the_codist()

Using OpenCL In Snow Leopard For A New Application

After much internal debate (with myself) I decided to skip working on another game and focus on a graphics application for Snow Leopard only, using OpenCL.

Sure, it seems suicide to only build an application for a new OS version only a few days old, using a brand new technology which is not 100% solid yet. It does make some sense, as OpenCL/Grand Central are something that Apple has built that could be really important in the future as hardware goes more and more parallel. Being able to take advantage of this not only give me the ability to ship something difficult to do otherwise, and might even attract some marketing assistance from Apple.

I am building a niche graphics application than can really benefit from scalable processing power. One could build this without OpenCL but it would take a lot of time and effort to make it scalable. The groovy thing about OpenCL is you can build an algorithm that works well for small images on slow Macs, and have it scale to ridiculous levels for large, complex images simply by having better hardware.

My main Mac is a quad Nehalem MacPro, with 4 cores and 8 processing units, along with a relatively wimpy GT120 GPU. Running various OpenCL benchmark apps, its obvious the CPU is faster than this GPU. However upgrade to a GT285 and the GPU blows the CPU away. Adjust your algorithm to split work between multiple devices (I believe you can use multiple GT285 plus the CPU) and the ability to process scales to silly levels.

The beauty is once you have the code working on a single device, adding more is pretty easy. Then without any changes on the programmers part the scaling just happens based on whatever hardware the user has. Image processing applications are perfect for this type of scalability since the work can be divided up and run in parallel without much trouble.

Currently the ATI support for OpenCL has some troubles (which will be fixed in 10.6.1 I believe) but the NVidia support seems a bit more mature. There are no shipping applications (that I know of, I could be wrong) using OpenCL, and I imagine it will take time for it to be widespread. So for once I in front of the steamroller.

The other great thing about supporting Snow Leopard only is I can use all the new features of Objective-C, Cocoa, XCode and new Frameworks and not care about Windows or anything. Who know, maybe there will be a iTablet, which would be a cool bit of hardware to run my new app on!

I expect once I have something working I will look for alpha testers to help round out the features. If nothing else it will be fun to play with.

It also gives me something fun to work on, since finding a fulltime contract or job seems to be taking geologic time.

What Did Steve Do For Apple

Why Not Agents For Programmers?

I always wondered how finding a job as a programmer was always so employer-centered. Sourcers find resumes and dump them on recruiters who dump the programmers on employers; yet no one is looking out for the programmer's needs. It seems so one-sided.

Athletes have agents. Movie, stage and screen people have agents. Musicians often have agents. Even house buyers can use a buyers agent if the local real estate people haven't put them out of business (real estate agents act like they are helping you, but they really are representing the seller).

There are millions of web sites that post jobs (some of them real, some of them not) and you can search all over the internet. If you find something interesting however you are on your own dealing with a recruiter (who acts like a friend but is really only interested in dumping you on the nearest employer) or HR person or whoever.

Sure, it doesn't cost the prospective employee anything but that's the rub, you don't get anything for the nothing you are paying. The employer pays everything and gets all the cards. If we were willing to pay 10% or whatever agents usually get, then they would be more motivated to find you something you wanted and get the best deal.

Sure agents can seem as unseemly as, well, recruiters; but if you maintain a relationship with them over time it is in their best monetary interest to find you long paying work that you are happy with, and also to understand your needs and abilities and make a good sale on your behalf.

In the long run an agent makes the most money if they can make their customers happy and employed long term; and thus will take the time to network and find the best opportunities. Rather than the employee having to hunt all over the country , trying to interpret stupid job ads, fight off nasty recruiters and sourcers, fill out myriads of forms and web sites (what is it with job sites that require 20 pages of forms just to look at a job), a programmer's agent would market your talents and find the best job for you.

I want someone on my side, the percentage I have to pay means I get real service, assuming the agent has a clue and a real network. In the end, the whole idea is to get the right person for the right job at the right price.

A nice dream but the industry seems to be set on having the employers pay for fishing and hope a good programmer falls in the net.

Kissing the App Store Goodbye

I give up, there is no way for a small developer to make it in the App Store anymore. Unless you are associated with a big publisher who can attract reviews and spend money on marketing, or win the lottery, the App Store is too big, too poorly organized, too low priced and run by some kind of Mafia-like organization.

Despite having a long list of products I'd like to make for the iPhone, it's not worth spending time (and what little money I have these days) only to disappear either into the Rejection black hole, or into the 200+ daily pile.

From a customer standpoint, the App Store is awesome, so much goodness you could download or buy apps every hour and never get bored. Most apps are either free or cost a buck. It's a real massive platform with great hardware, a mostly great SDK, and nearly 40 million strong market eager to get more apps.

From a developer standpoint it's a nightmare. An Orwellian review team which does not communicate to anyone (even inside Apple), uneven policies which change at random times leading to horror stories form Shaken Babies to Google Voices, and an apparent "so we treat you like crap, there are hundreds more suckers eager to try" attitude. Then there is the mouthpiece "Richard" whom I have been rejected by already (and appears in the Google Voice story) with a solid future in government stonewalling (at least the Iraqi Information Minister was entertaining). Add to the App Store Review Team problem the inability for anyone to find your applications (I always tell people that 2% of the applications sell 98% of the copies) in the app store. Unless you make it into a top 25 list somewhere, you may as well be selling out of your trunk at a flea market in Alaska in January.

I think the store surprised Apple in how big it grew, and from their standpoint the success came without having to do anything to promote the platform, so it makes sense (of a sort) to not care if most of the developers get disillusioned. In the mid 80's Apple was desperate to get developers on the Mac, and we were treated likely royalty in many ways, since the platform would die without applications. I still remember the 1986 WWDC, where the entire Mac developer community got to ride a boat in the harbor, talk, gamble for free hardware, and joke that IBM could kill the entire Mac market with a well-placed torpedo. Today Apple rules the world and thus the little guys no longer matter.

I'm not giving up on the iPhone, just developing apps for the store on my own. I would be happy to build apps for others, I like the SDK and the opportunities for amazing apps are still there. It's just the irritation of dealing with the App Store team, the poorly organized marketplace, and the sheer volume is too much.

My next app will be for OSX. Sure, the apparently market is less exciting but there are actually a similar number of devices running OSX and iPhone OS. Being able to update apps whenever I need to, not have to wait for anyone's approval, be able to directly communicate with customers, and also get to collect any revenue in something under 3 months time is much more appealing.

Sure, my apps aren't anything exciting. Quantum Pool has been doing well but with an eCPM of about 17 cents it's not even beer money. The more complicated interesting apps I would have done would take too long, and the risk is simply too great on my own thin dime.

The other phone markets don't interest me either, the devices and the SDKs don't really provide what the iPhone does. I've worked at Apple, used Apple computers since 1980, and really don't care to switch. It's just dealing with the App Store bunch is too much even for me (note the DTS folks are still great and do a wonderful job in support).

Apple as a big entity won't care what I think, and the other 14,000 iPhone developers either have given up, keep trying, made a million (very few) or gotten a job at Ngmoco. But there are another 14,000 out there hoping to hit the big time.

Good luck with that, hope you never hear from Richard.

Bailout Bucks Finally Released

After more than 3 months of rejections and waiting, I was finally able to release Bailout Bucks for the iPhone/iPod Touch.

Press Release