Home
| Calendar
| Mail Lists
| List Archives
| Desktop SIG
| Hardware Hacking SIG
Wiki | Flickr | PicasaWeb | Video | Maps & Directions | Installfests | Keysignings Linux Cafe | Meeting Notes | Linux Links | Bling | About BLU |
David Kramer wrote: > Before I waste more time tracking down problems caused by running > x86_64, I need to decide if it's actually worth it. From what I > understand, the only advantages of 64 bit over 32 bit are access to more > than 4GB RAM (not a problem for me, since I only have 2GB), and memory > mapping large files (not an issue for the same reason). Am I missing > something? On AMD-based systems, x86_64 also has a 10-20% performance edge over the 32-bit x86 build when running 64-bit applications on Linux. It mostly seems to come from the fact that the x86_64 architecture has twice as many general purpose registers, so the code needs fewer instructions. Code SIZE, on the other hand, stays about the same, because 64-bit instructions tend to be longer because of the bigger pointers. On the Pentium 4 and Pentium D processors with 64-bit support, x86_64 runs no faster than x86, suggesting that the bottleneck in those processors was instruction fetch rather than execution. I haven't yet seen any comparative benchmarks for the Core 2 series. 10-20% isn't a very big deal for most people. A desktop user would barely notice it, and it goes away if you spend your time running 32-bit applications. And if you run 3D applications, forget x86_64, because I don't think any of the proprietary video drivers come in 64-bit builds yet. All in all, it may not yet be worth the trouble for desktop Linux users, though I think x86_64 does have a place in the server room. -- This message has been scanned for viruses and dangerous content by MailScanner, and is believed to be clean. _______________________________________________ Discuss mailing list [hidden email] http://lists.blu.org/mailman/listinfo/discuss
BLU is a member of BostonUserGroups | |
We also thank MIT for the use of their facilities. |