Slashdot Mirror
Apple Responds to Adobe
Thargok333 writes "Apple calls out Adobe on the 'PC is Faster' article linked from the Adobe website. They state that it is an After Effects bug, which they are working close to Adobe to fix. With Adobe's idea of G4 optimization, I am not impressed that a 'single 1.25 GHz G3' gets beat by a P4 3 GHz."
I've heard that a good amount of the base code in their products is in Pascal. While I don't know if this is true, it would also imply a helluva lot of 68k code still lurking about in their software. Going through both 68k emulation as well as another compatibility API is just bad. I hope this is not the case.
Mac developers have had almost 10 years to get rid of the 68K code. Fat apps existed for a while, but I would be beyond shocked to find out there was a single 'mov' of 68K code in any Adobe product. It's not like they employ a bunch of lazy programmers -- they have a very talented lot.
Why not put some thought into making performance better rather than making gee-whiz features that most folks never asked for.
Again, I would doubt this is the case here. At a lesser development organisation, maybe, but not Adobe. Especially given that they are right up the street (so to speak) from Apple, they've got all the opportunity in the world to get help optimizing their products. And most of the gee-whiz features you find in a shrink-wrap program come about because more than a handful of customers asked for them. Sure, developers always find neat little things to do, but Usenet and other forums are far from short on ideas from everyday users. You may not use a cutesy feature, but plenty of other people will.
And that Apple has been able to tweak MUCH better performance and features out of products like the Final Cut series shows that it CAN be done.
The catch here is that Apple has one platform to target. They could write the whole of the app in PPC assembly and no one would care. Adobe has to maintain a code base that stretches across 2 platforms, and they have to weigh the cost of maintaining divergent sources with the benefit gained by increasing platform-specific code. It's not an easy trade, but in the end, they will probably have more shared code that is not 100% optimized for this reason.
Also, we don't know what Adode's Mac revenue stream looks like these days. You would think that they get a ton of Photoshop, Illustrator, etc. sales from the Mac side, but I see more and more Windows boxes on the desks of artists and designers. Again, this is a cost-benefit situation that Adobe must analyze, and maybe they don't see the need to pour hundreds or thousands on man-hours into optimizing tiny bits of their Mac code for a few more sales.
On the other hand, maybe Adobe realizes they need to tighten up the PPC/Mac code, but they are running short on resources for that. So they post a benchmark on their site saying Windows kicks Apple's ass. Apple panicks and says, "Not so!" The next day, 2 of Apple's crack engineers drive over to Adobe and help them optimize code for a week, writing a bunch of PPC stuff that Adobe didn't have time for. Voila! They get some free (or cheaper) help and Apple looks like the hero. Marketing bullshit at it's finest.
1) XML parsing is memory, cache, and possibly hard-disk *intensive* depending one exactly what you are doing. It is more I/O intensive than anything.
2) If you are getting 20% slower per MHz, either your math is screwy or your configuration is.
3) There are very fast XML parsers for the mac on the market. Find one.
4) That you claim they are 20% slower per MHz against both Intel and AMD processors, which do not run at the same speed per MHz throws your credability off.
Integrate Keynote and LaTeX
Off the top of my head I can think of the shorter pipeline and AltiVec. But, should the G4 have an "edge"? Who really cares?
Years ago, the Mac platform had the edge in performance and clock speed thanks to Power Computing's "Power Tower Pro 200". At that time nobody was whining that Windows has to switch to PPC in order to remain viable.
The "honest answer", the most relevant thing I can add to this, is that the PPC architecture and the Pentium architecture tend to leapfrog each other every few years, and right now it looks like PPC is losing by a small margin.
I'm not holding my breath, but IBM's 970 and other iterations of the "Power 4" line may well tip the scales slightly in the other direction. For a little while, at least.
Before Steve Jobs re-joined Apple he was interviewed in Rolling Stone (I think, or maybe it was Spin), and when asked how he felt about Apple's move to PowerPC architecture his response was that he was happy for Apple, because now they've got a Pentium of their own. Of course he was with NeXT at the time, but the point is that Apple, Motorola and Intel really don't care that much who is making faster machines, as long as they're marketable as being as fast as they can be.
