2.66 Quad vs 2.93 Quad Buyer's Guide (Includes Octos)

[jnc]

You can always work on HDD throughput with some sort of RAID implementation at a later time. It would give more of an overall performance boost than the higher clocked processors anyway, unless your applications are CPU dependent only.

Must_feed_the_CPU's.

I'm not too clued up on RAID yet, but maybe you can help. I remember not being to impressed with the access speed of the MacBook Pro 7200 drive. SSD on the Air impressed me when it came to startup and the like. `i briefly owned a Mac Pro last year, and I'm sure the desktop components are different as I don't recall any complaints, but nevertheless... SSD intrigues me.

Edit: Just noticed the price of the Mac Pro's RAID card - yikes - and that's before drives, how would RAID compare to, say, purchasing an SSD boot drive? With the stock HD as a secondary?

Thanks for any help you can give!

 

[nanofrog]

I'm not too clued up on RAID yet, but maybe you can help. I remember not being to impressed with the access speed of the MacBook Pro 7200 drive. SSD on the Air impressed me when it came to startup and the like.

Might it be a consideration to buy two 640GB internals rather than a sole 1TB to set RAID up, in addition to an SSD for boot? Is that possible/sensible?

It's quite possible.

You'd see even better performance by getting additional drives into the stripe (RAID 0), and place everything on it. OS, apps, and data. That's the up side.

The down side is that the failure rate of a single drive is multiplied by the number of drives in the stripe. So a 4 drive stripe is 4x more likely to fail in the same period of time. If this happens, ALL DATA IS LOST. Replace the drive(s), and restore the data from backups. So backup is absolutely essential, not an option.

If you choose this route, be sure you can spare the time necessary to rebuild the array manually, and re-perform any work that was lost, as there's always the possibility the backup is older than what you lost, even if it's only a couple of files. Those files may or may not be important, and if so, can potentially cost you hours or more of time to re-do the work. Bad for meeting deadlines if you're generating an income from it, or need to get assignments in to professors on time.

If you need redundancy, you'd want to look at other array types. OS X is capable of 0/1/01/10. Of these, 01 and 10 offer redundancy, but 10 is the better of the two. Check out Wiki's RAID page, and follow the links for more information. Then there's hardware solutions, which can offer other arrays, such as 5/6/50/60.

Keep in mind, even if the array offers redundancy, you still need a backup solution regardless, not just with a stripe. Data can be lost, and I've seen it happen with different array types. I've done it intentionally in fact, as I test an array before trusting data to it (drives and controller behavior, power outage,...).

 

[jnc]

It's quite possible.

You'd see even better performance by getting additional drives into the stripe (RAID 0), and place everything on it. OS, apps, and data. That's the up side.

The down side is that the failure rate of a single drive is multiplied by the number of drives in the stripe. So a 4 drive stripe is 4x more likely to fail in the same period of time. If this happens, ALL DATA IS LOST. Replace the drive(s), and restore the data from backups. So backup is absolutely essential, not an option.

Thanks. I was reading up on RAID and noticing that bit put me off a bit! Not only would i need the setup (hardware+drives), I'd essentially need a clone of it should things go bad. So even in a setup where I had, say, just 2 x 640 drives plus a 128 or 256 SSD, I'd need 1.5TB of additional storage for all of it to be backed up, eek... knowing me it'd turn into a mess.

What I'm thinking instead is perhaps a 256 SSD boot, backing up internally onto the stock 640 HD, all of which being backed up externally via Time Capsule or other method. If I can do such a thing. Should work out a lot less, after dropping £4000 on the Mac itself I want to be a bit less frugal beyond that...

 

[nanofrog]

Thanks. I was reading up on RAID and noticing that bit put me off a bit! Not only would i need the setup (hardware+drives), I'd essentially need a clone of it should things go bad. So even in a setup where I had, say, just 2 x 640 drives plus a 128 or 256 SSD, I'd need 1.5TB of additional storage for all of it to be backed up, eek... knowing me it'd turn into a mess.

What I'm thinking instead is perhaps a 256 SSD boot, backing up internally onto the stock 640 HD, all of which being backed up externally via Time Capsule or other method. If I can do such a thing. Should work out a lot less, after dropping £4000 on the Mac itself I want to be a bit less frugal beyond that...

I'm not quite understanding the reasoning behind a second backup. Are you still thinking of using a stripe? Or just doubling up backups on the SSD?

 

[jnc]

After reading this thread again...

* 2.66 quad plus 20% more cash = 2.93 quad = kind of some more speed, you may not notice
* 2.66 quad plus 14% more cash = 2.66 quad / Intel X25-M 80GB SSD = quite a lot of additional fun every day

Agree?!

