Trying to decide on Mac Pro setup, need advice please!

[righteye]

Quote "Ok, so if I understand you right I am better off with lower Ghz and getting the 12 core than I am with the higher Ghz and 6 core, right? that is what I am trying to figure out for rendering and program speed. thanks!"

If you are mainly doing Photoshop/Lightroom and other Photo software then the 6 core 3.33 Ghz is the sweet spot, 12 core is slower for this use (with lower or equal Ghz) but the 12 core is better for video and rendering ( this has been tested by Lloyd Chambers over at diglloyd.com
Something to look out for is an PCIe SSD from OWC which will have 2 esata ports built in so maximising your PCIe slot potential.
I personally doubt wether we will see a Pro with the same flexibility as the existing model, Apple policy of dumbing down will iam sure carry on to the Pro.
iam expecting a Mac Max, a big Mini, less drive options ,no optical drive, less PCIe slots, but do not worry it will have ThunderBolt so you can rush out and buy loads of peripherals to clutter everything up and spend your hard earnt dosh on.

 

[Anastasia72]

Don't get hung up on just x86 cores. Modern GPUs have computational cores also. It is easy to add 50-100 cores these days. The operative question is whether you will continue to invest in software over the next 2-5 years that ignores that capability.

You are fully free go down that path but if looking forward with statements the hardware has to last XX years, you should also be looking at where your software is going over the next XX years. Many folks get into a dubious position where the hardware is all about the future and the software is all about the past. Taking both positions at the same time doesn't make alot of sense when considering using a holistic system comprised of both.




Again this is a dubious position folks take where either something is totally configurable or totally integrated and not looking at the whole system.

Apple could add a embedded GPU to the Mac Pro and still offer four PCI-e slots. One of the x16 slots could possibly change to x8 but at the same time transition to PCI-e v3 doubling speed. There is zero net bandwidth loss in that context. This is really no different that integrating Wifi, Bluetooth, GbE, etc.

I'm sure there will be those who will wail about some "thing" they can't take out. That completely ignores what they can add.

Even if loose a slot ( doubtful if single and dual package systems will leverage the same core infrastructure motherboard ) there would still be at least one x16 slot available for upgrades. In that context, is it really worth throwing the baby out with the bathwater just because don't have 100% control over all the PCI-e lane assignments. The questions should be do you have enough. Not that it is maximized for some "maybe I'll later ..." notion of 'futureproof'. Nothing is 100% futureproof. Being in the future is the optimal position to solve future problems.

Thanks! I have really looked into all the specs of what the software I use and intend to use need and really the 6 core at the higher ghz is the way for me to go I believe. Your responses have helped me in my decision a lot. I personally tend to only upgrade my software every 2nd or 3rd release as I get used to working a certain way and hate change for the most part

Even when I decide to go to maya it seems the 12 core lower ghz vs the 6 core higher ghz did not make much of a speed difference in rendering other than a few seconds, which for me I can save the money and add a few seconds to rendering time.

I appreciate you taking the time to explain so much!

 

[snberk103]

....and really the 6 core at the higher ghz is the way for me to go I believe. ...

For the record, that is precisely what I did - and I love it. I use Lr, CS6, and Capture One. Sometimes all at once, with iTunes in the background. It just motors along.

Mine came with 1333MHz RAM as well which should help speed things up (vs 1066MHz also being offered). These don't come up often, but when they do I've seen them at a nice price. You just have to check often and then move fast. It seems to me you have the luxury of time since you still have a functioning system.

Good Luck.

 

[Anastasia72]

For the record, that is precisely what I did - and I love it. I use Lr, CS6, and Capture One. Sometimes all at once, with iTunes in the background. It just motors along.

Mine came with 1333MHz RAM as well which should help speed things up (vs 1066MHz also being offered). These don't come up often, but when they do I've seen them at a nice price. You just have to check often and then move fast. It seems to me you have the luxury of time since you still have a functioning system.

Good Luck.

