It has been a long time since the last CPU upgrade for my home server. Its previous CPU was an Intel Q9300, a 95 W quadcore processor released in Q1 2008 on a 45 nm process. Its new CPU is an Intel Xeon D-1518, a 35 W quadcore processor with hyperthreading released in Q4 2015 on a 14 nm process. The old CPU had a small advantage in clock speed, running at 2.5 GHz to the Xeon’s 2.2 GHz. The L2 caches were even the same size. Everything else favored the Xeon.

I decided to run some benchmarks to see what nearly eight years of progress could do. I wasn’t interested in anything that thorough, I only wanted to measure the CPU. One that I was familiar with, even with all its faults, was UnixBench. I tried to run all of the benchmarks but one involving pipes failed. I looked into fixing it but my C abilities are non-existent plus I did not care about the OS or system tests. That meant I would stick with the venerable Dhrystone and Whetstone. The before and after would be running the same OS and same binary, leaving the notable difference as the CPU. I ran single-core and quadcore tests on the old system and single-, quad-, and octocore tests on the new system.

Unsurprisingly, the new CPU bested the old in every test.

Processor Width Dhrystone Whetstone
Q9300 1 16779628 3305
D-1518 1 24531345 4900
Q9300 4 71008081 13269
D-1518 4 85922310 18530
D-1518 8 109194810 33433

Single-core shows the biggest gain with a 46% improvement in Dhrystone and 48% in Whetstone. The quadcore has a smaller improvement at 21% for Dhrystone but 40% in Whetstone. Not bad for eight years and a 60 W decrease in power usage.