Battling over processor speed is just what Intel/AMD/Motorola/IBM would like us to be doing, because of the few of us who are really qualified to say which architecture is faster, an even smaller percentage of those people realize just how moot the point is.
A slight Wintel performance edge is not going to have thousands of Mac users rushing over to the other side. And it isn't performance that makes Windows users switch to Mac.
Hmm.. I use to do billboard printing/design work. First off, Photoshop has a 2GB-3.5GB size limit. As far as the printing process is concerned, older large format printers could only do between 20dpi-70dpi ... using uncompressed TIFF CMYK files for output, on a standard 20'x50' billboard, this would yield a file around 219MB - 1GB+ per layer. Granted, given the viewing distance factor of a billboard, these graphics would generally be printed at only 10dpi yielding an effective TIFF size of 54MB.
However, given the current level of technology in this segment of the printing market, most (if not all) large format printers will now use Postscript files as input, so much of the design is done like traditional desktop publishing (quark, indesign, illustrator) and then rasterized on usualy a multiprocessor PC-based RIP (raster image processor).
In anycase, getting back to the original poster .. just because graphic houses are currently using Macs is not a representation of them being faster. It is simply a case of being what they have used for years. Many places that I do consulting work for (as well as where I previously worked) offload all of their heavy processing requirements to the fastest systems they can find, Mac or PC .. lately they have been buying PCs ..
Native != Cocoa, my friend. Or rather, not all native apps are Cocoa, nor are all Cocoa apps native (I doubt you'd call a Cocoa/Java app "native", for example).
Carbon is every bit as native as Cocoa. It is true that Apple was too lenient in its backward-compatibility measures; by not forcing developers to take advantage of new technologies while porting their apps, we've seen the rise of Bad Carbon Ports, epitomized by (ironically) AppleWorks but seen to lesser degrees in other apps as well. However, a Carbon app which actually uses OSX technologies can be not just every bit as fast as a Cocoa app, but even somewhat faster if it's done right.
Cocoa is good. Carbon can be good, though it does have some Bad Stuff that, regrettably, many porting jobs use. But an app is no better simply by virtue of being written in Cocoa or Carbon.
If you want proof of this, simply look at the fact that Apple is slowly re-implementing Cocoa on top of Carbon. Some of this work is already complete and is a part of Jaguar; more is coming. When it is finished, Cocoa will be little more than a layer of abstraction on top of Carbon. How can one be intrinsically better than the other, when this is possible?
I can only speak to multimedia content generation. I have never measured the speed of my PowerMac dual 1GHz with Final Cut Pro against a 3GHz Pentium running Premiere, but I do know that after 3 years of using Adobe's products with third-party hardware and drivers on Windows, I gave up and got a Mac with Apple software.
I bought 4 boxes: PowerMac, Final Cut Pro, DVD Studio Pro, and a 6-to-4 pin firewire cable for camera. I have never purchased another accessory, peripheral, or software package, and the system is so well designed and executed that I can start an editing session in the morning and mail a 1 hour tape or DVD to a client by 5.
With the Powerbook, I can shoot video and edit it on the plane home. If it's a long plane ride, I'll have the DVD burned before we land, while the guy in the seat next to me fights with his Vaio or Dell (I've been on many flights where some poor bastard gets no work done because Windows eats itself; it's happened to me too).
My experiences over the past 5 years convince me that the Megahertz mongers have got the issues backwards. If you can first show me any combination of PC and laptop hardware on the Intel platform that can do everything I describe above for the exact same price, I will look at the speed of filter application or transition rendering.
My point is that if Apple makes faster machines their superior systems will be better than they are now. If speed is the only improvement a PC with Windows and Adobe products offers, that system is still inferior to the Mac.
I disagree, it's used by my many graphics professionals, especially 'film gimp', it's very much a pro tool - for example it can handle file sizes much larger than Photoshop.
Aha! Film Gimp and GIMP serve two totally different purposes. Video files are WAY MORE HUGE than simple still image files. I can see where the scriptability and flexibility of Film GIMP can come in very handy as a tool in film and video post-production. Used in tandem with other post production tools, a very powerful combination can be made.
However, the GIMP as a stand alone productivity tool in a commercial graphic design environment just isn't up to snuff yet as far as featrues and usability go.
Pooty tweet