Navigate: Advice Beginners BIOS Guide CPUs Links Mainboards Memory Network Storage Video/Sound Cards Contact Forum SiteMap Sponsors WebNews Home	.	.
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:

Today marks the release of the eighth model of the Athlon processor, the CPU that has redefined the performance of the desktop platform in more than one way. After shrinking the die from the original 250 nm design to the intermediate 180 nm, the Model 8 or Thoroughbred core moves to a 130 nm process. As a consequence, the total die size shrinks from 120 mm ² to 80 mm². Aside from technical and advantages, the die shrink also moves the TBred core into a very cost-effective die class. Are there any hidden issues lurking between the down-scaled transistors? Does the performance scale up with the increased clock speed? Does the smaller die open up the way to the 2.5 GHz clock speed? We have the answers.

AMD "Thoroughbred" Athlon XP2200+

By now, the PIII is history, and the Athlon core goes into its 8th revision at approximately half of the original die process, that is 130 nm compared to the original 250 nm or 0.25 µm as it was called then. While the process has shrunk to ½ of the linear dimension, meaning that the die, everyting else being equal would be only a quarter, the speed has tripled from 600 MHz of the original Athlon release to 1800 MHz of the current AMD flagship. Note that we are referring to physical clock speed here. Since MHz or GHz are still perceived as the number 1 metric for performance, AMD with their high efficiency core got the short end of the stick compared to Intel's P4 geared towards high clock speed but performing much less efficient on a clock for clock or instructions per clock cycle. As a consequence, the so-called QuantiSpeed architecture was created to adjust the model number to a hypothetical performance of the Pentium4.

AMD "Thoroughbred" vs. "Palomino"
The Thorougbred core is substantially smaller than the Palomino core. Since all Thoroughbreds are unlocked, the die itself is not marked, instead, the CPU markings are on the packaging. Interesting is further that the surface mounted components have come back from underneath the CPU where they were hidden with the Palomino series.

So far, so good, there is really nothing wrong with this, except that the metric used as the basis for QuantiSpeed metric has changed itself. If the original QuantiSpeed idea was based on the Willamette core running off a quad-pumped 100 MHz bus interface, the latest offerings from Intel show twice the L2 cache and, as an additional performance bonus, throw in the increase in bus speed to 133 MHz for an effective 533 Mbit/ pin/sec data transfer.

Nobody will, at this point go back and reevaluate, particularly, since the rating was done very conservatively and, thus, it is not that far off anyway. In addition, nobody will blame AMD for not fighting will all power to further increase the performance of the Athlon core, after all, the next generation Hammer / Opteron or whatchamacallit CPUs are already around the corner or so it appears.

next page:    => Turning Copper Cores Into Gold =>