Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Trade In early 2008 for new Nehalem?
- Thread starter Bokes
- Start date
- Sort by reaction score
i'm in the same boat so very interested to know (although, my question is the difference between the both as I have a mp quad 2.66 and a G5, the latter being sold) My decision is on which one to buy.
Same type of work involving:
renders
mpeg2 compressions
DVD build compiling
Any speed increases in those areas would allow me to finish projects faster.
to the OP, I recently saw a nehalem mp on the refurb site. $600 (canadian) off the regular price.
Cheers,
keebler
Same type of work involving:
renders
mpeg2 compressions
DVD build compiling
Any speed increases in those areas would allow me to finish projects faster.
to the OP, I recently saw a nehalem mp on the refurb site. $600 (canadian) off the regular price.
Cheers,
keebler
Given all the hassle you'll have selling the old one, I wouldn't bother. Your render times will decrease 5% if that...and you wouldn't even know it unless you had the two machines side by side. There will be no perceivable difference.
No, not worth the money, and no real improvements. Nothing discernable at the very least. If it were used as a server, it might be a different story.
You'd be better off by improving other parts of the system with the money, such as disk throughputs (RAID in particular, if you haven't already).
I tend to agree with Nano and JS. In the overall it's not worth it. There will be a few video editing tasks that the new machine will be faster at. But the vast majority of VE tasks will be close to exactly the same and there will even be a few tasks that are slower. If the two machines were about the same price I think I would be tempted to say go for it. But the there's $2,300 difference for almost nothing.
.
.
The OP is using an Octad system (stated as dual Quad
), and assumed it was suitable for the purpose (not overkill either), but you do have a point, just in case it is. But if an Octad is needed/preferred, the higher clock helps it out, making the performance gap very minimal at best. And as Tesellator points out, the cost difference isn't inexpensive for such a minor difference. So I figure that cash could be better used to improve other areas of the system (bottlenecks). So I made the mention of RAID, as it's not indicated it's currently being used.
If you're considering such a move, I advise rather that you hold off for the next revision (or revision thereafter), when the technological advances in then-current Mac Pros have more of an impact on shipping software.
The OP is using an Octad system (stated as dual Quad
But if an Octad is needed/preferred, the higher clock helps it out, making the performance gap very minimal at best. And as Tesellator points out, the cost difference isn't inexpensive for such a minor difference. So I figure that cash could be better used to improve other areas of the system (bottlenecks). So I made the mention of RAID, as it's not indicated it's currently being used.
I'm sorry but if he is rendering something complex the 8-core Nehalem will be much faster than the early 2008 Mac Pro (Harpertown 8-core).
I've not seen enough evidence to support it from what I can tell (so far, and I don't place much confidence in Geekbench). The biggest architectural improvement is the IMC (IMO). Unfortunately, that doesn't seem to be utilized well ATM, as it operates in dual channel for most applications at best. The exceptions I'm aware of are with servers.
If you can show me something that's absolutely clear on it, it would be appreciated.
Apple said the the 2.93GHz is 1.1x the 2008 3.2GHz for rendering ONLY in FCS. So I guess it would be about 1.2x faster than the 2008 3.0GHz. At scrubbing, editing, and so forth I guess there is no difference at all or even that the 3.0 is faster.
The two machines are almost exactly 15% different at 3D rendering in C4D as tested in CineBench. That is to say that the 2.93 in a multi-core render test with all cores at 100% was 15% faster. So if one took 85 min. to render the other would take 100 minutes. That's not significant IMO. For single threaded it's even worse and the difference is only about 8%. If you were wise and such times were actually important to you then you might notice that you can purchase another complete (used) 2008 3.0GHz machine for just the upgrade difference between the 08 3.0 and the 09 2.93 at new Apple prices: $2,300. So you could spend about $2,500 for a 15% ~ 8% (max) increase by buying an 09 system (assuming you could sell the 08 for the same as it's new price!) - Or you could pay the same for a 100% increase by doubling up on the current machine. This means staying with an 08 model is between 92% and 85% smarter/thriftier than getting an 09 replacement.
The two machines are almost exactly 15% different at 3D rendering in C4D as tested in CineBench. That is to say that the 2.93 in a multi-core render test with all cores at 100% was 15% faster. So if one took 85 min. to render the other would take 100 minutes. That's not significant IMO. For single threaded it's even worse and the difference is only about 8%. If you were wise and such times were actually important to you then you might notice that you can purchase another complete (used) 2008 3.0GHz machine for just the upgrade difference between the 08 3.0 and the 09 2.93 at new Apple prices: $2,300. So you could spend about $2,500 for a 15% ~ 8% (max) increase by buying an 09 system (assuming you could sell the 08 for the same as it's new price!) - Or you could pay the same for a 100% increase by doubling up on the current machine. This means staying with an 08 model is between 92% and 85% smarter/thriftier than getting an 09 replacement.
