An analysis in Inside DSP by BDTI on the potential for the IBM Cell in embedded systems:  IBM’s Cell for Embedded?

Another interesting EE Times article on multicore.  Almost half of the embedded designers have no plans to use multicore devices: Embedded developers cautious on multicore

From the EE Times Europe:  Imperas chief scientist retires to pursue language interests  Imperas recently released some of their work as open source.

A New York Times article on the recent push to develop multicore software:  Race Is On to Advance Software for Chips

From an interview with Donald Knuth in InformIT:

To me, it looks more or less like the hardware designers have run out of ideas, and that they’re trying to pass the blame for the future demise of Moore’s Law to the software writers by giving us machines that work faster only on a few key benchmarks!

The rest of the interview can be found at:  Interview with Donald Knuth

After Spending $100M in venture funding, multicore chip company Montalvo has closed, with Sun buying its assets:  Sun salvages what’s left of a multicore CPU developer

Apple has bought low-power PowerPC multicore maker PA Semi.  Since Apple recently moved from PowerPC to Intel x86 devices in their desktop machines, there is still some speculation about this acquisition:  Apple buys chip designer PA Semi for $278m: lower power chips?

Steve Lebison’s blog at EDN continues to take a hard look at multicore, in particular software issues.  His more recent posting should be required reading for anyone interested in this area:  Multicore Parallel Programming: Can We Please do it Right This Time?

This is just a comment on a CNET story, but it is perhaps the most chilling scenario I have seen on the lack of multicore software.  Since it is a short posting, I’ll reproduce it all here.

Danger in the Mine: (reader comment from eightwings)

Posted on: April 15, 2008, 2:21 PM PDT

Story: Wall Street hopes Intel, AMD still singing

Interesting article. Intel may be still be posting record revenue but there is a real danger in the mine and it’s called the ‘parallel programming problem’. It’s an unforgiving toxic gas that threatens to suffocate the life out of even the healthiest multicore canary. The crisis is beginning to wreak havoc on Intel’s and AMD’s plans to transition from sequential computing to parallel processing. The reason is that the industry’s multithreading strategy for programming multicore computers is turning out to be a turkey. A multi-billion dollar turkey, I might add. [Google "Nightmare on Core Street" for more info on why multithreading is a turkey.]

I’m afraid that the good old days are soon coming to an end. Revenue growth in this business has always been more or less tied to increases in processor performance. Unless someone finds a solution to the parallel programming crisis, performance increases will slow down to a crawl and so will revenue. The industry is at a dangerous crossroad in its journey and there is BIG MONEY in the balance. This is the kind of stuff that revolutions are made of. My advice to all parties involved is, thread carefully (pun intended). Things are about to change rapidly.

PeakStream was bought quite suddenly by Google in mid-2007.  Since then little information has been available about PeakStream and their tools.  A presentation by the former CEO, Matthew Papakipos gives some information:  The PeakStream Platform for Many-Core Computing