I should note that I have no idea, what processor the analysts from c’t actually used for their testing. They state that the chip was the Athlon 64, but with 1MB of L2 cache. Either it is a typo, or they used a processor from the Opteron series, because the Athlon 64 incorporates only 256KB of L2 cache.
Here we go with the testbeds and test results:
| Benchmark Results of AMD Athlon x86-64 chip 1.20GHz, Athlon XP 1.20GHz and Intel Pentium 4 2.20GHz by c't | |||
| Processor | AMD Athlon 64 1.20GHz | AMD Athlon XP 1.20GHz | Intel Pentium 4 2.20GHz |
| Cache (L1+L2) | 128KB + 1MB | 128KB + 256KB | (8KB + 12 000 micro-ops Trace Cache + 512KB |
| Memory Type | PC2700 | PC2700 | PC2100 |
| Peak Bandwidth | 2306MB/s | 2041MB/s | 2080MB/s |
| Benchmark | |||
| 3D Mark 2001 | 8753 | 7601 | 9008 |
| Sysmark 2002 | 178 | 145 | 217 |
| Quake III /1024x768x16 | 225.3 fps | 172.4 fps | 218.9 fps |
| Aquamark | 49.5 fps | 48.3 fps | 51.4 fps |
| Commanche 4 | 36.6 fps | 28.4 fps | 36.3 fps |
| Cinebench 2000 Shading C4D | 18.8 CB | 12.4 CB | 17.2 CB |
| Raytracing | 20.4 CB | 16.9 CB | 22.4 CB |
| PovRay 3.5 | 251 PPS | 242 PPS | 333 PPS |
| PovRay 3.1 | 1037 PPS | 948 PPS | 1219 PPS |
| DivX 5.02 | 569 sec. | 709 sec. | 497 sec. |
| Xvid | 750 sec. | 929 sec. | 717 sec. |
| Linux Kernel 2.4 | 161 sec. | 222 sec. | 166 sec. |
| SpecFP_Base2000 | 674 | 504 | 677 |
| SpecInt_Base2000 | 739 | 532 | 774 |
The very important thing I would like to point out is the efficiency of the Athlon 64 1.20GHz processor compared to its Athlon XP 1.2GHz. The former manages to outperform the brethren in all the cases, both due to larger L2 cache and architecture improvements. Furthermore, AMD’s baby also functions not much slower compared to the Pentium 4 2.20GHz chip even despite 1.0GHz of the core-clock difference. In fact, in some cases the AMD’s novelty even leaves the Pentium 4 behind. Of course, we should understand that currently Intel’s processor was tested with a single-channel chipset, while AMD Athlon 64 will compete against dual-channel configurations when it finally reaches the market. Moreover, there is no information about performance scalability of the Athlon 64 processor with 256KB of L2.
So, to sum up, we can be sure that the Athlon 64 processors are a very powerful and with efficient architecture. On the other hand, their positions on the market are still uncertain, given that AMD constantly postpones the launch date, while Intel plans to strengthen its Pentium 4 processors with 800MHz Quad Pumped Bus and Hyper-Threading technology support in the second quarter 2003. In case AMD’s Athlon 64 CPUs are able to function faster than Intel’s Pentium 4 3.20GHz processors, AMD’s positions on the market will become more solid. Talking about market share, I should admit that AMD will need to supply a lot of its x86-64 processor in order to land more orders from major personal computer vendors. If it happens, AMD may eventually become the provider of the highest-performing desktop CPUs next year, since Intel also postponed the launch of the Pentium 4 “Prescott” chips until the Fall 2003 or even later.





