Navigate:

Advice
Beginners
BIOS Guide
CPUs
Links
Mainboards
Memory
Network
Storage
Video/Sound Cards

Contact
Forum
SiteMap
Sponsors
WebNews
Home

Xoxide Computer Mods
. .

Prices:

Mainboards

ABIT
ASUS
Chaintech
Shuttle
Soyo
Tyan

CPU
Intel
P4 2.4C-800
P4 2.6C-800
P4 2.8C-800
P4 3.0-800
P4 3.2-800

AMD
AthlonXP
XP 1700+
XP 2000+
XP 2400+
XP 2500+
XP 2700+
XP 3000+
XP 3200+

Athlon64
Athlon64 3200+
Athlon64 FX-51

Opteron
Opteron 240
Opteron 242
Opteron 244
Opteron 246

Memory

Corsair
Crucial
Kingston
Mushkin
OCZ

Search Prices:


























































































































LOSTCIRCUITS

SHORTCUTS:
Top Page
The Economy of Shrinking
Power Issues
Test Configurations
Quake3 Arena
3DMark, CodeCreatures
Ulead, Caligari
Winstones
Overclocking, Conclusions
AMD deals
 Athlon XP2200+   
The First of the Thoroughbreds
(Review by MS, June 10 2002)
The Economy of the Die Shrink

Changes in AMD's roadmap have occurred as the winds of the Hammer progress demanded them but the general outlines have remained the same and among those is the next die shrink. The importance of the die shrink goes far beyond making smaller transistors that will run faster, use less energy and run cooler. The one thing that where AMD has been hurting forever has been the wallet and here is where the greatest asset of the die shrink comes into play: manufacturing cost. Compared to a full wallet, Silicon-On-Insulator (SOI) is only eye candy, a bigger L2 cache as it is still on the current roadmap for the Barton will be helpful for keeping up with overall performance but the main priority has to be the profitability of the company. Let the hammers do the performance dance once they are available.


AMD "Thoroughbred" vs. "Palomino" Take 2
Bottom view of Throroughbred and Palomino. Where the Palomino had its surface components "inside the Socket A", the TBred is bald.

One of the golden rules in the IC business is that the die size where one can really start making money even if they yields may be low is 100 square millimeter. As strange as this sounds, since different dies will be sold at different costs, this rule holds surprisingly well even though Intel admits that they make money on the PIII at 106 mm2. Not everybody needs the very fastest processor at all times especially at the rate of obsolescence caused by the clock-race. However, a 33 % die shrink from 120 to 80 mm2 means a healthy cost reduction and if the overall pricing level remains at about the same level, a 33 % cost reduction can increase the net profit from e.g 10 % to 40 something %, or in this case (which is a totally arbitrarily chosen number) quadruple the net earnings.

The morale of this story is that we are not as much concerned with the question of whether the Thoroughbred will once again rip the crown of the fastest x-86 processor from Intel, that one might fall into the realm of fantasy at the moment. The more burning question is whether the shareholders are going to be happy and whether weather AMD stock is going to be a strong buy. For this to happen, the only thing that the Thoroughbred really needs, is to hold its ground, there are enough other system bottlenecks that are much worse than CPU performance. Of course, a bit of extra power has never hurt anybody either.

AMD "Thoroughbred" vs. "Palomino": Take 3
Die comparison between the old and the new CPU. The L2 cache has moved from the right side to the bottom of the die. Because of some of these rearrangements, the overall transistor count could be lowered from 37.5 Million to 37.2 Million transistors. In both cases, the L2 cache is roughly 15 million transistors.

With respect to overall system performance and particularly price / performance ratio, we need to factor in a few other parameters. Every system builder knows that the CPU is only a relatively small contributor to the total manufacturing costs of the entire system. To put it into perspective instead of numbers, a CPU can cost less than a keyboard or a wireless mouse and it can be as much as a high end monitor. One can build a system for $ 1000 where the CPU costs $ 35, meaning 3.5 % and one can double the performance by choosing a processor for $70. If $35 is all it takes to double the performance, it is a no brainer. If it is $400 to get an, er, measurable performance increase that may or may not even be noticeable by the consumer .... I guess, you get the drift. So much for the background that may or may not have been the rationale for the release of today's review candidate: Enter the AMD XP2200+ Thoroughbred.

next page:    => The Power Of The Shrink =>

Click here! All advice and educational articles on LostCircuits are free, but if you feel you can, please make a small donation to us!
Thank you!

General disclaimer: This page only reflects the author's personal opinion and assumes no responsibility whatsoever regarding any of the contents or any damages that may occur explicitly or implicitly from reading the contents of this site. All names and trademarks mentioned in this review are the exclusive property of the respective parent companies.
All contents of this site are protected by international copyright laws. Reproduction of the contents even in parts is not allowed except after written permission by the author and referral to this site.
Copyright 2002 - 2008 LostCircuits