AMD: Analysis And Forecast Of A Struggling Semiconductor Supplier

Before easing into the weekend, please reference my just-published analysis of AMD’s Phenom microprocessor, Spider platform and bigger-picture trials and tribulations. I welcome your thoughts!

Followup: Some additional comments beyond those made in the main writeup:

• I was disappointed, albeit not terribly surprised, to hear that AMD has done an about-face and won’t be supporting customers of the company’s Quad FX system (which formed the testing foundation of an external storage benchmarking project I tackled earlier this year) with Phenom processor upgrades. I say ‘not terribly surprised’ because it’s evident to me that the company is severely resource-constrained (funding, headcount, etc) right now and this particular project was an obvious ZBB candidate. And I say ‘disappointed’ because by virtue of the Quad FX systems’ substantial price tags, they were purchased by leading-edge enthusiasts and AMD advocates…exactly the audience that AMD doesn’t want to disappoint with broken promises right now.
• One of the aspects of the Spider platform that I’m most excited about, ironically, is a feature set evolution was notably overlooked by mainstream press coverage. AMD had actually launched the Radeon HD 3800 graphics family a few days before the rest of the platform. And a few days before that, AMD launched the FireStream 9170 professional GPU with double-precision floating point support, which I’m assuming (although AMD won’t officially confirm or deny) shares a common silicon foundation with the Radeon HD 3800 series. Long-time readers know that I’ve closely followed the topic of general-purpose GPU applications for quite a while, and I plan to continue doing so far into the future. Nvidia’s latest GeForce 8xxx series GPUs are restricted to only supporting single-precision floating point calculations, as I pointed out almost a year back, and AMD/ATI’s arithmetic advancement will undoubtedly be welcome news to supercomputing applications. Whether or not it’ll also be harnessed by mainstream apps remains to be seen.