![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
anyone willing to play thru CIV V all the way to some kind of an end game position deserves my undying respect - personally, I'd rather watch paint dry
(ps - FWIW, SMAC/X is always on my machines & still get regular play around here)
![Zarathustra[H]](https://cdn.hardforum.com/data/avatars/m/9/9298.jpg?1707758471){kind=avatar}
anyone willing to play thru CIV V all the way to some kind of an end game position deserves my undying respect - personally, I'd rather watch paint dry
(ps - FWIW, SMAC/X is always on my machines & still get regular play around here
Neat, except that's a canned benchmark, which means most of the CPU work is predone.
Details, details, details............... Both reviews do conclude, however, that Bulldozer still loses to i5 in ARMA2.... So, great purchase there buddy...... Enjoy....
That's what I get for browsing the boards so much from may phone. Post was meant as rebuttal for:
Kyle or Brent, do you have any sense at this point just how much of the performance deficit with Bulldozer you're seeing with dual and triple-GPU setups in games' FPS has to do with driver/platform issues and how much has to do with the FX architecture? I'm asking this because we're actually seeing those wild fluctuations in frame rates on the FX platform that often indicates poorly optimized drivers. I'm also aware that these fluctuations could be indicative of cache-thrashing, core-contention issues, you name it. In other words, if the FX platform's FPS were smooth, but simply lower than the Intel platform, then I'd buy that it's all the fault of the CPU, and it's simply slower than the Intel chip, plain and simple, but when I see those sawtooth patterns in the FPS graph, I'm thinking somebody just took an Intel-optimized graphics card driver [nVidia, I'm looking at YOU
I'm not disputing your findings in any way: it's clear the Intel chip is running these games much more smoothly and at higher FPS. What I'm trying to wrap my mind around is how much of the bad gaming performance of the FX chips has to do with real 'flaws' in the design, and how much has to do with code that is simply not compiled/optimized to run on such a different CPU architecture. I am VERY reluctant to use a car analogy, but I can't help myself here
Like a few others have pointed out, the much newer Battlefield 3 engine seems to like the FX chip, but so far, I haven't seen trustworthy multi-GPU Battlefield benchmarks run on the FX. I know this is a hard thing to nail down, but in your extensive experience (you've certainly seen drivers improve for newly released GPUs - perhaps this may also be true for graphics drivers running on such a radically different CPU) would you say that 'flaws' (eg. very high cache latency, FPU sharing between pairs of cores) in the FX CPU architecture are entirely responsible for the performance gulf with Intel, no matter how optimized the code is, or do you think improved/optimized GPU drivers (and possbily optimized gaming engines) for the FX platform might allow the Bulldozer and it's descendants make significant gains down the road? Just curious for your thoughts.