- Aug 11, 2008
- 10,451
- 642
- 126
Are you sure? The graph that I'm seeing showing the i3 "ahead of" Sandy Bridge is CPU load percentage, not FPS - meaning the i3's core is showing higher utilization under single and dual core benchmarks (lower is better in that graph). The graph above it which I believe does show FPS shows the Sandy Bridge CPUs still way ahead of the i3.
He was saying that the Sandy Bridge i3 was ahead of the 8350.
No arguing that the game is poorly optimized. But IMO it illustrates the very reason not to go with FX for gaming. The game "is what it is", the results still speak for themselves. Patches may change things somewhat, but I seriously doubt FX is going to come close to making up its deficit.
This illustrates perfectly the problem with FX: it is competitive at best in a game perfectly optimized for it s architecture, and in a game that is not, and there are still plenty of them, it gets totally destroyed, while using more power in the process.
Yeah, it does have a bit of a glass jaw. In highly multithreaded games it can run with the i7, but as soon as it runs into problems it just falls apart. It's a fundamental problem of the CMT design. A CMT module seems to have the same multithreaded throughput as an SMT core, but it can't devote as much of its resources to a single thread.
Its a cache latency and main memory access time problem, not related to CMT. Take a look at Kaveri vs Trinity, Kaveri should of been way faster but it has almost the same perf with Trinity when using a dGPU.