AMD versus INTEL

[Quazzimodo]

Ok please direct me to posts if i'm wrong.

But.... In the Blue corner we have the AMD and in the red we have the Pentiums.

now i am kinda a n00b at pc's for example i have been browsing around for parts on ebuyer and i keep having to buy this to this and this has to be bought with this, so to a novice (i.e me) it is quite confusing.

One question, keeps bugging me, which is better, AMD or Pentium, or something else if there is one.

I specifically would like to compare these two processors.
AMD Athlon64™ 3200+ / 2.0GHz and Intel Pentium 4E (Prescott) 3.0GHz

Now one first thing strikes me is that, the pentium is nearly 1ghz faster than the AMD or is it.

Help on this topic and various input would be much appreciated, for this pc N00B

[hockey man]

O.k. if you truly want to start going into detail between the differences between AMD and Intel read this (http://www.pcguide.com/vb/showthread.php?t=34050) thread. It gets deep! As a rule of thumb though, AMD is better at gaming wereas Intels are better at applications such as vidoe/ audio editing. If you read that thread you will see why AMD's can run at slower clock speeds and compete with Intel. Happy research!

[saphalline]

Neither Intel nor AMD is "better" than the other. Overall they are even. They each are supporting different architectures and so they trade blows in the benchmarks depending on the tasks. Intel CPU's have higher clock speeds, more SSE-class executions, and Hyper-Threading. AMD CPU's have higher IPC rates, more powerful FPU executions, and an onboard memory controller.

If you compare just the clock speed, you won't get anywhere. End-user performance is where it's at, and a P4 531 (3.0GHz) is about even with an Athlon64 3200+ (2.0GHz) at the end of the day. The P4 will pull ahead in SSE- and HT-optimized code (usually in encoding/decoding apps for video/audio editing) while the A64 will pull ahead in games (where the highly dynamic FPU-intensive code works to its advantage). For all other types of programs, like office apps and web surfing, they are both so powerful you'll never notice a difference. These same rules more or less apply to the Celeron D and Sempron, as well. And now we have a new arena for competition: the dual-core CPU's... :cool:

We can keep going if you want. Or you can just trust me that Intel and AMD are about even and that they just have differing ideas about how to execute instructions. ;) Everyone knows I love talking about this stuff, but it goes over most people's heads.

[Quazzimodo]

if you want to ellaborate more, just email me or PM me, to save people the brainache, i am keen to learn about how these processors, work, and other stuff, facts, and if it is allrite i would like to email you saphalline with a few questions if thats ok!
thank you

[Sempron]

I got a brief explaination of SSE, IPC & FPU (thanks for confusing me btw Saphalline:p )

Can you further explain how the onboard memory controller works?

[hockey man]

We can keep going if you want. Or you can just trust me that Intel and AMD are about even and that they just have differing ideas about how to execute instructions. ;) Everyone knows I love talking about this stuff, but it goes over most people's heads.

Hey, could you discuss the differences between a p4 and a D? Is a D nothing more than 2 p4 stacked on each other? Or is it a "core" issue?

[saphalline]

Geez, you guys never get tired of this stuff, do you?

Ok, onboard memory controller...

The K8 microarchitecture is AMD's latest one, used by all Socket 754, 939, and 940 CPU's - Sempron, Sempron64, Athlon64, Athlon64 FX, Athlon64 X2, and the Opteron family. It does not apply to the Socket A Sempron's, though (just making that clear right now). All K8 CPU's have an onboard memory controller.

The job of the memory controller is basically to translate memory addresses from the abstract (CPU & software side) to the physical (actual locations in RAM), and then to serve up that data. Typically, the memory controller has been housed in the Northbridge chip, but AMD's K8 CPU's changed that. The memory controller is now right on the CPU die (part of the core but not attached to the execution units). This has several advantages. First of all, the memory controller now runs at core speed, which decreases the L2-to-RAM latencies (L2 cache on the CPU to the RAM itself) by a significant amount and overall lets data flow more smoothly and faster.

Secondly, the CPU no longer has to rely on the Northbridge chip, but can communicate directly with the Southbridge chip. The Southbridge chip is a communication hub that deals with all expansion busses, from PATA and SATA to USB to the PCI and PCIe slots. This means there are less "middlemen" between a drive and the CPU, and additionally it all moves at a faster pace (this manifests itself in less CPU usage during transfers of data). This also has the confusing side-effect of eliminating the FSB. On all K8-based CPU's, there is no FSB. There's just the HyperTransport link, which connects the CPU directly to the Southbridge - and the HT link gets it timing from the system clock, which is also directly connected to the CPU now. The HT link is most often confused with the FSB because of the system timing and its relation to OC'ing a K8-based CPU (but more on that later).

Thirdly, because the memory controller is onboard and runs at core speed, the performance dividends increase with core speed. The faster your CPU runs, the faster the memory controller operates, which means less and less latencies are involved when accessing RAM from the CPU. So as the K8 CPU's get faster, and when DDR2 support gets added, we'll see more and more efficiency being squeezed out of the CPU-to-RAM bandwidth. Having the onboard memory controller is one of the main reasons that increasing the speed of a K8 CPU is more important than increasing the L2 cache size or single- vs dual-channel RAM. (It's also closer to being system-on-a-chip but more on that later).

