religious war : intel vs. amd

[Maximilien]

(I'm not an hardware person, so ... ) Is there really a difference ? I've been using intel CPU pretty much all my PC life, so I can't compare, but in what circumstances I should buy an AMD CPU ? Are there software issues that can happen by switching from one to another ? ( i know this is a naive question ) Thanks

---

Maximilien Lincourt Your Head A Splode - Strong Bad

[Mike Dimmick]

I've always used Intel apart from an AMD 286 in my parents' first PC (although at the time I think this was a licensed Intel design with AMD acting as a second source) and a Duron in my first PC at this company, which was terrible mainly due to a near-complete absence of L1 and L2 cache, and the support chipset from VIA. I've never had good experiences with non-Intel chipsets. Obviously Intel chipsets are not an option with AMD Athlon or Opteron processors, since Intel's and AMD's processor buses are completely different. Things may be different now that the memory controller is integrated in the AMD processors, but I still don't have that much faith in VIA, nVidia or ATI AGP/PCI bridge components nor, for example, IDE controllers or the other subsystems of the chipset which make up a complete PC system. Right now Intel's desktop processors suck. Their processing power per watt of electrical power, and per watt of radiated heat, is appalling, and because they've hit a wall in processor frequency, their poor computing power per clock cycle has allowed AMD to pass them. But I still think an Intel-based system will be more reliable than AMD. On the other hand my experience could be because I've generally known enthusiasts with AMD systems, who tend to overclock their systems and/or load them with every bit of junkware under the sun, which have tended to destabilise their computers. Both manufacturers try very hard to be compatible with their previous generations and with each other, but processor manufacturers make mistakes just as software engineers do. They both publish lists of errata (errors) in their processors and either put a fix into a future revision (stepping) of the processor or publish a workaround. One issue I'm aware of right now is that, if AMD's Cool and Quiet technology is turned on, the Windows high-performance counters don't count at a fixed rate - the rate is directly proportional to the current processor frequency. The processor drops its frequency to save power and reduce heat if idle or less stressed. Intel have had this problem on some steppings of the Pentium 4 as well. The current workaround is to turn Cool and Quiet off. The problem affects some games and other applications. Stability. What an interesting concept. -- Chris Maunder

[Stuart Dootson]

I'm in a reasonable position to answer this, I guess, I've experience of AMD's from K6 through Athlon and AMD64, and Intels from 486 through every Pentium (except Pentium M) up to Pentium 4. At present, AMD's desktop processors (I'm meaning AMD64 - X2 as well, as the core's pretty much the same) have (IMO) much better development (compilation etc) performance than Pentium 4s. I've built the same project from scratch on a 3GHz P4 and an AMD64 3400+ (clock speed is 2.2GHz) and the AMD took about 2/3 of the time of the P4 - other factors such as disk and RAM were approximately equal (1GB RAM, they both had 7200RPM disks, although the one with the AMD64 was a laptop drive, so might have been a bit slower). Other CPU bound tasks display similar speed differences. However, I think Intel are probably back on track with the Core Solo/Duo range - the Pentium M's were always pretty good for performance. Anyway - processor/chipset related software issues? Nah, didn't have any of those. HTH