Oh Dear, intel integrated graphics

[Eric5h5]

I agree that the GMA950 isn't exactly brilliant, but neither was the ATi Radeon 9200 that was in the PPC models.

All of these comparisons between the GMA950 and 9200 are missing the point. The 9200 is history. You should be comparing the GMA950 to whatever the low-end non-integrated graphics card is NOW. I admit I'm not entirely sure offhand since I've been following high-end for a while...the X300 maybe?

There's no possible way that this move can be spun into anything other than a mistake. Yes, Apple does make mistakes sometimes. The public perception alone makes it a mistake, never mind the actual technical performance. Which is bad enough...leaving aside the fact that T&L and vertex shading has to be done by the CPU, the system memory usage is a major problem. Most people getting the Mini won't upgrade the 512MB, so taking away 80MB (minimum) from that causes problems right there, and then add in the memory-hungry Rosetta--which will be an issue for quite a while--and it becomes clear that integrated graphics was a bad idea. It would actually have been more feasible to use integrated graphics a year or two from now.

It's obvious Apple only did it because of the Intel connection. Expect more of this sort of thing, unless Apple wises up. The "Mac Mini was never intended as a gaming machine" folks seem to be conveniently forgetting that Apple themselves were pushing the Mini as a machine that could play games. Not a power gaming machine to be sure, but at least somewhat capable, and the 9200 was (barely) capable when the Mini was introduced.

I know this has been beaten to death now, but:

I here all this integrated graphics is bad, well my imac dc has integrated graphic

Dude. It does not. As they say: "look it up!"

--Eric

 

[cube]

I agree that the GMA950 isn't exactly brilliant, but neither was the ATi Radeon 9200 that was in the PPC models.

http://www.extremetech.com/article2/0,1697,1821813,00.asp

Thats a comparison between the GeForce 6200 and the GMA950. Yes, the GMA950 sucks in comparison, but the 9200 wasn't that much better either.

This compares the Intel against a lousy 6200 TurboCache, not a 9200.

 

[cube]

The bench test results posted by Macintouch:

http://www.macintouch.com/#tips.2006.03.01

It looks like the Intel Mac mini handily beats the last version of the G4 mini in most of the graphics tests.

Yeah, right, like in the hardware-accelerated test which is the only one that matters for interactive 3D where the G4 is faster.

And the Intel graphics are so good that the machine is faster when the CPU does all the rendering instead of using the gfx chip.

 

[IJ Reilly]

I dropped out of this debate until we had some actual numbers to work with. As of today, we do, and they have been posted. So why is it, that the results of objective bench tests make absolutely no difference to some people?

 

[IJ Reilly]

Yeah, right, like in the hardware-accelerated test which is the only one that matters for interactive 3D where the G4 is faster.

And the Intel graphics are so good that the machine is faster when the CPU does all the rendering instead of using the gfx chip.

Only the Core Solo is slower than the G4 in the one test, wins easily in all of the others, and the Core Duo beats the G4 handily in all tests.

 

[jsw]

So why is it, that the results of objective bench tests make absolutely no difference to some people?

Your posts in - and reading of - various forums here lead me to believe you would not be shocked by people maintaining beliefs contrary to known facts.

 

[cube]

Only the Core Solo is slower than the G4 in the one test, wins easily in all of the others, and the Core Duo beats the G4 handily in all tests.

What happens when you actually have some interesting logic in the application, not just drawing a scene?

 

[tjwett]

definitely glad to see some real numbers. although they aren't the numbers i'm really interested in it's still good info to have. i don't plan to run any 3D games on this machine so i'd like to see some bench results for things like FileMaker calculations, Automator actions, PDF generation etc. how good is this thing at doing math? can someone whip up a monster calc in Grapher or something and let it chug away?

 

[IJ Reilly]

Your posts in - and reading of - various forums here lead me to believe you would not be shocked by people maintaining beliefs contrary to known facts.

True, but that doesn't mean I'm still not seeking an explanation.

 

[IJ Reilly]

What happens when you actually have some interesting logic in the application, not just drawing a scene?

As opposed to uninteresting logic? Bench tests are bench tests. They are meant to be as objective as possible. These are the only actual numbers that we've got right now, and they look pretty good to me. Not spectacular, but certainly nothing like the gloom and doom being predicted.

 

[jsw]

True, but that doesn't mean I'm still not seeking an explanation.

The explanation is simple:

Apple could have put the X1600 and a 2.0GHx Core Duo in the mini - really, because obviously they did it in the iMac.

All they would have needed to do in order to do so would have been to completely reengineer the mini, make it bigger, and make it even more expensive.

So... if they would have just made an entirely different system, not the mini, they could have called it the "mini" anyway and sold countless millions, instead of the maybe dozen or two units they'll sell to complete idiots in this stripped-down, useless configuration.

I mean, my God, the thing barely plays 3D games. Why would anyone buy a computer that doesn't play 3D games really well?

 

[cube]

As opposed to uninteresting logic? Bench tests are bench tests. They are meant to be as objective as possible. These are the only actual numbers that we've got right now, and they look pretty good to me. Not spectacular, but certainly nothing like the gloom and doom being predicted.

Of course. In a game or flight simulation you have a ton of physical, AI modeling and so on, not just traversing a tree with geometry. Your "fast" mini is just burning the cores in rendering.

 

[jsw]

Of course. In a game or flight simulation you have a ton of physical, AI modelling and so on, not just traversing a tree with geometry. Your "fast" mini is just burning the cores in rendering.

So buy a different Mac, or a PC. This, I know, is an innovative suggestion, one which no one else in this thread or others has yet offered.

 

[isgoed]

This compares the Intel against a lousy 6200 TurboCache, not a 9200.

