I decided to start this thread since i cannot fully understand the differences between thees two AMD and Intel.
It may seems to be very noobish thread but i cant sleep not knowing that differences.
As for an example i picked up two models for desktops: Intel i7-2600 and AMD FX-8150.
I went through some technical details. Between them two AMD is slightly faster on clock speed, 8 cores vs Intels threads,
memory controllers supports 1866MHz whilst Intel 1333MHz.
AMD 4 x 2MB shared L2 / 8MB shared L3 - Intel 4 x 256kB L2 / 8MB L3
Yet i do not get BUS speed differences (?)
Now according to PassMark from 29th of November 2011 and majority of forumers, i7 shows higher performance.
Can someone explain what is the difference and what makes that difference in general performance ?
Think about standing in front of a house. You want to go from the front yard to the backyard. So you open up the door and start walking. On the first house the floor plan of that house makes you turn to your left and start running up a set of stairs. Once at the top of the stairs you turn to your right and open a door and go through that room to open another door which leads to the bathroom and out another door to reach a set of stairs that go down and into the washroom and then into the kitchen before opening another door and walking out into the backyard. On the second house you can simply open the front door and walk into the front room then into the kitchen and then out the backdoor to the backyard.
While there is a lot more to it than this the idea is the same. Our current microarchitecture allows us to have a straight through floor plan that speeds up the job all the while allowing you to walk over running.
Basically what you need to know is that Intel 2600k (get the k version not just 2600), is better then anything AMD has in 99% of applications. There are a very few benchmarks I have seen that Bulldozer actually wins, but they are not things that an average user would be doing.
Especially if your a gamer go for the 2500k or 2600k. If all you do is very processor intensive and actually FULLY utilizes the architecture of Bulldozer (not much does yet), then that could be a viable option as well. Although i think that even the benchmarks Bulldozer does win, its not by much.
Instead of running fast through all floors and doors, walking straight through works faster...
I assume that internal architecture on Intel CPUs is better organised due to Sandy Bridge technology, processes takes shorter time internally ?
Could you come up with some technical details please ? ( I'm familiar with electronics and digital technology at engineering level )
What about external transfer through north bridge ?
One of the biggest factors is the pipeline. With the release of Intel® Core™ 2 Duo processors we dropped our pipeline to 13 deep; on the older Intel Pentium® 4 it had been at 20. The problem with this is that the processor spends more time waiting to get the information on the 20 packet pipeline then the on the 13. Originally it was thought that a smaller pipeline would lead to the processor getting incomplete information and leave it waiting to get the rest of the data. As the Intel Pentium 4 proved this wasn’t case. For some reason with the release of their new processors the other guys changed their pipeline to 20 deep.
There are a number of other factors that play into this but I am not an engineer that happens to understand how it works except for the pipeline part of the change.
Basically after a certain point specs do not matter. What matters is how the CPU was designed to execute instructions. Intel CPUs are more efficient (more instructions per clock cycle) than AMD CPUs so at the same clock speeds, Intel CPUs are faster than AMD CPUs.
It has to do with a lot of technical bits that few people outside of the engineers who design processors understand. As IntelEnthusiast already mentions a big part of that is the pipe line. The AMD FX processors also have much higher latency on their cache than Sandy Bridge, so despite having more Cache, the Cache is slower. Many variables go into how a processors architecture work. Core count, clock speed, amount of cache, those help determine performance but it's very hard to compare them directly between different architectures.