Help on solving a discussion about Pentium4

[imported_Aslund]

I had a discussion with a person. I must say that I know a bit about things inside a CPU, but he said things I had never heard about before and I would like your help to confirm if it is true.

First he said that the longer pipelines on Pentium4 are good because it helps on stalls in the CPU.
The reason for this is, that the longer pipelines on Pentium4 gives time to check and correct any problems on the way. Therefore has Pentium4 longer pipes, he says.

Also that there are a big difference on Athlon and Pentium4. He says that Athlon runs on a newer and smarter technology, and therefore can Athlon manage to have high performance on low speed.

He claims that he had seen a test of Pentium4, where it was mad to fail and then see if the pipes could solve the problem before the number of steps has run out.

Also that there is a very big difference on Prescott in socket 478 and socket 775, so much difference have I not seen.

Kind regards

Sebastian Aslund

 

[Sunbird]

Where to begin?

First he said that the longer pipelines on Pentium4 are good because it helps on stalls in the CPU.
The reason for this is, that the longer pipelines on Pentium4 gives time to check and correct any problems on the way. Therefore has Pentium4 longer pipes, he says.

The longer pipelines enables the CPU to run at a higher MHz speed, but at a penalty, if there is a problem (misprediction?) anywhere in the pipeline, it has to start all over again.

Also that there are a big difference on Athlon and Pentium4. He says that Athlon runs on a newer and smarter technology, and therefore can Athlon manage to have high performance on low speed.

Not newer, just smarter designed technology (Athlon 64 that is, and Athlon XP vs 400FSB P4).

He claims that he had seen a test of Pentium4, where it was mad to fail and then see if the pipes could solve the problem before the number of steps has run out.

People easely claim to see aliens/UFO's/god and many other things, not necesarily true though.

Also that there is a very big difference on Prescott in socket 478 and socket 775, so much difference have I not seen.

The big difference is........ they have a different socket! Thats about it.

 

[Frown66]

quote:
Also that there is a very big difference on Prescott in socket 478 and socket 775, so much difference have I not seen.



The big difference is........ they have a different socket! Thats about it.

True. Also, to my understanding at least, the socket 775 prescott's generally run a tad cooler as well. I could be wrong.

 

[hippotautamus]

Prescotts...running cooler...

Perhaps if you saw metric measurements lol

 

[Jeff7181]

Have him read these articles... in this order...

Pipelining Part 1
Pipelining Part 2
Multithreading, Superthreading, Hyperthreading
Prescott Architecture
AMD's Hammer (K8) Architecture

In case he won't take the time to read them... he's full of shit. He doesn't know what he's talking about... he's just feeding you a line, hoping he sounds convincing enough that you'll believe him.

*EDIT* For a long time I knew that a long pipeline paid a big penalty when the pipeline stalled, but I didn't know why. These Pipelining articles explain exactly why. The others build on the info given in the Pipelining ones and discuss the differences in architectures.

**EDIT** To respond directly to what he said though...

First he said that the longer pipelines on Pentium4 are good because it helps on stalls in the CPU.
The reason for this is, that the longer pipelines on Pentium4 gives time to check and correct any problems on the way. Therefore has Pentium4 longer pipes, he says.

That's backwards. Longer pipelines suffer more from pipeline stalls. Pipelines don't do any checking or correcting.

Also that there are a big difference on Athlon and Pentium4. He says that Athlon runs on a newer and smarter technology, and therefore can Athlon manage to have high performance on low speed.

True, there is a big difference between the Athlon 64 and Pentium 4 cores. But it has nothing to do with a newer or smarter technology. It's just two different ways to attain similar results.

He claims that he had seen a test of Pentium4, where it was mad to fail and then see if the pipes could solve the problem before the number of steps has run out.

Pipelines don't "solve problems." If he actually did read about a test like that, it sounds more like he's talking about the branch predictor, not the pipeline. But either way... there are no steps to run out of... he obviously doesn't have a good grasp on what's going on... refer him to the pipelining articles at the very least.

Also that there is a very big difference on Prescott in socket 478 and socket 775, so much difference have I not seen.

The difference is not that big. It's the same core, on the same process, in a difference package.

 

[Lithan]

The person you had this discussion with may be clinically retarded. He's also a godawful liar. Kick him in the nuts.

 

[Duvie]

Originally posted by: Lithan
The person you had this discussion with may be clinically retarded. He's also a godawful liar. Kick him in the nuts.

LMAO!!!!!

I also agree with Jeff...read those articles....

 

[imported_Kensai]

Hmm. I think I just lost 10 I.Q. points reading this, but gained 20 reading those links.

 

[Jeff7181]

Originally posted by: Kensai
Hmm. I think I just lost 10 I.Q. points reading this, but gained 20 reading those links.

Aren't they great? I LOVE those articles. I've probably read them each 5 times.

 

[imported_Aslund]

Thanks for confirming that what I knew was correct.
But the problem is that I am a crew member on a site and the person who said this was my boss. Therefore I was carefull about confronting him and I would be 110% sure that my knowledge is the correct, so I made this thread.

But my task on this site is to write about CPU architecture, so I hope he will give up his idee and learn from them.

 

[Zebo]

Point that dipshit to this thread if you're afraid to confront him, we arn't.

Originally posted by: Lithan
The person you had this discussion with may be clinically retarded. He's also a godawful liar. Kick him in the nuts.

LMAO!!

 

[Rankor]

What Jeff7181 said. If poopface said that with a knowledgeable tone, I'd walk away laughing.

 

[stevty2889]

Being the boss generaly means they don't really know how anything works, they just want it done..hope he doesn't force you to put out horribly incorrect information based on his lack on knowledge..