It’s interesting to read some of Intel’s product info on the P4. They stress the P4’s ability to provide optimal performance under Windows XP, Office XP, Java, web services, 3D rendering, video de/compression, audio (especially MP3), Digital video and editing, along with 3D gaming. That seems to be where the market, both business and home, is headed toward, but the problem is testing for performance under those specific applications. As you’ll see by the benchmarks below, in some aspects the P4 isn’t too impressive, but the question is whether it’s due to not using the “right” tests. Is the CPU here ahead of the applications that will truly take advantage of it’s architecture or did Intel design a CPU that missed the mark? Or could it even have something to do with my choice of video cards? That’s hard to tell. I think when Win XP ships it will be interesting to perform another set of tests. I also tried to run BAPCo SYSmark 2001, since it appears it’s tests are geared toward the P4, but so far I’ve been having trouble getting to run on any mainboard or CPU. Once I figure out what I’m doing wrong and get it up and running it will be interesting to check it out also.
Well, lets see how the P4 compares to the latest P3. As noted above, I used the P4 2.0GHz CPU, an Intel D850MD Mainboard with the Intel 850 chipset along with 2×128=256MB of PC800 RDRAM, a GeForce II GTS 32MB AGP card, WD 200BB ATA/100 7200 RPM Hard disk, 3.5″ Floppy disk and 50x ATA/33 CDROM. Both the AC 97 sound and the integrated NIC were enabled, Windows 98 SE was installed along with the .inf update for the chipset, the 12.41 nVidia drivers, sound and NIC drivers. For the P3 I used a Soyo TISU with the i815EP chipset and a Tualatin 1.2GHz CPU. Everything else was the same between the systems. I did do a complete retest of the TISU and P3 1.2 just to make sure there were no differences in the setup from the previous review that would affect performance.
I’m going to pick one of the tests that did show a significant advantage for the P4 first. Not that I feel it’s biased, but I think it shows the potential of the P4 over a P3.
Looks impressive, with a 32% advantage going to the P4. One thing ZD Labs has done is remove the CPUmark 99 scores from the Winbench 99 test series, though I still use it. As you’ll see below they may have a good reason to do so.
Here the P4 only has an advantage of about 17%, though still not too bad. But which is a better indicator of real CPU power with today’s applications? With current systems, programs and usage geared toward multimedia I would say the CPU results in the 3D Winbench test are a better indicator of overall system performance than the older CPUmark 99.
I found these next results interesting. They seem to confirm that the P4 is designed to enhance performance in different areas than the P3 does.
Here the P3 clearly shows it has the performance advantage – 22% in the Business test and almost 12% in the High-End test. But the question should be how much will this affect my overall system performance? Am I usually waiting for a screen to display or re-draw while using a spreadsheet or word processor, or is it 3D rendering and multimedia applications that cause more wait time for the screen to display or update? My guess would be the second.
So what about overall performance under Winstone – a business application based benchmark?
About a 6.5% advantage going to the P4 in both test, not much but we also may not be looking at what the P4 is designed to do best. One way is to see if SYSmark 2000 gives similar results.
The overall rating is a tie, with a slight advantage to the P4 (4%) in the Content Creation test but about the same percentage advantage to the P3 of Office test. So no, it doesn’t quite give the same results the Winstone tests did.
Be the first to discuss this article!