The Supermicro H11DSi Motherboard Mini-Review: The Sole Dual EPYC Solution

System Benchmarks

On a system like this, there’s not a whole lot to emphasize through benchmarking.

Wall Power

For our power testing, we're booting into Windows and letting the system idle in the operating system for 5 minutes, then taking a reading of the power at the wall. We then fire up Cinebench R20, which probes all the threads in our dual 7F52 setup, and take the peak power from the benchmark. For this review, we've also tested a series of DRAM setups, taking a minimum/minimum of 2 x 8 GB RDIMMs (1 channel) and 16 x 128 GB LRDIMM (8-channel).

For idle power, our RDIMM arrangement doesn't cause much extra power. With the LRDIMMs, we're looking at an extra 2W per module at idle.

For full load, again the 8 GB DIMMs only draw fractions of a watt a piece. Moving up to the large modules, and we're realistically seeing another 7 W per module on average. When we compare the min/max, there's an extra 100W dedicated just to the memory here.

Warm POST Test

For our POST test, we take a system that has been previously booted, shut it down until all the fans stop spinning, and then initiate a power on through the BMC. The time recorded is for the initial BIOS procedure until the OS starts loading. Again with this test, we've gone through with different DRAM configurations.

More memory means more training is required to ensure that each module will operate within a given set of sub-timings. The more capacity at play, and the more channels populated, means more time is required. At the quickest POST it takes 50 seconds, but our longest recorded POST was over two minutes.

DPC Latency

Deferred Procedure Call latency is a way in which Windows handles interrupt servicing. In order to wait for a processor to acknowledge the request, the system will queue all interrupt requests by priority. Critical interrupts will be handled as soon as possible, whereas lesser priority requests such as audio will be further down the line. If the audio device requires data, it will have to wait until the request is processed before the buffer is filled.

If the device drivers of higher priority components in a system are poorly implemented, this can cause delays in request scheduling and process time. This can lead to an empty audio buffer and characteristic audible pauses, pops and clicks. The DPC latency checker measures how much time is taken processing DPCs from driver invocation. The lower the value will result in better audio transfer at smaller buffer sizes. Results are measured in microseconds.

The DPC values for the Supermicro board are very impressive. Normally we consider here anything under 200 microseconds as a successful result, and a fresh system on the Supermicro goes well below that.