AMD CPU naming convention xxxx+ is it still valid?

[gramboh]

From memory, AMD started using the + naming convetion (3000+, 4200+ etc.) back in the Athlon XP days because their CPUs were more powerful clock for clock with Intel. Everyone bought CPUs based on MHz because that was the number that mattered most when arcitectures were similar.

With Core 2 Duo out, Intel is now better clock for clock, so does it make sense for AMD to be releasing 5000+ 5200+ etc? I know for marketing and lack of confusion they will continue to do it, but the number is meaningless now, who cares about a 5000MHz P4? For that matter, is a P4 @ 5GHz faster than an X2 5000+? Hehe.

 

[cmdrdredd]

The Athlon64 naming scheme came from comparing them to an AthlonXP CPU if I remember correctly. So an A64 3000+ was roughly equivelant to a AthlonXP at 3Ghz.

Does this sound right to anyone?

 

[Maximilian]

Originally posted by: cmdrdredd
The Athlon64 naming scheme came from comparing them to an AthlonXP CPU if I remember correctly. So an A64 3000+ was roughly equivelant to a AthlonXP at 3Ghz.

Does this sound right to anyone?

Wouldnt a 3ghz AXP maul an A64 3000+? I think it would.

The performance ratings from AMD havent been valid since the end of an athlon XP, most compared them with the P4 and assumed the AXP 3200+ is like a P4 at 3200mhz, but actually i heard these ratings were based on equivilent performance of amds old thunderbird core athlons. So an AXP 3200+ is the equivilent of a 3200mhz thunderbird cpu, if such a beast were to exist. Makes more sense since the P4 3.2 whooped the entite AXP line.

So, no the performance numbers havent been valid for some time

 

[TheRyuu]

Their just model #'s now. They stopped being a naming convention a LONG time ago.

 

[Aluvus]

Originally posted by: cmdrdredd
The Athlon64 naming scheme came from comparing them to an AthlonXP CPU if I remember correctly. So an A64 3000+ was roughly equivelant to a AthlonXP at 3Ghz.

Does this sound right to anyone?

Officially, they are a comparison to the original (no suffix) Athlons.

In practice, they are a comparison to Pentium competition.

 

[cmdrdredd]

Originally posted by: Aluvus

Originally posted by: cmdrdredd
The Athlon64 naming scheme came from comparing them to an AthlonXP CPU if I remember correctly. So an A64 3000+ was roughly equivelant to a AthlonXP at 3Ghz.

Does this sound right to anyone?

Officially, they are a comparison to the original (no suffix) Athlons.

In practice, they are a comparison to Pentium competition.

In reality they suck compared to a C2D.

Seriously tho I took a guess cause I remember some AMD CPU was supposed to be compared to an older AMD CPU.

 

[broly8877]

Yeah, the PR was in comparison to a tbird @ 1GHz.

So, for example, a tbird would need to be clocked at 2.1GHz+ to match a 1.73GHz Palomino (Athlon XP 2100+).

 

[MrUniq]

I see..looks like an old naming convention that has more served as model numbers now.

 

[Goi]

Originally posted by: broly8877
Yeah, the PR was in comparison to a tbird @ 1GHz.

So, for example, a tbird would need to be clocked at 2.1GHz+ to match a 1.73GHz Palomino (Athlon XP 2100+).

Yup, that's the story I've been told, but like wizboy11 mentioned, they stopped being relevant a long time ago, since nobody builds 3GHz Tbirds, and also this would mean an AXP, Sempron and A64 3000+ would perform the same, which is false.

 

[jiffylube1024]

Originally posted by: Aluvus

Originally posted by: cmdrdredd
The Athlon64 naming scheme came from comparing them to an AthlonXP CPU if I remember correctly. So an A64 3000+ was roughly equivelant to a AthlonXP at 3Ghz.

Does this sound right to anyone?

Officially, they are a comparison to the original (no suffix) Athlons.

In practice, they are a comparison to Pentium competition.

That was just the party line. Unofficially (and effectively) it served as a comparison to the Pentium 4/Pentium D.

But yeah, model numbers are worthless for comparing performance to Conroe now. Completely different architecture.

Instead, compare clockspeed and give the Conroe about a 5-10% advantage per clock (ie 2.4 GHz conroe 5-10% faster than 2.4 GHz X2). Considering Conroes overclock to 3+ GHz, AMD is a little behind...