I'm going to save myself a chunk of change further still by going Quad 2.66, not Octo With stock GPU, even.... for now. CTO Ram though, AppleCare'd plus no pointless 1GB sticks laying around then

I'm a light gamer, and an Adobe CS3 user... doesn't seem like I'd TRULY appreciate anything faster. Another way of saying I'd be paying too much before I'd notice any difference.

Hopefully Snow Leopard makes the machine perform even better. then I've still got room for an SSD boot and more recent GPU.

Plus now, I get to put a payment down on a car too

I'm not quite understanding the reasoning behind a second backup. Are you still thinking of using a stripe? Or just doubling up backups on the SSD?

Just figured I'd back it all up for added security, y'know - good to have a portable, external instance of all your data in a "my Mac Pro just spontaneously combusted" or "my house is on fire" scenario.

And here we have it:

£2,232.99 (formerly £4153!! nearly half the price of originally intended specs!)
One 2.66GHz Quad-Core Intel Xeon
6GB (3x2GB)
640GB 7200-rpm Serial ATA 3Gb/s
ATI Radeon HD 4870 512MB
One 18x SuperDrive
Apple Wireless Mighty Mouse
Apple Keyboard with Numeric Keypad (British) and User's Guide (English)
AirPort Extreme Wi-Fi Card with 802.11n

Future upgrades:
Snow Leopard
CS4/5
SSD boot drive
Replacing Dell 2709 and Cintiq 12WX with Cintiq 21UX + HDTV.

 

[Tesselator]

Final Suggestions:

As others have stated, the 2.66 is the sweet spot regardless if you are going Quad or Octo. The 2.93 on either end is a waste of money.

Simply, the difference between the 2.66 Quad and 2.93 Quad is imperceptible. If you ask, "But sometimes the 2.66 struggles at some tasks?" Well, if the 2.66 struggles then the 2.93 will struggle the same. The imperceptable speed bump is not going to solve it. Most of the time you can tell when a machine feels "peppier" or "snappier" than another. Not with the Quads. I have never tried the Octos first hand so I cannot say this with certainty, but looking at the scores it appears the story is unchanged.

Yeah, well I don't agree that the 2.93 "will struggle the same" but I guess it's pretty close. Anyway, yes it's clocks than pertain to "pep" and cores are throughput.

So like if you were editing 1 billion polygon 3D CAD objects the 2.93 would be the peppiest! At that point you could probably feel the difference between the 2.66 and the 2.93 quads.

Now if you were rendering those same objects over video at 1080p then that's the throughput or "bandwidth" where cores matter. The 2.93 quad let's say takes 3 minutes per frame then the 2.66 octad will take 1.8 minutes and the 2.93 octad will take 1.5 minutes.

If you're paycheck is determined even partially by these times then the 2.93 octad makes sense. If your paycheck is determined by the speed and/or number of those CAD objects you can produce then the 2.93 quad does indeed make sense. There really is an actual difference. It's just that you're not going to feel it if your system is improperly tuned (slow RAM, HDDs, etc.) or if you're not pushing the system to those levels - like you mostly work on CAD objects of 50K polys or less, etc.

 

[Flash SWT]

Quad vs. Octo

This is the thread that has my wife yelling at me to quit reading any more. I pulled the trigger on a 2.26 Octo last night, really wanted the 2.66 Octo but that $1,288 will go a long way toward stuff we need done at the house.

The developer of the main photo application I use says the Octo "should absolutely smoke the 4-core system for loading images. It will be slower in single threaded operations like [examples] since the clock rate is slower than the 4-core system."

So I'm torn, keep the 2.26 or cancel and get a Quad. Parts of my main application will benefit from 8 cores and parts would benefit from a higher clock rate. (Damn it, I've become one of those people I hate, talking about canceling something I've already ordered.)

 

[Mac Husky]

Damn it, I've become one of those people I hate, (...)

An experience more of us are going to have sometimes in life

 

[Gonk42]

Octo vs Quad Multi-tasking rather than multi-threading

All the bench mark comparisons I've seen on octos and quads (eg barefeats)
still seem to be based on running a single application (albeit a multi-threaded
one).

What I want to do is run multiple applications. For this purspose, provided
I'm careful about say using different drives to avoid bottlenecks I think
the octo should be pretty close to twice the speed of a same clocked
quad. Of course, given the pricing it is probably almost as cheap to
buy two quads instead!

It would be interesting to see a comparison between quad and octo
running a batch job of say 8 or even 16 programs at once.

(What I do is run multiple instances of a program, each on a different set of data. The last big run I did on a two core Linux machine took two weeks!)

