Major difference between AMD and Intel ?

[Organizer]

I'm confused. People are always getting AMD processors instead of Intel ones. The highest AMD I have seen was 2.4 Ghz while Intel has 3.6 at max I think. Is AMD more powerfull or something?
Anyone care to explain?

(I have a AMD 1700+ 1.4 ghz I think)

 

[Dead-Inside]

They count in different ways. AMD is essentially better then Intel for gaming purposes. The Intel HyperThreading makes Intel more suited for running several programs while AMD is, well, "the gaming processor" which runs one a lot faster. Well not *a lot* but, noticbly.

Anyway, that might be a bit blurry but basically, AMD launch their processors with a "3000+" or "3200+" or something tag because that's their equivilent to Intel. Hope that helped.

 

[Organizer]

Sweet :bounce:

 

[bosox188]

Yeah, see a 3000+ equals a P4 3.0ghz, a 3200+ equals a P4 3.2ghz. Although your 1700+ I'd say is better than a P4 1.7ghz because those were the early Pentium 4's and they weren't that good. And not to go off subject, but maybe you should think about upgrading.

 

[TheHitman5]

Yeah, see a 3000+ equals a P4 3.0ghz, a 3200+ equals a P4 3.2ghz. Although your 1700+ I'd say is better than a P4 1.7ghz because those were the early Pentium 4's and they weren't that good. And not to go off subject, but maybe you should think about upgrading.

Nope.

I'll get into the ratings later, basicly, AMD does more operations per second than intel, so it doesnt need a super high frequency rating. The Athlon XP line started getting a little crazy with thier ratings though. For example, an Athlon XP 3200 (2.2ghz) "matches" an Intel P4 3.2ghz, but the 3.2 intel totaly trashed the AMD. The 3200 was eq. to a 3.0C or 2.8C for the most part. The Athlon64 ( 754 ) is on about the same level as the P4C's but the 939's are extremely fast

 

[bosox188]

Yes well it's not really exact. Like a 2500+ is a bit better maybe than a P4 2.53ghz because the 2.53ghz is a P4B. But the P4C's, especially anything 2.8ghz or higher is better than the XP that's supposed to match it.

 

[riTuaL]

Wait for an answer from Asus, he is the Hardware God.

 

[Asus]

Wait for an answer from Asus, he is the Hardware God.

haha
/me decides not to post and walks away.


/me walks back
Sorry for the long read but here's the history from my point of view.
Starting with the Athlon XP and the Pentium 4 the differences between how AMD and Intel CPUs worked multiplied. Before that AMD and Intel chips were around the same clock speed and performed pretty similar as well. It should be known that AMD actually beat Intel to the 1GHz point with their original Athlon. You will see later why that matters.

The Athlon XP allowed AMD to do more work per clock while they continued to increase the freqency. The Pentium 4 changed how Intel's CPUs worked, to do less per clock which allowed them to increase the freqency quicker. They lost the race to 1Ghz, they didn't want to loose the race to 2Ghz and beyond. IMO Their main goal wasn't based on performance but numbers by this time.

Athlon XP's rating system is based on the clock speed the original Athlon would needed to be clocked at to have the same performance. The original Athlon would have to be clocked at 1.7Ghz for it to match performance with the Athlon XP 1700+ (1.47Ghz), for example.

As Intel increased it's bus speeds and frequency it put more pressure on AMD. Finally, with the 800FSB models Intel took the overall performance lead compared to AMD's Athlon XP models. AMD added L2 cache in an attempt to keep up and last until their new chip made it to market. By that time the rating system for the Athlon XP's was pretty complex. AMD dropped prices and they then matched the performance of comparably priced Intel 800FSB CPUs.

AMD finally released the Athlon 64/Opteron CPUs and they fixed most everything the Athlon XP lacked (Memory bandwidth, SSE2 and soon to be SSE3 support, A/V encoding performance etc) and improved apon everthing the Athlon XP did well (Games, Office apps, programing, scientific math and AI). Plus they improved the entire platform by putting the memory controller right onboard the CPU instead of having it a mile away on the board. That also cleared up any bottlenecks on the bus that connects the rest of the system to the CPU. This is a big reason why Opteron makes an excelent multiCPU platform with it's Hypertransport connecting each CPU to each other and other CPU's own memory controller.

