64byte packet forwarding rate how much more?

The operation in the interface is simple, a few dozen mouse clicks can copy out 1000 streams, and then MAC and IP, port number automatically incremented.

1000 streams, 64byte, XDR5430v1 test results are as follows: (wan to lan)

The packet forwarding rate is 0.221Mpps, and the bandwidth speed is only 149Mbps at this point.

1000 streams, 64byte, XDR5430v2 test results are as follows: (wan to lan)

The v2 version is a bit better, 0.254Mpps, with a higher broadband speed of 170Mbps.

How about wired band capacity?

Simulate 250 users, send data at the same time, and see how many are connected.

XDR5430 v1, showing 64, one of which is connected to my computer.

It does not mean that the maximum can only be connected to 64 units, I can still use the phone to connect. I do not know if it is a bug in the system or what, I use XDR6030 also shows the same 64 units.

XDR5430v2, can show 251.

XDR6030 can be. As follows.

This is full speed, although it is 1.455Mpps, a bit away from 1.488Mpps, which may be because my E5-2682v4 frequency is not high enough or other reasons.

1000 streams with 64 sachets can also be full, as follows.

 

The XDR5430v2 is a bit better than the v1 in terms of packet rate performance.

Does the same CPU have the same hardware forwarding performance in other brand models? No, it is not.

The 64 packet forwarding rate of H3C BX54, which also uses IPQ5018, is 0.956Mpps.

The 64-packet forwarding rate of Xiaomi AX6000, which also uses IPQ5018, is

BX54 and Xiaomi AX6000 would be close, but not up to line speed.

The small packet forwarding rate is less important in a home environment, so they say.

With the technological development of chips, hardware forwarding capacity is getting better and cheaper. This is what I think and not taken for granted.

In summary: v2’s small packet forwarding performance is better than v1’s.