Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Grimace

macrumors 68040
Original poster
Feb 17, 2003
3,572
250
with Hamburglar.
Looks like the 2.26 is doing better after all. Some users have even gotten Cinebench scores for the new 2.26 like: 3142 (single) and 20,138 (multiple) reported here.

CINEBENCH 10
This free benchmark app uses real world code from Cinema 4D to render a sample project. It stresses all available cores. In the case of the Nehalem, hyperthreading fools the app into thinking there are 16 cores on the 8-core models and 8 cores on the 4-core models. The graph below shows the Cinebench rating for "Multi-CPU" render test.

neh03_cin.gif


GEEKBENCH 2
It's not only multi-core aware, but it includes some memory tests which explains why the Nehalem based Mac Pros beat the older Penryn based Power Macs with higher core frequencies. The graph below shows the overall 32-bit score.


neh03_g32.gif


There is also a 64-bit version of Geekbench. Fewer results exist for it since, for some unexplained reason, consumers will gladly spend thousands for a new Mac but resist purchasing a $20 serial number in support of a starving Mac developer.
neh03_g64.gif
 
That looks very promising indeed. I look forward to them receiving their test units so they can do their own benchmarking.
 
There is also a 64-bit version of Geekbench. Fewer results exist for it since, for some unexplained reason, consumers will gladly spend thousands for a new Mac but resist purchasing a $20 serial number in support of a starving Mac developer.

If I had $20 to blow on a Mac Pro, I'd gladly help by donating $20 to the developers burial expenses -- since I'd likely be too late to keep him from starving to death.
 
Yeah I saw these a bit ago.. Looks like they're pretty on par with the 2.8's. They don't blow away the 3.2s, but I would expect that. I think the 3.2's are still beasts and shouldn't be blown away by Apple's bottom line MacPro.
 
Yeah, I think it is fair to assume that performance leaps aren't ever so large that the top of line (octo) in one generation is trumped by the bottom (octo) in the subsequent generation. But it is nice to see that the 2.26 is a real contender. I guess it all depends on which tests you use for benchmarking!

Single-threaded tasks are still very common, but over the life a machine, multi-threaded apps may become quite the norm. We'll just have to see!
 
Yeah I see where the single thread isn't as zooming, but I think we're really talking about old technology. Software has always tried to surpass the hardware and I think it will do it in this case too. Any single thread software will probably be left behind for multithread ones.

I would almost say it's similar to 32 bit software. Everyone knows 64 bit is the way to go even though it's been a while I think everyone else is finally catching on with that... At least the software that would be used a lot on OSx. (Adobe products, final cut.. the big name software).

I've really been highly tempted to buy a 2008 2.8.. but I think I'm going to bite the bullet and go ahead with the 2.26.. Besides.. You can most likely upgrade the processors in the future when they come down in price and aren't the "LATEST AND GREATEST THING!!" ;)
 
Yeap after seeing the benchmarks and that the 2.26GHz is actually much faster than the 2.8GHz 8 core 2008 model on everything (especially 2x overall faster memory test and 40% reduced latency will all add up in the later factor as well) and single thread being on par with the 2.8GHz 08 model its safe to say go ahead buy the new 2.26GHz beasts everyone!

Not to mention when Snow leopard comes the nehalem will be crazy fast. Quick Path Interconnect + 40% reduction in latency and bottlenecks overall + 2x memory speeds here we come!!
 
Snow Leopard is hype until proven otherwise. :(

The snappy quick GUI and Quicktime won't be hype. :p

The Snow Leopard benefit for apps, not so much until the developers do some recoding and for the laggards to hop fully into Cocoa.
 
Just ran Cinebench on my 2004 G5 liquid cooled Dual 2.5 (not dual core - dual processor). Results:

Single CPU Render Test
1696

Multiple CPU Render Test
2611


I'm weighing picking up a 2008 2.8 octo VS a 2009 Nehalem 2.26 octo. I work in Final Cut Pro primarily plus Compressor, After Effects, C4D, PS, etc. Currently working on a large 2TB HD documentary and this G5 is starting to draaaag.

I've never even used a Mac Pro with an Intel chip, so I'm assuming I will notice quite a boost with either option.

What should I do???
 
Just ran Cinebench on my 2004 G5 liquid cooled Dual 2.5 (not dual core - dual processor). Results:

Single CPU Render Test
1696

Multiple CPU Render Test
2611


I'm weighing picking up a 2008 2.8 octo VS a 2009 Nehalem 2.26 octo. I work in Final Cut Pro primarily plus Compressor, After Effects, C4D, PS, etc. Currently working on a large 2TB HD documentary and this G5 is starting to draaaag.

I've never even used a Mac Pro with an Intel chip, so I'm assuming I will notice quite a boost with either option.

What should I do???

Get the 2.26GHz, its about on par with the 2.8GHz 8 core model for single thread but on everything else its MUCH faster, not to mention 40% less latency, 2x faster in memory operations and especially the quick path interconnect. And MUCH more beneficial when Snow Leopard ships.
 
