Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
It's hardly Apple & Intel's fault that developers are lazy or incompetent

well, ya know, it IS possible to design CPUs optimized to the software people CURRENTLY have.

I've done that. I may be doing it again, soon.
 
Get SSDs?

Yeah, grab yourself 4 large SSD's (OCZ Core Series?) and a semi decent RAID card. That should boost the performance a little more :D Though I don't think you'd be paying $5000 on the system (with decent ram upgrade), more likely to be in excess of $6000~7000... Then there's the choice of al those beautiful displays, another $5000~6000 if you choose an EIZO ColorEdge CG301W.

Cha-ching!! $11,000 already!!

Large investment i it's just gonna be a home computer...
 
Sure it's possible, but with higher clock rates come higher heat, etc etc etc.

There are ways to not have to raise clock rate, too. And ways to keep power constant while raising clock rate. More importantly, "optimize" doesn't mean "raise clock rate."
 
Encoding 1080p materials, probably use the hdd bay to keep all my bluray movies and act as a huge server.

I also run vmware fusion for a few apps under windows, basically massive amounts of d/ling from unison, transmission, constant uploading huge files, handbrake once in a while, visual hub converts for iphone (hdtv episodes and such), plex player, FCP (just been into it since last year), etc..

I do know that I use up a good amount of RAM with my daily usage pattern something around 15GB worth from when I had the 2.8ghz from last year.

I havnt been able to do half as much as what I've listed on with my mac mini of course as it is a temporary solution until I choose the right mac pro. But for now I'm starting to think I should target either a 2.8GHz last gen model or a 2.26GHz OCTO.

I could probably get by fine with a Nehalem Quad but why settle for fine when you can get overkill fastness.


Man, I dont get it. I thought Nehalem would make a great torrent box.
 
Man, I dont get it. I thought Nehalem would make a great torrent box.

The mac mini makes a great torrent box. If I do decide on the Mac Pro, I would move the mac mini over to the sony bravia as an htpc.

I still think the mac mini would be over kill to use just as an htpc though.
 
After reading the thread I'm glad to see that the OP is coming to a fuller understanding of all of the benchmarks that we've been posting here. He's 100% correct too. The new machines in general use are between 5% slower and 15% faster. But this was generally true of each new wave of Mac Pro releases. And they're right on schedule with being 120% to 160% of the speed of the 2008 Macs at special purpose computing such as rendering, encoding, and etc. This follows the rules of pretty much every PC vendor - doubling the speed of their products about every two years. This year it's 150% and next year (or so) it will be 200% over the 2008 models.

There has been a subtle somewhat unnoticed shift however to the focus of such special purpose computing. For the 1st 20 years or so we doubled the clock speed of single core chips - now however, we're seeing the focus shifting as bandwidth (number of cores and core interaction) instead of the clock-speed, is being doubled every 2 years or so. I think it's interesting.

What's different and disappointing this year (from Apple) as we've discussed to death here on MacRumors, are the price hikes or as some people call them, the "Apple Premiums". I don't find this very interesting at all. Well about as interesting as a rock in my shoe. :D

Anyway I understand the OP's disappointment. I'm only surprised he isn't much more disappointed than he's saying here. I say this because he built very high expectations initially and when one expects a lot it's much more easy to be disappointed a lot.

The new 2009 machines are what they are though. I guess Apple isn't going to change policy mid-stream just for the users capable of discernment - even though I personally think they should regarding the new prices.

One saving grace to count is that we have such great users here! We know now how to get 4870 and 4890 ATI card working for less than 1/2 what Apple wants for one. And we know the processors are upgradable without too much fuss. We also know thanks to so many who contributed their time benchmarking the things, almost exactly how each machine profiles for everyday use as well as for special purpose computing - so anyone willing to spend 4 to 6 hours here reading, can discover all these things and more. Mac Rumors, what a collection of great users! :cool:

.
 
There has been a subtle somewhat unnoticed shift however to the focus of such special purpose computing. For the 1st 20 years or so we doubled the clock speed of single core chips - now however, we're seeing the focus shifting as bandwidth (number of cores and core interaction) instead of the clock-speed, is being doubled every 2 years or so. I think it's interesting.

.


More important than clock speeds, the increase in the number of transistors has increased at an unabated rate (in fact, it's increased).

Early on, CPU's were not pipelined. Then they were in-order, single pipeline. Then floating point was added. Then in-order, multiple pipeline. Then out-of-order issue, in-order retire. Then fully out-of-order. Then on chip caches. Then branch-prediction tables. Virtual memory tables. Saturating vector math. Cache tags for external caches. L2 caches. Multi-threading. Northbridges. Multicore. Soon graphics.

