Comment Strangely Linear Per Upgraded Component (Score 1) 264
About 10 years ago I tried an experiment. I had bought a new CPU and GPU. The CPU could fit in my current motherboard. (Nothing was *old* just a few years out of date.)
My theory was that if I upgraded the GPU first, I would get most of my performance improvement and the CPU wouldn't matter. So I upgraded them one at a time to be a proper scientist. I had FutureMark ready to go.
The GPU, first, gave me the 2x improvement I figured it could, based on the improvement in the specs. I had expected a little less because of the old CPU, but seemed to get the full performance boost.
Then I upgraded the CPU and, surprisingly, I got another 2x just like the the CPU specs promised. I expected maybe 20% or so since the GPU should have been most of the bottleneck.
I was so intrigued that I found a memory upgrade that also fit the board. Again the 40% or so I expected, based on specs, happened.
I'm surprised to this day.
While it has been a long time, the best I can guess/remember was that it was an AMD 3000+ CPU (1.9Ghz). And an NVidia GPU, I think GT6600 or something? The memory went from whatever to DDR, if I recall.