It's the Ahlon64 FX-51 that has the 940 pins. Next year a 939 pin
version will appear, and the 940 pin version will be discontinued. The
extra pin is used for ECC RAM - that is, the 940 pin Athlon64 FX-51
NEEDS ECC DDR memory!
Exactly! This is a major drag. Either we need cheaper (and available)
ECC 400DDR, or we buy the cheaper/slower A64, or we wait, or we're
very rich.
On this page there's a nice summation of relative performance of the
Athlon64 compared to the Athlon XP, the Athlon64 compared to P4, and the
Athlon64 FX-51 compared to the P4 EE:
http://www.tweakers.net/nieuws/28882
Averages:
Athlon 64 3200+ Athlon 64 3200+ Athlon64 FX-51
vs Athlon XP 3200+ vs P4 3,2GHz vs P4 EE 3,2GHz
Game 8,3% 6,0% 9,2%
Workstation 12,6% -10,5% -5,7%
Overall 14,7% 3,5% 3,7%
Conclusion:
Athlon64 is master in games.
The P4 still rules for serious work.
Well. No. - And there's no "still" about it either. - And this may be
a very serious misunderstanding. Which should not be nurtured in an
assuming and ignorant manner. Particularly not when choosing a system
for serious workstation use.
- Carefully check the benchmarks that are relevant for your
applications!
The P4 has an edge (significant?) on mediaencoding & mainstream
imageediting. Period.
If you're into media, the P4 is a nice option. If you're doing serious
conditional and numerical computing, the P4 is NOT an option! It is a
mistake. It's a damn dog! Even lowend AthlonXPs sometimes beats Intels
3.2GHz flagship.
Just look at some of the benchmarks in detail:
ScienceMark Cypher AES Athlon64 advantage on 3.2GHz P4 66,0%
AthlonFX advantage on P4EE 82,4%
(this is serious stuff, so it can't be SSE2 optimized)
(AthlonXP3200 advantage on 3.2GHzP4 is basically of the same order.)
(This is also the kind of advantage the Athlon has a lot of the time,
when you're using a non-P4-optimized app. Food for thought, eh?)
Plasma (P4-optimized) Athlon64 advantage on 3.2GHz P4 15,7%
AthlonFX advantage on P4EE 33.0%
What this website has made, is a sort of average on a large number of
benchmarks, with no thought on how they are weighted.
For instance, it includes three Kribi benchmarks, which is just
counting the same P4 benchmark tree times.
There's other multiple counts, like four P4-optimized 3DS benchmarks.
Just the simple expedient of reducing these multiple counts to one,
results in average advantage of the AthlonFX over the P4EE of 7.3%
and an advantage of the Athlon64 over the 3.2GHzP4 of 1.9%
....on "Workstation".
So paraphrasing:
"AMD still rules for serious work"
I'm not claiming this is a good way of comparing things. On the
contrary, I'm pointing out the big flaws in reducing a comparision
like this. You can proove anything you want. And this website HAS
CHOSEN the benchmarks from hundreds on the web.
....And why is not the AthlonFX compared to the 3.2GHz and the Athlon64
compared to the 2.8GHz?
The P4EE is a very bad cpu anyway. At $900 it's probably still sold at
loss for Intel, and the only things it runs faster than the 3.2GHz are
some benchmarks (which of course is exactly the game intended) but why
is the way this website compares, the right way? No absolute figures,
no global comparisions between A-XP, A-FX, A-64 and P4's?
Frankly, I think it stinks!
So I don't agree. This is not "a nice summation of relative
performance".
When
running (AMD optimized) 64-bit apps the P4 can't compete with the
Athlon64 tho.
That remains to be seen. I think the P4 will continue to hold its own
on mediaencoding, until data becomes larger than 2GB. The reason is
simply the high clockrates.
ancra