Thanks for the tip. Question, how long did you actually dig and wait for this refurbished one? I see typical specs seem to keep coming up. Even the 6 core one seems to not hit refurbished often. I would really like to get to work and be able to work well. I started looking at the pro because my iMac is not cutting it and is really starting to tick me off in my work flow so much so that I hate to work lol

 

[snberk103]

Thanks for the tip. Question, how long did you actually dig and wait for this refurbished one? I see typical specs seem to keep coming up. Even the 6 core one seems to not hit refurbished often. I would really like to get to work and be able to work well. I started looking at the pro because my iMac is not cutting it and is really starting to tick me off in my work flow so much so that I hate to work lol

Weeks, I'm afraid. You just have to be diligent. But there could be one up tomorrow.... you never know. The good news, is that they seem to ship from California (if you are in the US or Canada) so the delivery is pretty quick.

 

[righteye]

I looked for many months before a 6 core became available (UK) although in that time i missed one chance ,my finger was hovering over the buy button (i was not sure about spending the money first time round) and someone else purchased it! next time i was quicker off the mark.
Very happy with the Pro.
As Apple can not sell the Pro in Europe soon that might mean a few more refurbs on your side of the pond

 

[deconstruct60]

Well, don't you think that just like Apple sells the HD 5870 they could do the same with other cards?!

No. A good limited set of cards covers 90+% of demand. Apple isn't out to be everything to everybody. That is rarely a good business to be in over a long period of time and isn't part of Apple's objectives.

Redundancy can add just as much, if not more complexity( and costs) , as it adds to increased market penetration. 3 cards all with +/- 5% performance characteristics doesn't really solve any new problems.

They're not even supposed to design the card, they'd just have to add EFI and drivers...

That makes no sense is not how the industry or the OS X market works. Apple doesn't do any of the low level drivers. That's AMD's and Nvidia's job. That is a far more sensible division of labor. It is their proprietary hardware so it should be their drivers.

There is a much larger graphics layer written on top of that: CoreAnimation, much of OpenGL , etc. that Apple certainly is only in (or at least primary) the position to provide. To my knowledge that is how it is split.

So new cards need new software from the vendors.

I'd would partially buy into the notion that Apple probably should put seed money into AMD and Nvidia to at least get experimental drivers written for each graphics generation. If only to have cards Apple can bring in and use for design "bake offs" when selecting a focus subset of cards for the default configs and BTO options.

That isn't necessarily products that make it to market but Apple should be trying out all the new options ( since pragmatically it is really only two vendors here. ). If Apple doesn't give one of the vendors any business for a whole generation that vendor might reassign their Mac driver team to other, more profitable projects. Apple should at least see to it that they get an experimental shot at the business each round. That keeps the vendor competition up witch will help both Apple and customers in the long run.

10 hugely overlapping cards isn't competition in a positive sense. More likely to generate either "race to the bottom" or "minute spec chasing" market dynamics.

I'm pretty sure that with their money they could afford the investment,

This isn't a contest of how to give away a couple billions from the cash horde. It is about return on investment. Every dollar spent has to generate a long term return.

R&D seed money is nothing like ordering up 30K cards to sell (or not depending on what customers buy).

It is much easier to get a solid return on just 2-3 cards and sell them into the entire 100-200k base of Mac Pros than to fiddle around with higher overhead and marginal to no net increase in sales.

and maybe they'd actually make a profit selling them to all the Mac Pro owners who are now forced to flash PC cards...

Frankly this is an "round off" error business. If folks were collectively making 10's of millions a year selling flashed PC cards for Macs then one of the 3rd party vendors would be doing it. They don't because it is small and risky (as I outlined in previous post).

Frankly most card vendors go with the 'do nothing' and get sales anyway approach. They still sell the cards and do no support or R&D work at all. Flapping your arms about Apple when they are engaged in this practice is just misdirection.

I don't know about you, but that's not the kind of support I'd expect for a professional machine.

I expect people to support what they sell and certify. Apple does.

You are talking about something else. Apple as providing a subsidy to promote the ecosystem. No I don't expect professional vendors to do that. It never happens in reality. All vendors do is jack up prices elsewhere and use that pay. Customers bare brunt of paying either way.

(perhaps some sort of BIOS simulator) they could make it possible to use a normal PC card...

BIOS is a bonehead backwards looking position to take. It is well past when it should have been retired. Can't even deal with modern top end HDDs anymore. Apple is well position to leverage the on-coming EFI based future. It is bewildering why it has taking the rest of the PC market so long to get here.

