Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
I have never seen an entry-level Mac less expensive than an entry-level PC of the same hardware, I don't think at least. I think it's odd that HP's Z820 (This thing is absolutely top-of-the-line, I mean what it is capable of is absolutely insane) starts out at less than their entry-level workstation ($2339) but then again it comes with less powerful hardware, only 4GB of RAM and a 500GB hard drive. Aside from the processor, my HP Envy is more powerful. So it's not much comparison, even if it is extremely upgradeable and powerful when it is. I'd say these ones are probably more of a better buy, especially the Mac Pro because it costs the least and you could install Windows on it, so it'd run the same stuff as the other workstations.

That's a good idea for me because the only Mac OS X programs I'd use that could utilize the hardware is Final Cut Pro X and Adobe After Effects and such. I'd just use Windows most of the time because I play games quite often. And I still prefer Windows either way. But the Mac Pro now seems pretty cool. I love the lights on it; they remind me of TRON: Legacy.

EDIT: Unless you're a large corporation buying an absolutely top-end HP Z820, don't try and buy one. Fully configured (Two 12 Core E5-2697 Xeons, Two NVIDIA Quadro K6000s, 24GB total VRAM, 15TB storage, 512GB RAM, etc) this thing will run you $70,726. I just chose the configure option on their site. There's a bunch of other little things that add to that, but still, that's pretty crazy. The Mac Pro at most will probably run you only ~$15,000.
 
Last edited:
Did you read the article in full?

Also, I can't say that I know much about the computing needs of the scientific community, but if you have to run calculations that will take days, you're probably better off using something other than a Mac Pro.

I would love to hear the reasons for your conclusion. Why do you think anything other than a Mac Pro is a better solution? I suppose it depends what your doing but Matlab and R both using GPUs now. The Mac Pro has great thermal characteristics. What's the issue? Just don't like Apple?

I find it odd that someone would make such a bold claim about something they also say they know nothing about.
 
Not to mention an i7 is no match for a Xeon. They are not in the same league. And before you accuse me of being an Apple Fanboi, my day job is managing Linux servers so I know a thing or 2 about Xeon vs i7.

Which Xeon and which i7 did you compare? A desktop quad Haswell i7 beat the s*** out of quad Xeon E5 for less dollars.

Xeon is a huge markup from Intel, simple as that. For multithreading, sure a single hex Xeon would kill an i7 though not by much. But for what it's worth, I'll stick with i7.

So, I don't really know what do you mean by "i7 is no match for a Xeon", seriously? You maybe fed up by the brand? Xeon indeed sounds more cool and techy. Doesn't mean it's a lot better, though.
 
Ok, so it takes two raided SSD's to get the equivalent of one in the 2013 Mac Pro. That tells you of how good the Apple PCIe SSD's they are currently using.

They're not some Apple innovation.

Apple simply decided to adopt PCIe flash for all but their Mac Mini simply because OTHER COMPANIES started making it.

The PCIe flash in the Mac Pro is clearly Samsung PCIe flash that's been adapted to use a proprietary Apple connector.

It wasn't Apple who thought one day,

"I know, I'll take a PCIe RAID card with the silicon of 2 SATA 6Gb/s controllers and a caching scheme.

Add 2 SSDs with NAND flash memory and silicon for garbage collection, caching and the SATA 6Gb/s interface.

Then combine all the circuitry on 1 card to cut out interface bottlenecks and integrate the circuitry."

Apple DIDN'T suddenly think of this. It was an OBVIOUS move from the companies who make SSDs and SATA controller chips used in PCIe cards in the first place and shrinking the form-factor was a trivial matter when they can just use higher density NAND to save on circuitboard space.

Apple were just the first to start using the fastest kind as a standard feature in most of their range but it had to exist for them to source the parts!

----------

Which Xeon and which i7 did you compare? A desktop quad Haswell i7 beat the s*** out of quad Xeon E5 for less dollars.

Xeon is a huge markup from Intel, simple as that. For multithreading, sure a single hex Xeon would kill an i7 though not by much. But for what it's worth, I'll stick with i7.

So, I don't really know what do you mean by "i7 is no match for a Xeon", seriously? You maybe fed up by the brand? Xeon indeed sounds more cool and techy. Doesn't mean it's a lot better, though.

