Intel’s Prescott 3.6GHz CPU

Each system was initially configured with a clean install of Windows XP (service pack 1). Then all current critical patches were downloaded and installed from the Windows Update site. We also installed DirectX 9.0b and Windows Media Player 9. Virtual memory was set to a 2048KB fixed swap file.

We ran the systems with the best possible stable memory timings. The 875-based P4 testbeds required latency settings of 2-3-3-6 to remain stable in all benchmarks. We were able to achieve 2-3-2-6 on the socket 939 board. Note the slower latency, but higher clock speed, of the DDR2 memory in our 925X based Pentium 4 system.

The hard drives were defragged before any major test requiring significant hard drive access. VSync was disabled for all real-time graphics tests. We executed the following command before any test cycle: rundll32.exe advapi32.dll,ProcessIdleTasks. This completes any background idle tasks and improves benchmark score reproducibility.

Our benchmark suite has evolved over time. However, our suite covers a wide range of applications which are significantly affected by CPU performance. The suite consists of a mix of synthetic and real-world applications, but is heavily weighted towards real applications. Here are the tests we ran for this processor preview:

Business Winstone 2004

Business Winstone 2004 is the latest version of VeriTest’s Winstone benchmark suite. It consists of a variety of common desktop applications, run in a scripted sequence that resembles actual user usage patterns. Most of these consist of Microsoft Office applications, including Microsoft Project and Access. Also included are Norton AntiVirus Professional 2003 and WinZip 8.1

New this year is a second set of four inspection tests designed to yield information on multitasking performance. The first runs Outlook and Internet Explorer in the foreground while performing a file copy in the background. The second runs Excel and Word operations while WinZip runs in the background, archiving files. The final test runs a Norton AntiVirus scan in the background while Excel, PowerPoint, Project, Access, FrontPage and WinZip perform foreground chores.

Multimedia Content Creation Winstone 2004

The latest release of Content Creation Winstone updates most of the applications to recent versions. It also shifts away from Windows Media Encoder 7.1 to the current Windows Media Encoder 9. Sound Forge has been replaced with Steinberg’s WaveLab. One note: LightWave is currently running as a single threaded application.

Both Business Winstone 2004 and Multimedia Content Creation Winstone can be ordered from VeriTest and delivered on CD-ROM for a nominal shipping charge. They cannot be downloaded.

Adobe After Effects 6.0 Professional

This is an updated version of our earlier After Effects 5.5 test, using the newer version from Adobe After Effects is a professional video compositing and editing tool. This test runs a scripted set of typical After Effects composting and filter operations and generates a log file with elapsed time data at the end.

We use the freeware VirtualDub and the latest DiVX 5.1.1 codec to compress a 330MB AVI file extracted from the DVD The Rock and originally encoded at full resolution with Indeo 5.1 to about 80MB. The AVI file offers both rapid action and high contrast scenes, making it a challenge for any compression scheme. The same file is also used in our other video compression tests.

WMV 9 Test

We use Windows Media Encoder 9 to encode the above video to a 60MB WMV file. Audio is compressed to 70kbps and the total bitstream is encoded at roughly 2050 kbps.

QuickTime 6.5 / Sorenson Test

We used QuickTime Professional 6.5 and the highly regarded Sorenson 3 codec to compress our 330MB AVI file to about 75MB, using its highest quality settings.

MusicMatch 8.2 MP3Pro Encode

We use the latest version of MusicMatch to encode a 248MB WAV file to an 11.8MB MP3Pro file at 64 kbps and note the time.

WMA 9 Encode

We use Windows Media Encoder to compress a 248MB WAV file to 11.3MB at a 70kbps data rate and record the time.

Cinebench 2003

We run Cinebench 2003 to test the software 3D rendering performance using Maxon’s Cinema4D engine. Cinebench also allows us to see how performance varies with multiple CPUs, virtual or real.

LightWave 7.5
LightWave is a highly popular 3D modeling and rendering applications used extensively in Hollywood and elsewhere. We run three different renders from LightWave’s benchmark folder to hammer on the CPU. All LightWave renders take place with two threads enabled.

Discrete 3D Studio Max 5.1

Another popular professional 3D modeling app, 3D Studio Max, is multithreaded. We run a variety of rendering tests and report several results. We rendered five consecutive frames and recorded the rendering time.

PCMark 2004

The latest iteration of FutureMark’s suite of synthetic tests has expanded on the limited repertoire of the original. FutureMark has added several multithreaded tests, as well as expanded to include storage and graphics. We focus on the memory and CPU tests here.