Intel released it's long awaited 'saviour', the Prescott core, which turned out to actually perform worse in many cases than the established Pentium 4 Northwood core. Again this was because Intel wanted to hit another MHz mark, 4Ghz. They had to reduce the core's performance by adding pipelines to do so and they slapped a 1MB L2 cache on top of it in an attempt to cover up any performance loss. There is a flaw in Prescott, which is to believed to be a design flaw and not a proccess issue, that makes it give off a lot of heat from the high power consumption.

 

[riTuaL]

see what I mean?

 

[Asus]

By the way, the reason the above doesn't matter to the average consumer is because they buy a product, a PC. A dell, a HP or Gateway. They don't care what is under the hood. Besides, the average consumer doesn't need a 3GHz chip. But what does make me itch is that they buy Celerons...
One reason Intel is the #1 CPU maker is because they sell volume to the big OEMs (Dell, HP, Gateway etc) and generally at a discounted price.
Also they make enough of their chips to meet the demand for Desktop PCs world wide. AMD does not.

But most who do care, care because they are knowledgeable and they know they have a choice.

 

[Ruuustar]

Presumably, Intel have been giving their chips deeper and deeper pipelines (Prescott) in an attemp to reduce clock cycles and ramp up clock frequencies. But this means less and less work is being done in each clock cycle, and also increases branch penalties and feed forward issues.

 

[Gajdycz]

all threas have been owned by asus, he needs to save that and paste it in any topic, regardless of subject

 

[Cyanide]

Yes, Intel does have a much larger pipeline in its processors. I believe it's 20 segments actually. Where the AthlonXPs have a 7 segment pipeline. See, all modern CPUs are built to try to anticipate what the next operation is going to be and then start on it before it's actually issued. When the operation is predicted correctly this gives the processor a head start and increases overall performance. However, when the next operation is predicted incorrectly, the pipeline has to be flushed clean so the chip can start over again. The prediction is usually determined to be correct or incorrect about 3/4 of the way through the pipeline. Obviously, 3/4 of 20 is much bigger than 3/4 of 7. So it takes longer for the P4 to recover from an incorrect prediction, but it can have higher clock speeds because of the longer pipeline and thus performs better in predictable applications like games, multimedia stuff, and encoding/decoding operations.

For the most part, P4s perform better in the applications that gamers use. Athlons perform better in office apps like CAD and other things that are not as predictable. However, it should be noted that AMD almost always has a better price/performance ratio than Intel, and the differences between comparable chips are not noticable to most users.

Here's a benchmark that compares the P4 3.2ghz to the AthlonXP 3200+. I suggest you take the time to read it all if you're really interested in knowing what's best for your purposes.
http://www.tomshardware.com/cpu/20030623/index.html

You should also know that Athlon64s are not operating at their best performance levels right now. Once more software is written for 64-bit processors, the Athlon64 will out perform any 32 bit processor by leaps and bounds. This is because it's capable of working on 64 bits of data at a time, rather than 32. Which creates and exponential increase in performance. 64-bit processors are also capable of managing more than 4gigs of memory, which seems like a hell of a lot now, but by the time you're in need of a new processor upgrade, it will be a little bit above average. The next version of windows (currently code-named Longhorn) will be 64-bit and is being built with Athlon64s in mind. Once the operating system is working in 64-bit, things will get much better for people who decided to go with an Athlon64.

Also, you might want to know that CPUs are getting so fast now that unless you can double your clock speed, it's not really worth upgrading. At this point, even a doubled clock speed (1.5ghz to 3ghz) only yeilds about a 20% performance increase. It's noticable, but might not be worth hundreds of dollars, depending on how much money you have. That is of course assuming you're staying with the same chip. Moving from something like a celeron to a P4 with double the clock speed is going to give more than a 20% peformance gain.

Also, companies like Dell only use P4s because Intel has a contract with Dell to pay for their advertising as long as Dell puts the Intel logo at the end of their commercials. AMD would definately like for Dell to use their processors also, but Dell won't do that because of that contract.

