Some thoughts about geekbench results
Looking at the geekbench results http://www.geekpatrol.ca/browse/2006/, one sees that the intel's conroe chip's, from E6600 to X6800, performance is really in it's own class compared to everything else, all AMDs but also woodcrest. The question is, why didn Apple take conroes to the pro line. Maybe there will be some big benefits in having two dual cores, that we don't see yet. Or was it so important to have a quad as the previous top end model of g5. And if you take a quad there, you cannot take for the lower end a single conroe because it is too fast!
Another geekbench result is, that windows is a "better" os than OS X. All the best result you have with mac pro are made on windows. Also in some test (don't remember where i saw it) there was on a intel mac used same program's windows and universal versions, the windows was much faster. Is it possible that macs could get some speed increase when leopard comes, optimized for intel dual core?
When you compare mac pros of different configurations, the 2.66 with 2 gig RAM seems to be clearly the best value for money. More than 2 gig brings only a minimal increase in performance, at least in this test. For a similar further jump you got with adding 1 gig RAM to a 2.66 Ghz, you need 3.00Ghz and a lot of RAM.
It would be interesting to see some comments about these things. Some other also, I hope, than just that the whole geekbench doesn't tell about the real life performance.
Another geekbench result is, that windows is a "better" os than OS X. All the best result you have with mac pro are made on windows. Also in some test (don't remember where i saw it) there was on a intel mac used same program's windows and universal versions, the windows was much faster. Is it possible that macs could get some speed increase when leopard comes, optimized for intel dual core?
When you compare mac pros of different configurations, the 2.66 with 2 gig RAM seems to be clearly the best value for money. More than 2 gig brings only a minimal increase in performance, at least in this test. For a similar further jump you got with adding 1 gig RAM to a 2.66 Ghz, you need 3.00Ghz and a lot of RAM.
It would be interesting to see some comments about these things. Some other also, I hope, than just that the whole geekbench doesn't tell about the real life performance.
Comments
Looking at the geekbench results http://www.geekpatrol.ca/browse/2006/, one sees that the intel's conroe chip's, from E6600 to X6800, performance is really in it's own class compared to everything else, all AMDs but also woodcrest. The question is, why didn Apple take conroes to the pro line. Maybe there will be some big benefits in having two dual cores, that we don't see yet. Or was it so important to have a quad as the previous top end model of g5. And if you take a quad there, you cannot take for the lower end a single conroe because it is too fast!
Another geekbench result is, that windows is a "better" os than OS X. All the best result you have with mac pro are made on windows. Also in some test (don't remember where i saw it) there was on a intel mac used same program's windows and universal versions, the windows was much faster. Is it possible that macs could get some speed increase when leopard comes, optimized for intel dual core?
Windows XP may be faster than OS X, but faster doesn't tell you everything. I don't have huge problems with Windows, but the problems I do get are generally more tedious to fix than the problems I get on OS X.
There is some, but not necessarily a lot of benefit to a quad, it's really limited by the software right now. Pro software is generally the type of software that is most likely to take advantage of a quad. Hopefully this will improve as more software is designed to take better advantage of it. I think Apple needed to justify the price range, which does the job very well, IMO. Frankly, the predecessors to the Mac Pro were a lot more in the class of a workstation type machines than consumer desktops so it makes sense that it's using workstation parts. The hope is that there will be a conroe-based consumer Mac tower, but I'm not that optimistic. I would not expect an Extreme Edition Conroe to be put in Apple's line, they seem to have an odd tendency to make sure there is no performance overlap between their model lines. That tendency is not present in any other computer brand that I am aware of.
First off, Conroe, is a dual core desktop-class chip. It cannot be used in a MP (multiple processor) setup, this is why it is not in the Mac Pro.
While Woodcrest OTOH, is a dual core server/workstation -class processor, that is meant to be used in an MP setup.