Brace yourself, blather about cache size, memory protection and ECC RAM is coming to try and make a Quad Xeon CPU that benches lower than a Quad i7 still be the better option when all people care about is how much CPU power they're getting over their current system :D
 
Last edited:
Mac Pro is definitely in line with workstation pricing (even cheaper than some premium brands like Boxx and HP). But it definitely can't beat the cost of DIY, even on the high end. That FutureLooks article was total nonsense. No one building a PC (even for high end work) would put in two of those FirePro cards which make up the majority of the cost in their "test". Two GTX Titans would run circles around those ATI cards and come in at a fraction of the price.

Those Titans are not officially supported and can be broken with the next .x Mavericks upgrade.

The Titans are incredible cards. The people that use them in oMP are enthusiast/professional users who have squeezed the oMP for every drop of performance.

I have read the threads, I follow the Mac Pro MR forum. Adding auxiliary power supplies and hacking them to boot with the main power supply, soldering larger ROMs on some PC cards to make them work with MacOS, hitting eBay to find deals on server processors...great jobs at tinkering for those that are interested in doing that type of stuff, but enthusiasts are a minuscule part of the buying public, even for workstation-level iron. Enthusiasts can take a 2009 Mac Pro and tinker up a very fast computer, but one that lacks the modern expansion of TB2 and will be farther behind the more time advances.

I would guess that Apple would at least like to sell the nMP in the tens, better yet hundreds of thousands. The ship has sailed on the $1,499 PowerMac, you are not going to see those prices anymore, barring the next paradigm shift in computing.
 
Seems fair!

One of the silly things about the "I can build it cheaper" is that it really shows those folks don't know what they are talking about. I don't own a Mac Pro. I don't want a Mac Pro (well, I do.. but...) I have a homebuilt PC as my desktop. BUT, my homebuilt PC is not a 'Mac Pro' killer by any means! It's a gaming PC.

Sure you can slap a 6 core CPU, lots of RAM and a nice GPU, but try and price it with a Xeon CPU, high end RAM, and FirePro workstation GPU's? Suddenly things change. And for what the Mac Pro market uses it for, those things are a must. Comparing a Mac Pro to a homebuilt gaming PC is like telling a contractor that his $60,000 diesel dually truck he uses to haul materials to job sites was a stupid purchase, and a Mercedes CLA would've been a much better way to spend that money because it's fast, sexy, and fun; and costs a bit less than his big truck! What they fail to realize is, the CLA won't do the WORK he needs done!

The point being is that DIY builders can build a pc that does task x as fast or faster. You do realise that not everyone needs xeons or workstation GPU. Also high end gaming gear/enthusiast hardware is excellent.

If you need workstation grade , Mac Pro is an excellent choice. Don't judge DIY pc as cheap and crap. High end gaming PC, running upto 4xgpu have serious power .

----------

How is a 3.7GHz Ivy Bridge 4-core i7 CPU faster than a 4-core 3.7GHz Ivy Bridge 4-core Xeon? I'm really interested to know how a 3.4Ghz 6-core is faster than a 3.5GHz 6-core on the same architecture too.





Apparently not that much, they are the same technology, specifications and price on a good few models.

http://ark.intel.com/compare/77781,75779,77780,75780

Overclocking

Every motherboard for DIY builds allows for overclocling. Is so simple these days.

Can't OC a Xeon
 
I would love to hear the reasons for your conclusion. Why do you think anything other than a Mac Pro is a better solution? I suppose it depends what your doing but Matlab and R both using GPUs now. The Mac Pro has great thermal characteristics. What's the issue? Just don't like Apple?

I find it odd that someone would make such a bold claim about something they also say they know nothing about.

After happily using Mac Pros for many years, I've just asked by employer for a Dell workstation (on which I'll run linux) rather than a new Mac Pro. I'm a mathematician whose research involves a lot of computation. I use Mathematica a bit, but much of my work makes use of heavily threaded code that I wrote myself. To me, cores are more important than GPUs. Here are some specs for my soon-to-arrive workstation. 16 physical cores (but no SSD, no OS X and just one video card) for about $5000.

