This summer I came to France, Begium (and Neerland too), to see our family, and moreover discover the next generation, that is already promising, let me introduce them…

First there was the Christening of my nephews Mathieu & Thibaud (5 years old both), that Christophe & France adopted 2 years ago in Haïti. They are 5 years old and packed of positive energy (and some malice too!).

Then there’s Louison (3 years old) with her mother Elizabeth, that I didn’t had a chance to meet in France, she has the eyes of her mother and the kind smile of her father. A little shy at first, but playful and lovely!

The last one is Quentin (3 month), brother of Louison, that I almost ever see sleeping or drinking milk! Seems able to sleep while all family is having a dinner and speaking loud, laughing out very loud, and enjoying it!

This was a big trip for us, from Montreal to Paris, then 2500km (1600 miles) of road-trip, including Amsterdam, Namur, Lyon, and back to Paris! We would like to see our family, discover the new generation, and celebrate our parents!

CUDA is now disabled on my MacBook Pro!

Sorry Apple, but you screwed my development environment!

PS: As stated, this is the newest graphic drivers from Apple but they are not compatible with actual CUDA Drivers!

Go read it’s interesting, and written by someone that thought at first that Adobe Flash Player might succeed on Android platform…

On RAID-5 & RAID-6, it’s much more complicated and in fact not manageable from a controller point-of-view without having insight into filesystem structures. That is normally disabled as RAID (hardware or software) are on a different layer than FileSystem handlers, on the existing major OS.

The simplest way is just to implement 0-write detection on SSD, and treat it as TRIM information:

• To TRIM data, send a 0-block (either 512bytes or a multiple)
• SSD detect this as “trimmed” (unused and flag it)
• on read, trimmed blocks just send back 0’s, as it was written. To avoid confusion between TIMMING and a block filled with 0’s
• On RAID-5 or RAID-6 (or any-else), TRIMMED data will be treated as 0’s and thus drive status will be maintained on RAID crc calculation
• When reconstructing a failed drive, each trimmed sector will result on 0’s (due to CRC calculation) and then considered trimmed AGAIN on the new drive

No more no less… The only difference is that the filesystem handler have to send a block with 0’s to the drive controller, and SSD drive firmware updated to handle it as a TRIM command.

It will limit BIG deletion to approximately 300MB/s or 600MB/s depending on SATA version, but it’s not a limitation on real life use, albeit being able to use any RAID, wether hardware or software, transparently, to TRIM data on drives, will be a huge step forward for SSD in enterprise world

The main idea is to get ride of CPU-based memory protection and IO protection, everything is running without any protection, the protection itself is based on static code analysis and dynamic checking. For this to work, any code must in form of an intermediate representation (often incorrectly called bytecode), that is analysed at load, interpreted or compiled on-the-fly with security checks.

The interesting thing is that code is “naturally” protected from buffer overrun, or any kind of attack, as any code is suspect by default, and on the other way switching from one thread to the other is lightning fast as no context change must be done. A very interesting project, that might be bases for future Microsoft or Open-Source OS.

SingularityRelease 2.0 is available here, with it’s source-code.

I like the explanation of the “intent” mechanism on QTKit (QuickTime framework), as “intent” for me is a key on future libraries and frameworks to communicate application future behavior and needs to Operating System, to help it optimize resources (including power consumption) on-the-fly knowing the real needs of each application/process/threads. This is an interesting trend!

Naturally you will need technical background to understand it, but for me it’s worth reading for anyone developing or planning to develop, maintain, deploy Mac OS X 10.6 Snow Leopard.

PS: don’t forget to read Ars technica review of Mac OS X 10.5 leopard before, it’s worthwhile too and many technologies included in Leopard might surprise you, such as LLVM!

As a meta-platform (cross-hardware platform), Flash evolve a different speed depending on the target patform (Mac OS X, Windows, Linux, …), it’s Adobe that decide which platform will be supported and what will be supported on each platform.

Adobe doesn’t support major distributions of Linux for example. That’s Adobe choice, not Linux user choice, or Linux distribution maker choice.

Adobe choose when to implement new features, including hardware acceleration, optimizations, or security patch, depending on the platform.

With this long-awaited feature, hardware acceleration of H.264 HD videos on Mac OS X, with a limited set of Mac supported, Adobe upgraded Flash with limited support of thsi feature (compared to PC with supported video cards, some unsupported on Mac OS X version of Flash), and lately. This give a competitive advantage of the Windows platform for Flash users.

Apple doesn’t want a third-party to decide if and when features must be added, optimizations should be done (they are NOT on done correctly for Flash on Mac OS X), and moreover a meta-platform maker deciding to give competitive advantage to another platform.

I fully understand the choice of Steve Jobs concerning Flash on iOS, and support it. Even if Flash was fast (it isn’t), reliable (lol), or adapted to multi-touche screen, accepting it is letting Adobe decide of the future of the iOS platform!

From my iPad, I tried to use geolocalization of my iPhone and it didn’t work. So I decided to check if I could Lock it with a password. And thus didn’t worked either!

To have Geolocalization of an iPhone or Locking (also Wiping/resetting) in foreign countries, you absolutely need DATA ROAMING that is the best way to end-up with a montruous bill from your cellphone carrier! These features, presented as really great, and mandatory in some countries, are juste unavailable on default setting (that is safe for your wallet!).

Hey Apple, rework it, it should work by SMS too, even if I have to use a special password on MobileMe to use these features!

PS: Locking was effectively done 10 days later, when I was back on Canada, on Trudeau Airport. Totally useless, and false security for travellers!

Previous iMac with reduced frequency and low-cost graphic memory (VRAM), laptops with down-clocked GPU, premium price but low-cost parts. And performance lagging far-behind half-cost PC!

This time, it’s the Radeon HD 5670, that ATI specify for GDDR5 with 64GB/s bandwidth. Only GDDR5. Apple put GDDR3 on it to attain 32GB/s, diminishing performance on a 1.5X to 2X factor! That might be a 10$ or 20$ economy for Apple, 1% of the iMac, but a real performance loss for gamers and anyone using GPU, for OpenCL, CUDA, Photoshop CS5 Suite, …

Notice that the ATI Radeon HD 5750 of the Quad-Core 27″ iMac is in fact a HD 5850 Mobility, offering approximately same level of performance, but less bandwidth (as usual).

Steve, sometime you are really cheap!

PS: Preceding iMac 27″ Core i5 Quad-Core had ATI Radeon HD 4850 with GDDR3, but offering 64GB/s bandwidth, having 256bits bus instead the 128bits bus of the HD 5670.

I tried to encode a 720p concert (mkv/h264/ac3), with Turbo.264 HD, and it was real fast, 3X to 4X faster than handbrake encoding for the same input file. AppleTV preset in both case. Wow!

But the resulting file is still 2.4GB long, that is real big for a 87minutes footage, and will translate to 3GB per full-length movie….