Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Those are not fair tests.

I'm happy to have this discussion with you, but you've got to start holding up your end. If you think that the benchmarks I provided are flawed, then do some research and find some benchmarks that support your contention about the viability of the Atom 330.

The article you got them from is comparing the Atom with an intel GMA950 graphics chip to a 3.16 Core 2 Duo chip with an NVidia 7800GT

The E8500 Core 2 Duo CPU was released almost a year ago and the XFX 7800GT graphics card was released over three years ago. We're not talking cutting edge here. It's not like they compared it to a quad core i7 965 with a state of the art SLI video subsystem.

The MP3 encoding tests are closer to the mark, but an Ion platform will be faster as it will have faster RAM, etc.

The memory controller is built into the Atom 330, not the motherboard chipset, so why would the RAM be significantly faster with the Ion platform? And, really, how fast is it going to be when the RAM bus speed is 533mhz? The MP3 encoding was slow because of the pathetic floating point performance of the Atom 330, not the RAM speed.

Here's proof that the floating point sucks:

san7cpu.png


And here's The Register telling you just what I've been saying throughout this exchange:

http://www.theregister.co.uk/2009/01/16/mac_mini_ion/

From that article:

The Atom is no speedster - but, then again, the Mac mini hasn't lived its life in the fast lane. The Atom 330 runs at 1.6GHz, has a 1MB L2 cache, and sits on a 533MHz frontside bus (FSB). Compare those specs to those of the current Mac mini and you'll see what we mean: a 1.83GHz or 2GHz Core 2 Duo, 2MB or 4MB L2 cache, and a 667MHz FSB.

But comparing the dual-core Atom 330 to a dual-core Core 2 Duo involves much more than mere speeds and feeds. The architecture of the Core 2 Duo is far more sophisticated than that of the Atom. For one thing - one major thing - the Atom 330 (neé Dual Diamondville) is an in-order processor while the Core 2 Duo uses Out-of-Order (OoO) processing.

Simply put, OoO execution allows a processor to intelligently manage its workflow in a way that prevents precious processing cycles being wasted while instructions wait for data. In-order processors are too dumb to do this, so they can frequently twiddle their digital thumbs while waiting for something to do.

Again, Atom, in-order; Core 2 Duo, OoO. Or to put it another way, Atom, archaic; Core 2 Duo, modern.
...

The Atom was designed for low-end products such as netbooks, Ultra-Mobile PCs (UMPC), Mobile Internet Devices (MID), and the like. Asking it to carry a Mac on its back would be like asking Verne Troyer to give Yao Ming a piggyback ride.


But what does he know? He was only the Executive Editor of MacUser and, later, the Editor in Chief of MacAddict magazine. And what do I know? I've only been working with, and designing, microprocessor and microcontroller hardware for about 30 years.
 
I'm happy to have this discussion with you, but you've got to start holding up your end. If you think that the benchmarks I provided are flawed, then do some research and find some benchmarks that support your contention about the viability of the Atom 330.

Nah, I was just saying those benchmarks aren't fair. I agree that the Atom is a big CPU downgrade compared to the Core 2 Duo's found in the current Mini.

The E8500 Core 2 Duo CPU was released almost a year ago and the XFX 7800GT graphics card was released over three years ago. We're not talking cutting edge here. It's not like they compared it to a quad core i7 965 with a state of the art SLI video subsystem.

Yes, but the 7800GT card was nearly top of the range for three years ago, and the GMA 950 sucks hard. The 9400 will be way way faster, and will provide transcoding benefits for zipping, etc.

The memory controller is built into the Atom 330, not the motherboard chipset, so why would the RAM be significantly faster with the Ion platform? And, really, how fast is it going to be when the RAM bus speed is 533mhz? The MP3 encoding was slow because of the pathetic floating point performance of the Atom 330, not the RAM speed.,

My mistake.

But what does he know? He was only the Executive Editor of MacUser and, later, the Editor in Chief of MacAddict magazine. And what do I know? I've only been working with, and designing, microprocessor and microcontroller hardware for about 30 years.

Yes, but I'm a developer/tester and I know an unfair test when I see one. That was my point. I'm not disputing that an atom is worse than a core 2 duo. It's a lot worse, but it just isn't that much worse.

However, saying that, I thought it was an Atom 230 in the Ion? That'll make it worse!

