AMD versus Intel

[Pez D Spencer]

Why is it that a CPU from AMD that has the same performance as one from Intel usually runs at a lower clock speed? From what I understand the naming of AMD CPU's such as 3000+ and so on were put in place to give people an idea of how that CPU perfroms in contrast to an Intel CPU.

So based on this, an Athlon XP 3000+ should perform similarly to a Pentium 4 3.0GHz correct? Why is it that the Athlon XP uses such a lower clock speed than the Intel but performs roughly the same?

I know that the easiest explanation is that the two companies use a different internal architecture, but what is the major difference in the two architectures that cause this?

Thanks.

 

[Capt Caveman]

Search is your friend - http://www.anandtech.com/cpuch...el/showdoc.aspx?i=2748

 

[Pez D Spencer]

Thanks for the link. I searched Google and couldn't find a decent article but I didn't bother searching here.......

 

[coldpower27]

Originally posted by: Pez D Spencer
Why is it that a CPU from AMD that has the same performance as one from Intel usually runs at a lower clock speed? From what I understand the naming of AMD CPU's such as 3000+ and so on were put in place to give people an idea of how that CPU perfroms in contrast to an Intel CPU.

So based on this, an Athlon XP 3000+ should perform similarly to a Pentium 4 3.0GHz correct? Why is it that the Athlon XP uses such a lower clock speed than the Intel but performs roughly the same?

I know that the easiest explanation is that the two companies use a different internal architecture, but what is the major difference in the two architectures that cause this?

Thanks.

Yeah, for quite awhile on the desktop AMD had an architecture which had a lower clock speed and performed better then higher clocked Intel CPU's, currently the situation is reversed on the desktop with Intel having CPU with lower clockspeed outperforming higher clocked AMD CPU's.

 

[pm]

I'm not going to go into an AMD vs. Intel kind of issue - it's a not a company vs. company thing. It's more of a philosophical debate - and I call it "philosophical" because both sides can find data to support their arguements. The philosophy of which is better - high IPC or high clock rate - has been around for as long as there's been pipelined CPU's. The debate has been often titled the "brainiac" (high IPC) vs. "Speed Demon" (high clockrate) and there are numerous articles on the subject.

Here's an article from 1999 in the Microprocessor Report: http://www.mdronline.com/mpr_p...torials/edit13_17.html

You can find many more on the subject by Googling: "speed demon" brainiac microprocessor

Intel interestingly enough has had designs that are solidly in both camps - all of the Itanium family are (thus far) undeniably along the "brainiac" approach, while the Pentium 4 was clearly a "speed demon" approach. The current Core 2 Duo design is, to me anyway, a blend of both.

 

[MarcVenice]

How is this a philosophical debate? It has clearly been won by AMD when Intel released your so called brainiac processor in the form of conroe/core 2 duo and stopped pushing up the clockrates and went for higher IPC. They had 1.8ghz cpu's outperforming their 3ghz p4's who als consumed twice the power ... The fact they've now got 3ghz core 2 duo's is because of finetuned producing techniques and such. I think penryn is going to top out at 3.33ghz, so yeah, speeddemons went extinct.

And with AMD has won I mean they showed that pushing up the clockrates wasn't the way to go.

 

[bryanW1995]

penryn will intro at 3.33 with the QX6950. It will only stagnate at that speed of AMD doesn't deliver the goods with barcelona/phenom.

 

[aigomorla]

Originally posted by: MarcVenice
How is this a philosophical debate? It has clearly been won by AMD when Intel released your so called brainiac processor in the form of conroe/core 2 duo and stopped pushing up the clockrates and went for higher IPC. They had 1.8ghz cpu's outperforming their 3ghz p4's who als consumed twice the power ... The fact they've now got 3ghz core 2 duo's is because of finetuned producing techniques and such. I think penryn is going to top out at 3.33ghz, so yeah, speeddemons went extinct.

And with AMD has won I mean they showed that pushing up the clockrates wasn't the way to go.

correct me if im wrong but wasnt the mobile p3's on laptops earily arch, of c2d?

and your statement about AMD being won when intel released there C2D. Im lost on this statement. Did you mean the mhz war was over? Because i still see a MHZ war going on between the two companies. The C2D is faster per clock, and more efficient in terms of power, unless you count in the 45W versions.

Also, i hate to be a pessimist, but i dont have high hopes for AMD. I dont think there Phenom will be as breakthough as the C2D was. And people are overclocking kents to a modest 3.2ghz with very little difficulty. 3.2ghz Kent will wipe the floor of a 2.0ghz Barcelona.

in terms of power, i think a T2100 would be more efficent in power and performance then anything amd could throw.

So can you explain this statement a little bit more? im kinda lost on this.

 

[MarcVenice]

Well the decrease in heat production from Penryn over Conroe is relatively small and each micro architecture has it's limits. If it's not 3.33ghz then it will be 3.66ghz or those cpu's will top out at. That's my ineducated guess though.

 

[bryanW1995]

I think he meant that amd showed intel that the highest possible clock speed isn't necessarily going to yield the best cpu. Basically, AMD was right. The only problem is, now that intel has embraced this philosophy they are stomping the crap out of AMD, so it looks like AMD won the battle but lost the war. The next salvo is fired tomorrow...