More than one way to skin a cat. :) But parallelism started early on, with superscalar architectures. Arguably still a more efficient way to go, but the software/hardware interface is a little tougher to get right to take full advantage of it. Also gets increasingly expensive to keep track of deep speculative state, and the penalty for speculating wrong goes up commensurately.

We ran into a clock speed wall both because power scales as a square, because clock distribution (and hence skew) became a huge fraction of the cycle time, because juicing the process results in massive leakage, because interconnect distributed caps became more important than gate gaps, etc.
 
The 2009 Mac Pro is my first Mac tower ever... (Always been an owner of MacBook Pros & PowerBooks) So i cannot know my impression with this and the old 2008 Mac Pros.
 
I honestly believe though that if I opt for the 2.26GHz or the 2.66GHz Octos, I wont notice much of a difference from the 2.8GHz model that I owned last year.

I guess its my own personal fault for believing that Nehalem was this magical architecture that will actually boost performance considerably in everyday tasks.

But in reality its so small in speed bump that I'd actually feel like I'm using a last year's Mac Pro while paying $500 more.

Man I should have jumped on that $1899 2.8GHz 8 core model when I had the chance. :mad:
 
I honestly believe though that if I opt for the 2.26GHz or the 2.66GHz Octos, I wont notice much of a difference from the 2.8GHz model that I owned last year.

I guess its my own personal fault for believing that Nehalem was this magical architecture that will actually boost performance considerably in everyday tasks.

But in reality its so small in speed bump that I'd actually feel like I'm using a last year's Mac Pro while paying $500 more.

Man I should have jumped on that $1899 2.8GHz 8 core model when I had the chance. :mad:

You can probably still find a used one for that. And yeah, the 2.66 isn't going to be a speed increase over the the 2.8 that will make you sit up in your chair with excitement and cheers. ;) I bet dropping an ATI HD4890 into a 2.8 would come closer to inducing that feeling tho!

I think if you take a 2008 2.8 and add 3x or 4x RAID, put 16GB of RAM, and drop that 4890 card in you're going to spank Apple's 2.66 Nehalem octad every which way!!! (except for a few 3D rendering application's renderers) And still have paid $1,500 to $2,000 less than Apple's bare bones (base) price.

The crapper part of all of this is that none of this could be said and wouldn't be brought up at all probably if they had just kept the same price structure pacing.

.
 
Yeah, grab yourself 4 large SSD's (OCZ Core Series?)

Save the money.

So far the only decent Solid State Disks are the Intel X25-M, Samsung made drives and the OCZ Vertex series.

Of those the best clearly is the Intel X25-M.

The rest will be latency galore.
 
The crapper part of all of this is that none of this could be said and wouldn't be brought up at all probably if they had just kept the same price structure pacing.
I was going to say something like this.

If Apple released the SP Mac Pro at $2199 (similar to the lowest config 2006/2008 Mac Pro) and the DP Mac Pro at $2799, there probably wouldn't be many complaints.

If Apple just used 2.67/2.93/3.2 GHz CPUs instead of 2.27/2.67/2.93 GHz CPUs, things would be better.
 
I was going to say something like this.

If Apple released the SP Mac Pro at $2199 (similar to the lowest config 2006/2008 Mac Pro) and the DP Mac Pro at $2799, there probably wouldn't be many complaints.

If Apple just used 2.67/2.93/3.2 GHz CPUs instead of 2.27/2.67/2.93 GHz CPUs, things would be better.

You pretty much stole my thunder with this one. The Octad machine is priced very much in line with Windows workstations but the Quad is not. I know Dell/Lenovo/HP give you crap at the base model, but their bases start between $1,100-$1,300 dollars. After adding 3rd party ram, hard drives, warranties and video cards for the windows machines the Dell T3500 was $800 cheaper and the Lenovo S20 was $1200 cheaper. Oh and they have 6 ram slots =)

I just keep thinking "keep the iMac until the next revision" and hopefully Apple will put more ram slots to match triple channel and/or give us at least a real video card for the money but who am I kidding? :rolleyes:
 
I honestly believe though that if I opt for the 2.26GHz or the 2.66GHz Octos, I wont notice much of a difference from the 2.8GHz model that I owned last year.

I guess its my own personal fault for believing that Nehalem was this magical architecture that will actually boost performance considerably in everyday tasks.

