When we are comparing single digit framerates, I would say that it doesn't matter. 6 vs 9 FPS? for such simple / old games? who cares?
I think you start to see all kinds of weird behavior from the drivers when the framerates aren't even regularly hitting 30. It comes down to the overhead costs associated, which, on actually capable hardware (e.g. a discrete ATI/Nvidia from the past 5 years), are made negligible because the fill rate of the card is enough to keep up with the game.
I think the conclusion to take away from this post is not that the Mac OS X or Linux drivers for Intel graphics suck. The conclusion is that the Intel 945 graphics chipset sucks. Immensely.
I have one in a ThinkPad with twice as much RAM as your Mac, and trust me, it is no better on Windows 7. IE 9 is unusable with hardware rendering, and still quite painfully slow with software rendering. Only Chrome is fast enough to provide a reasonable experience on that thing.
I also have a 965GM chip. It's ... faster, but since it is based on largely the same tech, it's still underwhelming.
And based on the latest and greatest out of Intel, it seems like their IGPs still have no real performance to speak of.
My recommendation is, if you use such a tiny IGP, limit your screen resolution to no more than 1024x768 if you intend to do anything involving 3d accel. The squaring effect quickly makes these chips useless at a high res, not to mention multi-monitor. I guess that's why both my ThinkPads have 1024x768 -- they didn't want to push a higher res and risk that people would complain about the performance.
In other words: get a laptop with a low res LCD and an IGP, by all means. Saves battery life and your laptop won't be multi-monitoring very often anyway (if you truly intend for it to be a laptop, i.e. portable, i.e. not a desk fixture). But if you are thinking of getting any sort of non-portable desktop of any kind, for the love of god, go with discrete. Or at least AMD Fusion.
Mesa Intel Linux Driver
Linux Intel driver performance should not be tested against stock driver, because they have many performance and feature bugs.
I do not know why Mesa even releases official driver with having so many Shader related bugs. For example,
I had to use Mesa 7.7.1 for Ubuntu 9.10
for Ubuntu 10.10 all Official Mesa 7.8, 7.8.1, 7.8.2 have bugs, if I compare them with Mesa 7.7.1. Same is true for Ubuntu 9.10
Official Mesa 7.9, 7.9.1 and 7.10 all carrying bugs from 7.8 series as well as new bugs, if I compare them with Mesa 7.7.1
So, if u r using Ubuntu 9.10 use Mesa 7.8.3-rc1 which fixed many bugs from Mesa 7.8 series and as reasonable as Mesa 7.7.1
And if u r using Ubuntu 10.10, use Mesa 7.11-devel from git which fixed almost all the bugs from Mesa 7.9 and Mesa 7.10 series (still has some bugs related to GLSL when u run game by WINE) or stick to bugs proof 7.8.3-rc1 (I do not recommend, as it has less feature than 7.11-devel)
Thnx for reading
Originally Posted by Michael
i've a mac mini (november 2010) with OSX, Windows 7 64bit and Ubuntu 10.10 64bit...
and everything works fine...???!!!
It's the older 2008 (or 2007) era Mac Minis that have known issues with 64-bit Windows.
Originally Posted by insane74
I think the best comparison for the 2 OS's is if only Ubuntu Linux can be installed on a Mac computer natively (i.e. dual boot) without the use of virtualization, then we can see the real deal between these 2 OS's.
Or use the same hardware specs as the Mac computer for the PC platform and install Linux on it and we see the difference.
Bottom line is that a Mac computer is where OS X really belongs.
IIRC you CAN install Ubuntu in Mac, otherwise I don't know why they used to give a PPC image.
Originally Posted by jybumaat
You have been able to do this for years and this is how this article was done. No "virtualization" were used in these benchmarks.
Originally Posted by jybumaat
Tags for this Thread