Marvell Buys Over Intel's XScale Processors!
Marvell is a relatively small company established in 1995 and headquartered in Santa Clara, California. They are most well-known for their Gigabit Ethernet and SATA controller chips, although they also produce W-LAN, VOIP, optical storage and power management chips.
On June 27th, 2006, the low-key Marvell soared to attention when they bought Intel's cellular communications and applications processor division for a cool $600 million plus liabilities. In a stroke, Marvell acquired Intel's considerable patent portfolio of cellular and application processor technologies.
More importantly to us, Marvell now owns Intel's XScale application processors. Yes, those are the famous XScale processors that power practically all of the PDAs in the market. As a result of this purchase, it is Marvell, instead of Intel, who will be announcing the release of the new PXA300 series of application processors.
Formerly codenamed Monahan, the new PXA3xx processors promise better performance and battery life for cellular and handheld devices. But since when has any mobile processor not promised that? Let's take a look at what the fuss is all about!
The Marvel PXA3xx Application Processors
The new Marvell PXA300-series will consist of three processors - the high-end PXA320, the mid-range PXA310 and the value-segment PXA300.
Both PXA300 and PXA310 will be capable of running up to 624MHz, but the PXA310 will feature hardware video acceleration. This allows it to decode H.264 video at full VGA resolution with a framerate of 30 fps while the PXA300 is only capable of doing that at the QVGA resolution of 320x240.
The high-end PXA320, on the other hand, will be capable of speeds of up to 806MHz. It also comes with a 32-bit memory interface for even better performance. All three processors come with a 64KB L1 cache and a 256KB L2 cache.
Only the Marvell PXA320 is available in volume right now. Marvell will only ship the PXA300 in January, 2007 and the PXA310 soon after that.
Let's take a closer look at each processor.