But in reality its so small in speed bump that I'd actually feel like I'm using a last year's Mac Pro while paying $500 more.

Man I should have jumped on that $1899 2.8GHz 8 core model when I had the chance. :mad:

So what was the turning-point for you? You were such an advocate last week?!
 
I don't know about the nehalem's, but I definitely love my 2008 2.8GHz dual quad MP! It's a beautiful machine, and fast for our uses. And I have 10 GB in it right now. I just love that I can kick it up to 32GB if I want to, or anywhere in between.
 
I'm not satisfied with OP's unsupported assertions. I am going to wait for the next generation of assertions.

You win!



Encoding 1080p materials, probably use the hdd bay to keep all my bluray movies and act as a huge server.

I also run vmware fusion for a few apps under windows, basically massive amounts of d/ling from unison, transmission, constant uploading huge files, handbrake once in a while, visual hub converts for iphone (hdtv episodes and such), plex player, FCP (just been into it since last year), etc..

Maybe you should spend your lucre on an eye-patch and a peg leg. Next generation parrots are supposed to be hot too, but you could save a bundle on the last gen ones though.
 
Just thinking out loud, but does it make sense to do a cost analysis based on the Geekbench points?

For example, take the price ($ or £) of the 2008 2.8GHz Octo (stock model) and divide by the geekbench score. That would give you $ or £ per point.

Do the same for the 2009 2.66 Quad, 2009 2.26 Octo, and so forth.

Is there a website out there that already does this?
 
I just keep thinking "keep the iMac until the next revision" and hopefully Apple will put more ram slots to match triple channel and/or give us at least a real video card for the money but who am I kidding? :rolleyes:

If they took away slots to be a multiple of three, would that make you happy?
 
If I might throw this historical perspective out there:

My last Mac was a G5 Quad with 8gb RAM, 2 internal 500gb HD's and the 7800GT video card. For a little over $1k more, I will have a Mac Pro (tomorrow) which is 2x faster (using Photoshop as a measure which isn't overly efficient), has 4x the storage, 2x the RAM, much better video card (by how many times I don't know), much faster RAM, 2x the cores (4x if you count hyper-threading), runs cooler, more energy efficient, will be supported for 64 bit and will run Snow Leopard.

So for those of us who keep our Macs for 4 years (or more), compared to what we paid then, the new Mac Pro's really aren't a bad deal.

As has been mentioned before, Apple probably made a sweet deal with Intel for the first few generations as part of the transition and now in all likelihood the deal has expired and we will end up paying higher prices for Macs going forward.
 
Just thinking out loud, but does it make sense to do a cost analysis based on the Geekbench points?

For example, take the price ($ or £) of the 2008 2.8GHz Octo (stock model) and divide by the geekbench score. That would give you $ or £ per point.

Do the same for the 2009 2.66 Quad, 2009 2.26 Octo, and so forth.

Is there a website out there that already does this?

Is there any doubt that the 2008 provides better value? This would merely prove what we already know. One snag is that I believe Geekbench scores scale proportional to CPU cores/threads... which is not true of most application performance... thus such a metric would be misleading as actual value would be less at the top end than this metric might suggest.

Here's another reason why this is probably both obvious and not that helpful... A 2001 BMW M5 with 400 Horsepower can now be had for $20K while a 2009 BMW M5 with 500 Horsepower can be had for $100K... it's clear that the 2001 provides 4x more horsepower per dollar in value, but that's not a factor most people use to make a purchase decision like this. BTW, you can imagine the uproar that occured in 2006 when the new model debuted with only 20% more power but cost nearly twice what the previous model could be had for (sound familiar?). :p

Life on the bleeding edge is not cheap. :p
 
If I might throw this historical perspective out there:

My last Mac was a G5 Quad with 8gb RAM, 2 internal 500gb HD's and the 7800GT video card. For a little over $1k more, I will have a Mac Pro (tomorrow) which is 2x faster (using Photoshop as a measure which isn't overly efficient), has 4x the storage, 2x the RAM, much better video card (by how many times I don't know), much faster RAM, 2x the cores (4x if you count hyper-threading), runs cooler, more energy efficient, will be supported for 64 bit and will run Snow Leopard.

So for those of us who keep our Macs for 4 years (or more), compared to what we paid then, the new Mac Pro's really aren't a bad deal.

As has been mentioned before, Apple probably made a sweet deal with Intel for the first few generations as part of the transition and now in all likelihood the deal has expired and we will end up paying higher prices for Macs going forward.

