As some of you may have read from my microblogging in the comment threads, I recently broke my i5-3570K. ・゜・(ノД`)
Intel replaced it for me, and I'm back up and running (at 4.4Ghz on stock voltage!! Intel XTU is amazing!) now.
BUT, in the interim, I used a backup machine, an A10-5800K.
I knew it was slower, but I overclocked the CPU bits to 4.4Ghz (the fastest my cheapie Biostar board would go), and figured it would be okay with my GEFORCE GTX TITAN installed.
I was WRONG! The gaming experience on that thing with a discrete GPU was TERRIBLE! 。ﾟ(ﾟﾉД｀ﾟ)ﾟ｡
Spiky frametime graphs ahoy! My framerates were similar to my usual under light load (nothing happening ingame), but in combat, it was horrible! The recorded framerate would remain relatively steady around 30-40 FPS, but with constant stutters and hitches, sometimes 200ms or longer! Obviously this made the subjective experience pretty bad.
I saw this behavior in these games:
- Saint's Row 4
- Lost Planet 2
- Borderlands 2
- Crysis (Warhead / 2 / 3)
- Blacklight: Retribution
- Warhammer 40,000: Space Marine
Framerates were good, mostly -- Crysis 3 hated the chip (as expected) and Blacklight also ran very poorly (I had no idea it was so RAM-intensive! 。ﾟ(ﾟﾉД｀ﾟ)ﾟ｡ )
I did some messing around with settings and profiling using Afterburner and Process Hacker, and as far as I can determine, I think the issue, across the board, was memory performance, even though I was running 16GB DDR3-1866 CL9 1T RAM in it. I rarely was fully loaded on the CPU, even on a single core, and my GPU load never went over 50%. (Usually hovering around 25%! Sad! Although I rarely break 60-70% even with my i5 without enabling SSAA.)
Soo, TL;DR, I'd like to see another article exploring and explaining this kind of behavior.
The old one is, well, pretty old, and doesn't include dual-cores or APUs.