Still cant get AMD's Ratings - Socket 754

[FatRakoon]

Ok, here is one that has bugged the nips out of me recently.

Forgetting my Socket A and 939 systems, I am looking at tjust the 754 CPUs here, but I have 4 of them, each are completely different specs:-

Sempron 3300 ( 2.0Ghz & 128k )
Sempron 3100 ( 1.8Ghz & 256k )
NewCastle 3000 ( 2.0Ghz & 512k )
ClawHammer 3700 ( 2.4Ghz & 1mb )

Now, the ClawHammer is a fine performer, nowhere near as quick as its S939 counterpart from my tests on my own PCs, so lets look at the other 3.

One thing that bugs me, is that the 3300 being given a 3300 rating, should be the fastest shouldnt it? - it is in fact the slowest... On the desktop its not too hot, showing a few areas where it slows down, most noticeably when you have something like Office Open, and then you open Outlook Express, it gets hit for 6, while the Newcastle handles this much better, even thats not a speed demon to be honest.

So, where the hell does AMD get off in giving these CPUs the ratings they do?

Especially when comparing the Newcastle 3000 against the Sempron 3300 when the newcastle has 4 times more cache, and the Sempron has a 166FSB as opposed to the Newcastles 200 - HOW?

It does not even compare to the 3100 either.

Yesterday, I decided to get my daughter up to 64bit, and I gave her the 3300 in a DFI LanParty Mobo instead of her XP2400 in the Abit KD7R and after just about half an hour, I decided not to be so cruel, and gave her the XP back.... Dotn get me wrong here, the 3300 felt a lot snappier for a short while, but once I started to throw stuff at it, it was falling over all the time with pauses every few seconds and even kids games were jerky.

My next plan is to compare the Smepron 2600 against an XP2600

 

[ACESHIGH]

Ill try and answer some of you questions.


Firstly in order of fastest to slowest are ....

ClawHammer 3700 ( 2.4Ghz & 1mb ) - True Athlon 64
NewCastle 3000 ( 2.0Ghz & 512k ) - True Athlon 64
Sempron 3300 ( 2.0Ghz & 128k ) - budget CPU (to match Celeron)
Sempron 3100 ( 1.8Ghz & 256k ) - budget CPU (to match Celeron)

Im sure their respective prices when purchased (if new) would bear this out as well.

Ther ratings really should only be used comparing A64 v A64 (v Pentium4) & then Sempron v Sempron (v Celeron).

As for your comment "One thing that bugs me, is that the 3300 being given a 3300 rating, should be the fastest shouldnt it?"

No it should be 3rd fastest. Same way Intel sells 2.8 Gig Celerons & yet they are trounced by P4 2.4 Gig processors or an Athlon XP 2400+.

Semprons are budget CPUs like Celerons, & marketed & priced as such (you can compare Sempron v Celeron, though Semprons are still generally a bit faster).

As for "On the desktop its not too hot, showing a few areas where it slows down, most noticeably when you have something like Office Open, and then you open Outlook Express, it gets hit for 6, while the Newcastle handles this much better" ... the Newcastle likely handles this better due to the larger L2 cache. Hence why the Sempron 3100 may seem faster, as it has more L2 cache than the Sempron 3300 & when doing tasks that require cache it will actually perform better. But some tasks require raw power & there the Sempron 3100 should be slower than the Sempron 3300 (which is clocked faster). When AMD tested the processors (cant remember how many tests they do - 45 - 60, i think) the Sempron 3300 was on average faster as more tasks require raw processing power & hence the nomenclature.

The Athlon XP 2400+ is 2.0Ghz, 256k L2 cache & 133MHz FSB. The Sempron 3300 has less L2 cache, but 200MHz FSB actually (im pretty sure). So should be pretty close. But if the tasks you tried need more L2 cache the XP will run quicker, but then again when it was released it was the top of the line CPU, put up against the P4 (same as the A64 CPU).

BTW i expect the Sempron 2600 (whether it be socket 754 Oakville, Palermo E3, Palermo E6 or socket A) to be slower than the XP2600+ (but which one exactly do you have as there are 4 different cores - Thoroughbread B 133MHz FSB, Thoroughbread B 166MHz FSB, Thorton & Barton).

Note: thats why the socket A Sempron 2600 (1833MHz; 166FSB & 256MB L2 cache) & socket 754 Sempron 2600 (all 3 different cores/revisions are 1600MHz; 200FSB & 128MB L2 cache) have the same numerical value though are quite different, because AMD expects them to run about the same speed as both are budget CPUs & thus can be compared. (AMD expects the slower clocked 754 CPUs to make up ground to due the benefit of the socket 754 platform). The nomenclature AMD uses is generally pretty good, but remember only compare budget CPU v budget CPU & mainstream CPU (A64) v mainstream CPU (A64).

So yes it can get confusing, but makes sense for those who keep up to date with info. Hope this helps.