• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Are A64s twice as good as XPs? AMD's marketing apparently says no

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Terminat.

Member
Joined
Jan 11, 2004
As I understood it, A64s were HUGELY better than Athlon XPs - because of their integrated memory controller etc etc.

But then I read this marketing blurb about the Athlon 64 FX55 S939 chip:

AMD's new top of the line CPU aimed squarely at the performance enthusiast brings levels of performance unheard of before to the desktop market. With its revolutionary 64bit architecture this CPU is fully compatible with future 64 bit operating systems and offers blistering performance in current 32 bit applications. Clocked at 2.6GHz and utilising AMD's new HyperTransport with a massive 1MB L2 cache, the Athlon FX 55 offers an average performance increase of 50% over the AMD Barton XP3200+ CPU.

Taken from: http://www.overclockers.co.uk/acatalog/Socket_939_cpus.html

Did you read that underlined bit? 50% increase? IS THAT ALL? And this is the top-of-the-range chip, that costs £500 ($950) So someone is expected to pay five times as much for a 50% increase?

And what does that mean for all the other S939s, and what about the S754s? Do they give 20% increases over the 3200+ Barton or what?

I thought A64s would thrash Athlon XPs, especially with their high price tag - but that means people are paying hundreds of dollars/pounds more than an XP 3200+ for a 20-50% increase - at least according to the marketing blurb!

What's going on here?
 
Yeah this really is a huge leap, just think about how long it took to get from a 2.4 P4 to a 3.6GHz P4.

In addition to this, this number is just an avarage, most probably even conservative - They don't want to speak bad about their AXP, while they're still selling it ;) .. just my thought.

In games, compiling and other tasks, that make use of the large bandwidth and latency improvement, the FX will be even faster.

My gentoo install was more than twice as fast with my 2.6GHz winchester compared to my 3200+ barton.
In windows, surfing and stuff is not much faster on the other hand.

edit: it's almost insulting to use the words AXP and A-FX in the same sentence ;).
 
Typically, people were saying that A64's were 20% faster clock for clock than XP's. After all, the K8 is a lot like the K7.

This is why I've waited to upgrade to Athlon 64. An overclocked mobile XP is a damn nice little chip. Unlike some folks, I like to see a difference outside of a few benchmarks.
 
What is not computed in this formula is overclocking. Just for the sake of argument say the average Xp core could be clocked to the top of the line 3200xp speeds and the average A64 could be clocked to FX55 speeds, then the average increase would be 50% across the board. I know this model is considerably less than perfect, but the intent is to show that average gains are probably in the 40% range. Cost increases vary according to what chip you bought, but I paid 50% more for my A64 3000 versus my old 2500XP. Considering that prices go up over time I don't think that is a poor price performance ratio.
 
I know that my itty bitty A64 2800+ will be on par with the AXP 3200+ for about $25+ or so less. Most of all cooler. Now, that is what brought me into the bottom end of the A64 lineup. Plus the future of x86 will be 64bit. If I read my numbers right, I will get an edge over the AXP 3200+ in some areas. That is at 1.8, but I overclock, so it is even more unequal.
I seen a healthy boost in gaming with an A64. Plus everything else, but right now my interests go with gaming.
 
Playing games such as Rome: Total War and America's Army which are very cpu intensive, my framerates have made dramatic increases going from a 2.3Ghz barton to a 2.65Ghz newcastle. By dramatic, I mean my average framerate in America's army jumped from 40 to 60. My max frame rate went from 55 to 95! I am very pleased. Also my framerates never drop below 40 no matter what. They used to drop pretty low sometimes.
 
why are you complaining? 50% would mean something like a 2GHz winchester matching a 3GHz barton and frankly my 2GHz winchester does just that. I have a 23.6k 3dmark score with my winny at 2GHz while my friend cant even come close on his prometeia cooled 2.76GHz barton :( He would need 2.9 to 3GHz to match my 2GHz winny
 
The FX CPU has a megabyte of cache vs 512 KB of cache on the non-FX Athlon64's. People claim that the cache difference makes up a couple hundred MHz. So it is likely that the clockspeed difference isn't that dramatic OC550. It would be more like a 2 GHz winchester vs a 2.4 GHz XP. I'll even go so far as maybe 2.6 GHz XP, but not 3 GHz.
 
Last edited:
I can show you right now my 2GHz winny just slaughtering my friend's 2.76GHz barton. I would take a 2GHz winny over even a 3GHz barton, seriously
 
Overclocker550 said:
I can show you right now my 2GHz winny just slaughtering my friend's 2.76GHz barton. I would take a 2GHz winny over even a 3GHz barton, seriously
Were the other parts the same? (other than the mobo and CPU).
 
Sentential said:
You gotta keep in mind that 20% and 50% is a huge leap. They are faster...just not 2X faster


Agreed, 50 percent is quite an improvement. If they were 100 percent faster or twice as fast, that would mean it would take a 4GHz Barton to keep up with a 2GHz FX-55. Seems a bit far off to me. If anything, that 50 percent number is a little generous, but it is roughly in that range I would say.
 
Overclocker550 said:
I can show you right now my 2GHz winny just slaughtering my friend's 2.76GHz barton. I would take a 2GHz winny over even a 3GHz barton, seriously
I want SuperPi times to prove that. The best way to fairly compare is by taking graphics out of the loop.
 
Overclocker550 said:
I can show you right now my 2GHz winny just slaughtering my friend's 2.76GHz barton. I would take a 2GHz winny over even a 3GHz barton, seriously

Oh yes, I could beleive it in some very memory intensive apps like gaming and such. Of course allot of it would depend on the settings and the rest of the setup that these CPU's are running off of. Things like FSB speed, memory latencies, etc.
 
mitro said:
I want SuperPi times to prove that. The best way to fairly compare is by taking graphics out of the loop.


I think a fair way to compare overall, would be to take every single benchmark available and average them together. Of course that Barton could get "slaughtered" in some tests like mostly gaming, and I am sure there are other things that aren't memory or instruction sensitive (just the types of things that require raw power), that Barton would more than own that Winchestor having another 733 MHz and a 2 stage shorter pipeline.
 
I know that on an A64 even small gains are significant over previous chips. Efficiency has its benefits.
 
OC550 routinely exaggerates. Take what he claims with a grain of salt. Anyone that uses catchwords like, "slaughtering, destroying, owning, kicking the ____ out of" can't be taken seriously.
 
Let me be of some help here :D
Based on newegg prices right now.
XP3200 is 137
FX55 is 842

Price difference: 615% difference
Perfomance difference: 50%

Summary, you pay 123% more for every 10% difference between the two chips
Guess some people are made of money.
 
mitro said:
I want SuperPi times to prove that. The best way to fairly compare is by taking graphics out of the loop.

SPMod.JPG


Theres a super pi shot at 2.75 on a A64...Don't think any 3G barton will be able to do that...you would have to run a barton around 3.5G or so to do around the same...and how many people have OCed a Barton to 3.5G ;)...
 
Back