November 3rd, 1997

PERFORMANCE MISCONCEPTIONS
CPU speed is often touted as the end-all-be-all of computer performance. This has been exacerbated by the speed wars ignited by Power Computing and other clone efforts. In reality, the CPU can often stand around waiting for information from other aspects of the system. This is why the adoption of PCI slots over NuBus slots was so important as the PCI architecture communicates much faster. In the same way, it has become imperative for the system bus to achieve some sort of speed bump past the 50-MHz that until recently was the fastest in the Mac world. PCs have enjoyed a 66 MHZ system bus for some time and now it is arriving for us too.

Heretofore, Power Computing had the fastest shipping system bus speed of 60-MHz on the PowerCenter Pro model. Tatung, the Asian cloner awaiting CHRP certification from Apple, had a 75 MHz system bus system ready to ship. As Motorola and IBM bid farewell to the clone market Tatung, an IBM sub-licensee, left too.

Power Computing was rumored to have 100-MHz systems around the corner and even Apple supposedly has an 83 MHz system they might release sometime after the new year. All of these increases would have meant major performance increases for Mac OS systems that would continue to allow the advantages of the PowerPC chip show its superiority. As it is, Motorola was ready to ship the first 66-MHz system, and now Apple will have that honor with the upcoming release of the Gossamer based G3 systems on November 10th.

F E A T U R E

Yesterday's Future Tomorrow
Motorola's StarMax Pro 6000/266 offers a glimpse at the future of Macintosh hardware technologies, some of which will be seen with Apple's G3 introductions next week.

November 3, 1997
by: Bryan Chaffin ([email protected])

Apple's new G3 systems are to be released in one week on November 10th. Webintosh Labs is going to give you a sneak peak at what we might expect from these systems as well as future CHRP and G3 implementations. There has been much controversy in recent months over CHRP so we are taking a close look at what should have been the first CHRP machine, the StarMax 6000/266. The purpose of this review is to explore what the Mac world might expect in the future if Apple, or possibly Umax, deliver a shipping CHRP system.

Things that make you go Hmmm...
The StarMax 6000/266 is not as beefy as its bigger brother the StarMax 6000/300. Its Backside Level 2 cache is 512k, set in a 1:2 (in this case that means the Level 2 cache is operating at 133-MHz) ratio that puts it behind such erstwhile competitors as the PowerTower Pro 275 with its 1:1 implementation and the StarMax 6000/300 which has a 2:3 (or 200 MHz) ratio. This 1:2 setup is faster than the in-line cache systems in the PowerMac 9600s and 8600s. The StarMax 6000 would have brought two firsts to the Mac market, and that was the 66 MHz system bus and the use of SDRAM. The upcoming Gossamer based machines coming from Apple next week also uses a 66-MHz system bus and SDRAM; neither of these officially stem from CHRP technology.

SDRAM is the fastest widely available RAM for PCs. It has a 10 nanosecond access speed that is much faster than the common Mac RAM speeds of 60-70 nanoseconds. This represents another major stepping stone in eradicating the gains made by PC manufacturers in recent years. Bringing these technologies to the Mac world was part of the original CHRP mission.

Let it not be said that the Motorola engineers lacked a sense of humor. When we first started up the StarMax 6000 we were greeted with a sort of anti-DOS screen. This was a white screen with black letters that gave boot information at the same time as a very Windows-like start-up trill graced the speakers. This was a very chilling moment for everyone at Webintosh Labs that quickly erupted into gales of laughter. For those of us with heart conditions, the friendly, smiling, happy Welcome To Mac OS screen soon took over and from there on out it was nothing but the usual Mac OS that we all know and love.

An aspect to this CHRP system that may or may not factor into the future of Apple branded machines was the presence of PC serial, PS2 and parallel ports. These ports allow the use of PC mice, keyboards, and other input devices. The reason this may be important is because of the ability to use much cheaper PC standard devices. It is important to remember that the ability to use a 2-3 button PC mouse on your Mac does not mean that the additional buttons will do anything on your Mac OS system. That sort of feature is only to be found when built into the OS or controlled by software such as found with the Kensington brand mice (made possible by Kensington MouseWorks). The PC ports also represent the possibility of PC peripheral devices, including printers, but as with input device, Mac specific drivers would have to be developed.

