Originally posted by: Fox5
Originally posted by: Cerb
Originally posted by: remagavon
Originally posted by: Eug
Originally posted by: Fox5
What's the matter? The core duo does beat it...
Except you're talking a significantly more powerful cpu(from the cpu rendering test, about 3x the power for the core duo over the g4) and it just barely edges out the G4 with 9200. Still, it does look to be in the same ballpark, though game performance will be what really matters.
What's the matter? OpenGL performance on the Radeon 9200 is abysmal, and everyone has been complaining about that for just about forever. Now the new purdy Intel Mac mini comes out, and the OpenGL performance is still abysmal, with one model even worse than before (at least according to this one OpenGL bench).
Even the lowly Radeon X300 HyperMemory (32 MB on-board and up to 128 total with shared memory) would have been a humungous improvement.
The GUI is significantly faster. I'm using 1920x1200 right now and it feels somewhat like a powermac as far as the responsiveness.
🙂
...that's because the CPU and RAM give it all the power (except alitvec, of course) of a low end Power Mac.
I doubt altivec was used in low end powermacs.
I didn't say anything about whether it was
used. The G4 and G5
have it, and it clearly outperforms uses of SSE and SSE2. The social and economic aspect has kicked in in the last few years, though: with Apple representing such a small market, it is becoming increasingly less beneficial for them to have their code taylored to make the PPC chips perform very well.
In many ways, it's more powerful than a low end powerMac.
Nope. The lowest one is a dual-core 2GHz, with a real video card. Which is faster at CPU/RAM-bound stuff, I don't know (I'm sure Barefeats and the like will get around to it
🙂), but they're on par, with the HDD being main limitng factor on the Mini.
Faster memory performance (maybe higher bandwidth, definetely lower latency),
Arguably of no real use, though.
larger and better cache, and probably higher integer performance as well. Floating point and SIMD/Vector are likely to be horrible, but SIMD is rarely used anyway
That depends on the task and software. Vector performance is not horrible--just not great. Even the G4 chips, once given a moderately fast FSB and DDR RAM, were no slouch in vector performance. The G5 still beats anything else we've got. However, while SSEx may not offer as much performance, the capability of using pretty much the same code on Windows, *n*x, and OS X will make taking advantage of it much more appealing.
and as for floating point..well it's still an improvement over the G4, though since Macs are supposed to be content creation machines and the Mac Mini doesn't even have a decent gpu (for 3d rendering or video encoding or whatever else a gpu can do)
Gaming cards can't keep up with their Pro-flashed brethren. The GPU means nothing, compared to the CPU's raw FP and SIMD. It's changing, but not fast. Until everyone has a video processor like in the newer Geforces, or a PPU to help out, it likely won't really catch on, except for use in common video codec decoding.
then I guess the Mac Mini isn't much more than a stylish, easy to use PC that can't play very many games. On the other hand, it can do 100% of what at least 2/3rds of the population would want it to do. Oh, and it can make a decent number cruncher/server box I suppose.
It's also smaller and quieter than other SFF boxes with similar capability, and doesn't look bad next to non-computer hardware.
