6/17/2008
I know you all want to get married today, but if you can stomach a few more years you will have the ability to purchase cheaper diamonds.
No, de Beers isn’t opening their warehouses and flooding the market with their inventory.
Rather, modern-day alchemists have grown flawless diamonds in the lab, some of which have hit the marketplace and most of which are 15% cheaper than their natural counterparts.
If you have a chance, be sure to read Diamonds on Demand from the latest Smithsonian. It is a great update to the 5-year-old piece from Wired and one of my favorites: The New Diamond Age.
Among other quotes in the new piece are statements from de Beers who lobby intensely against synthetic competition. However, all of their arguments boil down to the same nonsensical diatribes used by the anti-GMO crowd.
For instance, here is a whopper from de Beers: “Diamonds are rare and special things with an inherent value that does not exist in factory-made synthetics. When people want to celebrate a unique relationship they want a unique diamond, not a three-day-old factory-made stone.”
Au contraire, I’m fairly certain that specific consumers like guys buying an engagement ring would much rather buy the cheaper alternative if the two items are virtually indistinguishable.
See also: Three cheers for Wired mag and many more for genetic engineers
Since everyone is building their own multi-node high-performance computing cluster, you may be interested in a new product by Microsoft.
Aside from using Ubuntu, trying to do anything on Linux usually seems like a chore, this includes setting up and maintaining an HPC environment.
Ah, but after three years and several betas, according to a recent piece by ComputerWorld it looks like Microsoft HPC Server 2008 hits the g-spot for non-geeky researchers.
Think it’d work on the new PetaFLOP shattering Roadrunner?
The last two days have been great for researchers, scientists and just about every consumer (unless you refuse to buy anything but abacuseses).
Yesterday ATI announced their new GPGPU’s. And as the rumormill predicted, these single-card solutions not only hit the 1.2 TeraFLOP marker (single-precision), but they scale easily (thus a 4-way SLI configuration brings you up to 5 TeraFLOPS). Double-precision is quoted at around 200 GFLOPS.
And according to their press release, you can grab one of them for $999 sometime in the Q3. Be sure to check out the coverage from TG Daily because they have some interesting numbers (historical performance versus Pentium Pros).
The same day, Nvidia announced its own refresh of the GPGPU line, the expensive Teslas. In addition to being able to buy the traditional PCIe card you can also buy a spiffy blade model that crams 4 of these GPGPUs under one roof. The only drawbacks from these solutions is that in addition to slightly slower speeds (900 GFLOPS versus the ATI 1.2 TFLOPS), they cost twice as much ($1999). And the double-precision penalty is even harsher (90%) amounting to around 100 GFLOPS. Ouch.
On the other hand, not only can the consumer grab a Tesla today, but apparently there is a substantially larger community effort behind the Tesla solution, so there are more tools at the developers disposal. (Yes, I know about CTM, but it’s still not as mature).
I wonder when the FASTRA guys will upgrade.
