Jump to content

AMD Chips vs Intel Chips


Recommended Posts

I've always been a fan of AMD, I like their products and their price accordingly.

However, there are tons of people that choose Intel over AMD. 

 

To start I'll compare my CPU that I've owned for almost 3 years now, to an Intel chip that runs at just about the same speed. Here.

 

AMD Phenom II965 Black Edition Deneb 3.4GHz (OC'd to 3.8 GHz) - $99.99

vs.

Intel i7-4820K Ivy Bridge 3.7GHz (Turbo 3.9 GHz) - $324.99

 

Why is there such a huge price difference if supposedly they run approximately at the same speed?

 

note; I've played on some Intel Core built PCs that are way more powerful than I imagined. What am I missing here?

Share this post


Link to post
Share on other sites

  • Replies 37
  • Created
  • Last Reply

Top Posters In This Topic

Its been a long while since I used an AMD Phenom chip, but looking back at http://www.overclockersclub.com/reviews/intel_corei7_3820/3.htm this review you can see the clock per clock performance is just not there. I have the Phenom 980 and 1090T in that review to show performance from close to that generation. Strictly for gaming unless pushing some of the new games that more effectively use the CPU you will be GPU bound. For rendering or CPU intensive tasks Intel is the faster processor. 

Share this post


Link to post
Share on other sites

Clock speed is just one element of how fast a CPU is. AMD and Intel cpu's have different architectures. Intel has the superior architecture, thus per clock Intel is generally faster.

 

Modern cpu's are far to complicated for me to understand but I can provide an example from the days of old. For example the original Z80 cpu would execute one instruction about every 4 clocks. 

However later revisions of the CPU could do most instructions, in one clock.  These revisions of the architecture were nearly 4 times as fast per clock. This is just one example of how clock can be irrelevant.

 

In modern times you have single core cpu's witch can do more than one instruction per clock. You have some real smart people wizards designing these things and currently it seems Intel has the best people wizards.

 

Not saying AMD is bad, after all I am running a 1090T but per core Intel is definitely faster.

 

Edit: Thanks for the delete ccokeman, that was embarrassing   :unsure:

Edited by Mysion

Share this post


Link to post
Share on other sites

Thanks for the replies, modern CPU architecture is just a mind blowing idea for me, and I'm trying to understand it a little more. 

 

When I went on to think about all this (months ago), I thought it HAD to be some in-depth architecture that I wouldn't entirely understand. 

I suppose I was on the right track, after reading your replies ccokeman & Mysion.

 

I thought GHz was sort of a measurement of clock construction rate.

 

What does GHz actually measure then? To my knowledge Hertz is some sort of electrical measurement - and I'm sure I could figure out more via research, but since we're talking about it already..

Share this post


Link to post
Share on other sites

Thanks for the replies, modern CPU architecture is just a mind blowing idea for me, and I'm trying to understand it a little more. 

 

When I went on to think about all this (months ago), I thought it HAD to be some in-depth architecture that I wouldn't entirely understand. 

I suppose I was on the right track, after reading your replies ccokeman & Mysion.

 

I thought GHz was sort of a measurement of clock construction rate.

 

What does GHz actually measure then? To my knowledge Hertz is some sort of electrical measurement - and I'm sure I could figure out more via research, but since we're talking about it already..

It's still the clock speed, but it's a lesser factor of concern behind architecture complexity/excellence and core count. It's only really useful to determine which CPUs will do better within the same family and generation.

Share this post


Link to post
Share on other sites

Ok I will admit that ipc Intel beats Amd Which is why single threads applications are better performed on Intel while multi threaded applications are better performed on AMD the problem is most programs don't use multithreaded applications much. However one the things that HSA HUMA in kavarie are supposed to do is make increases in ipc efficiency for amd processors. But if I'm not mistaken in order for that to work correctly the programs have to be written in a language that takes advantage of that. So once more programs are written for multithreaded tasks and single threaded program are written in the language needed for this which I believe is open CL or gl can't remember which I believe we will see some changes but for now Intel tends to outperform and because its architecture is more effective for single threaded applications and its ability to handle more ipc

Share this post


Link to post
Share on other sites

In real world terms it takes an overclocked Phenom II X4 or FX-6300/6350 to compete/beat an Intel 13 dual core with hyper threading that does not overclock.

