Macs vs. PCs

Although I am a long-time Mac User, I recently bought a Wintel machine as a supplement to my Power Mac 7300, since the Eastern Michigan University Computer Science department recently decided to switch to a slower, larger, newer Microsoft C++ compiler that doesn't work well under RealPC. Pretty soon, we'll be switching to Java as well, but I will hopefully have graduated.

I've stripped various 7x00 series machines down to the bones several times, so I'm comfortable inside a computer. After consulting with some friends, I decided that the build-it-yourself approach was the best and potentially the least expensive way of acquiring a PC. Some problems arose when I failed to realize that a PC floppy drive, unlike a Mac, requires a seperate power cable. Otherwise, the assembly was simple enough. After recieving two largeish boxes from the UPS guy, I was playing MechWarrior4 quite happily within three hours. Admittedly, it took me a while to get the Windows CD booting right, since I was a tad confused by the hard drive format utility. Gladly, I should never have to do that again.

Since I had some spare time and have both machines sitting next to each other, I decided to run some benchmarks. I admit freely that these are not scientific by any stretch - the machines are configured differently, and I wasn't about to go spending money so that the PC had the same software as the Mac. I simply compared the time it took for the two computers to perform the same tasks using the software that was available. Since I had always thought of PCs as the brute-force machines and Macs as "slower" but more elegant, the results surprised me somewhat.

The Winner

Working strictly by points, the Wintel box won 3-2. (Assuming both machines got a point for the Web browsing test, which was a draw by any measure.)

In terms of price/performance, however, I think the machines are a wash. Macs have historically cost about 25% more than equivalent PCs. Adjusting for that, the Mac was actually $40 less expensive than the Wintel box. Even without the adjustment, the prices were pretty similar, and it is difficult to say that a price/performance winner emerged, since the two machines hand off superiority at different tasks.

On the other hand, I didn't do any tests where the extra $170 I spent on the Mac's video card might come into play, such as Unreal Tournament of Quake III. While I have demos of these and other 3D programs for the Mac, I am not about to download the Win9x versions over a 56k modem.

Also, the Mac is a hot-rodded version of an entry level business computer from 1997 which, if it was lucky, might be considered almost-kinda cutting edge for 1999. (Except for the Radeon.) The Wintel, on the other hand, is all new components, some of which didn't exist a year ago.

Consider what these results imply about my machines potential performance vs. a 1999 vintage Pentium II or III at 500 Mhz.

Finally, it should tell you something that I am writing this on the Mac. The AMD is sitting here as well, crunching a SETI work unit. A $4 keybord, in short, sucks and a $6 mouse isn't much better. Adding components of the same quality as the Apple keyboard and mouse would have cost me another $100. Finally, Office XP for Windows is really, really expensive - and I already have AppleWorks. Each according to ability, right?

Technical Information

	7300/G4	1.0 GHz AMD
CPU	PPC 7400 (G4)	AMD Athlon
Speed	350 MHz	1 GHz
CPU Bus	50 MHz	133/266 MHz
Storage	Fast SCSI (10 MB/s)	ATA/100 (Running @ ATA/66)
Main hard drive	9 GB 7200 rpm IBM	15 GB 7200 rpm Seagate
Secondary hard drive	2 GB 7200 rpm Seagate	none
CD-ROM	12x CD	52x CD
RAM	160 MB DRAM	128 MB DDR-RAM
Max RAM	512 MB	3 GB
RAM Speed	50 MHz	133 MHz (266 MHz)
Video Card	ATI Radeon PCI	Generic TNT2 M64 AGP4x
VRAM	32 MB DDR	32 MB SDRAM
Expansion	3 PCI - 1 Free	5 PCI - 3 Free, 1 CNR / AMR (shared, free), 1 4x AGP (used)
Price As Configured (approx):	7300 - $1000	AMD - $840

Both machines have 56k modems, 2 USB ports, 17" VGA monitors, ethernet, and floppy drives.

Tests

MP3 Encoding

A 07.10 duration audio CD track was encoded directly from the computer's internal 12x SCSI CD-ROM.

