I've observed everyone discussing the Qotom Q20332G9 s10 on the forums, and I also purchased one via Amazon. After approximately 10 days, I received my shipping notification and eventually my unit.
Several main points have been raised, including temperature, auto-boot jumpers, and interfaces. After a few days of use, I have some experiences to share:
1. Temperature: If you live in an area where the average temperature exceeds 25 degrees Celsius, you might want to consider adding a fan to the unit to prevent overheating, which could result in frequency reduction and performance impact.
2. Jump-start: There has been limited sharing on this topic, and I was unable to locate the exact jumper position. However, in the BIOS North Bridge chipset settings, you can set the power status to either "Last State" or "Auto Start".
3. Interfaces: The device comes with various ports. If you have a console wire, you can even get quasi-IPMI capabilities in a pinch. Just a single RJ45 console wire is needed, no need for a monitor or keyboard for system setup and BIOS configuration.
Moving on to my main point and questions, it seems like the bus bandwidth of this device is unable to accommodate the combined speed of four 10G SFP+ interfaces running simultaneously:
I performed some simple iperf loopback performance tests (127.0.0.1) to evaluate whether the C3758R can run at the theoretical limit speed of 10G SFP+ *4 simultaneously.
Code:
W1290 iperf 127.0.0.1
[ ID] Interval Transfer Bitrate Retr Cwnd
[ 6] 0.00-1.00 sec 1.79 GBytes 15.4 Gbits/sec
[ 8] 0.00-1.00 sec 1.79 GBytes 15.4 Gbits/sec
[ 10] 0.00-1.00 sec 1.79 GBytes 15.4 Gbits/sec
[ 12] 0.00-1.00 sec 1.79 GBytes 15.4 Gbits/sec
[SUM] 0.00-1.00 sec 7.16 GBytes 61.5 Gbits/sec
i3-12100 iperf 127.0.0.1
[ ID] Interval Transfer Bitrate
[ 6] 0.00-1.00 sec 3.43 GBytes 27.2 Gbits/sec
[ 8] 0.00-1.00 sec 3.43 GBytes 27.2 Gbits/sec
[ 10] 0.00-1.00 sec 3.43 GBytes 27.2 Gbits/sec
[ 12] 0.00-1.00 sec 3.43 GBytes 27.2 Gbits/sec
[SUM] 0.00-1.00 sec 13.7 GBytes 109 Gbits/sec
N5105 iperf 127.0.0.1
[ ID] Interval Transfer Bitrate Retr Cwnd
[ 6] 0.00-1.00 sec 694 MBytes 5.81 Gbits/sec 0 1.12 MBytes
[ 8] 0.00-1.00 sec 694 MBytes 5.81 Gbits/sec 0 1.12 MBytes
[ 10] 0.00-1.00 sec 694 MBytes 5.81 Gbits/sec 0 1.12 MBytes
[ 12] 0.00-1.00 sec 694 MBytes 5.81 Gbits/sec 0 1.12 MBytes
[SUM] 0.00-1.00 sec 2.71 GBytes 23.3 Gbits/sec 0
ATOM C3758R iperf 127.0.0.1
[ ID] Interval Transfer Bitrate
[ 6] 0.00-1.00 sec 526 MBytes 4.41 Gbits/sec
[ 8] 0.00-1.00 sec 526 MBytes 4.41 Gbits/sec
[ 10] 0.00-1.00 sec 526 MBytes 4.41 Gbits/sec
[ 12] 0.00-1.00 sec 525 MBytes 4.40 Gbits/sec
[SUM] 0.00-1.00 sec 2.05 GBytes 17.6 Gbits/sec
I know my tests were too simplistic, but the results have given me some insights. In computer networking, the overall data transfer rate is usually the total speed that all ports can reach under full load.
Take the ATOM C3758R, for example. Despite having four G SFP+ ports, if all four ports are running at full speed simultaneously, their total data transfer rate would only reach 17.6 Gbits/sec. As such, it seems that it can only operate: one 10G SFP+ port + three 2.5G RJ45 ports simultaneously.
I would like to hear from anyone who has carried out practical tests and is willing to share their findings.
Coming up, I will purchase a Mellanox MCX4121A-ACAT to add to my PC, and use a DAC copper cable to interconnect two devices. Some people have noted issues with DAC copper cables. Once it arrives, I'll ascertain for myself. I hope it supports and doesn't give me any trouble.