On Nov 27, 2007 4:34 PM, Michael Tinsay <[EMAIL PROTECTED]> wrote: .. > A 64-bit processor means that it can fetch memory 64 bits at a time compared > to a 32-bit machine which fetches 32 bits at a time. Offhand, fetching from > main memory to cache should take less time.
Yes -- but -- ultimately you are limited by how wide is the bus width to main memory. Even 32-bit Intel CPU's have a 64-bit bus width to the main memory. Also, working on 64 bits at a time means that a machine word consumes 8 bytes instead of 4 bytes. So you can only store half the number of words in the cache. This is why I said the cache is effectively halved. Also, all current FPU's use 80-bit math (IEEE 754 standard). Having 64-bit integer registers won't do anything to improve floating-point performance as the floating-point unit is independent of the integer units. Fast floating point with SSE / SSE2 / SSE3 does benefit from wider registers, because these That said, almost all contemporary CPU's except the entry-level ones are 64-bit capable. Even the entry-level Intel Pentium E2140 (the cheapest Intel with dual cores @ 3150 pesos) is EM64T capable. To truly leverage 64-bit, you'd need to buy a ton of RAM, which although cheap these days (1400 pesos for 1GB) most main boards can't take >3GB. > On the math side, It means it can handle 64-bit integer addition or > subtraction in one step instead of several. Same thing goes for floating > point with >32 bit precision. On strings, it can compare 4 bytes at a time > instead of just 2. But this only means something depending on what > application you're going to run. _________________________________________________ Philippine Linux Users' Group (PLUG) Mailing List [email protected] (#PLUG @ irc.free.net.ph) Read the Guidelines: http://linux.org.ph/lists Searchable Archives: http://archives.free.net.ph