The reason Apple didn't put Conroe in the Mac Pros is because, as stated above, it is merely a desktop class proc ,albeit a good one. Woodcrest is a stronger processor than Conroe, and because you can use 2 (or more?) of them on the same mobo, you get speed increases, even if your apps aren't multithreaded.
And I mean, I don't know about you, but wouldn't you much rather have two dual core Woodcrests (2x2 cores) than a single Conroe (1x2 cores). Especially when you are paying ~$2499 + for a computer?
Cheers,
Noah
I was only asking WHY it is so important to have a quad setup only. All I need is a very powerful mac for my work, to me the desktop and workstation are only concepts. Of course it's great, that there are so called workstations, that are on their best in multitasking and with multithreaded apps, but faster (and cheaper having only one unit) core 2 duo or extreme systems seem impossible for marketing (!) reasons.. I think JeffDM got the point.
I was thinking before august 7, a fast conroe-based tower for my "worksation". With a huge wacom 12"x19" tablet, everything on my desktop is build to get more efficiency to my workflow. I'm a pro illustrator, and use mainly Painter (and photoshop, of course). Painting digitally it's hard to imagine a computer too efficient. For example, If I'm painting an image in 300ppi res (I work always for print) with multiple layers and want to use big brushes of "watercolor", that blends dynamically to previous strokes, and the color spreads depending on how "wet" I want the layer to be. The strokes get slower when increasing the brush size a lot, making working impossible with those adjustments. (my one cocern is also the role of gpu in this kind of 2D processes).
I'm going to get a mac pro, and I'm happy to do so. It will be an enormous upgrade from g4!. And as said in my previous post, I hope the real benefits of the four core are on their way...
Cheers
It would be interesting to see some comments about these things. Some other also, I hope, than just that the whole geekbench doesn't tell about the real life performance.
It most definitely doesn't. Here are the explanations to the discrepancies.
1. geekbench tests a lot of things...not only the CPU. If you got Conroe hooked up to faster HDDs than Woodcrest, the Conroe will have a score that comes closer to Woodcrest scores.
2. SMP has diminishing returns and rarely will performance scale linearly.
#2 is the main factor that "doesn't tell about the real life performance". If you're only testing the multi-threaded capabilities of one program, you might be disappointed. These apps that don't fully use all CPUs available isn't always a negative thing though...it just means you've got CPU to spare for other apps.
I can guarantee you that an app that uses two cores effectively (but more less effectively) will make a Core 2 Duo iMac crawl while that same app + other apps will chug along fine on a Mac Pro.
your reply was answering for what I was asking (in the title of this thread I already tried to say it's mainly a question about geekbench results).
One thing still I think is always worth of noticing: maybe none of us will ever want profiling of the product lines and pricing politics becoming dominant over getting the best processors.
But the REAL question is, when they are getting these aparatos here, almost north pole!
But the REAL question is, when they are getting these aparatos here, almost north pole!
I think only Santa Claus delivers in that area. You're gonna have to wait for Christmas. Heh...just kidding.
Geekbench's results are skewed by:
1) Lack of utilization of all 4 cores or higher FSB bandwidth.
2) Probably optimized to test on memory latency, which FB-DIMMs hurt.
3) It's probably less optimized on OS X than on Windows.
One thing to keep in mind when looking at Core 2 Duo (Conroe) scores is that a lot of the top scores are from systems with over-clocked processors (e.g., the top score is from a Core 2 Duo running at 4.4GHz). You have to dig quite a bit to find Conroe-based systems running at the default clock speed, and when you do, they Geekbench scores are on-par with the Power Mac G5, not the Mac Pro.
Geekbench is designed to use all available cores (when running multi-threaded tests, of course), and isn't optimized for any one particular platform (although it is developed on Mac OS X, then ported to other platforms, so if it's biased towards any one platform, it'll be Mac OS X). Geekbench also doesn't measure memory latency, just memory bandwidth.
I hope this clears up some of the confusion when it comes to Xeon vs. Conroe performance!