Sadly, I'm a student so arguments about time is money don't apply in the same way - letting the family starve whilst I buy a 2.93 octo is not really an option

 

[tome viewer]

Your posts would be more valueable if you were in tune with the needs of your audience and not just pushing your own values on everyone in 18pt text.

To make a statement that these machines are a waste really just proves you don't understand the full extent of the audience in these forums.

So like if you were editing 1 billion polygon 3D CAD objects the 2.93 would be the peppiest! At that point you could probably feel the difference between the 2.66 and the 2.93 quads.

Now if you were rendering those same objects over video at 1080p then that's the throughput or "bandwidth" where cores matter. The 2.93 quad let's say takes 3 minutes per frame then the 2.66 octad will take 1.8 minutes and the 2.93 octad will take 1.5 minutes.

If you're paycheck is determined even partially by these times then the 2.93 octad makes sense. If your paycheck is determined by the speed and/or number of those CAD objects you can produce then the 2.93 quad does indeed make sense. There really is an actual difference. It's just that you're not going to feel it if your system is improperly tuned (slow RAM, HDDs, etc.) or if you're not pushing the system to those levels - like you mostly work on CAD objects of 50K polys or less, etc.

So on a job that takes 30 seconds to complete you will shave off 1.5 seconds by going with the 2.93 at a premium of $500 plus tax. I could see how this would be worth it in some rare cases where every second is worth another $100 to you. But for how many is this true? I would say less than 0.01% of us here.

I am merely trying to share information with the average person in the hope that it might help them make a wise decision. Especially people that could use the money for other things, like family, etc. Their priorities are elsewhere, and rightly so. Our economy dictates that people should be prudent and not spend frivolously. I feel this is the majority. And yet, having lurked here for awhile, who are the ones giving advice to these professionals? Looking at the information they discuss, it appears to be topics much like a tweaking site for Windows. Where they are trying to squeeze the very last GHz out of their boxes. They get off on tweaking their systems, and brag and boast about how much faster their systems are after the tweak. Are these the professionals that you are speaking of? The one's that make $100 per second for each second that they shave off of their time? That mentality is typical of gamers. Running cross fired 4070's, SSD's in a RAID setup, etc. They say they are not all about performance, and cite one or two examples where they have recommended the MacMini instead. But those who take their advice would probably make better use of the MP than they would. Seeing that they are ultimately the individuals who are using their purchases for work and school. It is a means to an end for them, not an end in itself, as it is for the gamer or tweaker.

Ultimately the tweaker is giving the advice and the professional is listening. And if you ever point out that the one who tweaks his machine doesn't know what he's talking about, that he is wrong to tell others the 2.8 will thoroughly smoke the 2.26, or that the 2.93 he owns is a waste of money, watch out. They will now cite one or two examples where these machines are worth every penny. Why? Because they are pissed off that their machines are made to look like something disgusting. And yet the argument they are using to increase the value of their machines back to the profound, godly status that it deserves, does not apply to them. Every second saved is not worth $100 to them as a professional, but only to them as a gamer or tweaker, and because they are one, they would pay anything for it.

Now I am being told that I do not understand the “audience” here, that the audience is really made up of professionals who would pay any premium for the fastest machine possible. I thought that was the gamer, not the majority?

Honestly, I am tired of this. This will be my last post.

 

[JediMac]

I hope not, because you changed my mind.

 

[jnc]

Well, thanks, it was helpful to me. I about halved my CTO price when I considered the testimony here.

 

[AppleWorking]

yeah, poor guy. lots of useful insight. like watching myth busters unfold. i had my popcorn out. i think they gave him the good ol "we're professionals" line to justify purchasing the high end models after they felt dumb for wasting money. should have just said we're gamers, so we do pay more for the little we get. i think his advice is good for the gamer and pro too, gives us more money to play with. more money = more fun

 

[Tesselator]

I am merely trying to share information with the average person in the hope that it might help them make a wise decision.

What??? Someone actually reads threads here other than their own? Hmm, you wouldn't know it by looking at the thread titles and seeing the same questions posted over and over and over and...

EDIT:
BTW, imho, spending prudently would automatically disqualify ALL of the 2009 Mac Pro models! ALL OF THEM! They're all WAY overpriced for an almost unperceivable speed bump over the 2008 or even the 2006 models. In one of your blind comparisons I bet that the user also would NOT be able to tell if they were using a 2006 machine vs. a 2009 machine given similar clock rates.

EDIT: EDIT:
Also I think the "audience" here is not "really made up of professionals who would pay any premium for the fastest machine possible" I think they are mostly Apple fanatics that will pay a premium for whatever Apple tells them is "good" for what they do. Most of them anyway. You can tell the ones who actually think for themselves as they ask questions instead of announcing their newest purchase.