See: Warning: PDF Link! 1.96 Mb
 
The E8500 Core 2 Duo CPU was released almost a year ago and the XFX 7800GT graphics card was released over three years ago. We're not talking cutting edge here. It's not like they compared it to a quad core i7 965 with a state of the art SLI video subsystem.

and when was the GMA950 chipset released ?

comparing once computer equipped with a gt 7800 and one equipped with a intel 950 and then drawing conclusions on the CPUs from that is just one thing: bad benchmarking
they should have put that intel the cpu on a motherboard with a similar chip-set than the atom and then compared...

But what does he know? He was only the Executive Editor of MacUser and, later, the Editor in Chief of MacAddict magazine. And what do I know? I've only been working with, and designing, microprocessor and microcontroller hardware for about 30 years.

and yet he goes around calling in-order execution archaic while there are/were plenty of modern CPUs designs which are defacto only in-order (Cell and the sadly failed Itanium) all while calling an x86 with more bells and whistles than it's own weight added, modern

it could easily be that they didn't recompile the benchmarks for the atom and that alone might explain quite a bit of difference

don't get me wrong an atom in any 500 buck computer is wrong and rubbish but so are such ridiculous lop sided benchmarks from the beginning
why didn't they compare the atom against intels current low end range in desktop (E5200/E5300) or even better: low level mobile cpus
 
Nah, I was just saying those benchmarks aren't fair. I agree that the Atom is a big CPU downgrade compared to the Core 2 Duo's found in the current Mini.

If you look at all of the benchmarks in the article (and elsewhere), it's pretty clear that the Atom 330 is weak in all areas. You have to remember that the GMA950 can perform at two or three times that frame rate when coupled with a modern CPU, so let's not be too quick to pin all of the blame for the dismal gaming on the GMA950.

The 9400 will be way way faster, and will provide transcoding benefits for zipping, etc.

Not sure I'm following you there with regards to file compression.

As far as transcoding, it's only going to help if someone writes software for an Ion-based Mac that uses the 9400 for transcoding. It's also unclear to me whether the Ion has the capability to, for instance, go from MPEG2 to h.264 or whether it's just capable of some pre-programmed stuff for preparing video for 1080i/p presentation.

That was my point. I'm not disputing that an atom is worse than a core 2 duo. It's a lot worse, but it just isn't that much worse.

My point is that it's so much worse that it isn't viable as a replacement for the Core 2 Duo in the current Mac Mini.

We can quibble about details of benchmarks, but if a consumer, or professional reviewer, saw a system that could not come close to matching the performance of a 2006 Mac Mini, it would be a marketing disaster for Apple.

However, saying that, I thought it was an Atom 230 in the Ion? That'll make it worse!

The Ion will support an Atom 330, but there's not a huge performance difference between the two parts.

If an Atom CPU and Ion chip is used by Apple, it will be used in an updated Apple TV -- which was really the kind of market for which the Atom 330 was aimed.
 
There is no doubt the Atom is weaker than a standard processor (duh, it's supposed to be) but the point is that bad benchmarking is bad benchmarking, and is bad science to boot :D
 
Remember the Performa Apple!

Apple tried a watered down processor in their Performa line (mid-90's) and unsuspecting buyers were terribly disappointed with their purchase.

Please Apple, don't repeat your mistake again! :confused:
 
Of course it's a rumor and maybe other models will be offered, but it's quite simple for me:

Atom processor - No purchase. Core-Duo processor - I buy 2 or 3.
 
but the point is that bad benchmarking is bad benchmarking, and is bad science to boot :D

It measured exactly what it was supposed to measure: How well an Intel D945GCLF2 motherboard with an Atom 330 processor performs. That was the purpose of the article: To review that motherboard with that processor.

It sucked for gaming. If you put a $500 video card on it, it would have still sucked. It would also have been stupid since the CPU/motherboard combo they tested costs less than $90 from Newegg.

Other CPUs get several times the gaming performance of the Atom 330 when paired with the same GMA950 graphics processing unit.

You're getting way too hung up on the gaming benchmarks. Look at all of the benchmarks taken together. The CPU failed in the floating point test, failed in the MP3 encoding test, failed in the video encoding tests, and failed in the memory bandwidth tests. It fell flat on its face over and over.

Their choice of a year-old CPU and a 3 year old video card for comparison was a good one. Not state of the art and not terribly outdated. Probably pretty typical of the kind of system that many of their readers own.
 