 

[bosox188]

K so really you can't say that 2500+ really means 2.5ghz or that 3200+ really means 3.2ghz. I did know that once Intel made the Pentium 4 C's they were better than the Athlon XP's that were supposed to match them. It really does get complicated, but as long as you understand that the clock speed doesn't matter and that there are a bunch of other things that affect performance besides clock speeds you can get a good idea.

 

[Asus]

I agree with everything you said except perhaps this.

For the most part, P4s perform better in the applications that gamers use. Athlons perform better in office apps like CAD and other things that are not as predictable. However, it should be noted that AMD almost always has a better price/performance ratio than Intel, and the differences between comparable chips are not noticable to most users.

Athlon CPU generally perform more favorable in tasks that are based on user input, like gaming and like applications.
Intel CPUs generally perform more favorable in tasks that are not based on an unknown input.
Obviously there are always possible exceptions for both CPUs.

Here's a benchmark that compares the P4 3.2ghz to the AthlonXP 3200+. I suggest you take the time to read it all if you're really interested in knowing what's best for your purposes.
http://www.tomshardware.com/cpu/20030623/index.html

I only brought up Athlon XP to show the past and give an idea of direction for AMD. At 149$ for an Athlon 64 2800+, there is hardly any reason someone would be looking at the Athlon XP series. Besides, their prices match the competition with compariable performing Pentium 4 CPUs.

 

[Cyanide]

Athlon CPU generally perform more favorable in tasks that are based on user input, like gaming and like applications.
Intel CPUs generally perform more favorable in tasks that are not based on an unknown input.
Obviously there are always possible exceptions for both CPUs.

Well, that's not evidenced by the benchmarks I linked to. Nor any other benches I've seen. In fact, you're the first person I've ever heard say that. Anyway, not going to argue with you. Whoever is buying a processor can make their own decision.

I personally own 2 AthlonXP based systems, most of the systems I've built in the past were AMD based, and the next system I build is going to be an Athlon64. AMD's price/performance ratio is just better. So it's not like I'm an Intel fanboy or something. I'm just reporting what all the evidence I've seen has shown me.

 

[Asus]

I made that comment too generalized and it may be out of place.

But look at this chart.
Look at the things that the Athlon leads in and Intel leads in.

Rendering, Media Encoding, and Image Editing. Do these have any user input as far as during the proccess? Not really. The user tells the PC what it wants to do and then it takes a certain amount of time for it to be completed.

Does Gaming, Buisness Applications, Content Creation, and Science wait based on user or results input? Yeah.
It's not a hard line but there is grouping of like tasks there.

Software Development, Data Compression and Multitasking results are so for different reasons.

 

[Cyanide]

Well, the benches you posted and the benches I posted seem to directly contradict each other. So there's something wrong somewhere.

Anyhow, user input really isn't a factor. Remember that the time scale's you're talking about are in nano-seconds. Human input just doesn't happen often enough or fast enough to really make a difference. Especially in a benchmark test where everything is automated anyway. The mouse and keyboard are polled at a constant speed, so the same number of clock cycles are used for those in each second, no matter what the user is doing. On a chip that's capable of performing well over 3 billion operations per second (~3ghz), human interaction makes little difference because it just doesn't happen fast enough. The real difference is the type of algorithms involved.

 

[Kristafon]

Whats the equivalent of a AMD Athalon 64 3000+?

 

[Asus]

I was thinking more along the lines of pipelines being flushed if it guessed wrong because of user input. Not only that but the way they use their L1 and L2 cache is different. Pentium 4 stores a copy of their L1 in the L2 which takes up space, while the Athlon does not.

And the benchmarks you linked to are quite old and only show the older battleground, Athlon XP VS Pentium 4. Also if you try to compare the Athlon XP scores between the sites, the exact program and settings used can make a world of difference.

 

[Dead-Inside]

Wait for an answer from Asus, he is the Hardware God.

He's not the only one who knows his stuff. I decided to post a short answer. I hate dragging it out.

 

[Gajdycz]

what about the difference betweem the fx's and the basic 64's, is the performance that divided

 

[Subatomic]

The FX's have a 1MB L2 cache (64's have 512KB), and are generally faster than the regular 64's. They also have the FSB and multiplier unlocked for more possibilities when overclocking.