 

[Lancer]

I only waited for the new iMac because I didn't have enough money until recently and knew a new design was on it's way, otherwise I probably would have got the 2011 about a year ago and been perfectly fine with it.

But we don't know what is being introduced this year. Regardless, I wouldn't buy it in the first year no matter what it was because the one thing I am sure of is that it is radically different than the existing Mac Pro. And no matter how stupendous it is, it will have teething problems. Anyone want to bet against me? Getting a Mac Pro this month means that you only miss the first 2 and bit years of the new iStupendous - or 1 maybe 2 upgrade cycles - before your 3 years of AppleCare expires. Perfect timing in my books.

I say if you can afford it then get the current Pro, new or refurbished. Maybe to cut costs get a 3rd party LCD, there are load cheaper than the Apple ones and will work just fine. Then yes in 2-3 years when the need arises get the new Mac Pro.

I'm of the opinion unless you can wait and know there is something new coming very soon then buy when you need it, you can wait a life time for the next model which will always be faster/better/cheaper.

 

[Zwhaler]

what exactly are you running on a current iMac? Are you using 3d software and rendering out textures or using 2 intensive 3d programs at once? I am just curious of why you would go from a Mac pro to iMac? I find my iMac's go down so fast. I realistically have not been able to work efficiently on my iMac for the last year, which would really make an iMac for me last about 2 years, not a long life for the price in my opinion. I make mine go to 3 years, but not working well.

If only using it for small tasks then yeah I would go with an iMac again. My biggest issues have been the graphics card going extinct, the ram, the cores. Forget memory I tend to use external memory a lot anyway.

----------



Do you think? See this is what I am on the fence about. There are assumptions with Apple trying to make everything smaller that they just might screw up the mac pro too, which I am sure you have read, then all of us needing a pro will be scrambling to get the one out now lol

Look, I own the 12 core 2.93GHz with 24GB of RAM (8 from Apple), HD5870, 2+3+4+4TB HDDs and 2 Super Drives. The machine is an effing beast. You can run a commercial production company on this mac pro EASY. In fact, many big companies do, and it shows why. BUT, I got it in June 2011 making making the opportunity cost much lower to own it.

Buying it now would cost 10 grand when in possibly 6 months something new will come out making (especially a 4-5k purchase) much faster. It might be out in less than 6 months, in which case it would definitely be worth waiting, trust me. If it is longer than 6 months you have to ask yourself if the opportunity cost is too high to wait. It would have been for me !!!!

What I'm saying is that the 6 core is good, but the 12 core is really rocking (2010 gen) basically the new Mac Pros will likely come in 4 core, 8 core, 12 core, and 16 core arrangements. That means you will probably end up with an 8 core for the same price you're paying today for a 6 core.

Finally: IF the software you use won't benefit much from a change from 3.33GHz 6 core to say more 8 or more cores then you will be FINE buying a new 6 core today. Then don't get mad if a new one comes out in 3 months, learn how to use the full power of that machine which is A LOT

 

[Anastasia72]

Look, I own the 12 core 2.93GHz with 24GB of RAM (8 from Apple), HD5870, 2+3+4+4TB HDDs and 2 Super Drives. The machine is an effing beast. You can run a commercial production company on this mac pro EASY. In fact, many big companies do, and it shows why. BUT, I got it in June 2011 making making the opportunity cost much lower to own it.

What I'm saying is that the 6 core is good, but the 12 core is really rocking (2010 gen) basically the new Mac Pros will likely come in 4 core, 8 core, 12 core, and 16 core arrangements. That means you will probably end up with an 8 core for the same price you're paying today for a 6 core.

ahhhh you are killing me here lol, I was almost ready to pull the trigger on the 6 core!

 

[thekev]

I agree with everything you have replied here above. I see lots of people moving to plenty of Nvidia cards in this forum. I see plenty using different kinds of RAM, I just do not see that MANY limitations, unless I am missing something.

PC's even have limitations on upgrades depending on what type of machine you build.

My biggest issues were those that I listed in my original post, most concerning specs for my use.

At this point ANY Mac pro is better than my iMac 2009 obviously, just wanted to make sure I can make it last 6 yrs or so.