It measured exactly what it was supposed to measure: How well an Intel D945GCLF2 motherboard with an Atom 330 processor measures in standardized benchmarks. That was the purpose of the article: To review that motherboard with that processor.

You're getting way too hung up on the gaming benchmarks. Look at all of the benchmarks taken together and it's obvious that the GMA950 was not crippling the CPU. The CPU failed in the floating point test, failed in the MP3 encoding test, failed in the video encoding tests, and failed in the memory bandwidth tests. It fell flat on its face over and over.

Their choice of a year-old CPU and a 3 year old video card for comparison was a good one. Not state of the art and not terribly outdated. Probably pretty typical of the kind of system that many of their readers own.

Again, it's bad benchmarking. Because they put too many variables out there (motherboard, chipset, GPU, etc.) and ran standardized programs. To get true comparisons, you have to minimize the variables and they didnt.

I've read about Atom long before, and I know it isn't a top performing CPU. It has a very set focus and goal, and Apple would do a bad job if they tried to make it do more than it was intended to.
 
Again, it's bad benchmarking. Because they put too many variables out there (motherboard, chipset, GPU, etc.) and ran standardized programs. To get true comparisons, you have to minimize the variables and they didnt

If they ran the game benchmarks at low enough settings than even those are pretty indicative, and the rest of those benchmarks aren't going to be affected to any huge degree by the chipset. Sure, maybe a few percent or whatever, but not the kind of difference those showed (and indeed the gaming benchmarks seemed to reflect the other benchmarks).
 
Again, it's bad benchmarking. Because they put too many variables out there (motherboard, chipset, GPU, etc.) and ran standardized programs. To get true comparisons, you have to minimize the variables and they didnt.

You are really starting to frustrate me. It's not "bad benchmarking" so stop saying that it is.

It is a system-level test, not a component test.

The product being evaluated is an Intel D945GCLF2 motherboard. It has a non-removable Atom 330 CPU and an integrated GPU.

The reader is trying to decide whether to buy that product (the motherboard/CPU combo). They don't care how fast the CPU would be in some other motherboard. They don't care how fast the motherboard would be with a different CPU, GPU, or chipset. They want to know how that unfamiliar motherboard/CPU combo compares to a normal PC that uses a mainstream CPU, chipset, and GPU.

It's a black box test: They plug in a keyboard, mouse, display, and disk drive into each system and perform the benchmarks. Since they used the same model disk drives, the benchmarks were done properly and fairly.

When a car magazine compares the performance of a BMW to a Lexus, they don't start with two identical BMWs and then swap the Lexus engine into one of them.
 
It is a system-level test, not a component test.

The product being evaluated is an Intel D945GCLF2 motherboard. It has a non-removable Atom 330 CPU and an integrated GPU.
I definitely agree here, although a component test would need other variables to be equal, this isn't a component test, and even if it was, it wouldn't be realistic.

And since the Atom is not removable, the only possible test is a system test, and that shows clearly that the Atom is very slow compared to the Core 2.

And I like your car analogy. :cool:
 
The benchmarks that fmaxwell provided might not be the best but it's still a bad idea. I've seen benchmarks of a hackintosh wind vs a 1st gen mac mini and the g4 mini is still faster. It's the atom 270 but it doesn't make that big of a difference. Atom would be the worst idea ever unless they plan to make the most energy efficient desktop that will fit in your pocket. If they use an atom CPU then it better be the size of an iPod classic with an adapter to plug in a mouse, keyboard and display and run leopard.
 
The benchmarks that fmaxwell provided might not be the best but it's still a bad idea.

I'd love to be able to provide benchmarks showing the exact rumored "next gen" Mac Mini hardware configuration (CPU, chipset/GPU, but I have not able to find such benchmarks.

Lacking that, the system benchmarks that I provided make it pretty clear that the Atom 330 lacks the processing power that users expect and demand.
 
I'd love to be able to provide benchmarks showing the exact rumored "next gen" Mac Mini hardware configuration (CPU, chipset/GPU, but I have not able to find such benchmarks.

Lacking that, the system benchmarks that I provided make it pretty clear that the Atom 330 lacks the processing power that users expect and demand.

But it would probably be more then enough in a new Apple TV if coupled with a hardware video decoder chip.
 
