|
|
Advertisement:
|
|
AMD AthlonXP Processor 2600+ |
|
Join the community - in the OCAU Forums!
|
Performance Comparison
Testbed Details
AMD kindly provided everything needed to test this CPU, including an Epox EP-8K3A+ motherboard and 2 sticks of TwinMOS PC2700 CAS2.5 DDR memory. The high multiplier (16X) required a custom BIOS from EPoX but a production BIOS with support is coming soon - this is probably true for all KT333 and KT266A boards at least.
 
AMD AthlonXP 2600+
EPoX EP-8K3A+ motherboard (beta BIOS)
TwinMOS PC2700 CAS2.5 DDR (1x 256MB)
IBM 75GXP 30GB ATA-100 HDD
Enermax EG301P-VE PSU
X-Micro GF4 Ti4600 (review here)
Intel PentiumIV 2.53GHz
Asus P4B533-E motherboard (1004 BIOS)
TwinMOS PC2700 CAS2.5 DDR (1x 256MB)
IBM 75GXP 30GB ATA-100 HDD
Enermax EG301P-VE PSU
X-Micro GF4 Ti4600
Software Configuration
Windows 2000 SP3
DetonatorXP 30.82 video drivers
i845E drivers version 4.00.1013 (P4)
Intel Application Accelerator 2.2.2 (P4)
VIA 4-in-1 (KT333) drivers version (AthlonXP)
The P4 CPU, motherboard and the GF4 video card were all kindly provided at short notice by Altech Computers.
Memory Considerations
There are quite a few possible configurations for memory in today's market. When considering the AthlonXP, the best currently is DDR333 and, given AMD provided EPoX's EP-8K3A+ motherboard for testing - which is a KT333 motherboard supporting DDR333 memory - that part was easy. On the P4 side we primarily have DDR266 with some more recent chipsets supporting DDR333. The P4 also has the option of RAMBUS, which does provide a small speed advantage but carries a cost penalty both when purchasing the RAM and the motherboard. Also, the availability of RAMBUS-based solutions in Australia is extremely limited at the moment and I was unable to arrange one for this comparison. Altech Computers were kind enough to rush over an i845E-based P4 motherboard (ASUS's P4B533-E) and a 2.53GHz P4 CPU. This chipset does not support DDR333 natively, only DDR266. We have seen that going to a DD333MHz-capable chipset - SiS's 645DXA2 in that example - does not make much difference to overall P4 performance. However, to level the playing-field I will test the AMD system at DDR266 as well as at DDR333. Timing in all cases was (CAS-RAS-CAStoRAS-PRE): 2-3-3-6. The default 4X interleave was left on the KT333 board - the i845E board does not report the interleave setting.
SiSoft Sandra 2002 SP1
SiSoft's "Sandra" suite is a widely-used standard for CPU, multimedia and memory bandwidth benchmarks. We used the latest version available using their "WebUpdate Wizard" at time of writing.
The CPU and Multimedia benchmarks isolate the CPU, cache and memory subsystem:
In either memory configuration, the AthlonXP has the edge in those benchmarks. The Memory Bandwidth benchmark is self-explanatory, stressing the memory subsystem:
3DMark2001SE
MadOnion's 3DMark2001SE is the standard DirectX benchmark. We used the default settings (Hardware T&L):
Things are looking pretty even overall. On the next page we have more "real-world" benchmarks..
|
|
Advertisement:
All original content copyright James Rolfe.
All rights reserved. No reproduction allowed without written permission.
Interested in advertising on OCAU? Contact us for info.
|
|