The problem with that theory is I'm not sure it will be pleasant for 6 years. Also if you're working on your own and budge constrained, why the desire to use maya? They killed their cheaper education upgrade path a while ago. If you go that route it's over $4k + annual maintenance after that. Viewports are OpenGL dependent, so yeah heavy scenes hammer the gpu. I wouldn't be surprised if some of the solvers are eventually written to leverage OpenCL once it matures further. Zbrush so far makes no use of OpenGL. Photoshop isn't noticeably faster in actual use regardless of what gpu you use. In my opinion if it takes 6 years to justify the hardware purchase, you're spending too much. Right now the hardware came out in 2009 to 2010. The quad model still uses a 2009 cpu. 6 years from now will be 2019. I suspect that is quite a stretch unless computing demands remain incredibly flat.

 

[Anastasia72]

The problem with that theory is I'm not sure it will be pleasant for 6 years. Also if you're working on your own and budge constrained, why the desire to use maya? They killed their cheaper education upgrade path a while ago. If you go that route it's over $4k + annual maintenance after that. Viewports are OpenGL dependent, so yeah heavy scenes hammer the gpu. I wouldn't be surprised if some of the solvers are eventually written to leverage OpenCL once it matures further. Zbrush so far makes no use of OpenGL. Photoshop isn't noticeably faster in actual use regardless of what gpu you use. In my opinion if it takes 6 years to justify the hardware purchase, you're spending too much. Right now the hardware came out in 2009 to 2010. The quad model still uses a 2009 cpu. 6 years from now will be 2019. I suspect that is quite a stretch unless computing demands remain incredibly flat.

It's not about justification, it is more the idea that an iMac may last a super long time, I have an iMac from 2006 still working great, but ram, gpu, etc forget it, it is dead in that area. I also have my 2009 iMac working perfectly well, but again gpu, ram limitations, duo core, forget it, it is junk. Both of these machines are good for nothing, but surfing the net and small work basically. Furthermore, neither of these were great for 3d work to begin with.

 

[deconstruct60]

It's not about justification, it is more the idea that an iMac may last a super long time, .... Furthermore, neither of these were great for 3d work to begin with.

If your workload is flat then an iMac with each iteration will get better. The 2012 models are better. The 2015 models will be better still. A 2012 + 2015 can cost you about as much as a Mac Pro.

Likewise if your computational workload is increasing 30% per year and your software is going to ignore GPGPUs , then a Mac Pro 2012 isn't going to last 6 years. It will barely last 3-4.

Over time an iMac will become a better hobby 3d machine. Especially with a time horizon of 6 years into the future.

people who have solid plans to stick with a machine for 6 years are not planning to grow much. That is a plan for a relatively fixed in stone workload. The trap he is highlighting that folks use that fixed workload to buy stuff way out of their budget by just extending the service lifetime of the machine. Can't afford a $4,200 machine.... just tad a year or two more onto the service time time. So instead of 4,200/4 it is not 4,200/6 or 4,200/10. Oh yeah 420/year is much more affordable. For a house, relatively fixed in stone functionality this is fine. For technology grounded in transistors doubling every 18-24 months and software with major updates on a similar or shorter basis it is not so good an idea. Computers are more a consumable capital expensive than a very long duration durable one.

 

[Derpage]

Six core bro.
You have no need for 12 cores.
Save the money and buy yourself some booze to make the sticker shock go away when you realize how much you are spending for some old tech.

 

[righteye]

I will certainly not be an early adopter for the new Mac Pro or MacMax, mini's big brother(just my guess at how the new Pro will turn out) i will wait till all the glitches have been ironed out and live with the one i got for a while longer, the MacMax may not even be worthwhile, fingers crossed on that one!