 

[aigomorla]

Originally posted by: MarcVenice
Well the decrease in heat production from Penryn over Conroe is relatively small and each micro architecture has it's limits. If it's not 3.33ghz then it will be 3.66ghz or those cpu's will top out at. That's my ineducated guess though.

ahhh i see...

But if Intel was using there arch, in the mobile P3 platforms. Then how could they have lost the war? The hypertransport was absolutely crap. We all know that, but i think thats because intel couldnt get the chip to a set speed in which we would see a benifit from hyper transport no?

 

[bryanW1995]

Originally posted by: MarcVenice
Well the decrease in heat production from Penryn over Conroe is relatively small and each micro architecture has it's limits. If it's not 3.33ghz then it will be 3.66ghz or those cpu's will top out at. That's my ineducated guess though.

Don't forget they are going in 1/2 steps now, too, so they could get 3.5, 3.66, and 3.83 too.

I heard the other day that intel is going to pull the best cpus for quad cores in the future. If that is true it could explain why the quads are clocking as well as/better than the dual cores. That also gives them a tremendous advantage over amd in the quad core arena imho. I wonder if amd might not be more competitive in dual cores for this generation?

 

[pm]

How is this a philosophical debate? It has clearly been won by AMD when Intel released your so called brainiac processor in the form of conroe/core 2 duo and stopped pushing up the clockrates and went for higher IPC.

It's not a company vs. company competition like you are characterizing it. Intel had been shipping a vastly more "brainiac" design for over a decade - the Itanium 2 tops out at 1.6GHz and beats benchmark scores of designs that are clocked more than twice it's frequency. And it's not my "so called brainiac" idea - this whole concept has been debated in academic and conference forums for literally decades.

 

[dmens]

Originally posted by: pm
It's not a company vs. company competition like you are characterizing it. Intel had been shipping a vastly more "brainiac" design for over a decade - the Itanium 2 tops out at 1.6GHz and beats benchmark scores of designs that are clocked more than twice it's frequency.

What'd you think about power 6?

 

[coldpower27]

Originally posted by: bryanW1995

Originally posted by: MarcVenice
Well the decrease in heat production from Penryn over Conroe is relatively small and each micro architecture has it's limits. If it's not 3.33ghz then it will be 3.66ghz or those cpu's will top out at. That's my ineducated guess though.

Don't forget they are going in 1/2 steps now, too, so they could get 3.5, 3.66, and 3.83 too.

I heard the other day that intel is going to pull the best cpus for quad cores in the future. If that is true it could explain why the quads are clocking as well as/better than the dual cores. That also gives them a tremendous advantage over amd in the quad core arena imho. I wonder if amd might not be more competitive in dual cores for this generation?

Well that depends for Dual Core's AMD has the option of disabling Quad Core, or having a Native Dual Core die, which is the approach they are most likely taking, no matter how good their process is there is no way holding all other factors constant, that they could clock their Quads as high as Dual, Intel and AMD are both only limiting the clockspeed of their Dual Core so they can have Quad as the high end product and not have the situation, where in some cases a Dual outperforms a Quad.

 

[dmens]

Originally posted by: MarcVenice
How is this a philosophical debate? It has clearly been won by AMD when Intel released your so called brainiac processor in the form of conroe/core 2 duo and stopped pushing up the clockrates and went for higher IPC. They had 1.8ghz cpu's outperforming their 3ghz p4's who als consumed twice the power ... The fact they've now got 3ghz core 2 duo's is because of finetuned producing techniques and such. I think penryn is going to top out at 3.33ghz, so yeah, speeddemons went extinct.

And with AMD has won I mean they showed that pushing up the clockrates wasn't the way to go.

That is ridiculous. The team that can hyperpipeline and manage complexity and power wins the day. You'd be foolish to think clock speed is no longer a consideration in CPU design.

 

[ViRGE]

Originally posted by: bryanW1995

Originally posted by: MarcVenice
Well the decrease in heat production from Penryn over Conroe is relatively small and each micro architecture has it's limits. If it's not 3.33ghz then it will be 3.66ghz or those cpu's will top out at. That's my ineducated guess though.

Don't forget they are going in 1/2 steps now, too, so they could get 3.5, 3.66, and 3.83 too.

I heard the other day that intel is going to pull the best cpus for quad cores in the future. If that is true it could explain why the quads are clocking as well as/better than the dual cores. That also gives them a tremendous advantage over amd in the quad core arena imho. I wonder if amd might not be more competitive in dual cores for this generation?

Wait, Intel is moving to half-multipliers? Do the current 3x-series chipsets even support this?

 

[bryanW1995]

I would assume that they will for penryn, perhaps with a bios update. they have already announced that there will be penryn server chips at 7.5 & 8.5.

 

[ViRGE]

Originally posted by: bryanW1995
I would assume that they will for penryn, perhaps with a bios update. they have already announced that there will be penryn server chips at 7.5 & 8.5.

Sweet. 400x8.5 here I come.

 

[MarcVenice]

I never said clockspeed is no longer important, it is no longer the key to producing the fastest cpu's out there though. C2D is at 3ghz or so now, where the p4's were years ago. It's obvious they stopped pushing up the clockrates just for the sake of pushing up the clockrates, to make the public think they were faster then any other cpu.

Neither did I say Intel lost the war Intel did get bitchslapped when AMD released their athlon 64 cpu's though, and they soon followed suit with C2D for desktops and xeons for servers. And when they did, Intel hit back hard. I'm no fanboy of either company btw. Competition is good in a market like this one.