Second Life on Core 2


Linden Lab has a great product in Second Life and now their running it on another great product, Intel processors.



I disclosed Linden Lab on Intel's new processors (mobile/desktop/server which all share the same core micro-architecture). I essentially told them why the new processors are so great (and will get even better in the future), for the following reasons:



· Higher Performing OOO Engine


· A full 4-wide super scalar pipeline that can decode, execute and retire instructions at a sustained rate of 4 instructions per clock


· Deeper buffers allow the processor to more effectively find instructions that can be executed in parallel


· A very short and efficient 14 stage pipeline that attributes to higher frequency and delivers higher instructions per clock


· Macro-op fusion allows for certain instruction combinations (e.g. cmp & jmp) to be executed as a single instruction


· Advanced Power Capabilities


· Components in the processor are only powered on when needed


· Many buses and arrays are split so that in some modes of operation the unused portions can be placed into a low-power state


· Enhanced Cache System


· Implements a shared L2 cache that can enable better performance by: 1) enabling memory intensive apps with the opportunity to see a larger cache when running concurrently with less memory intensive apps in the other core as well as 2) enabling the ability for data and code in the L2 cache to be shared between the cores without having to go out on the FSB, lowering FSB utilization and finally, 3) in single threaded environments the single threads running on a single core will have the ability to utilize the full cache


· Improved L2 cache to core bandwidth


· Improved Memory Access


· A variety of new HW prefetchers both within the cores and within the L2 allowing many array and vector codes to continuously stream data from memory without suffering from memory delay or requiring specific SW pre-fetch instructions


· Uses memory disambiguation to get more of the benefits from OOO execution to be felt during memory operations by breaking false dependencies between loads and stores


· Improved SSE/SSE2/SSE3 Performance


· 128-bit wide SSE instructions now have a throughput rate of 1 per clock


· Added the Supplemental Streaming SIMD Extensions (SSSE3) instruction set



In the end they got the message that it's overall the fastest processor available, doesn't cost a whole lot more, and saves them some green with their power bill. :-)



I'm confident that the performance/watt of Intel processors will continue to improve as the processor architects, designers, etc. find more ways to conserve power even as the performance of the chip increases. As this performance/watt trend continues so will Linden Lab's confidence in Intel's products.


Per informazioni complete sulle ottimizzazioni del compilatore, consultare l'Avviso sull'ottimizzazione