Are you buying a 2.8GHz 8 core model? I noticed you said only a $1k more, because there is no way you could sell that Powermac rig and pay a $1k difference to buy a 2.26GHz Octo model. But my guess is either a 2.8Ghz octo from last generation model or a 2.66GHz Quad.

My point is that I'm not comparing a powermac to the Nehalem Quads or Octos, I'm comparing last generation Octos against this generation Octos. And its a downgrade for $500 more.
 
Just thinking out loud, but does it make sense to do a cost analysis based on the Geekbench points?

Is there a website out there that already does this?

No it doesn't. And yes, that's what barefeats and all of the other extreme loyalist sites are doing - or seem to be doing.

The pattern has always been that machines get faster and faster for roughly the same price point. If not we would be paying millions of dollars for a machine today. Because it's X Times faster than the Apple ][, X Times faster than the 68000, 68010, 68020, 68030, 68040, 68060, G1, G2, G3, G4, G5, Woodcrest, Hampertown, etc. etc.

So if it made sense to bump prices based on those progressions we would all still be using the Mac IIfx.

Yet this is the change-up logic that Apple and their loyalist support sites are trying to feed us. I'm not that dumb myself.

But since I don't think (didn't think) Apple was evil or greedy I keep looking for alternate reasons. I mean there has to be a reason they jacked the price structure up across the board between $1,000 and $2,000 right? The only thing I've been able to come up with is that they know better than us common folk what the affects of the recent economical collapse are going to be and are pricing their systems based on real dollar devaluation.

But if that were true then wouldn't IBM, Gateway, HP, Sony, DELL and others be doing the same thing? None of them are. Some have raised prices $200, $300, on previously similarly priced units but nothing close to Apple's jump.

Anyway, I still looking for an explanation meself.
 
My point is that I'm not comparing a powermac to the Nehalem Quads or Octos, I'm comparing last generation Octos against this generation Octos. And its a downgrade for $500 more.

Yup, and that is if you allow them to slide their entire line backwards one notch. I contend that :

the new 2.93 replaces the old 2.80,
the new 2.66 replaces the old 2.66,
the new 2.26 replaces the old 2.00 (from 2007)
And they don't have a replacement for the old 3.20 this year.

Arguably I guess we could say that in the 2.93's case it replaces last year's 3.00.

The single chip quads are a new low-end offering. They don't replace the Quads from 2006 because those were already replaced with Octads in 2007 and for the same basic price point.

So really saying that it's only a $500 increase when it's actually $1,000 to $2,000 is being very kind - too kind.

Here you can see the progression over the past 4 iterations:
2006
Mac Pro Quad 2.0GHz $2,199 NEW
Mac Pro Quad 2.66GHz $2,499 NEW
Mac Pro Quad 3.0GHz $3,299 NEW

2007
Mac Pro Quad 2.0GHz $2,199
Mac Pro Quad 2.66GHz $2,499
Mac Pro Quad 3.0GHz $3,299
Mac Pro 8-core 3.0GHz $3,997 NEW

2008
Mac Pro Quad 2.8GHz (2008) $2,299 NEW
Mac Pro 8-core 2.8GHz (2008) $2,799 NEW
Mac Pro 8-core 3.0GHz (2008) $3,599 NEW
Mac Pro 8-core 3.2GHz (2008) $4,399 NEW

2009
Mac Pro Quad 2.66GHz $2,499 NEW
Mac Pro Quad 2.93GHz $2,999 NEW
Mac Pro 8-core 2.26GHz $3,299 NEW
Mac Pro 8-core 2.66GHz $4,699 NEW
Mac Pro 8-core 2.93GHz $5,899 NEW​
We can also take this same progression back to the early 80's and it follows along perfectly.
Here's that on just one machine base:
1998 Apple Releases a 1 core G3 266MHz (AV) $2500 ... Speed increase from previous = more than 2X
2000 Apple Releases a 2 core G4 450MHz $2500 ... Speed increase from previous = more than 3X
2002 Apple Releases a 2 core G4 1.00GHz $2500 ... Speed increase from previous = more than 2X
2004 Apple Releases a 2 core G5 2.00 GHz $2500 ... Speed increase from previous = more than 2X
2006 Apple Releases a 4 core 2.66 GHz $2500 ... Speed increase from previous = more than 3X
2008 Apple Releases a 8 core 2.8 GHz $2800 ... Speed increase from previous = more than 4X
2009 Apple releases a 8 core+HT 2.66 GHz $4700 ... Speed difference = between 0.9X ~ 1.68X = on track for odd year.​

BOOM! What happened to that last entry?

.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.