3D Mark 2003

The latest version of 3D Mark has had its share of controversy. However, it’s useful for gauging how a processor might fare in real-time 3D applications.

Viewperf 7.1.1

Viewperf is a mostly graphics-bound test that uses datasets from OpenGL-based professional applications. Note that Viewperf is not necessarily a good CPU or systems test. We’re including here because it does move a lot of data from the system to the graphics hardware and we wanted to see if PCI Express offered any tangible performance improvement on this particular test suite.

3D Gaming Tests

Perhaps no application exercises the system more than current-generation 3D games. We use the following to test the processors’ performance levels. All results are reported at low resolutions and, in most cases, low detail. While you’d never play a game at these resolutions, running that way serves to isolate CPU performance and negate any potential impact from the graphics card. The games we use include:

• Halo for the PC
• Dungeon Siege
• Flight Simulator 2004
• Comanche 4
• Code of Duty
• Unreal Tournament 2003
• Splinter Cell

Multitasking Tests

One of Intel’s key value propositions for its new generation of Pentium 4 CPUs is simultaneous multithreading, or what Intel calls Hyper-Threading. We wanted to examine multitasking performance carefully, so we looked at the results of several tests:

• Business Winstone multitasking tests

• An applications-based test that runs both a Norton AntiVirus 2003 scan and a scripted Photoshop Elements 2.0 filter sequence.

• PC Mark 2004 multithreading tests

• A test where we encode video using Microsoft Windows Media Encoder 9 and run the Flight Simulator 2004 test.

With these tests in mind, let’s look at actual performance data.

Business Winstone 2004/Multimedia Content Creation Winstone 2004

Interestingly, the 3.6GHz Prescott pulls ahead of the 3.4GHz Pentium 4 Extreme Edition in Content Creation Winstone 2004. The Business Winstone tests, however, are almost identical. Note that the performance improvement over the 3.4GHz Prescott, however, is pretty substantial.

Media Encoding

Video processing, including applying filters, video compression, and transcoding are increasingly important applications in today’s media-rich computing environment.

It looks like the combination of Prescott and the 925X chipset makes for a good marriage. The new CPU picks up quite a few seconds in all the benchmarks vis-à-vis the 3.4GHz Prescott and even outpaces the P4EE. Note that AMD’s 2.4GHz FX-53 does edge out the 3.6GHz P4/560 in the QuickTime and DivX tests, but not by wide margins.

Audio, on the other hand, continues to be the bane of Intel’s 90nm desktop processor. We’re at a loss to understand exactly why Prescott fares so poorly in these tests. It will be interesting to see if any of these results shift with updates to the applications.

3ds max, LightWave, and Maxon’s Cinebench 4D are professional modeling and animation tools. Here, we test rendering performance, enabling multithreading where needed (3ds max is multithreaded by default).

As with audio encoding, software 3D rendering is not Prescott’s strong suit. The increased clock rate and 925X chipset puts it within spitting distance of the Northwood 3.4GHz processor, but it still trails. It lags the P4EE by an even wider margin. However, it now splits the results with the Athlon 64 FX-53, posting a better score in our 3D Studio test, but trailing AMD’s best in the LightWave renders. Similarly, the Cinebench test also reveals Prescott’s shortcomings in this arena.


In several of the Viewperf tests, both the LGA775 CPUs fare very well. This is perhaps our first clue that PCI Express may actually be an improvement over AGP, though it’s hard to be definitive. The differences in memory, chipsets, and clock rates may also have an impact. However, the graphics subsystem is as nearly identical as possible in all these systems, so PCI Express does seem to have a positive impact.

These are synthetic tests, but can reveal the behavior of key subsystems.

PC Mark is one of the few benchmarks that loads code specifically optimized for the processor under test at runtime. The 3.6GHz Prescott packs a pretty good punch here, essentially tying the P4EE in the memory tests and leading in the CPU test. Note that the Athlon 64 FX-53’s integrated, 128-bit wide memory controller gives it a big edge in the memory tests.


The 3DMark 03 tests are also interesting. The new kid on the block does pretty well in 3DMark using software vertex shaders, which is quite CPU-intensive. It fares less well when the test is mostly graphics bound.

We’d expect the results of any memory test that completely fits in the L2 cache to be a tie between the two Intel CPUs. So instead, let’s look at the PCMark 4MB tests, which are bigger than the total onboard cache.