.

 

[AppleWorking]

What??? Someone actually reads threads here other than their own? Hmm, you wouldn't know it by looking at the thread titles and seeing the same questions posted over and over and over and...

Yeah, and with your crummy advice over and over and over...

 

[Tesselator]

Crummy??? What, I should add more eggs?

 

[Flash SWT]

BTW, imho, spending prudently would automatically disqualify ALL of the 2009 Mac Pro models! ALL OF THEM! They're all WAY overpriced for an almost unperceivable speed bump over the 2008 or even the 2006 models. In one of your blind comparisons I bet that the user also would NOT be able to tell if they were using a 2006 machine vs. a 2009 machine given similar clock rates.

Then again, how many folks buying the 2009 Mac Pros are like me? I bought mine to replace a Windows XP machine, running an AMD Athlon XP 1900+ (1.6 GHz) that I built off Newegg a little over 7 years ago. I'm not new to Macs, been using them since collage and at work, but I've keep this old PC going as an email/web/gaming machine for years because there was no real reason to upgrade.

I'm finally tired of dealing with a custom built rig and Windows day-to-day (plus it is starting to feel slow with newer things like HD video online). I reached the point that my free time is worth more than the money I'd save building a box myself. I could obviously get a similarly powered machine for much cheaper by building a Hackintosh, but I'd rather spend the time with my family or fixing up the house. (Damn mom would be proud of how mature I sound.)

I can see that for someone who already has a Mac Pro, upgrading to the 2009 model might not make much financial sense. But for many of us who don't already have a Mac Pro, buying "old" or "outdated" tech doesn't make very much sense either.

Just throwing a different viewpoint out there!

Oh! Just got my shipment notification while in the middle of typing this!

 

[Neliskens]

Forgive me if this is a dumb questions, but isn't it in the future not an option to (with all this fuss about the limitation of the 4 memory slots of the 2.66 Quad), just get bigger memory cards (say 4 GB ones) and install 4 of these to get more then the 8 GB limit now?

 

[Coyote2006]

I got my 2.66 quad/4870 one week ago and I can say that it's so extremely fast that a 2.93 upgrade is not worth spending another $600.

 

[Neliskens]

What I am missing in this thread is discussions with types of workflow. Will the quad be more suited for gamers or video editers etc. I am a video editer and work on Final Cut Studio 2.

My workflow is mostly this:
I edit a project in FCP 6 and run it through compressor, while compressor is running I make the DVD-menu in DVD Studio Pro and make the DVD-cover in Photoshop.

What about this workflow, will I benefit from a 8-core but with a slower 2.26GHz?

 

[Mac Husky]

Forgive me if this is a dumb questions, but isn't it in the future not an option to (with all this fuss about the limitation of the 4 memory slots of the 2.66 Quad), just get bigger memory cards (say 4 GB ones) and install 4 of these to get more then the 8 GB limit now?

Good morning

 

[Neliskens]

haha told you it probably is a stupid question, but I guess the obvious answer is yes. What about the other question about the workflow and which version to buy. Any thoughts on that?

 

[Mac Husky]

haha told you it probably is a stupid question, but I guess the obvious answer is yes.

The answer is YES - as it has been posted a hundred times during the last weeks
And of course not only for the future. If you have the money, you may get the 4GB sticks right now.

Look here and here.

What about the other question about the workflow and which version to buy. Any thoughts on that?

I would say, that this has been discussed as much (or even more) than the 4GB RAM stick
You will find a lot of threads in this forum and a lot of critical benchmarks discussions elsewhere, too.
You don´t even have to use the search funtion. Just scroll down the threads a little bit.

It is not exactly your workflow, you will find, but a lot of comparable applications discussed.

Have a deeper look in here.
Especially this.

 

[Tesselator]

Then again, how many folks buying the 2009 Mac Pros are like me? I bought mine to replace a Windows XP machine...

Just throwing a different viewpoint out there!

Oh! Just got my shipment notification while in the middle of typing this!

Yeah, conversation is all guud! The POV I was interjecting to use the familiar car metaphor yet again, is:

You can buy a 2009 Porsche with mag rims and pay $1.5 million or you can buy an otherwise identical (new, but last years model) 2008 Porsche with aluminum rims and pay only $1 million dollars.

To me the 2009 is just not worth the extra money whether I currently own the 2006 model or not. YMMV - tho I don't see how it could.

 

[Neliskens]

Thanks a lot Mac Husky for that! I've been searching on a lot of forums for that information and read this whole thread two times all ready.

I am still in the learning phase with this computer technology stuff and I don't want to make a mistake now, so I am a bit paranoid

But I appreciate your answer buddy!