Now it is time to look at benchmarks that test memory bandwidth. For the testing today we are using Aida64 and SiSoftware Sandra.
Let’s start with Aida64, it has separate tests for memory read, memory write and memory copy.
This is where the advantage of running at 3600MHz comes into play on the Ryzen platform. Memory read is right where it should be at 50GB/s on both AMD CPUs, there is no difference between them. This beats the memory read bandwidth of 38GB/s on Intel Core i5-9400 and 34GB/s on Intel Core i3-9100F.
However, oddly enough, the tables turn on memory write tests. We don’t actually know what is causing this result, but the memory writes tank on the AMD platform. They go down to 28GB/s each CPU. Whereas on the Intel platform the memory writes stay pretty much the same as the read rate, which is what should happen. We don’t know why writes are slower on the AMD platform, all memory settings were verified.
Things go back to normal on the memory copy tests, with performance at 46GB/s copying data. Intel copy performance is also down a little bit on both CPUs, but this is to be expected. AMD is much faster copying and reading, but something about writing is slowing it down.
In SiSoftware Sandra these are the aggregate memory performance results we saw. Both AMD CPUs are around 38GB/s and Intel CPUs at 25 and 24GB/s.