The new P4/560 essentially ties the 3.4GHz Northwood and P4EE CPUs in 4KB block reads and beats them in 192KB reads. But check out the write performance — the slightly higher cache latency inherent in Intel’s latest design has a noticeable impact on write performance. Random access performance suffers as well.

When we move to the 4MB test, which breaks the L2 (and L3) caches on the various CPUs, Prescott performs a bit better versus it Intel brethren. Note how close all the Intel CPUs are in block write performance for 4MB data blocks. However, running with data that’s larger than the cache really plays to the strengths of AMD’s FX-53.

Our 3D game tests offer a mix of CPU-intensive and memory bandwidth hungry tests. We keep the resolution low, so that the graphics card doesn’t unduly affect the CPU impact.

The standard games tests are run at low resolutions, in order to minimize the impact of the graphics card on the test. But we’re also posting the results at 1280×1024.

It looks like the new Pentium 4 model 560, at 3.6GHz, can finally outpace the older Northwood CPU in game benchmarks. It does trail a bit behind the P4EE. Of course, AMD’s Athlon 64 FX-53 still is the king of the PC gaming realm–though all these CPUs can certainly handle the CPU requirements of modern games quite well.

The combination of the 3.6GHz Prescott and the 925X chipset seems to be a potent one for multitasking. The higher clock rate really seems to give Hyper-Threading a boost, something Intel has suggested would be the case. We see a dead heat with the P4EE in the Business Winstone multitasking tests and substantial leads in the PCMark tests. The Norton AntiVirus/Photoshop Elements test sees the Northwood actually edging out the higher clocked Prescott. Of course, the Athlon 64 fares poorly in these tests.

One interesting tidbit about the Flight Simulator 2004/WMV encoding test is that the average frame rate during the multitasking test was higher than the P4EE, even though the P4EE outperformed the 3.6GHz Prescott running FS2004 standalone. We also see how poorly AMD’s latest and greatest does under two simultaneous, severe loads.

Good performance doesn’t come without cost. In the case of the new Prescott CPU, the cost is both high thermal output and high dollar outlays. Let’s take a look at the pricing for the new processors:

Processor	Frequency	Packaging Technology	Front Side Bus (FSB)	On-Die Cache	Price in 1,000-Unit Quantities
Intel Pentium 4 processor Supporting HT Technology Extreme Edition 3.40 GHz	3.40 GHz	LGA775	800 MHz	512K L2, 2M L3	$999
Intel Pentium 4 processor Supporting HT Technology 560	3.60 GHz	LGA775	800 MHz	1MB L2	$637
Intel Pentium 4 processor Supporting HT Technology 550	3.40 GHz	LGA775	800 MHz	1 MB L2	$417
Intel Pentium 4 processor Supporting HT Technology 540	3.20 GHz	LGA775	800 MHz	1 MB L2	$278
Intel Pentium 4 processor Supporting HT Technology 530	3.00 GHz	LGA775	800 MHz	1 MB L2	$218
Intel Pentium 4 processor Supporting HT Technology 520	2.80 GHz	LGA775	800 MHz	1 MB L2	$178

As we can see, Intel’s pricing for the new P4/560 is north of $600, in 1,000 lots. We need to also note that these probably won’t be shipping in volume until late July into the retail channel.

So, as has been the case in the past, the most cost-effective way to get a new Prescott (assuming you want one) is to buy a system from Dell, Gateway, HP, or other large system OEMs. If you are interested in systems using the new Pentium 4 coupled with the new 900 series core logic, check out PC Magazine’s reviews .

The new Pentium 4 offers surprisingly good performance in a number of areas, including video encoding, content creation, and multitasking performance. It’s not even a bad gaming CPU, though it does get trounced by the Athlon 64 FX-53 (and likely the 3800+ as well). We’re not convinced, however, that it’s worth $650 or more to someone building their own system. Nevertheless, if Intel has resolved its manufacturing issues and can build high-frequency Prescott’s in volume, the company’s strategy seems to be at least partially vindicated. It’s just taken longer to get there than Intel would have liked.

The combination of Prescott and the new 925X and 915 chipsets appears to be a potent one, albeit not without some limitations. As a general desktop processor, the 90nm Pentium 4 CPUs are just ramping up into their product cycle. However, we feel that users would be better served had Intel moved immediately to a 1066MHz effective frontside bus clock, particularly with DDR2/533 memory. And like other observers, we feel a bit like we’re in a holding pattern, waiting for the “P4e” versions, with 64-bit instruction support. Since that’s not likely to happen until Autumn, at the earliest, it’s going to be a long, hot 32-bit summer for Intel.