technical question about intel

[Guest]

I can't help but wonder, I know that my money is all on AMD but I have a technical question. Now if The T-bird has 384K with 200MHz FSB, why does a lousy coppermine beat it clocked at the same clock speed? In the report about the 1.2G Thunderbird which beat the pricy 1Ghz, the p!!! got higher scores when both of them were clocked at the same clock rate. Now why is that? The Athlon Core is better and all of the other stuff, is it about the SSE vs. 3dnow!? Thank you in advance.

 

[Guest]

The main reason for a performance difference is the more advanced cache system of the P3. The P3 has a much wider connection to the on chip cache (256 or 128 bit i think) than the athlon (64 bit i think). Apart from this the relative performance of each CPU is not as different as you may think. In business apps, all things equal, the Althlon and P3 are pretty close (The Athlon would probably win out with a better cache architecture). The Athlon does however have a superior FPU (Floating Point Unit) to the P3. This gives the Atlon a noticable advantage in Floating Point intensive apps such as scientific programs, 3D-modelling and maybe some games (if they were as optimised for atlon as they are for P3). I hope this has helped

 

[SoulReaper]

All depends on the test. If the program testing the cpu's doesn't support 3dnow, then the speed tests are kind of invalid.

"upgrading is no longer an option...it's a necessity"
Visit www.elitehunters.com
--SoulReaper =)

 

[Guest]

The Athlon has raw power. The P!!! has optimization. Through SSE and all the advanced cache system, the P!!! is able to keep good performances. If you want to see the true power of the Athlon's FPU, just look at 3d studio benchmarks...

 

[Guest]

The P!!! has a shorter pipeline than the athlon, This means instructions take less clock cycles to finish executing on the P!!!.
The athlons long pipeline means that while it is theoretically possible to start several instructions per clock cycle, it often can't because the next instructions rely on the result of earlier ones - and the result is not yet available.

The athlon also has a crappy Branch Prediction Unit(BPU) while the P!!! has a good one. If a program is going to branch depending on the result of an earlier instruction, it is not known which way it will branch until that instruction has finish execution - 5 to 10 clock cycles. Instead of sitting doing nothing in that time, the CPU guesses which way it will go and starts executing. Obviously, if it guesses wrong it has to cancle all the instructions it's done and the time's wasted anyway. The BPU is responsible for making these guesses.
The next generation of Athlon is expected to have a better BPU.