There's that. But no, there's just not that much difference between the 08 and 09 machines. It's really just that simple.
The cores themselves, haven't actually changed by much (if at all). It's primarily the IMC and QPI vs. FSB & DMI of the older architecture that's changed, and where the speed differences derive from. But that's only useful in servers for the most part (move data faster to the cores). It's a wonderful idea, but is something else in practice. Simply put, software that can use it is almost non existent (full throttle).
This will change in time, but who knows when that will happen. Pun intended but we still do not know which applications the original poster uses.
There will be circumstances in heavy rendering (not Final Cut that heap of **** when it comes to multithreading) but actual rendering of complex scenes (and compiling) where the Nehalem will easily be quite a lot faster. Even the slower 2.26Ghz Nehalem Mac Pro will be faster in these cases. I will agree that it will require specialized cases until the software catches up.
And to say the architecture haven't changed that much...that would be an understatement
There will be circumstances in heavy rendering (not Final Cut that heap of **** when it comes to multithreading) but actual rendering of complex scenes (and compiling) where the Nehalem will easily be quite a lot faster. Even the slower 2.26Ghz Nehalem Mac Pro will be faster in these cases. I will agree that it will require specialized cases until the software catches up.
And to say the architecture haven't changed that much...that would be an understatement
Yeah, but I think we're hitting the realm of special case of a special case,...
In simple terms, not very likely. My thinking is fairly simple. If the OP's situation where such an exception, the question wouldn't have been posted. A consultant would have been hired to design out a system, or at least the software developer would have made some recommendations, if not explicitly listed in the System Requirements.
Thanks all.
I use- FCS- (all apps)
Sometimes Avid and After Effects, Photoshop.
My main concern was the long renders in FCP and Compressor.
Doesn't sound like I will see any improvement.
I'll wait for the next rev.
What about the Graphic card?
I have heard good things about the ATI 4870 being better than the 8800Gforce-
but I also heard the ATI drivers will not take advantage of what SL has to offer. (?)
I use- FCS- (all apps)
Sometimes Avid and After Effects, Photoshop.
My main concern was the long renders in FCP and Compressor.
Doesn't sound like I will see any improvement.
I'll wait for the next rev.
What about the Graphic card?
I have heard good things about the ATI 4870 being better than the 8800Gforce-
but I also heard the ATI drivers will not take advantage of what SL has to offer. (?)
Graphics card will not have a large impact on any of the software you use except Motion, Color, and After Effects. I suggest you wait until SL matures a bit more before buying a graphics card so it is more possible to make an informed decision.
In all honesty, I have a hard time believing that CPU speeds are going to get all that much faster in the next few years, and massive software efforts are needed to tap the GPUs. I think all the really substantial speed increases are due to come from SSDs and eliminating the storage bottleneck. Intel's already taken care of the RAM bottleneck in Nehalem, so it seems like you'll get the most speed from your cash for SSDs now, which will be twice as fast and have twice the capacity in a year or two.
Update: Barefeats just posted OpenCL test benchmarks, showing what most of us suspected: The GTX 285 is the fastest OpenCL card, but the 4870 is the second fastest and performs quite well indeed. He also posted links to the software used to test.
In all honesty, I have a hard time believing that CPU speeds are going to get all that much faster in the next few years, and massive software efforts are needed to tap the GPUs. I think all the really substantial speed increases are due to come from SSDs and eliminating the storage bottleneck. Intel's already taken care of the RAM bottleneck in Nehalem, so it seems like you'll get the most speed from your cash for SSDs now, which will be twice as fast and have twice the capacity in a year or two.
Update: Barefeats just posted OpenCL test benchmarks, showing what most of us suspected: The GTX 285 is the fastest OpenCL card, but the 4870 is the second fastest and performs quite well indeed. He also posted links to the software used to test.
I call bollocks!!
https://forums.macrumors.com/threads/779241/
Still havent seen a single test on that subject.
I am starting to think that the "motion,color and aftereffects" is a evil mantra concieved by AAA (apple,adobe,autodesk).
LIES!!! IT´S ALL LIES!!!
Uh...nobody mentioned Autodesk, who doesn't make any of the software mentioned. Second, Barefeats has tested the various graphics cards and shown an actual difference in Motion. Go look at their pages if you don't believe me.
Also, linking to one post you authored with nothing substantive to back up your opinion just makes you look foolish.
Also, linking to one post you authored with nothing substantive to back up your opinion just makes you look foolish.
Dude,dont get your panties all wedged up. It was all in jest,if you didnt understand?
I have tried to search for clear,plain performance charts on the gfx cards,but they seem to be harder to find than uncorrupt politicians.
And yes,i have checked out the bares test,and it tested only motion but failed to clear what and how they tested it.
Let alone shake,color or lightroom/aperture.
So,once again i´ll put a
in here so you dont all revved up!