We were able to easily connect a PS2 mouse and keyboard up to the StarMax 6000. The mouse worked just as a mouse should with nothing remarkable to mention (which, in itself, is interesting). It was a tad bit disconcerting for some of our testers to use a keyboard with the Windows logo on it. Many people will not find a PC keyboard an acceptable option since there is no Command (or "Apple") key for keyboard shortcuts. If CHRP were to become a reality however, the free market would likely respond quickly with some sort of fix for this.

Form Factor
The inside of the machine is a bit different from what you might expect. Frankly, the inside design is not as elegant or easy to work in as a typical Apple design. To put things in perspective however, the StarMax 6000 was not built for form factor considerations. The footprint of the G3 chip is quite small and the cache is built onto a card that connects directly onto the chip. This is part of what makes Backside Cache possible. There are two IDE connectors on the motherboard, as well as Ultra SCSI connectors including the Zip Drive, CD-ROM, and 4 GB hard drive. The PCI slots are on a riser card coming off the motherboard that maximizes space. This feat is made easier by the low wattage and heat output from the G3 chip. Our StarMax was equipped with an IMS Twin-Turbo 8 MB graphics card in addition to the 1 MB of on board VRAM. On a critical note, it is necessary to unscrew the bottom plate of the casing structure in order to screw or unscrew any cards onto the PCI slots. While some people may not mind leaving their cards unsecured, this was a really poor design shortcoming to an otherwise functional case design.

The industrial design of the outside case is one of the best we have seen on any of the clone offerings. It marks a positive departure from earlier Motorola models but Apple still has the best industrial design in our book. The back of the machine is marked and labeled very carefully, which is important because of the new additions of the various PC ports.

Performance
One of the best aspects of working in Webintosh Labs is the opportunity to play with the latest/greatest technology. Fast computers certainly fall into this category and if the StarMax 6000 is any indication of what we might expect from Apple in the future, we (the Mac world) are in for a wild ride. No matter how they are implemented, the new G3 chips are nothing short of awe inspiring. They are fast. Webintosh Labs has had the pleasure of testing two of these systems now (see "Fastest Computer You'll Never See"), and they both smoked. They perform much faster than any Intel offering and they use a fraction of the power while generating very little heat at the same time. If the Mac OS was not saddled with so much legacy emulation code, there would be almost no way for the PC industry to keep pace. To be blunt, the AIM group has produced a fantastic design in the G3 architecture.

We performed a battery of tests in Photoshop on a 19 MB image to put the StarMax 6000 through its paces and we were not disappointed. As the point of this review is to discuss the possibilities of CHRP in the future, we will not get into specifics. Every single filter and effect we applied was processed in a remarkably short time.

Copying, transferring, and downloading files was equally brisk. We downloaded multiple files from the Internet including large and small file sizes on a 128k connection and were delighted. We transferred files between partitioned hard drives and across an AppleTalk network individually and at the same time. Even large collections of small files were speedily handled.

More of the Hmmm...
We received the StarMax 6000 with System 7.6.1 for CHRP installed on it. 7.6.1 for CHRP was the last system specially modified for CHRP. Motorola's engineers made the StarMax 6000 not dependent upon ROM-less System 8 CHRP code in view of Apple's stalling tactics on certifying CHRP systems. As such, we were able to install a normal version of System 8. After running a utility program Motorola engineers wrote to bypass Apple's ROM code that deactivated the Level 2 cache, the machine performed flawlessly. Once again, this was on a normal version of System 8.

Interestingly enough however, the MacBench scores we ran under both systems were different. The results obtained under 7.6.1 for CHRP were noticeably higher, most specifically on the Floating Point scores. The scores were as follows:

Test
7.6.1 for
CHRP
Mac OS 8
Processor
906
887
Floating Point
849
812
Disk
751
745
Publishing Disk
283
280
Graphics
789
755
Lo-Res Graphics
810
775
CD-ROM
505
500

Conclusion
The promise of CHRP has been long in the coming and even still might never reach the open market. Apple has vacillated on whether or not they will pursue their own CHRP technologies, but the signs look good. Umax was rumored to be demonstrating a CHRP system in Europe in early October, but quickly squashed any hint that they would actually release it. The StarMax 6000 from Motorola was the very first machine incorporating CHRP standards, and their implementation of this untried technology is impressive. While the StarMax will never see the commercial light of day, it shows the promise of what we can hope for in the future. Hopefully Apple too will incorporate some CHRP standards in the near future.