386DX 40 MHz 😀

I prefer to keep away from AMD stuff for retro. From Pentium times they often positioned hardware as cheaper, so the higher chance to get lower quality.

K7 and P6 are not the same or similar in any way, Ok so maybe they have some common ground. But if your looking at Die charts and pipeline charts a lot of cpus from the same time frame will look a lot alike. That's mostly marketing.

Jade Falcon wrote:

K7 and P6 are not the same or similar in any way, Ok so maybe they have some common ground. But if your looking at Die charts and pipeline charts a lot of cpus from the same time frame will look a lot alike. That's mostly marketing.

Erm, they are similar in a number of ways which I pointed out above, including overall structure, pipeline length, choice of pipeline stages, but also latency, throughput etc.
Most other CPUs (x86 or non-x86) from that era are nowhere near as similar, not even in marketing material.
If you're going to argue, at least argue the facts I've already presented above, instead of some blanket dismissal without any arguments other than 'mostly marketing', which isn't even true. I mean, drag up charts of a contemporary PowerPC for example, and you'll see it it's far different from P6. In fact, Apple specifically marketed those differences (such as shorter pipeline, less speculative execution), as the 'Megahertz myth': http://www.macworld.com/article/1016953/megahertz.html
http://www.macworld.com/article/1017000/myth.html
http://www.macworld.com/article/1017080/myth.html

That's PPC, of course its going to be different.

Are you saying if pipe lines are similar the cpus are similar ?
If that was the case netburst and Nehalem (and newer i7s) are about the same. Oh and the pipelines of the P3 and core2 are very similar. If that's the case the p4 and i7 are similar, very similar and the same for for the P3 and core2.

But it doesn't matter what we say, your not going to change your mind, your just arguing for the sake of arguing here.

Good day.

Jade Falcon wrote:

That's PPC, of course its going to be different.

What other x86 CPUs are there (and why would we limit ourselves to x86 in this context? Moving goalposts again)? P6 and K7 were the only contemporary x86 at the time (aside from some weird VIA or Transmeta stuff for example).
Other than that you'd only have older x86 architectures, and do we really need to argue that P5 and P6 are considerably different?

Jade Falcon wrote:

Are you saying if pipe lines are similar the cpus are similar ?

For some values of 'similar' perhaps.

Jade Falcon wrote:

If that was the case netburst and Nehalem (and newer i7s) are about the same. Oh and the pipelines of the P3 and core2 are very similar. If that's the case the p4 and i7 are similar, very similar and the same for for the P3 and core2.

Not for this value of 'similar' at least. First you'll have to explain how in the heck Netburst and Nehalem are even remotely similar (again, you're making blanket statements, no argumentation whatsoever).
Nehalem has a 12-14 stage pipeline, where Netburst is more like 28-31 stages. So by that logic, 'similar pipelines means similar CPUs' doesn't apply in the first place, since the pipelines aren't similar.

P3 and Core2 are reasonably similar, however, a major difference is that the Core2 has an extended pipeline. It has been argued in the past that Core and Core2 were basically a 'revamped Pentium 3', eg http://www.zdnet.com/article/inside-intels-co … e-architecture/:

the Core architecture is mostly a mix of Pentium M (Banias and Dothan) ideas with those from the Pentium Pro/P6. It's a major departure from Pentium4's NetBurst design.

Of course, that is somewhat oversimplified, but in general, Core/Core2 are indeed closer to P6 than Netburst was.

Jade Falcon wrote:

But it doesn't matter what we say, your not going to change your mind, your just arguing for the sake of arguing here.

I'm arguing with facts and logic. If you want me to change my mind, you'll have to prove my facts and logic wrong first. You're not doing any of that so far. You're only making blanket statements and using flawed logic, which is almost too easy to poke holes into with just a few simple facts.

The facts were already posted and you did not care.

Jade Falcon wrote:

The facts were already posted and you did not care.

What facts? You claim that Nehalem is similar to Netburst because of similar pipeline length, while there's a huge difference in pipeline length.
I pointed that out, and once again you fail to answer to that.

Scali wrote:

Jade Falcon wrote:

The facts were already posted and you did not care.

What facts? You claim that Nehalem is similar to Netburst because of similar pipeline length, while there's a huge difference in pipeline length.
I pointed that out, and once again you fail to answer to that.

We are talking about p6 and k7 here, not netburst and Nehalem. and I never said anything about the pipeline length being the same or similar with netburst and Nehalem, but there are a few Nehalem cpus with 20 steps and the same goes for the Northwood p4.
going by your stamens they are very similar, oh and they both have hyperthreading 🤣

Jade Falcon wrote:

We are talking about p6 and k7 here, not netburst and Nehalem.

Say what now?

Jade Falcon wrote:

Are you saying if pipe lines are similar the cpus are similar ?
If that was the case netburst and Nehalem (and newer i7s) are about the same. Oh and the pipelines of the P3 and core2 are very similar. If that's the case the p4 and i7 are similar, very similar and the same for for the P3 and core2.

You still haven't explained that.

As for P6 and K7, it was already posted that the pipelines of both are 10 stages long. So at least that part does not require discussion. It's similar, it's more than similar, it's equal.
Which leaves the other things I posted earlier as similarities, which nobody has responded to yet.

Edit:

but there are a few Nehalem cpus with 20 steps

Erm, no there aren't.
Core2: estimated 12-14 stages
Nehalem and beyond: estimated between 14-19 stages worst-case, depending on the use-case.
P4: 20 stages best-case (not including the decoding of x86 to trace-cache, so you are skipping a number of stages there).
See also: http://www.agner.org/optimize/microarchitecture.pdf
And: http://www.hardwaresecrets.com/inside-pentium … architecture/2/

Scali wrote:

You still haven't explained that.

Ok I'll bight and keep jerking your chain abit lounger, you been fun anyway. 😎

Ok both Nehalem and northwood have 20 steps pipeline, both have hyperthreading, both run hot and both Nehalem and netburst have a max TPD of about 130w
They use quad pumped buss's as well and are both made by Intel.

And way I grow tiered of you, I got a few things that need done, Have fun.

See my edit above, your figures and understanding of Nehalem and Willamette pipeline are wrong (you forgot to factor in trace cache of P4, among other things).

Doesn't matter, I was just spouting off nonsice and you took the bait. Rare moment for me. I tend not to do this. 😐
Anyway proves that you like to argue 🤣

If you want to do this again shoot me a PM 😊

EDIT:
And it proves that I'm to lazy right now to even try to prove a point...

Jade Falcon wrote:

Anyway proves that you like to argue 🤣

No, it proves that I care about providing people with correct information, so that they may learn something.
I don't get why people spread misinformation, especially when they do it deliberately, as you claim (unless you are just covering up the fact that you were wrong, which is no less excusable given how aggressively you were pushing your 'facts', without bothering to verify them first).