SYSTEM COMPONENTS

Precisions Workstations T5610 Dell Precision T5610 CTO Base
Processor Dual Intel® Xeon® Processor E5-2650 v2 (Eight Core HT, 2.6GHz Turbo, 20 MB) 252650 [338-BCSD]
Memory 16GB (4x4GB) 1866MHz DDR3 ECC RDIMM 16GB4 [370-
Video Card 1 GB AMD FirePro V3900 (DP, DVI) (DP-DVI, DVI-VGA adapters) V3900 [490-BBRO]
Hard Drive 1TB 3.5inch Serial ATA (7.200 Rpm) Hard Drive 1T72 [400-AAWN]
2nd Hard Drive 1TB 3.5inch Serial ATA (7.200 Rpm) Hard Drive 1T72A [401-AACJ]
Optical Drive (Hard Ware only) 16X Half Height DVD +/- RW DRWHH [429-AAIW]
4th Hard Drive No Additional Hard Drive NOHDDA [401-AADF]
3rd Hard Drive No Additional Hard Drive NOHDDA [401-AADF]
Chassis Options Dell Precision T5610 685W TPM Chassis 685WTP [329-BBNL]
Hardware Support Services 4 Year Basic Hardware Service with 4 Year NBD Limited Onsite Service After Remote Diagnosis U4OS

TOTAL: $4,951.59
 
Not to mention an i7 is no match for a Xeon. They are not in the same league. And before you accuse me of being an Apple Fanboi, my day job is managing Linux servers so I know a thing or 2 about Xeon vs i7.

I don't doubt your experience but for day to day use the performance isn't that much different.

The difference is that the Xeons are designed for heavy lifting. They run cooler and are spec'd for 24/7 reliable continuous usage - such as encoding video and graphics, servers, scientific tasks, etc. ALL DAY LONG. In a typical desktop, there's be periods where the machine may do occasional intense work but a lot of the time only moderate work or may be idling much of the time.

Xeons are also, of course, able to be used in multi-socket motherboards, where i7s are not. Not that Apple gives this option now.

Some people will probably buy the MacPro as a sort of expensive headless iMac for general purpose desktop computing. A Xeon really is huge overkill in this situation.
 
They're not some Apple innovation.

Apple simply decided to adopt PCIe flash for all but their Mac Mini simply because OTHER COMPANIES started making it.

The PCIe flash in the Mac Pro is clearly Samsung PCIe flash that's been adapted to use a proprietary Apple connector.

How does, with what you've just said, have anything to do with the fact that it takes two raided SSd's to equal the speed of one of Apples SSD?


"I know, I'll take a PCIe RAID card with the silicon of 2 SATA 6Gb/s controllers and a caching scheme.

Add 2 SSDs with NAND flash memory and silicon for garbage collection, caching and the SATA 6Gb/s interface.

Then combine all the circuitry on 1 card to cut out interface bottlenecks and integrate the circuitry."

Perhaps I've missed something with the nMP breakdowns, but were are you getting your information that Apples raiding two SSD's to get these speeds?
 
Last edited:
Some people will probably buy the MacPro as a sort of expensive headless iMac for general purpose desktop computing. A Xeon really is huge overkill in this situation.

Huh? How is a Xeon overkill? i7's are available up to 4 cores. Do you really want the Mac Pro to be limited to 4 cores?
 
What is it that you (and others like you) get out of the whole finger-pointy, I-told-you-so, denouncing of so-called 'haters' that makes you want to fill threads up with it? What does it achieve that makes behaving so childishly acceptable?

Everyone has an opinion, sometimes in the fullness of time the opinions turn out to be wrong. Why dwell on it? Why take up space in threads with it? What does it achieve?

Why am I wasting my time finding out?

Because it irritates us?

And I like to kill ideas that irritate me. :eek:
 
Huh? How is a Xeon overkill? i7's are available up to 4 cores. Do you really want the Mac Pro to be limited to 4 cores?

Not saying that. I'm saying that a Xeon is of little benefit over an equivalent i7 for general desktop computing. But it costs lots more.
 
Quite right, most other workstations come with 3 years out of the box (not to mention on-site, NBD) compared to 1 year from Apple with 90 days on the phone.

Yeah, this too. Except I don't know if the warranties are equal.

Pricing is off.

Since the others include the warranty, the Apple price should reflect the added cost of getting Applecare. That's more of an Apples to Apples comparison.

I think Anandtech pricing included the price for AppleCare. But some of the other warranties may still be better in terms of what they offer.
 
How does, with what you've just said, have anything to do with the fact that it takes two raided SSd's to equal the speed of one of Apples SSD?




Perhaps I've missed something with the nMP breakdowns, but were are you getting your information that Apples raiding two SSD's to get these speeds?