The snappy quick GUI and Quicktime won't be hype. :p

The Snow Leopard benefit for apps, not so much until the developers do some recoding and for the laggards to hop fully into Cocoa.

I'm betting I still have to use cp -R in the terminal to do batch copies, though.
 
Its a shame Barefeats left out the single threaded runs of Cinebench where the 2.23GHz Nehalem is massively outpaced by the 2.8GHz model from the previous line. Hopefully Barefeats will add some more single threaded examples. From the benchmarks that I've seen so far, it doesn't look good for the 2.23GHz Mac Pro in comparison to its direct (but cheaper at launch) predecessor of the 2.8GHz Octo. :(
 
Snow Leopard is hype until proven otherwise. :(

I had a chance to use a Snow Leopard developer version - pretty awesome! There were definitely bugs, like folders in the dock not opening properly, but it was very snappy, and the install was only 4GB!!! I'm a fan!

Also, in relation to this thread: that's fast!
 
Yes, these benchmarks are looking excellent for the 2.26. Im even more excited about receiving my Mac Pro now!!

I am also very interested in seeing the Photoshop CS4 results for the 2.26 on Barefeats, as this is one of my most used apps.
 
cp -R? How arcane!
I much prefer tar pipes ;)

I'll be the first to admit there are probably better ways to do it, but I'll do anything to avoid having to use the Finder. 10,000 files to copy and one fails, it'll just stop. Junk. A copy function coded by someone who's not an idiot would say "Hey, File.MXF is broken. I'll copy the rest though!"
 
I'll be the first to admit there are probably better ways to do it, but I'll do anything to avoid having to use the Finder. 10,000 files to copy and one fails, it'll just stop. Junk. A copy function coded by someone who's not an idiot would say "Hey, File.MXF is broken. I'll copy the rest though!"

Indeed! I wish default GUI apps would catch up to the usefulness of CLI apps.. while maintaining the usability aspects ;)
 
I'll be the first to admit there are probably better ways to do it, but I'll do anything to avoid having to use the Finder. 10,000 files to copy and one fails, it'll just stop. Junk. A copy function coded by someone who's not an idiot would say "Hey, File.MXF is broken. I'll copy the rest though!"
Precisely. I don't know why that's not already there.
 
Its a shame Barefeats left out the single threaded runs of Cinebench where the 2.23GHz Nehalem is massively outpaced by the 2.8GHz model from the previous line. Hopefully Barefeats will add some more single threaded examples. From the benchmarks that I've seen so far, it doesn't look good for the 2.23GHz Mac Pro in comparison to its direct (but cheaper at launch) predecessor of the 2.8GHz Octo. :(

Not really, the single thread is about the same in the 3100 point area for the 2.8ghz 8 core 2008 model and the 2.26ghz octad nehalem.

Also dont forget to consider the turbo boost when that kicks in it should see about 400mhz more so bumping the 2.26ghz up to 2.66ghz and when you take that to consideration, throw in 40% less latency + 2x faster in memory operations, the new nehalems are miles faster than the previous gen.

I'm finally excited!
 
Not really, the single thread is about the same in the 3100 point area for the 2.8ghz 8 core 2008 model and the 2.26ghz octad nehalem.

Also dont forget to consider the turbo boost when that kicks in it should see about 400mhz more so bumping the 2.26ghz up to 2.66ghz and when you take that to consideration, throw in 40% less latency + 2x faster in memory operations, the new nehalems are miles faster than the previous gen.

I'm finally excited!

Are you quite so sure about that? Link and link.

I'm sure the 2.26 Octo will redeem itself in some ways but its performance is hugely disappointing. Actually I'll phrase that differently, the performance of Apple's entry level Octo model is hugely disappointing in comparison to the previous (cheaper) entry level Octo model. The chip's fine, the choice of chip and the cost of the machine aren't. Had Apple reduced their margins a little (although would have still been higher than the previous model) and started the Octo line off at 2.66GHz at the price point of the 2.26GHz model then people wouldn't have been as perturbed. The price rise would have thrown some people off but it could have been easily justified by the boost in performance. As it stands, my ageing 2.66GHz 2006 Quad Mac Pro looks to be faster at single core stuff than the 2.26GHz Octo. Not by much admittedly, but certainly not enough of an improvement for me to have justified the purchase of an entry level Octo Mac Pro. :(
 
I agree that the new machines aren't worth upgrading to if you bought a mac within the past year or two.. but some of us, this is our first Apple purchase and the argument between last years model and this years model can be confusing and also discourage people from buying/switch to an Apple.

All the argument on here aren't helping the community. Honestly the price is just the price. Inflation is apart of this crazy, messed up economy. Prices go up, especially on brand new hardware that's just been released to the public. I doubt Apple is losing any sleep over the arguments here on this board. Like I said.. if I bought last years mac, then I wouldn't be thinking upgrade so soon, unless I just had a ton of cash laying around and nothing better to do with it :)
 
At the very least, they could have used the 2.4 GHz chips for $530 x 2 instead of the 2.26 GHz chips for $373 x 2 for the entry 8-core model. 2.26 GHz just sounds so low today. If they did that, it would be a done deal for me.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.