For gaming the I5 (2500K, 3570K, 4670K) outperforms AMD

4670K is 22nm 84 watt (with HD4600 graphics) 1.4/1.6 Billion transistors depending on count metodology

Phenom II X4 965 is 125/140 watt, no graphics, 45nm, 748 million transistors

FX-8350 125+ watt, no graphics, 32 nm, 1.2 Billion transistors

 

Anyway the Phenom II x4 is 4 cores, the FX-8350 is 4 module/8 core, some programs can use all 8 core, some only 4, some one or two.

Most games give the 4670K/3570K 35 TO 70% advantage to Intel, 4670K $240, FX-8350 $180-200.

Generally, with a few specific game exceptions, real world gamers do not notice a performance difference. Overclockers like the FX ability to reach 4.4-5+ GHz given adequate cooling, motherboard and RAM. Intel absolute overclocks have decreased with each die shrink, but so has power consumption (though Haswell VRM control on die and increased graphic performance sort of minimizes Ivy Bridge to Haswell "jump."

Intel also has enough market share software is taking advantage of hyperthreading more than AMDs 1 module/ 2 core design.

 

Phenom II x4 overclocked remains a good gaming chip, though 6 core are preferred. FX Chips are one die with functions removed, ie 8350>8320>6350>4350. Mostly? 8320 can overclock to 8350 levels. 63xx and 43xx are 95 watt, there are 83xx 95 watt variants.

Share this post


Link to post
Share on other sites

  • 2 weeks later...
  • 1 month later...

Ever since my first Intel I retired my AMD's and never looked back.

 

get a video conversion program and run the exact file on an AMD system and then on an Intel system, nuff said.

Share this post


Link to post
Share on other sites

Ever since my first Intel I retired my AMD's and never looked back.

 

get a video conversion program and run the exact file on an AMD system and then on an Intel system, nuff said.

 

Just don't forget that not TOO many years ago the roles were reversed, An Athlon X2 (as in the first dual cores) would stomp a Pentium D with no problem.  The Athlon X2 was a model of efficient design and performance while the  Pentium D was two P4s welded together in crude fashion requiring even OEMs like dell to make huge cooler and even prompted the "BTX" Motherboard/case standard to help with cooling.  

 

Funny how years later it's almost a 100% role reversal.  With AMD making the 125W+ CPUs and Intel relaxing with low power consumption and elegant design.  AMD's brute force method isn't doing much good in the short term, and I don't see it changing in the next couple years, but no one saw the domination of AMD in the mid 2000s coming form years before either.  

 

On a side note one of the fundamental changes in AMD's architecture that hurts it badly is what people call it not having 8 "real" cores.  it's a VERY different approach than Hyper Threading on the intel side. AMD's design has only one FPU (Floating point calculations) per two Integer units on the CPU, so if the code is integer heavy then AMD's CPUs work as what you would consider a true 8 core design, but if the FPU is being loaded down the CPU effectively becomes a quad core.  This steps down the line with the number of FPU units being half the INT units. The older Phenom II X6 CPUs are actually faster clock per clock than the new CPUs because of this architecture change when compared to the first FX-6000 generation. Hyper Threading on the other hand is a way of using the 4 full function CPU cores and when a certain piece of the core is idle feeding it another thread to work on beside whatever the rest of that core is already doing.  So it's not doubling the core and it does NOT rely on software to function but you can get some nice gains with the right work load.  That's a VERY basic run down of how each works and misses a LOT of info.  

 

The only way clock speed is important is when you compare CPUs of the same architecture. 

Share this post


Link to post
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now

×
×
  • Create New...