7300/G4, SoundJam 2.5.2, Faster, 128 kbps, Joint Stereo, 66 Seconds (6.5x)

A 06.28 duration audio CD track was encoded directly from the computer's internal 52x ATA/33 CD-ROM.

1.0 GHz AMD, FreeRIP MP3 1.0.0.1, 128 kbps, 84 seconds (4.6x)

Conclusions: I selected FreeRIP because it was free and because the archived download was about 700k. However, there were no options for adjusting quality settings, except for bitrate.

In addition to being slower than the Mac, the Wintel-produced MP3 file was quite a bit muddier sounding than the original - a problem I don't generally have with SoundJam.

SoundJam would be my preferred method of ripping because of sound quality, even if the Wintel box was faster.

SETI@Home

• 7300/G4: WU1 = 17hrs 55min 04.4s, WU2 = 18 hrs, 38 min, 00.5s, WU3 = 19 hrs, 10 min, 30.1s, WU4 = 22 hrs, 28 min, 59.3s, Average = 19 hrs, 33 min, 8.575s
• 1.0 GHz AMD: WU1 = 9hrs, 08min, 52s, WU2 = 7hrs, 55min, 45s, WU3 = 12hrs, 16min, 26s, WU4 = 7hrs, 31min, 00s, Average = 9hrs, 13min, 0.75s

Conclusions: Obviously, since these weren't the same work units, I can't guarantee more than an approximate measure of relative performance. Still, the numbers are such that you probably have a good idea of the relationship between the two machines. This new AMD box, besides playing a mean game of MechWarrior4, is going to vastly improve my average SETI crunch times. (Team MacAddict!)

Web Browsing

Given a blank browser window, the URL is typed in and the enter key is pressed. Time is taken from the final key press until "Done" message appears or "Stop" button is greyed out.

7300/G4:

http://www.yahoo.com =

• Netscape 4.7: 8s
• IE 05. 7s
• IE 2.1: 12s

http://eshop.macsales.com =

• Netscape 4.7: 25s
• IE 05. 22s
• IE 2.1: 30s

http://www.lowendmac.com =

• Netscape 4.7: 19s
• IE 05. 17s
• IE 2.1: N/A

http://www.macromedia.com =

• Netscape 4.7: 19s
• IE 05. 18s
• IE 2.1: 25s

1.0 GHz AMD:

• http://www.yahoo.com = IE 05. 6s
• http://eshop.macsales.com = IE 05. 41s
• http://www.lowendmac.com = IE 05. 8s
• http://www.macromedia.com = IE 05. 30s

Conclusions: Considering the mighty reputation of Wintel machines when it comes to HTML rendering, I was honestly expecting better.

I was, however, using a stopwatch - I couldn't tell you exactly how many clock cycles or processor ticks were used, or if components other than the CPU affected performance.

The stopwatch merely told me what my eyes had been saying for a long time: the machines are too close for it to matter which one is faster.

RealVideo/Audio Encoding

Unfortunately, SoundJam is the only program I have which uses AltiVec, so it's the only program I have which would grant the 7300 any kind of edge (or, perhaps, any kind of chance) here.

Convert 11.32.06 Sorensen Encoded Quicktime Video to RealVideo:

Convert 12.26 Duration (125.6 MB) WAV file to RealAudio:

Quicktime Movie - Normal Motion Video, Voice Only, 56k Single Rate Stream

• 7300/G4 - RealProducer G2 - 59 min, 15 sec
• 1.0 GHz AMD - (Windows version of RealProducer 8.5 does not support compressed Quicktime.)

WAV File - No Video, Stereo Audio, 56k Single Rate Stream

• 7300/G4 - RealProducer G2 - 3 min, 12 sec
• 1.0 GHz AMD - RealProducer 8.5 - 37 sec.

Given the results of the MP3 tests, the similarity in the web browsing tests, and the relatively modest 125% speed boost in SETI@Home, I was expecting a slightly closer finish.

On the other hand, while Quicktime 4 for Windows played the video just fine, the Windows version of RealProducer 8.5 couldn't make heads or tails of the Sorenson video (an old Macaddict staff video.) so I guess the Mac gets an honorable mention here.