The stats are in (historical price/performance graphs)

[avemestr]

So, I spend some time doing a couple of graphs.

The first graph shows Geekbench performance for the weakest iMac, strongest iMac, weakest Mac Pro and strongest Mac Pro over time (higher is better):

Three observations:

The strongest iMac really took off, when they switched from a Core 2 Duo to an i7.

The first powerful Mac Pro ($3299 in 2006) was not surpassed by the strongest iMac before 2011, and it still beats the weakest iMac by quite a margin.

The weakest current Mac Pro ($2499) is beat by last years best iMac ($2199).

Another graph showing the $/Geekbench Score (lower is better):

Two observations:

The iMac shows by far the best price/performance ratio at the moment. But it wasn't always so.

Even though it's "common sense" that the current low-end Mac Pro is far to expensive, it actually seems to provide a better price/performance than a current CPU-maxed out Mac Pro.


In conclusion:

Of course the Mac Pro has a lot going for it; expandability and all that.

But since 2006 the price per Geekbench point on the strongest Mac Pro has only declined from 34 cents to about 28 cents. In comparison, the price on the strongest iMac has gone from 62 cents in 2007 to about 19 cents in 2011.

It's obvious from the graphs, why the Mac Pros have been such great propositions in the past.

Something drastic took place in 2009 with the iMac. To make the Mac Pro competitive again, something similar has to happen to that line with Sandy/Ivy Bridge. The forthcoming 2012 iMac will just increase the price/performance gap, and make the Mac Pro an even worse proposition that it already is.

Bottom Line: We need a new Mac Pro

Sources:
http://browser.primatelabs.com/mac-benchmarks
http://www.everymac.com/global-mac-prices/mac-prices-us-usa-united-states-america.html
http://en.wikipedia.org/wiki/Intel_iMac
http://en.wikipedia.org/wiki/Mac_Pro

P.S.
I know there's more than geekbench points that matters, when it comes to performance. I.e. Intel AVX provides "performance that you feel in your daily work", but those kinds of technologies is also only present in iMac, making the Mac Pro look even worse.

P.P.S.
If I've made an error, it's a mistake an not on purpose. Never attribute to malice, what can adequately be explained by incompetence.

 

[tamvly]

I think it would be interesting to include the laptops in your graph. Many would argue that it's not just the Mac Pro that will ultimately fade away, but most desktops as well.

I have a 2012 MacBook Pro (9,1) with a 2.3 GHz CPU + an SSD and 16GB of memory. I also have a 2008 Mac Pro (2 x 2.8 GHz) with an SSD boot and SSD user data disks and 12 GB.

In everyday use, the laptop feels much faster. I think their GeekBench scores are just about the same. If it weren't for the multiple disk drives I'd ditch the MP.

 

[avemestr]

You have a valid point - I might look into it next weekend

On the same note, El Awesome couldn't understand why his 2009 MP performed worse in a CPU-demanding game than a MBPr. The stats speaks for themselves:

https://forums.macrumors.com/posts/15710139/

 

[gpzjock]

In your $/geekbench points ratio how do you balance an iMac with a high spec IPS panel against a Mac Pro without one? Would it be more representative if you added the price of an equivalent screen to the Mac Pro?
Do you balance this with the expandability of the tower?

 

[avemestr]

In your $/geekbench points ratio how do you balance an iMac with a high spec IPS panel against a Mac Pro without one? Would it be more representative if you added the price of an equivalent screen to the Mac Pro?
Do you balance this with the expandability of the tower?

I haven't added a $999 display to the MP calculation. It would make it an even worse proposition (move the $/GB-point to 32 cents for the strongest and 36 cents for the weakest Mac Pro), and hence increase the gap to the iMac.

Worth noting is the fact, that adding a monitor makes the stronger Mac Pro a better "GB value" than the weaker one.

OTOH, I haven't placed a value on the expandability of the Mac Pro. For people that require PCI-E slots, it's priceless. For some, they're virtually worthless. Impossible to put a $-tag on.

So the calculations are based on the machines power and price alone.

 

[thegreatdivorce]

Agreed, I'd love to see laptops (MBP specifically) added to this. While we're making requests, a legend of what exact models the "strongest" and "weakest" were.

 

[ScanPro]

avemestr,
Thank you,
excellent post, very informative.
SP

 

[avemestr]

Agreed, I'd love to see laptops (MBP specifically) added to this. While we're making requests, a legend of what exact models the "strongest" and "weakest" were.

The "strongest" and "weakest" are defined based on these benchmarks:

http://browser.primatelabs.com/mac-benchmarks

So basicly I search for i.e. "Late 2009" and look for iMacs. The first (strongest) I find is the i7-860 2800 MHz (4 cores) with 8330 points. The last (weakest) I find is Core 2 Duo E7600 3060 MHz (2 cores) with 4175 points.

Those numbers are averages for the specific model, because some might have upgraded RAM etc, but it's nevertheless a good measure of the performance people owning the specific model have "at their fingertips" on average.

 

[Pressure]

Heh, based on Geekbench ...

 

[avemestr]

Heh, based on Geekbench ...

Please, do tell. What metric should be used instead?

The Geekbench scores actually favors the MP regarding performance. A lot of applications can't utilize all the cores of a system, so the Geekbench score makes an MP look better than it is.

And the MP doesn't look good, so that only reinforces the conclusion that the MP is a hard sell these days.

But please let me know which metric you'd use to quantitatively compare machines going back 5 years, and at the same time take into account configuration choices made by individuals. I'd love to see the stats you provide, and how they differ from the ones I've provided.

"Never Argue With An idiot. They'll Lower You To Their Level And Then Beat You With Experience!". It's hard to argue with an idiot, when he makes a stupid, uninteresting remark and doesn't provide any arguments.

 

[Torrijos]

The thing is benchmarks are usually well paralleled, while everyday software isn't.

The question then is how well real life work is going to fare, knowing that you might not be able to leverage the 12(24) cores and better memory throughput of the Mac Pro.
This might be another negative for the Pro.

The big positive for the Pro that no iMac has or will be able to match is GPUs or GPGPU performances.

In order to have a better picture a GPU focused benchmark should be performed and a composite score made. Then the Pro might be clear winners, as long as you use Pro software fully optimized for GPGPU and lot of cores.

 

[avemestr]

The thing is benchmarks are usually well paralleled, while everyday software isn't.

The question then is how well real life work is going to fare, knowing that you might not be able to leverage the 12(24) cores and better memory throughput of the Mac Pro.
This might be another negative for the Pro.

Yep, agree!

The big positive for the Pro that no iMac has or will be able to match is GPUs or GPGPU performances.

In order to have a better picture a GPU focused benchmark should be performed and a composite score made. Then the Pro might be clear winners, as long as you use Pro software fully optimized for GPGPU and lot of cores.

True. But I find it impossible to determine how a composite score should be calculated to be fair for all.

GB isn't an answer to all, I fully agree. But it provides a scoring system that have been consistent for many years, and provides averages across a number of customizations and tens of thousands of machines.

If you have a Quadro card and you use software that utilizes CUDA day and night, the GB score for your specific machine doesn't tell how well it performs on the tasks you use it for. But thats statistics for you: There'll always be data points not following the over-all trend. That doesn't invalidate the overall stats altogether.