Slashdot Mirror
AMD Introduces the Athlon XP 2200+
NevDull writes "AMD introduces the Thoroughbred core in the Athlon XP 2200+. Tom's Hardware Guide has a review of the new CPU based on the 0.13 micron core, and subsequently declares the current CPU war to have been won by Intel." Update: 06/10 12:48 GMT by T : DavoHH writes "To add to the list of reviews and benchmarks around the net for the new Athlon XP 2200+,
HotHardware.com has one and also
and also Anand's
and AMDMB." Update: 06/10 13:45 GMT by T : One more: Johan contributes a link to an Ace's Hardware review which tries to answer the question "Does the 0.13 Athlon XP run well an on older motherboard, and does it provide good value as an upgrade?"
Another report can be found at tech-report.
Personally, I'll just wait for the price cuts to take effect, then buy an XP.
review at Ace's hardware
Much info about upgrading older boards to the new AMD.
At least here the reviewer make sure that both CPU work with the same memory.
Tom's gives the P4 PC1066, while 95% of the P4 systems are sold with DDR.
Review at Anandtech
http://www.anandtech.com/cpu/showdoc.html?i=1635
Their conclusion in short:
Thoroughbred is more of an evolution to the Palomino core than a revolution. In other words, nothing new except minor speed increases to the end user. No special architecture changes, except decreased transistor amounts to allow higher clock frequencies and perhaps a bit lower prices as well.
After attempting to overclock their Thoroughbred @ 1.8 GHz, they observed there was almost no overclocking potential at all, leading to some doubts to whether AMD will keep up with Intel that well until their Hammer processors is ready.
So the Thoroughbred core seem to extend the Athlon XP lifetime with perhaps a few more 66 MHz jumps from the current 1.8 GHz, but will probably never get more than a 10-20% performance increase above the Athlon XP "Palomino" 2100+. From Anandtech's analysis, I'd think the best Thoroughbreds will end around a "2600+" performance rating.
Beware: In C++, your friends can see your privates!
You're wrong about that. AMD made it quite clear when they introduced the PR figure that it related to the performance/Mhz of their previous generation Athlon (the Thunderbird) rather than any Intel CPU. Therefore an Athlon XP 2200 should perform equivalently to an Athlon T'Bird @ 2200Mhz - which would not have been possible seeing as the T'Bird 1400 cranked out even more heat than an XP 2100 (1733Mhz). It's still a dumb system, but at least it has a consistent INTERNAL logic.
That was classic intercourse!
It's a Pentium-equivalent naming scheme. Since typically AMD chips are more efficient with the use of their cycles, an AMD chip can accomplish more in 100 cycles than a Pentium chip can accomplish in 100 cycles. Lets say that the AMD chip gets 100 units of work done in 50 cycles, while the Pentium chip only gets 50 units of work done in 50 cycles. In order for them both to get 50 units of work done per second, AMD only needs to produce a chip that operates at 25 cycles per second, while Intel needs to produce a chip that operates at 50 cycles per second. However, a lot of people don't understand that the AMD chip uses the cycles more efficiently and simply think "Oh, the Pentium operates at 50 cycles per second and the AMD operates at 25 cycles per second, 50 is more than 25, so the Pentium is better" (even though they both accomplish the same exact amount of work. So to counter this uninformed behaviour, AMD decided to use a rating system to sell their chips instead of simply using the cycles per second (Hz) that their chips operate at to sell the chips. The rating system does closely resemble a Hz rating, but they are using it as a Pentium-relative measurement to how much work they get done, saying that "Even though this chip only operates at 25 cycles per second, it gets 50 units of work done. The pentium operates at 50 cycles per second, and it gets 50 units of work done. We are losing sales as a result of the uninformed public, so instead of rating our chip as 25 Hz and having people think it is inferior to to the 50 Hz Pentium, we will rate our chip with the number 50 (which is not the cycles per second of the chip, is not false advertising, but may seem deceptive to some (I tend to agree with AMD's rating system, but I look at benchmarks anyhow)) and show that it is equivalent to the Pentium chip."
Hope this helps.
Intel P4 2.53Ghz: $535
AMD Athlon XP 2200+: ~$250
Intel price is from Pricewatch, AMD is a guess based on two sightings here and here, and past release prices.
HardOCP got theirs up to 2025MHz (which they say would be a 2500+ part)
Belief is the currency of delusion.
Gee, tell that to the simulation folks I work with! We've got a good half-dozen recent-vintage Linux servers also in use as desktops. At least once a week someone comes by to ask why my machine is running so slow. The answer is that I have a genuine Intel 2GHz processor. The "1.7+" and even "1.6+" AMD machines kick my Intel's butt. These AMDs have clock speeds at least 25% slower than my Intel's, but their throughput is 20-40% better. I never really believed in "clock equivalence" bullshit before, but I do now!
Why on earth did you feel the need that you had to upgrade a PIII-1GHz machine?? Unless you're one of gaming-lusers who thinking squeezing out an extra 2 fps at a cost of several thousand dollars is worth it, you're crazy.
Starting with the Athlon XP series, AMD's strategy to deal with extreme heat conditions is to simply have the information available but leave it up to the motherboard and ultimately the user to decide what action, if any, to take.
The P4's thermal diode will automatically trigger some throttling if excess temperatures are generated. The Athlon XP's thermal diode will simply alert the motherboard as to what is going on so that some action will be taken. Ultimately, the Athlon XP can still have the same kind of protections as the P4 but the processing power and logic necessary is offloaded to the motherboard.
Perhaps this is actually a cost-cutting measure and it is part of the reason why AMD chips on the whole seem to cost less than intel. Either way, if you want to build a good desktop around one of these chips, you will pay for good thermal protection in the CPU if you get intel or in the motherboard if you buy AMD.
What most non-programmers (and even some "hand-coders" like yourself) don't realize is that most software runs in loops. All you have to do is make your L1 cache big enough to hold a typical inner loop (less than 100 instructions) and you have yourself a 60-80% hit rate. Increase the size of the cache more, and you can enclose the typical outer loop and maybe a few often-called libraries, event handlers, or system calls. From then on, you run into severely diminishing returns. Increasing your cache beyond a hit rate of 95% to 100% (theoretically impossible of course) to compensate for a mere 2-cycle access L2 only buys you an additional ~5% of performance.
Doesn't sound like much of a "tremendous problem" anymore, does it?
From Intel's point of view, approximately 0% of the buyers out there care about the cache size. I'm sure Intel performs due diligence when modeling and selecting an appropriate cache size. When the diminishing returns set in, they know when to draw the line. I don't know about you, but I would prefer Intel spend an appropriate amount of resources on L1 cache and an appropriate amount on L2 cache, then spend the rest of their resources increasing the clock speed and validating the chip to make sure what I buy is bug-free. Surely Intel can better spend those resources on removing speed paths and reducing their cycle time. This has the potential to increase the performance well over the 5% needed to compensate for the additional L1 cache misses.