 

[righteye]

Quote:
Originally Posted by Zwhaler
" Look, I own the 12 core 2.93GHz with 24GB of RAM (8 from Apple), HD5870, 2+3+4+4TB HDDs and 2 Super Drives. The machine is an effing beast. You can run a commercial production company on this mac pro EASY. In fact, many big companies do, and it shows why. BUT, I got it in June 2011 making making the opportunity cost much lower to own it.
What I'm saying is that the 6 core is good, but the 12 core is really rocking (2010 gen) basically the new Mac Pros will likely come in 4 core, 8 core, 12 core, and 16 core arrangements. That means you will probably end up with an 8 core for the same price you're paying today for a 6 core.
ahhhh you are killing me here lol, I was almost ready to pull the trigger on the 6 core! "


For certain tasks only!! it depends on what you are doing, For Photography use PS,LR etc the 6 core is the sweet spot and its FASTER.
Go to diglloyd.com if you are a doubter (his blog is free but the rest of the site is subscription but it will provide you with good buying advice and much more)
I have no connection with diglloyd.com i just appreciate his work to provide sound Mac Performance advice.
Also; Why are you mixing RAM from different manufacturers ? not recommended nor different HD sizes (or have i misunderstood your post ?)

 

[deconstruct60]

but the 12 core is really rocking (2010 gen) basically the new Mac Pros will likely come in 4 core, 8 core, 12 core, and 16 core arrangements. That means you will probably end up with an 8 core for the same price you're paying today for a 6 core.

Nothing indicates that so far. Sandy Bridge Xeon E5 1600 series capped out at 6 cores even though the E5 2600 capped out at 8. It is shrinking but there is still a chance Apple will roll out with currently available E5s sooner rather than later.

It is likely Intel will hold the core cap on the 1600's with 1600 v2 ( Ivy Bridge) versions also and use the process shrink primarily to boost clock rates and perhaps lower TDP slightly.

The reason why the 6 core version is being named as a sweet spot is primarily because the balance between core count and clock speed is largely a generous helping of both. If increase to the core count to 8 the base clock speed is going to go down (if keep the TDP the same).

It is more useful for folks with about an even mix of legacy single threaded stuff and new explicitly parallel workloads. That is still going to be true with the transition to v2 versions.

What will likely be a bit different though in v2 is that the dual package 12 core entry model will be clocked higher and may have a broader turbo range. That will negate somewhat the size of the "sweet spot" the 6 core version now sits on with Westmere and Sandy Bridge models.

Longer term Intel is probably positioning the Xeon E5 1600 series to pick up a integrated GPU or more pure GPGPU. At some point the additional transistors at a shrink will get assigned to providing that. That will allow much more economical use of the PCI-e lanes being provided. (either that or they will need to bump up the lane count by about x8 ).

P.S. going from a 32nm to 22nm process is a -31% decrease. Going from 6 to 8 is 33% increase. it isn't an even trade-off if trying to stick to the same sized dies, prices , and profit margins. It would take another shrink to get some "free room" for cores when cores come with 2.5MB of L3 cache attached. That is huge chunk of transistors that tend to soak up lots of die space.

 

[Anastasia72]

The reason why the 6 core version is being named as a sweet spot is primarily because the balance between core count and clock speed is largely a generous helping of both. If increase to the core count to 8 the base clock speed is going to go down (if keep the TDP the same).

What will likely be a bit different though in v2 is that the dual package 12 core entry model will be clocked higher and may have a broader turbo range. That will negate somewhat the size of the "sweet spot" the 6 core version now sits on with Westmere and Sandy Bridge models.

Well, this just made up my mind. I am going with the 6 core now. I really do not see the point of a 12 core highest ghz for what I am doing, so no reason for me to wait if I am going to stick to 6 core anyway. I mean what is the worst thing that can happen? I get a 6 core and see that they upgrade the 6 core slightly? In this case does not seem like a big deal to me if they do.

 

[righteye]

Quote:
Originally Posted by righteye
but the 12 core is really rocking (2010 gen) basically the new Mac Pros will likely come in 4 core, 8 core, 12 core, and 16 core arrangements. That means you will probably end up with an 8 core for the same price you're paying today for a 6 core.

This not my quote my quote starts at the paragraph below not a big deal i just wanted to put the record straight.

 

[deconstruct60]

Quote:.....

This not my quote my quote starts at the paragraph below not a big deal i just wanted to put the record straight.

You're right. Sorry about that.