Why would you downgrade? They might as well use a dual core g4 (if they invented that)

Playing devil's advocate here: The argument could be made that a $399 Mac Mini might sell to people who would not be willing to plunk down the current $599.

That said, I don't see Apple releasing such a product.

First, Apple has profit margins that make Michael Dell green with envy. While you and I pay a pretty penny on eBay for Core 2 Duos, Apple is getting them for a fraction of that, so going to an Atom 330 doesn't cut $200, or even $100, off of Apple's cost to produce a Mac Mini.

Second, Apple is all about user experience and the Mini is the first Mac purchased by many users. If the user is frustrated by a lethargic OS and can't even get satisfactory performance out of bundled apps like iMovie, then such a system would be likely to drive users back to Windows PCs and result in bad word-of-mouth about how slow OS X is (yeah, we know that's not fair, but people give bad online product reviews because UPS lost their packages -- go figure).
 
But it would probably be more then enough in a new Apple TV if coupled with a hardware video decoder chip.

Agreed. That's what I predicted a few messages back.

I believe that the "source" of this rumor saw a prototype Apple TV and mistook it for a prototype next-gen Mac Mini.
 
Agreed. That's what I predicted a few messages back.

I believe that the "source" of this rumor saw a prototype Apple TV and mistook it for a prototype next-gen Mac Mini.

Especially since a lot of people keep saying that the Mac mini and the :apple:TV/Time Capsule share the same "form-factor".

Yes they have the same shape and design, but a Mac mini is 6.5x6.5" while the :apple:TV/Time Capsule are 7.7x7.7".

I do hope to see a new 7.7x7.7" Mac mini with a 3.5" drive in it though.

Knowing Apple, however, I would not be surprised to see a 5.47x5.47" Mac mini (same size as the MacBook Air SuperDrive) with a 1.8" drive (again, same as MacBook Air).

And while it would not make sense to put a 1.8" drive in a desktop computer, remember that the current Mac mini uses a 2.5" laptop drive.
 
based on the quote in one of the other threads about low end iphones and netbooks it sounds like apple would not want to put a netbook cpu in a desktop computer. It wouldn't be "the kind of experience people want."


"We're watching that space, but from our pov the products are based on hardware that's much less powerful, software technology that's not good, cramped displays. We don't think that people are going to be pleased with those type of products. It's a category we watch, we have some ideas here, but we think the products there now are inferior and won't provide the kind of experience people want."

Good point... :)

Maybe we can take the updated White MacBook as the example??? :eek:

Since when is creating a WinRAR file or converting audio to MP3 "high-end gaming"?

If you read my later posts, the quote from bluedevils convinced me...

Yes, I don't think Apple will be too happy if the Atom takes a long, long time to rip an Audio disc...
 
You people are crazy! its been almost two years since the last Mini update and not a single word from Apple yet this web site like others host millions of lies and rumors regarding the Mini not a single one has any credibility at all.
 
Especially since a lot of people keep saying that the Mac mini and the :apple:TV/Time Capsule share the same "form-factor".

Yes they have the same shape and design, but a Mac mini is 6.5x6.5" while the :apple:TV/Time Capsule are 7.7x7.7".

I do hope to see a new 7.7x7.7" Mac mini with a 3.5" drive in it though.

Knowing Apple, however, I would not be surprised to see a 5.47x5.47" Mac mini (same size as the MacBook Air SuperDrive) with a 1.8" drive (again, same as MacBook Air).

And while it would not make sense to put a 1.8" drive in a desktop computer, remember that the current Mac mini uses a 2.5" laptop drive.

I think there is enough of a cost premium with 1.8 drives that it will remain a 2.5 drive. 3.5 would just make too many people happy, that would never happen! :)
 
You people are crazy! its been almost two years since the last Mini update and not a single word from Apple yet this web site like others host millions of lies and rumors regarding the Mini not a single one has any credibility at all.

nah im actually pretty sure that there was a fictional story on here somewhere from Apple, they talked about the status of the Mini and the iMac. it wasnt that long ago and i dont remember what they said hahaha.
 
I'd love to be able to provide benchmarks showing the exact rumored "next gen" Mac Mini hardware configuration (CPU, chipset/GPU, but I have not able to find such benchmarks.

Lacking that, the system benchmarks that I provided make it pretty clear that the Atom 330 lacks the processing power that users expect and demand.

Hey I totally agree with you.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.