 

[myocardia]

The model numbers served as a very good comparison tool for the Pentiums, at least up until the A64 X2 dual-cores came out. AMD adding 600 to their A64 3200 was a real laugh, at least to me, just because it was a dual-core. It should definitely have been called the X2 3200, since that's exactly what it was-- two 3200 cores, NOT two 3800 cores. You have no idea how many people used to go to the forums, on the AMD website, and ask "Why does my X2 3800 feel/seem so much slower than the single-core 3800 that it replaced?"

And, for comparing a dual-core AMD, just multiply it's "number" by .45, to see how fast it is compared to a C2D. For instance, an X2 5000 is almost exactly the same speed as a 2.25 Ghz C2D. Lastly, the C2D's are more than 5-10% faster, clock for clock, than an A64. They are 10-20% faster, depending on the application that's using them.

 

[Gikaseixas]

Originally posted by: cmdrdredd
The Athlon64 naming scheme came from comparing them to an AthlonXP CPU if I remember correctly. So an A64 3000+ was roughly equivelant to a AthlonXP at 3Ghz.

Does this sound right to anyone?

Not at all. Correct me if am wrong but what AMD did was try to match the performance of the A64 3000+ to that of a Pentium 4 3.0 since the XP 3000 failed to achieve that. AMD released the XP 3200 but that chip too came short of it's goal and then the K8 was born.
A XP3000 running at 3.0 ghz would destroy a A64 3000+ @ stock speeds

 

[customcoms]

Yeah but any AMD processor and the C2D at 3ghz+ is really fast (obviously the C2D is faster). These naming schmes mean JS really, you have to look at the benchies. Comparing an AXP to an A64 is like comparing an X2 to a C2D; theirs a 20% (20-30% in the AMD 1st case) speed increase clock for clock in the newer architecture.

 

[Gikaseixas]

I just hope AMD can at least match the C2D overclocking capabilities when they introduce their 65nm chips.

 

[harpoon84]

Originally posted by: Gikaseixas
I just hope AMD can at least match the C2D overclocking capabilities when they introduce their 65nm chips.

You mean 80 - 90% overclocks ala E6300?

Hmm, that would mean 3.6 - 3.8GHz from an X2 3800+.

If that's what you're after, good luck to ya. I think you're gonna need it.

Anyhow, back on topic, I think AMD PR system should be totally scrapped, it's long past it's use by date. The P4 ceased to use the clockspeed rating ages ago, so that can't be used as an excuse anymore. AMD should either get back to using the real clockspeed or just a model number like everything else on the market at the moment.

I'm not sure how they calculated the dual core ratings anyway, so a dual core is apparently worth '600+' performance points over a single core. Right...

With dualcore, it's either a LOT faster (with apps that can take advantage, or multitasking), or not at all (single threaded apps).

 

[Some1ne]

The model numbers served as a very good comparison tool for the Pentiums, at least up until the A64 X2 dual-cores came out. AMD adding 600 to their A64 3200 was a real laugh, at least to me, just because it was a dual-core.

Why? Having a second core can come close to doubling performance under some situations, so if they really wanted to, AMD could have even marketed the 3800+ as a 6400+. Only bumping the performance rating 600 points for doubling the core count seems conservative, if anything, since basically the second core is only being counted as 3/16th's of a core for the purposes of determining the rating. Basically, it's like AMD is saying that having a second core will only boost performance by < 20% on average over a comparably clocked single core setup.

 

[harpoon84]

Originally posted by: Some1ne

The model numbers served as a very good comparison tool for the Pentiums, at least up until the A64 X2 dual-cores came out. AMD adding 600 to their A64 3200 was a real laugh, at least to me, just because it was a dual-core.

Why? Having a second core can come close to doubling performance under some situations, so if they really wanted to, AMD could have even marketed the 3800+ as a 6400+. Only bumping the performance rating 600 points for doubling the core count seems conservative, if anything, since basically the second core is only being counted as 3/16th's of a core for the purposes of determining the rating. Basically, it's like AMD is saying that having a second core will only boost performance by < 20% on average over a comparably clocked single core setup.

You can't hide the fact that with the advent of dual core the rating system has become even more ridiculous than before.

So basically a '3800+ X2' will perform like a 3.8GHz Thunderbird will it? Will a '10000+' released 3 years from now be equivalent to a 10GHz Thunderbird?!

Geez, when will the craziness end?

 

[SparkyJJO]

the things you people argue about is amazing sometimes.... Short and simple answer:

It is JUST a model number now and doesn't mean the same that it used to in comparing whatever to whatever.