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?"

Comments

It's nice...

[eyegor]

to see that Intel has viable competition.

In spite of AMD "losing" the so-called CPU wars, they're still a winner in my eyes.

Geek life would be much different if we had only one viable CPU vendor (shades of Micro$oft, Batman!!!).

I've been using AMD chips in my x86 boxes since early days of the K6-2 and I've been very satisfied. The only reason that CPU prices are anywhere CLOSE to reasonable is that Intel has real competition.

AMD L1 cache is huge

[mrm677]

Smaller memories are always faster (when comparing similar technologies). Registers, being the smallest memory, are the fastest. Followed by the L1, then L2, then main memory, and then disk.

AMD's huge L1 cache probably contributes to the difficulty in ramping up the clock rate. An L1 cache must be able to respond to a data access within usually 1-2 clock cycles. Many computer architectects believe that the size of the L1 cache should be less than 10% the size of the on-chip L2 cache. AMD's chips have L1 caches on the order of 25% the size of the L2. Such a large L1 probably cannot keep up with increasing clock frequencies.

Intel chips have very small L1 caches as compared with AMD. T

Re:AMD L1 cache is huge

[gillbates]

Intel chips have very small L1 caches as compared with AMD.

Which is a tremendous problem, from a performance standpoint. Having hand-coded assembly for both processors (Intel and AMD), I know that seldom do these processors live up to their claims of being able to execute 2 instructions per clock cycle. Actual benchmarks that I've done indicate an instruction throughput of about 1 instruction every 2 clock cycles; coincidentally, the L2 cache runs at 1/2 processor clock speed. Which means, of course, that the processor is operating mostly out of the L2 cache because the majority of instructions are resulting in L1 cache misses. Thus, any increase in L1 cache size will have a large impact on actual system performance.Intel correctly recognizes that most PC buyers make decisions based on processor speed rather than actual system throughput, and this is why they can use smaller L1 caches - because the typical user will never notice the difference in actual performance. Fortunately, the result is that AMD has had to make up for the MHz gap with processors that perform better in actual computing situations, and a large L1 cache can work wonders for system throughput.

when heat is important, this is a good chip

[dgym]

I spend more of my time trying to make my computers quiet than worrying about performance. To be able to upgrade from my 1GHz thunderbird to a 1700+ thoroughbred and see a 9% decrease in heat dissipation is good news.

The only time I am soley conserned with performance is when upgrading my server, and for that I will be waiting for the hammers. A recent hammer review at THG showed an 800MHz hammer out-perform a 1600Mhz pentium 4, and that was just for 32 bit tasks.

As soon as the hammer is available at a decent frequency (AMD might start with a ~1500 model) the race for performance will be on again, so for now I am not too surprised that AMD arn't doing that much to keep the XP on the bleeding edge (more cache will help, but it is hardly revolutionary).

Right now VIA are winning more of my CPU money for their excellent C3, but that is purely a heat thing. Unfortunately they aren't suitable for my server or games box.

Still, with all these companies comming to market with different viewpoints the choices have never been so good, I think these are interesting times for CPUs and as much as I like AMD, I am glad that neither they nor any other company is at the top for too long because complacency always puts a damper on things.

Not quite yet

[RayChuang]

If people think Intel has won the CPU war, they've kind of deluding themselves.

Remember, the AthlonXP 2200+ is essentially a shrunk-down CPU core based on the current Palimino core design. That means it still has the same 256 KB of L2 cache. What happens when AMD's new Barton CPU core with the 512 KB L2 cache arrives later this year? I think AMD CPU performance will take a major jump once that happens, and will become competitive with the Intel Northwood-core Pentium 4's with their 512 KB L2 cache.

Is it small wonder why Intel is spending large amounts of money to develop the Prescott core Pentium 4 on the 0.09-micron process and 1024 KB L2 cache? At 1024 KB L2 cache, that's reaching Xeon-class server CPU territory.

Re:AMD lost?

[UncleFluffy]

It depends on what your simulation code is doing. (Surprise !)

If you look at the SPEC CPU2000[1] scores for AMD vs. Intel you'll see that for, say, an XP2000 vs a 2G P4, the AMD is faster at integer work and the Intel is faster for floating point. Note though, that the Intel scores are with the Intel Reference Compiler, which will probably be generating SSE code. If you're running non-SSE code, then Intel stuff is considerably weaker for floating point work.

The second thing to consider is how the pipeline length affects the execution of your code. The longer pipeline in the P4 means that, roughly speaking, the P4 is faster in a straight line, but Athlons corner better.

For the simulation stuff I'm doing, which involves huge amounts of conditional integer and bit-twiddling operations and next-to-no floating point, I use a dual MP1800 box[2]. Getting the equivalent performance - for my code - would have cost a huge amount more with a P4-based solution and may not even be possible with the current P4 range.

[1] If you're doing real computing rather than fragging folks, SPEC is probably a better place to get your information from than Tom's Hardware.

[2] One SETI work unit every 80 minutes. Yummy.