Memory Performance
For our memory tests we use the built-in memory benchmarks in AIDA64 Engineer and SiSoft Sandra. For more details on each of the benchmarks please see here and here respectively.
AIDA64 Engineer
SiSoft Sandra
Combined Latency Test
Generally the swap between SODIMM and traditional DIMM memory formats does make a negative impact on performance, but the CPU bottlenecks the performance here.
It would be interesting to see the power consumption tests on units like this using something realistic like a pico psu unit. The unit you’re using here is complete overkill and I’m a little dubious about the efficiency at these demands. It’s only pulling about 3.3% of peak output. The efficiency isn’t measured at anything below 20% I think for this classification.
I can understand wanting to standardize the test rig as much as possible but question how much sense that really makes with such disparity in the platforms themselves. Nobody in the real world is going to use a $200 800+ watt PSU for an n3700 board. The attraction is in the low power and for many in the silence. I suppose if you have it in a closet as a NAS you might use some larger power supply you had on hand. But for htpc duty almost everyone is going to go for a pico psu unit.
It would be interesting to see the power consumption tests on units like this using something realistic like a pico psu unit. The unit you’re using here is complete overkill and I’m a little dubious about the efficiency at these demands. It’s only pulling about 3.3% of peak output. The efficiency isn’t measured at anything below 20% I think for this classification.
I can understand wanting to standardize the test rig as much as possible but question how much sense that really makes with such disparity in the platforms themselves. Nobody in the real world is going to use a $200 800+ watt PSU for an n3700 board. The attraction is in the low power and for many in the silence. I suppose if you have it in a closet as a NAS you might use some larger power supply you had on hand. But for htpc duty almost everyone is going to go for a pico psu unit.
It would be interesting to see the power consumption tests on units like this using something realistic like a pico psu unit. The unit you’re using here is complete overkill and I’m a little dubious about the efficiency at these demands. It’s only pulling about 3.3% of peak output. The efficiency isn’t measured at anything below 20% I think for this classification.
I can understand wanting to standardize the test rig as much as possible but question how much sense that really makes with such disparity in the platforms themselves. Nobody in the real world is going to use a $200 800+ watt PSU for an n3700 board. The attraction is in the low power and for many in the silence. I suppose if you have it in a closet as a NAS you might use some larger power supply you had on hand. But for htpc duty almost everyone is going to go for a pico psu unit.