The relative performance in the diglloyd.com blog isn't going to hold up as well when it comes to Xeon E5. With the E5 1600 series Intel finally uncorked the 4 core entry model. The base clocks go:

1620 3.6GHz ( 4 cores max Turbo 3.8GHz )
1650 3.2GHz ( 6 cores max Turbo 3.8GHz)
1660 3.3GHz ( 6 cores max Turbo 3.9Ghz)
http://ark.intel.com/products/series/63197

When it comes to drag racing on single threaded apps there isn't going to be a big difference. The high priced model does have a wider dynamic range.

Just saying so that folks don't take "6 cores" as some generic rule of thumb. It is just what happens to be true with Nehalem vs. Westmere models which is quirk more so than a general rule of thumb.

 

[righteye]

You're right. Sorry about that.

The relative performance in the diglloyd.com blog isn't going to hold up as well when it comes to Xeon E5. With the E5 1600 series Intel finally uncorked the 4 core entry model. The base clocks go:

1620 3.6GHz ( 4 cores max Turbo 3.8GHz )
1650 3.2GHz ( 6 cores max Turbo 3.8GHz)
1660 3.3GHz ( 6 cores max Turbo 3.9Ghz)
http://ark.intel.com/products/series/63197

When it comes to drag racing on single threaded apps there isn't going to be a big difference. The high priced model does have a wider dynamic range.

Just saying so that folks don't take "6 cores" as some generic rule of thumb. It is just what happens to be true with Nehalem vs. Westmere models which is quirk more so than a general rule of thumb.

I have to admit i was only thinking 6 vs 12 core and forgot about 4 core, glad you mentioned it

 

[flat five]

I'm looking at the quad core and the 12 core and trying to figure out a few things first before I take the plunge. I currently use Zbrush, PS CS4 extended (want to go to CS6 extended), Blender (eventually to move to maya), and Topogun.

if you use blender/maya/other 3Dmodeling apps a lot then go with the fastest clock speed as opposed to #of cores..

those types of apps require linear processing so if you have 12 cores, 11 will sit idle when modeling..

the flipside is that rendering loves multicore.. the more the merrier(faster)

but for me personally, i'd rather have high performance modeling environment while i'm physically at the computer.. then the renders can be slower while i sleep or whatever..

dunno, im getting to that point of needing a new computer but i'm going to hold out for a while.. if i had to buy now though, that 3.33 6core is looking real nice..

 

[flat five]

I want a machine that will take me through the next 5 yrs or so and will allow me to switch out and upgrade parts as needed. I can no longer see the value in an iMac that lasts me just 3 yrs, which both of my iMac's have done.

about this part.. just my own experience with mac pros..
i have two of them.. both 1,1.. i bought one in 2006 then the other refurbed the next year..

and they're great.. to this very day.. almost 7 years later..

i've had a couple of hard drives go bad then 3 video cards (both original 7300s and an 8800) but besides that.. awesome. very dependable

but they've finally lost OS support and i (maybe wrongfully so) see that as the beginning of the end for them..


i've looked fairly seriously at the imacs and they seem perfectly fine for me but i just don't know if i can expect 6+ years out of them.. and for whatever reason, you saying you lost two after 3 years just seems like something i would expect to happen to me if i go that route..

 

[Macsonic]

about this part.. just my own experience with mac pros..
i have two of them.. both 1,1.. i bought one in 2006 then the other refurbed the next year..

and they're great.. to this very day.. almost 7 years later..

i've had a couple of hard drives go bad then 3 video cards (both original 7300s and an 8800) but besides that.. awesome. very dependable

but they've finally lost OS support and i (maybe wrongfully so) see that as the beginning of the end for them..

My friend is also using 2006 Mac Pro. He told me he is fine with Snow Leopard and not in a hurry to have upgrade and have Mountain Lion. The new features in ML was not needed with his work for now. This is his opinion though.

 

[deconstruct60]

if you use blender/maya/other 3Dmodeling apps a lot then go with the fastest clock speed as opposed to #of cores..

those types of apps require linear processing so if you have 12 cores, 11 will sit idle when modeling..

What about 3D modeling is inherently linear? Manipulating/translating points is a likely a matrix transform. (e.g., http://en.wikipedia.org/wiki/Transformation_matrix#Examples_in_3D_graphics ) Maxtrix transforms can be parallelized. There is a difference between legacy assumptions built into the code and a requirement for serial, linear processing.