You clearly fail absolutely to understand what PCIe flash actually is and think it's some kind of Apple magic. Try doing some research. A child could probably understand the breakdown of what PCIe flash consists of VS a PCIe card with dual SSDs and if that's still to difficult to comprehend, it's never going to sink in with you anyway.

I'll try again.

IN HARDWARE (get this? on a chip. SOLID STATE) is an arrangement of NAND chips.

IN HARDWARE (see above). is a controller chip that connects these together in an array of NAND, combining the throughput of several NAND chips to increase I/O speeds (just like RAID 0 does with HDDS/SSDs via multiple SATA channels).

IN HARDWARE (see above). It presents this as a single PCIe volume in a compact form-factor because it doesn't need the bulk of a PCIe connector, SATA connectors or 2.5" HDD enclosures because guess what? (and this is REALLY obvious) IT'S IN HARDWARE.

Is none of this sinking in at all?

Watch this:

http://www.youtube.com/watch?v=Lk4Ii_ZjBXE

Isn't learning fun!!!!
 
Last edited:
Overclocking

Every motherboard for DIY builds allows for overclocling. Is so simple these days.

Can't OC a Xeon

I agree that a self-builder can get much better performance, performance that is suited to their needs, for the price they might pay for a Mac Pro, but as far as an "professional" work is concerned the number of people who overclock is minute and arguing a self-built system overclocked is better value opens up so many holes in both sides of the argument it isn't worth pursuing.

I just get fed up of the "Apple should have used i7s" nonsense in every Mac Pro thread in the news section when since 2009 the Xeons used in single CPU Mac Pros have been the same or better than the i7s with no cost difference and have allowed people to use, first 8GB, and later 16GB DIMMs when those weren't available on i7s. Also ECC for those that wanted it. These ones probably allow 32GB and 64GB DIMMs if people wanted them, Mavericks at least supports 128GB.

----------

Not saying that. I'm saying that a Xeon is of little benefit over an equivalent i7 for general desktop computing. But it costs lots more.

Wrong.

http://ark.intel.com/compare/77781,75779,77780,75780

Xeon gives support for 16GB, 32GB and 64GB DIMMs as I mention above and ECC functionality.
 
How does, with what you've just said, have anything to do with the fact that it takes two raided SSd's to equal the speed of one of Apples SSD?

it doesn't

it takes 2 raided together SSD's on the SATA channel to hit those speeds due to SATA's 6gbps limitation

there are standard PCI-E slot SSD's with similar performance that can be used in any standard PC.
 
You clearly fail absolutely to understand what PCIe flash actually is and think it's some kind of Apple magic. Try doing some research. A child could probably understand the breakdown of what PCIe flash consists of VS a PCIe card with dual SSDs and if that's still to difficult to comprehend, it's never going to sink in with you anyway.

I'll try again.

IN HARDWARE (get this? on a chip. SOLID STATE) is an arrangement of NAND chips.

IN HARDWARE (see above). is a controller chip that connects these together in an array of NAND, combining the throughput of several NAND chips to increase I/O speeds (just like RAID 0 does with HDDS/SSDs via multiple SATA channels).

IN HARDWARE (see above). It presents this as a single PCIe volume in a compact form-factor because it doesn't need the bulk of a PCIe connector, SATA connectors or 2.5" HDD enclosures because guess what? (and this is REALLY obvious) IT'S IN HARDWARE.

Is none of this sinking in at all?

Watch this:

http://www.youtube.com/watch?v=Lk4Ii_ZjBXE

Isn't learning fun!!!!

As usual, you avoid answering my question. Instead you just post trivial information that has nothing to do with what I asked. So basically your just guessing without really knowing.
 
I'm a mathematician whose research involves a lot of computation. I use Mathematica a bit, but much of my work makes use of heavily threaded code that I wrote myself. To me, cores are more important than GPUs.
Have you looked at OpenCL? You might find that moving your code to the GPU could result in a bit of a speedbump..
 
You clearly fail absolutely to understand what PCIe flash actually is and think it's some kind of Apple magic. Try doing some research. A child could probably understand the breakdown of what PCIe flash consists of VS a PCIe card with dual SSDs and if that's still to difficult to comprehend, it's never going to sink in with you anyway.

I'll try again.

IN HARDWARE (get this? on a chip. SOLID STATE) is an arrangement of NAND chips.

