The CPU's have a different architecture, Intel processors have hyper-threading, etc.... All that can change performance. The clock speed only matters when comparing the same architecture of a processor. A i5-3570k at the standard 3.4 GHz clock will be slower than the same processor (i5-3750k) overclocked to 4.5 GHz. I can't effectively explain this myself, perhaps someone else who knows more about this can explain it better than me.
No, they aren't a different architecture, they're both x86_64. A different architecture resulting in incomparability would be like trying to compare an i3 to the Xbox 360's PPC architecture processor. AMD and Intel can easily and validly be compared. The main difference is that Intel tends to have more L3 cache in their CPUs, which is probably the biggest performance affector after clockspeed and core count.
I think that you can call both things "architecture". In this case both CPUs are based on the same base architecture and use the same instruction set (x86-64). But while they are still x86 compatible different CPUs do have a different microarchitectures which result in different efficiency. You for sure can compare Intel and AMD cpus but not by just comparing the clock frequency or amount of cache. For example both the old Phenom II x4 965 and the i53570k have four cores, 6MB of L3 cache and a clock frequency of 3.4Ghz but still the i5 is way faster because of its more efficient architecture.
Yes, however the price range difference is significant, however my 8350 overclocked performs equally to the 4770 at stock. It performs equally at stock to my old i7 2600 and performs similarly in multi threaded applications. https://teksyndicate.com/videos/amd...s-3820-gaming-and-xsplit-streaming-benchmarks For some reason people seem to have it in for AMD for no particular reason, they have a high price to performance ratio and perform similarly to the i5, and you cannot disagree with that. It isn't much different to the i5 in most multi threaded games. http://www.youtube.com/watch?v=uD6UcsVG56g I'm not saying the 8350 is better, but AMD doesn't deserve the bashing that it gets from a lot of people, just saying.
Yeah, I personally prefer AMD. If I'm just playing games, I don't need to spend so much money on an Intel processor when I can get just as amazing performance in games with AMD.
There's also die size, transistor count and transistor size which affect the speeds of the CPU. Seems like a bad CPU, like it's just OC'd to 5GHz without any real improvements.
You guys do know that ADM cpu's with 8 cores actually have 4 cores and 4 virtual cores, so it will still only show the performance of a 4 core CPU, and being that it is AMD = Shit
are you talking about hyperthreads? intel quad cores will also show 8 cpu's but really it's hyperthreads
I know a Computer that has 3,120,000 cores! That's what noone seems to understand about AMD CPUs. Hyperthreading = vritual cores
Both die and transistor size don't affect the speed of the CPU. The main advantage of smaller transistors is that they're more efficient and run on lower voltages but smaller transistors do not automatically result in a faster CPU. With smaller transistors you can pack more of them on a single die of the same size which actually does increase the speed but die size itself has absolutely no influence. The AMD 8 core cpus don't have 4 actual and 4 virtual cores. They are based on a modular design consisting of 4 modules for the 8 core CPU. On each of these modules some components are shared between its two actual cores but still the main components are there eightfold. If you just want to brag about the number of cores get a AMD Opteron 6xxx, it has 16 cores. Only thing is that it is quite useless for regular desktop usage, it's a server CPU for a reason.
This. Each bulldozer module has two integer cores, but with shared L1, fetch, decode, FPU and L2. This means that an 8-core CPU has 4 modules, which means it has 8 integer cores and 4 FPUs, with cache, fetch and decode hardware shared between the two conventional x86 cores in each module.
It's not the same as increasing clock speed, but having a smaller die size means the electricity in the CPU can get around quicker, so the CPU would be faster... And transistor count is like you said. About the 4 FPU and 8 integer units: wouldn't it be better to have 8 FPUs and 4 integer units? since integer arithmetic isn't really that hard, and you would want parallelization in heavy computing tasks; which mainly involve maths, which involve fractions. I would understand if it was an ALU though, not a Integer arithmetic unit, that would make sense, and I think that's what you meant?
I guess they chose to add more integer/logic performance as it was "cheaper" than including FP performance. Each Integer cluster has two AGUs and two ALUs, just take a look at some block diagrams on Wikipedia.
