We first learned the basic details of Larrabee early last year, with much more information discovered earlier this year at the spring Intel Developer Forum. Well, Intel has gone ahead and released even more information about their upcoming graphics processor, and we of course have the goods.
Included in our look is information on what will set Larrabee apart from the competition and we also take a look inside the processor itself to see how things are structured. As it stands, we can have high hopes for the new architecture, but the real proof will be in the pudding when we can give the new GPU a hands-on test. Of course, there’s no sense of thinking about that since it’s a ways off…
Regardless, why Larrabee is so interesting is that it uses numerous Pentium-derived cores to deliver overall performance, rather than sticking to one or two massive GPU cores, as seen on current ATI and NVIDIA graphic cards. What will prove more effective won’t be known for a while, but in our article, we can at least keep you up to date and give you a better understanding of how Larrabee is designed.
The L2 Cache in Larrabee is designed a little differently than how it’s implemented on a normal desktop CPU. Rather than being ‘banked’, the Cache is divided into sub-sections where each section is directly connected to a specific core. If one core is reading data not being written by the other cores, it’s stored in its local cache, which improves latency and also bandwidth.
Source: ‘Intel Opens Up About Larrabee’