News and other tidbits that Chad Cloman finds interesting enough to share
According to the interface specifications, all of my hard drives should be running at about 160MB/s. After using the HD_Speed utility, however, I found that actual, sustained transmission times are a fraction of the theoretical maximum. While this wasn’t a huge revelation, I was surprised to find a drive running at one-third the speed of the others.
Link: http://www.steelbytes.com/…
(via Lifehacker)
Why go to all the hassle of a dual- or triple-boot system when you can mechanically switch between hard drives? The linked article explains how to build the switch—or you can purchase one from the article’s author. One caveat: do not swap drives while the computer is powered on. Ever.
Link: http://www.thesataswitch.com/
(via Lifehacker)
The SheevaPlug is definitely the smallest PC I’ve ever seen. It’s literally the size of a power adapter and just plugs in to the wall.
Link: http://www.marvell.com/…
(via engadget)
Apple has been making changes to the standard 1⁄8” audio jack. Take a look at the picture in the linked article to see what I mean. My (older) iPod came with the normal three-part headphone adapter.
Link: http://www.appleinsider.com/…
(via digg)
A company named pureSilicon revealed a new line of fast, high-capacity SSDs at last month’s Consumer Electronics Show. The drives, which have storage capacities up to 1TB and a maximum transfer rate of about 300MB/s, are scheduled for release in early 2009. No word on the cost, but it’ll probably be on the high side.
Thanks to Josh for this topic.
Turns out that wiping data from an SSD is much easier than from a traditional magnetic hard drive.
Link: http://www.computerworld.com/…
(via digg)
Several Seagate hard drive models, but especially the Barracuda 7200.11, have been freezing up and failing. The problem is with the firmware, and Seagate says that the data is still intact on the drives (but just not accessible). See the linked article for instructions on how to update the firmware.
Link: http://seagate.custkb.com/…
Update: The first fix released by Seagate had a bug in it. They’ve released a second fix.
The speed at which a program is executed on a multi-core system can vary by up to 10 percent, depending on a variety of things from how interrupts are handled to the way the program is coded. Chad’s news has previously discussed the “memory wall” problem that occurs in multi-core computers.
Link: http://gcn.com/…
(via Slashdot)
It seems like the new thing in computer hardware is to get the most cores. If two is better than one, then why not four, or eight? But the truth of the matter is that processing power doesn’t necessarily increase at the same rate as the number of cores. One of the major problems is the “memory wall”, where the cores still share common memory and you’ll run into a situation where one core is waiting for another to finish using the memory bus. The linked article mentions the case where, for certain types of problems, a 16-core CPU has about the same processing power as a dual-core CPU. Yes, you read that right. The recommended solution is stacking memory on top of the CPU—I’m not sure what exactly that means, but I imagine we’ll be hearing more about it in the coming years.
Link: http://arstechnica.com/…
Chad’s News has previously discussed the concept of using graphics card GPUs to do regular computing tasks. Now NVIDIA has released a card that’s full of GPUs yet doesn’t have a video output—it’s meant to be used solely for heavy computational tasks.
This paradigm could radically change the way we use computers. And it’s coming soon.
Link: http://blogs.computerworld.com/…
(via digg)
Solid state hard drives (SSDs) aren’t just for laptops anymore—they’re making inroads into the server market. IT managers like them because they run faster and don’t use as much power, which compensates for the high price. And as far as capacity goes, IBM is testing a 4 TB SSD array. Before you switch out your hard drives, however, be aware that not all is perfect in SSD-land. Regular hard drives are still more cost-effective for systems that perform lots of non-sequential read operations or lots of write operations.