IN HARDWARE (see above). is a controller chip that connects these together in an array of NAND, combining the throughput of several NAND chips to increase I/O speeds (just like RAID 0 does with HDDS/SSDs via multiple SATA channels).

IN HARDWARE (see above). It presents this as a single PCIe volume in a compact form-factor because it doesn't need the bulk of a PCIe connector, SATA connectors or 2.5" HDD enclosures because guess what? (and this is REALLY obvious) IT'S IN HARDWARE.

Is none of this sinking in at all?

Watch this:

http://www.youtube.com/watch?v=Lk4Ii_ZjBXE

Isn't learning fun!!!!

The person you quoted wasn't even saying it was an Apple innovation. He was offering a comparison. Every point you made had nothing to do with that.

You're kind of just being a tool to him.

Isn't learning fun?
 
Not saying that. I'm saying that a Xeon is of little benefit over an equivalent i7 for general desktop computing. But it costs lots more.

No? A similarly specced Xeon costs the same as an i7. Xeons are not more expensive, that's a common misconception. The base model nMP uses a 350$ Xeon, which is in the same ballpark as 3.7 Ghz i7's.
 
Huh? How is a Xeon overkill? i7's are available up to 4 cores. Do you really want the Mac Pro to be limited to 4 cores?

I've had my 6-core i7 chugging away at 4.2ghz quite happily for a couple of years :)
Personally I don't really see the benefit in xeon unless you need multiple cpus and have no other choice and ECC doesn't benefit me (nor do I think in my naive little world that it really has much of a place in anything that's not churning away in a rack for months on end). Not to say that some people might require this, but regarding the cpu itself - a 'cheap' consumer model will run day and night at full load.

Call me crazy but just to expand on this 'omg workstation stability' thing, for a personal machine I'd take a enthusiast board (not that you've really been able to buy anything else as a separate for years) over a low-end workstation machine any day of the week. Assuming it's quirk-less (again, most are nowadays and obviously you shouldn't expect anything less), they may have silly names, but they're all quality boards, thick tracers, solid state and covered with kilos of copper. I can't say the same for many workstation models.
 
As usual, you avoid answering my question. Instead you just post trivial information that has nothing to do with what I asked. So basically your just guessing without really knowing.

No. I gave an concise explaination of something you're failing to grasp, providing a link that explains the advantages of PCIe flash just incase it was still too difficult. You've now decided to back track so there's no point in continuing this discussion because you're incapable of reading for comprehension.
 
Hardware costs only. The big factor not included in OS X. That's where the true power comes from. So many people ignore that when determining the value of a system. Any Mac is far more valuable than any "comparable" PC.

In fact, OSX adds nothing in terms of "power"... you're just naive.

I like and prefer OSX, but repeating marketing nonsense doesn't make anything true.
 
The person you quoted wasn't even saying it was an Apple innovation. He was offering a comparison. Every point you made had nothing to do with that.

You're kind of just being a tool to him.

Isn't learning fun?

No, I think you'll find he specifically DID say the PCIe Flash in the Mac Pro ARE an Apple innovation and then failed to grasp what PCIe Flash actually is, regardless of manufacturer after several concise explainations from not just me.

He originally posted posted this.

Ok, so it takes two raided SSD's to get the equivalent of one in the 2013 Mac Pro. That tells you of how good the Apple PCIe SSD's they are currently using.

followed by the following explainations from other people which they also refused to acknowledge.

In sequential speed yes - the benchmark generally least relevant as such speeds will pretty much only be attained during copy operations. The CPU/GPUs won't be able to keep up with anything else.

The real benchmark lies in its random I/O performance. This is what make SSDs 'feel' fast. Here the results are not so convincing. The Mac Pro pulls out a 95mb/s read, but earlier benchmarks by Anand have shown the Samsung 840 Pro to beat it with 102mb/s random read.

Not only that, other PCI-E drives such as the Revodrive R4, or Intel 910 - absolutely annihilates it. In the same benchmark, the 910 does 592mb/s or 373mb/s depending on capacity and the R4 356mb/s. Thats 6x respectively 4x faster in a benchmark far more important.

(Which is wrong because the Mac Pro benchmarks are 950Mb/s not 95Mb/s)

it doesn't

it takes 2 raided together SSD's on the SATA channel to hit those speeds due to SATA's 6gbps limitation

there are standard PCI-E slot SSD's with similar performance that can be used in any standard PC.
 
Last edited:
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.