Comments on: A New Mac Pro http://www.elharo.com/blog/mac/2010/07/26/a-new-mac-pro/ Ranting and Raving Thu, 09 Feb 2012 04:38:18 +0000 hourly 1 http://wordpress.org/?v=3.1.3 By: Dolan Halbrook http://www.elharo.com/blog/mac/2010/07/26/a-new-mac-pro/comment-page-1/#comment-877458 Dolan Halbrook Tue, 27 Jul 2010 22:33:46 +0000 http://www.elharo.com/blog/?p=1003351#comment-877458 LR3 seems pretty responsive on my i7 iMac, and it's a killer screen. Admittedly it would be a lot better machine if it had an eSATA port, but it's worth considering. LR3 seems pretty responsive on my i7 iMac, and it’s a killer screen.

Admittedly it would be a lot better machine if it had an eSATA port, but it’s worth considering.

]]> By: Chris Adamson http://www.elharo.com/blog/mac/2010/07/26/a-new-mac-pro/comment-page-1/#comment-877445 Chris Adamson Tue, 27 Jul 2010 20:42:18 +0000 http://www.elharo.com/blog/?p=1003351#comment-877445 I'll be surprised to see OpenCL put to much use, but GCD is harder to avoid. Also, GCD is messing up iOS 4 apps in surprising ways. I was maintaining some code that was setting up network connections in NSOperations. In iPhone OS 3, those were always called on the main thread, but in iOS 4, they come from whatever thread GCD feels like calling you back on. And as it turns out, Cocoa's networking classes really want to be set up on the main thread. We pretty much ignored threads in our iPhone OS 3-targeted book last year, but if we were doing a new version, I think we'd have to cover them, and do so early. Another big question is if any of these concurrency-oriented languages really take off, will they get Apple's blessing for Mac and iPhone use? Or is shoehorning "blocks" into C and Apple's various C/Obj-C libraries really going to work? I’ll be surprised to see OpenCL put to much use, but GCD is harder to avoid. Also, GCD is messing up iOS 4 apps in surprising ways. I was maintaining some code that was setting up network connections in NSOperations. In iPhone OS 3, those were always called on the main thread, but in iOS 4, they come from whatever thread GCD feels like calling you back on. And as it turns out, Cocoa’s networking classes really want to be set up on the main thread.

We pretty much ignored threads in our iPhone OS 3-targeted book last year, but if we were doing a new version, I think we’d have to cover them, and do so early. Another big question is if any of these concurrency-oriented languages really take off, will they get Apple’s blessing for Mac and iPhone use? Or is shoehorning “blocks” into C and Apple’s various C/Obj-C libraries really going to work?

]]> By: Elliotte Rusty Harold http://www.elharo.com/blog/mac/2010/07/26/a-new-mac-pro/comment-page-1/#comment-877416 Elliotte Rusty Harold Tue, 27 Jul 2010 14:37:21 +0000 http://www.elharo.com/blog/?p=1003351#comment-877416 Yes, that is strange. Also annoying that I can't price out the options yet. I expect I'll go with a 6 core system at maximum 3.33 GHz clock rate, the lower end video card, and buy a passle of 3rd party RAM. I'll have to see what Apple's prices are on the 512 MB SSD drive before deciding whether or not to get that preconfigured or just buy a third party SSD drive. I suspect we'll never see the performance gains that OpenCL and Grand Central Dispatch were supposed to bring to multi-core machines with discrete graphics CPUs outside of a few very special purpose applications. Performance gains are a lot easier to realize when they come from components that are invisible to the application programmer. E.g. an SSD disk speeds up applications because they don't need to do anything different to take advantage of the faster drive. However if apps have to be rewritten to take advantage of potential speed gains, then by the time programmers have learned how to use the new features, much less shipped software based on them, the classic model hardware has caught up an exceeded the performance gains promised by the new model. I've seen this happen time and time again. Think RISC vs. CISC, for example. In 2010, most application programmers still don't know how to take advantage of multiple cores to run their program. Yes, that is strange. Also annoying that I can’t price out the options yet. I expect I’ll go with a 6 core system at maximum 3.33 GHz clock rate, the lower end video card, and buy a passle of 3rd party RAM. I’ll have to see what Apple’s prices are on the 512 MB SSD drive before deciding whether or not to get that preconfigured or just buy a third party SSD drive.

I suspect we’ll never see the performance gains that OpenCL and Grand Central Dispatch were supposed to bring to multi-core machines with discrete graphics CPUs outside of a few very special purpose applications. Performance gains are a lot easier to realize when they come from components that are invisible to the application programmer. E.g. an SSD disk speeds up applications because they don’t need to do anything different to take advantage of the faster drive. However if apps have to be rewritten to take advantage of potential speed gains, then by the time programmers have learned how to use the new features, much less shipped software based on them, the classic model hardware has caught up an exceeded the performance gains promised by the new model. I’ve seen this happen time and time again. Think RISC vs. CISC, for example. In 2010, most application programmers still don’t know how to take advantage of multiple cores to run their program.

]]> By: Chris Adamson http://www.elharo.com/blog/mac/2010/07/26/a-new-mac-pro/comment-page-1/#comment-877409 Chris Adamson Tue, 27 Jul 2010 13:00:42 +0000 http://www.elharo.com/blog/?p=1003351#comment-877409 No Mac Pro update today. The <a href="http://www.apple.com/macpro/" rel="nofollow">product page</a> says 6/12-core models are coming in August. Apple rarely pre-announces revs like this (since it will kill sales of 4/8-core MPs for the next few weeks), so I wonder if something slipped at the last second.I'm not a Lightroom user, so I can't answer your performance questions there, sorry. What I do wonder about is when, or if, we'll ever see the performance gains that OpenCL and Grand Central Dispatch were supposed to bring to multi-core machines with discrete graphics CPUs. No Mac Pro update today. The product page says 6/12-core models are coming in August. Apple rarely pre-announces revs like this (since it will kill sales of 4/8-core MPs for the next few weeks), so I wonder if something slipped at the last second.I’m not a Lightroom user, so I can’t answer your performance questions there, sorry. What I do wonder about is when, or if, we’ll ever see the performance gains that OpenCL and Grand Central Dispatch were supposed to bring to multi-core machines with discrete graphics CPUs.

]]>