Fourth, and probably very big on AMD's list, is the fact that the onboard memory controller is helping the K8 CPU's pull ahead of Intel's NetBurst2 CPU's. Ironically enough, Intel's own P4 would benefit tremendously from having an onboard memory controller, but instead AMD developed it. At the same time that AMD CPU's are scaling well with faster core speeds, Intel's are scaling worse.

Ummm... ok, now the P4 vs PD?

The Pentium D is an odd name, I think. Instead of calling it the Pentium5, Intel decided that "D" would be a more appropriate suffix for their new dual-core offshoot. Well, it would be if they hadn't already had Celeron D on their menu! :rolleyes: Oh the confusion that must be causing...

Anyway, the Pentium D is nothing more than a dual-core Pentium4. But of course that in itself is a big thing! This dual-core stuff is cool, and both Intel and AMD are already talking about quad-core and multi-core for the future (but more on that later). Dual-core is a neat way of stuffing a dual-processor system into your very own home! The performance increases are quite significant, as demonstrated by Intel when they released the 820 version at only 2.8GHz. But that's 2.8GHz times two, so you can see the faith they have in the dual-core revolution. At 2.8GHz, it won't win any single-core awards (such as gaming performance) but it excels at multi-tasking and in general keeping your computer from ever feeling "slow". No longer are you shackled by opening a program and having to wait for it to load! Now the workload can be shared by two full processors, so that one can load the newly opened program while the other is free to let you continue working in your already opened programs! Or you can go to the other extreme: you can encode a home movie and still be able to use your computer without fear of slowing down the encoding or suffering through a bogged-down system.

The other cool thing about dual-core is that it's actually cheaper than a dual-processor system. Instead of buying two CPU's, a mobo with two sockets, and RAM for each CPU, you can now buy a normal-looking mobo, one CPU (with two full cores inside it), and your normal amount of RAM. It looks just like a classic single CPU system, and that's the idea. You get the extra power without the extra cost or complexity.

AMD's own dual-core CPU's are the Athlon64 X2's, and are very similar to Intel's PD. The X2's have two A64 cores in them. And of course Intel and AMD have their server dual-core CPU's. Take a look in the PC World section for a recent thread on Intel's Xeon DP's. And you can check online for more info about the Opteron Italy.

[neilt]

Nice description Saphalline! After reading that, I asked myself "Why am I buying Intel?" Do you have a rebuttal for Intel single cores? Specifically the 6xx series processors.

[saphalline]

Do you have a rebuttal for Intel single cores?Photoshop and audio/video encoders/decoders. SSE-optimized code and in particular any type of codec (whose code is by definition very predictable) just flies on the P4! It excels at using the same instructions over and over and over again.

[Sempron]

Now the workload can be shared by two full processors, so that one can load the newly opened program while the other is free to let you continue working in your already opened programs! Or you can go to the other extreme: you can encode a home movie and still be able to use your computer without fear of slowing down the encoding or suffering through a bogged-down system.
Will you be able to decide which processor does what?
Example. Tell Processor 1 to work on regular computer usage and tell Processor 2 to work on specific jobs? How does that work? or does it work like that?

Geez, you guys never get tired of this stuff, do you?Thats apparent, Im just glad we have someone with your patience and knowledge to teach us this stuff.

Thanks Saphalline!

[saphalline]

Will you be able to decide which processor does what?Well, not you personally. The OS doles out processor time. And the NT kernel is pretty good at that, all things considered. If one CPU gets full (or even before it goes to 100%) the other CPU will get things to do. It's on a program-by-program basis, though, so you can't have both CPU's working on one program unless that program is SMP capable.

[Sempron]

Ok understood.

Thank you almighty guru saphalline
Ohms Ohms Ohms

[odannyboy000]

Saph, you make p4's sound like gaming crap. amd isn't vastly superior to pentium in my opinion. I have a p4 w/HT tech, and every piece of software right now from games to video editing can be done without the slightest glitch (not 64bit programs though...). I don't think AMD has much over pentium at all. now, if you want to go the overclocking path, then i would CERTAINLY choose AMD, but if you want something that can fly right out of the box choose pentium. you dont even need dual core when you have hyper threading, anyway. its not the same as two processors, more like 1 1/2, but with just that you can save money rather than going amd. thats just my opinion though, lol.

[saphalline]

No, the P4 isn't "gaming crap" but it doesn't perform nearly as well as the Athlon64 per clock cycle, and it doesn't perform quite as well per dollar. But that's just on gaming, and that assumes all else is equal.

But all this may be moot now! Check this (http://www.pcguide.com/vb/showthread.php?p=251632#post251632) out!! :D

[neilt]

Photoshop and audio/video encoders/decoders. SSE-optimized code and in particular any type of codec (whose code is by definition very predictable) just flies on the P4! It excels at using the same instructions over and over and over again.

I have seen these benchmarks, and have been impressed by the way Intel kicks in this area. I guess I'm wondering how this would apply to recording software (like Cakewalk Sonar), where I would be recording a live track, and then recording another track while the previous track is playing, ect. until there were several tracks playing (wav files), while recording, and maybe applying effects at the same time. Would the P4 be better at handling all this processing, or would an AMD system be better at it? Would this type of processing fall into the same category where the P4 does so well?

[jcnoernberg]

I think you've all scared off the "n00b" :)

[saphalline]

Would the P4 be better at handling all this processing, or would an AMD system be better at it? Would this type of processing fall into the same category where the P4 does so well?The P4. This type of code uses the same instructions over and over and over and over again. These types of apps are the very reason Intel invented SSE! And when it comes to SSE, Intel is the best!