Zacate lacks balls

AMD is preparing the launch of its Zacate processors for Q4 2010, and Zacate is said to deliver quite the punch for such a low-power CPU.

Despite AMD having nudged us about giving us the grand tour in San Francisco, we couldn't actually make it this year to the hub of Intel developments. So we'd like to point out the next best thing. Anand Lal Shimpi's account of how Zacate is actually performing.

Anand managed to get himself scooped away by AMD pixies and magically transported to a hotel suite where he was treated to a full volley of frenzied Bobcats, lucky him.

So, first impressions about Zacate, the 18W dual-core Bobcat-based processor that will power notebooks and nettops around the $500 tag: it's tiny. Not like Atom-tiny, but smaller, even. The small footprint is also explained by the low ball-count and from which Anand deduces is a single DDR3 memory channel.

Fooling around for a bit in CoH, Anand was left with an undisclosed Core i5 Arrandale system. The scribe was able to discern that the Zacate processor is quite thrifty and more than keeps up with the competition at a much lower consumption... one of the clear advantages of integrating the GPU into the CPU. In fact, Zacate almost doubled the Core i5 in-game performance in City of Heroes, which seems very strong indeed. AMD then proceeded to demo IE9 and the HTML5 Psychedelic benchmark, where Zacate really tore into the struggling Arrandale.

Although AMD is said to be aiming at "mainstream notebooks", this kind of technology does seem like an instant lotto winner for netbooks, if the right form factor and price is found. AMD's greatest challenge might be, of all things, avoiding head-on comparisons to Sandy Bridge, something the press are bound to do, due to the nature of both technologies. For now, and until Llano comes out, AMD isn't aiming to compete with Intel's Sandy Bridge, but give a good sample of what it'll be able to do later in 2011.

AMD has the manufacturing advantage, we think. It has a 380M transistor Fusion processor built on a 40nm process, which will provide good yields and a ton of chips per die due to the small die size, while Intel is challenging itself with a 32nm part that is considerably larger and harder to make.

On a side note, we can't help but think back to the famous "can o'whoopass" that Nvidia threatened Intel with a couple of years back. Said argument was based on Nvidia's belief that the GPU was something that would evolve to an equal status as the CPU in a computer system's hierarchy, and take over many tasks.

Well, Nvidia was right after all, but we're left thinking it wishes it wasn't.

With Zacate, Ontario and Sandy Bridge, both CPU companies have all but made redundant the discrete graphics parts in the value and low-end computing segments. Sure you'll still need a fiery noisy dustbuster-type graphics cards for the high-end gaming and computing experience, but anything rated as "discrete" on the sub-$500 radar will simply shrivel up and die a slow death on a shelf somewhere... Ion included.

You can give it a glance, here.