You are confused. The geForce 6200 is up to 4 times as fast as the 9200 (based on 3DMark03 scores).

What happens when you actually have some interesting logic in the application, not just drawing a scene?

In the Hardware OpenGL Cinebench test the Core Duo Mini trounces every other mac in the test. First you use the test as a way to prove how poor the core solo is and once the core duo mini is added to that same test and blows out a 50% lead over its closest competitor you say the test is unrepresentative? Gimme a break!

edit:
Whoops: I misread "iMac Core Duo" for "Mini Core Duo". My bad. I posted in the edit of this post of mine some corrected numbers.

 

[cube]

So buy a different Mac, or a PC. This, I know, is an innovative suggestion, one which no one else in this thread or others has yet offered.

The only alternative is to buy the much more expensive Power Mac that will be available in Q3.

 

[NeuronBasher]

The only alternative is to buy the much more expensive Power Mac that will be available in Q3.

Then I guess you don't have anything to talk about until Q3. We'll miss you.

 

[cube]

In the Hardware OpenGL Cinebench test the Core Duo Mini trounces every other mac in the test. First you use the test as a way to prove how poor the core solo is and once the core duo mini is added to that same test and blows out a 50% lead over its closest competitor you say the test is unrepresentative? Gimme a break!

The Intel does not have the hardware to do all the rendering in it. It's doing important parts in software. A ton more work than the PPC is doing.

 

[jsw]

The only alternative is to buy the much more expensive Power Mac that will be available in Q3.

Or, say, an iMac. The 17" runs circles around the mini graphically and is not all that much more expensive - it's just bigger.

 

[isgoed]

The explanation is simple:

Apple could have put the X1600 and a 2.0GHx Core Duo in the mini - really, because obviously they did it in the iMac.

All they would have needed to do in order to do so would have been to completely reengineer the mini, make it bigger, and make it even more expensive.

So... if they would have just made an entirely different system, not the mini, they could have called it the "mini" anyway and sold countless millions, instead of the maybe dozen or two units they'll sell to complete idiots in this stripped-down, useless configuration.

I mean, my God, the thing barely plays 3D games. Why would anyone buy a computer that doesn't play 3D games really well?

You are being cynical right?

 

[jsw]

You are being cynical right?

Just a wee bit.

 

[cube]

Or, say, an iMac. The 17" runs circles around the mini graphically and is not all that much more expensive - it's just bigger.

I have a better monitor than the iMac has, and I don't have the space for another one. Plus, I don't want the monitor to be obsoleted together with the computer.

 

[LethalWolfe]

True, but that doesn't mean I'm still not seeking an explanation.

"Everything for nothing"

How is that for an explanation as to why there is always ample amounts of largely illogical whining after every Apple product release?


Lethal

 

[peharri]

So buy a different Mac, or a PC. This, I know, is an innovative suggestion, one which no one else in this thread or others has yet offered.

Why on Earth do people think they're winning an argument when they say "So it's crap? Well, then buy a PC!"

The Mac mini used to have acceptable graphics. It had useful hardware acceleration. Now the hardware accelleration is worse, despite being compared to three year old graphics cards, such that most games, etc, will actually need to do most of the rendering in software, competing with CPU time required for everything else. This is a *bad* thing. The performance of the Mac mini for a certain, major, class of applications is now *worse* than it was on its predecessor. That major class of applications *is* important for a computer aimed at the home.

How people can turn around and say "So? Buy a different machine!" and not realise that they're merely confirming the point, not brushing it off. What next? "You need your Mac mini to run Word and Excel at a decent speed? Well, buy a PC!" (Yes, the $600 version of the Mac mini will be unable to run Office X at a speed comparable with anything other than the last of the G3s until Microsoft comes up with the universal binaries, which it doesn't appear to be rushing out of the door.)

With the low-end version of the Intel Mac mini unable to:

* Run most games
* Run most productivity apps for several months, many even for years, at a decent speed.
* Run Classic Mac apps back from the days where MHz was at a premium (no OS 9 compatability)

...just what is it supposed to run? Are buyers supposed to be happy with the iLife suite and TextEdit.app? Who is this computer aimed at? It's not switchers. It's not students. Grandma doesn't have a keyboard or mouse. Who?

 

[peharri]

You are confused. The geForce 6200 is up to 4 times as fast as the 9200 (based on 3DMark03 scores).
In the Hardware OpenGL Cinebench test the Core Duo Mini trounces every other mac in the test. First you use the test as a way to prove how poor the core solo is and once the core duo mini is added to that same test and blows out a 50% lead over its closest competitor you say the test is unrepresentative? Gimme a break!

You have a link to anything to back that up? cube was responding to the Macintouch benchmarks which most certainly do not test the Core Duo Mini. There's also little reason to believe the Core Solo's hardware acceleration are remotely different from the Core Duo, unless far more is being done in software, even when, supposedly, using hardware acceleration, than should be being done.

I hope you're wrong and the Core Duo and Core Solo have equivalent "hardware" OpenGL speeds, because if you're not, it's worse than we thought.

 

[IJ Reilly]

You have a link to anything to back that up? cube was responding to the Macintouch benchmarks which most certainly do not test the Core Duo Mini. There's also little reason to believe the Core Solo's hardware acceleration are remotely different from the Core Duo, unless far more is being done in software, even when, supposedly, using hardware acceleration, than should be being done.

I hope you're wrong and the Core Duo and Core Solo have equivalent "hardware" OpenGL speeds, because if you're not, it's worse than we thought.

To clarify: (1) the Macintouch benchmarks most certainly do test the Core Duo min.; and (2) it might be worse than you thought, not we. So far the numbers we're seeing look fine.