High-End P4: Better, Not Best

Today's Best Tech Deals

Picked by PCWorld's Editors

Top Deals On Great Products

Picked by Techconnect's Editors

Take that, AMD: Intel's Pentium 4 Extreme Edition.
Take that, AMD: Intel's Pentium 4 Extreme Edition.
Photograph: Marc Simon
With the unexpected announcement of the company's first gaming-oriented processor--the Pentium 4 Extreme Edition--Intel obviously hoped to steal some thunder from AMD's Athlon 64 launch. But while PC World tests show that the new P4 EE outruns a standard Pentium 4, AMD's shipping Athlon 64 FX-51 outperforms both.

We tested the P4 EE, a 3.2-GHz CPU with a whopping 2MB L3 cache, on an Alienware PC originally equipped with a standard 3.2-GHz P4. That system included 1GB of memory and an ATI Radeon 9800 Pro graphics card with 256MB of graphics memory--as did each of three Athlon 64 FX-51-based systems (from Alienware, Falcon Northwest, and Voodoo).

On PC WorldBench 4, the average score of the three Athlon 64 FX-51 systems was 142, about 8 percent higher than the P4 EE system's score of 131 (the same system with a standard 3.2-GHz P4 scored 126). The FX-51-based PCs also dominated the P4- and P4 EE-based PCs in our AUGI Gauge and Premiere 6 tests. Other scores were closer, but the only test in which the P4 EE unit actually beat the AMD competition was the one for Musicmatch 7.1 encoding. (See the full test report below.)

The average cost of the FX-51-based systems we tested for our November issue was over $3300. Intel expected to begin shipping vendor and retail processors in early November. Volume pricing for the P4 EE is higher than for the Athlon 64 FX-51, but the AMD chip requires both a more expensive motherboard and costlier memory. Analyst Kevin Krewell at Reed Business says that he expects P4 EE systems to cost about the same as comparably configured FX-51 PCs.

Gamers should be pleased with either chip, but the Athlon 64 FX-51 has two notable advantages: It's 64-bit-ready, and it's easier to overclock.

Test Report: Pentium 4 EE Challenges Athlon 64 FX-51

Massive L3 cache boosts new P4, but AMD's top-end chip reigns supreme.
SYSTEM PC WorldBench 4 AUGI Gauge 15 (seconds) Musicmatch Jukebox 7.1 (seconds) Adobe Premiere 6 (seconds) Adobe Photoshop 7 (seconds) VideoWave 1.5 (seconds) Return to Castle Wolfenstein1 (frames per second) Unreal Tournament 20031 (frames per second)
Alienware Area-51 (modified)
3.2-GHz Pentium 4 Extreme Edition
131 227 139 213 265 81 145 340
Intel P4 standard
3.2-GHz Pentium 4
126 246 140 217 269 82 133 292
Average of three systems
2.2-GHz Athlon 64 FX-51
142 171 152 174 255 75 145 379
1Game tests performed at 1024 by 768 resolution at 16-bit depth.

How We Test: We asked vendors to provide PCs of identical configuration, including 1 GB of DDR 400 memory and an ATI Radeon 9800 Pro graphics card with 256MB of memory. All systems ran Windows XP Home except the Alienware Aurora, which ran Windows XP Professional. All systems included RAID-striped 7200-rpm hard drives, except the Voodoo Fury, which included 10,000-rpm drives. All PCs tested with PC WorldBench 4, PC World's application-based benchmark. In the AutoCAD test, we time AutoDesk Group International's benchmark, AUGI Gauge v15, which runs CAD tests. In the Musicmatch Jukebox 7.1 encoding test, we time the conversion of six .wav files into 160-kbps MP3 files. We further resample all six MP3 files to the lower bit rate of 64 kbps to simulate preparation for use on a portable player. In Adobe Premiere 6, we time the preview rendering of two video tracks and one audio track, scrolling through frames and exporting the finished movie to various video formats; then we scroll through the output to verify the encoding.In the Photoshop 7.0.1 test, we time the rotation of a JPEG file by set amounts, and save it as a thumbnail and an EPS file. We also apply a wide variety of filters, undoing the process between filters.In the VideoWave 1.5 test, we time the process of taking a raw AVI video file, adding standard effects and music, and then converting it into an MPEG2 file using the DivX codec. In Return to Castle Wolfenstein, we measure frame rates at the given resolution generated during a recorded demonstration. In Unreal Tournament 2003, we measure frame rates at the given resolution during a fly-through of one game level. On PC WorldBench 4 and the two game tests, scores higher is better; on all other tests, lower is better.Tests developed and conducted by the PC World Test Center. All rights reserved.
Note: When you purchase something after clicking links in our articles, we may earn a small commission. Read our affiliate link policy for more details.
Shop Tech Products at Amazon