Our test systems consisted of the following hardware:
The main focus of this review will be on the benefit of adding a second core to the existing Pentium 4 architecture. We have thrown in results for our 3.6 GHz Prescott, as well as a Pentium 4 670 underclocked to 2.8 GHz. On the AMD side, we have included results for our just-reviewed Athlon 64 X2 4200+, as well as a regular Venice-cored Athlon 64 running at 2.2 GHz, which is equivalent to a 3500+ rating. We feel that this set of products will conclusively illustrate both the benefits and drawbacks of the Pentium D 820. Please note that our underclocked Pentium 4 670 has 2 MB of L2 cache, whereas the Pentium D 820 only has 1 MB per core.
To recap, the processors against which the Pentium D 820 will be compared against are:
- Pentium 4 560 (Prescott, 3.6 GHz, 1 MB L2, 800 MHz FSB)
- Pentium 4 670 @ 2.8 GHz (2 MB L2, 800 MHz FSB)
- Athlon 64 X2 4200+ (2.2 GHz, 512 KB L2)
- Athlon 64 3500+ Venice (2.2 GHz, 512 MB L2 per core)
At this time, the results for our combination benchmarks are not available for the Pentium 4 560. Time allowing, we plan to supplement these results at a later date.
Software used during testing consisted of the following:
- Windows XP Service Pack 2
- NVIDIA ForceWare 71.84 drivers
- NVIDIA nForce 6.53 drivers
- CineBench 2003
- LAME MP3 Encoding
- PC Magazine Business Winstone 2004
- PC Magazine Multimedia Content Creation 2004
- POV-Ray 3.6
- POV-Ray 3.7 beta4
- RightMark Memory Analyzer
- SiSoft Sandra
- TMPGEnc MPEG2 Encoding
- WinRAR
- Call of Duty
- Comanche 4
- Doom 3
- FarCry
- Halo
- Half-Life 2
- Jedi Knight: Jedi Academy
- Unreal Tournament 2004
- X2 Rolling Demo
Since dual-core processors are being targeted at the multi-tasking crowd, we thought it would be prudent to attempt to simulate some common multi-tasking scenarios. This is where our "combination benchmarks" come in. We talked about these benchmarks in detail in our Athlon 64 X2 4200+ review, though a quick re-cap is in order. The combination benchmarks performed look as follows:
- WinRAR compression (using the benchmark code path) + LAME MP3 Encoding; recorded MP3 encoding time
- WinRAR compression (using the benchmark code path) + Doom 3; recorded the Doom 3 FPS
- TMPGEnc MPEG2 Encoding (single-threaded) + LAME MP3 Encoding; recorded both encoding times
- TMPGEnc MPEG2 Encoding (single-threaded) + Doom 3; recorded the Doom 3 FPS
- Windows Media Encoding + streaming to two clients + Doom 3; recorded the Doom 3 FPS
There is no particular reason for why I selected Doom 3 as the "de-facto" game test for these combination benchmarks, except perhaps for the fact that Doom 3 has proved to be a very clock-speed oriented and "single-threaded" benchmark.
Let's see what this baby can whip up, shall we?