My recent call for reader responses to Apple Computers latest generation of Power Mac G4 towers had two paradoxical effects: While the wealth of articulate opinion (both pro and con) proved manifestly worthy of a column, its sheer volume left me busily exercising my correspondence skills in an effort to answer each e-mail individually—instead of actually delivering the planned follow-up.
Meanwhile, some additional controversy over the new machines threatened to render anachronistic my initial argument that the systems many architectural improvements outweighed the modest performance improvements in the twin Motorola PowerPC G4 chips powering each new model.
Specifically, early benchmarks posted on the Bare Feats site just hours after my column hit eWEEK suggest that those Moto CPUs are unable to take full advantage of the DDR memory in the new machines—a finding that many readers cited as a deal-breaker.
On the other hand, I heard from plenty of readers who have already shelled out for a new dual-processor system; while some of them expressed mild disappointment in the rate of PowerPC evolution between Januarys Power Macs and these models, they argued forcefully for the position that these systems represent a great leap forward for price and performance.
Along the way, readers joined me in musing about the effect of Mac OS X on professional markets and the widely discussed prospect of Apples abandonment of Motorola processors in favor of a Mac-friendly version of IBMs 64-bit Power4 chip.
Without further ado, heres a sampling of those responses:
I just ordered a new dual-867MHz G4 desktop, along with a 17-inch Apple LCD display. Ill admit I was hoping for more … at least 1GHz on the low end, with 1.4GHz or better on the high end.
Even so, the other architecture changes to the hardware platform make the lower-than-expected CPU speeds worth the money. Dual CPUs, combined with improvements in Mac OS X 10.2 (Jaguar) and the move to ATA/100 and DDR SDRAM will pack a big punch.
I can also (finally) add a second 5.25-inch device to the second media bay, and even better, can now have up to four hard drives in a RAID array. My point: Theres plenty of room for growth, something Apple typically doesnt accommodate (at least not since the days of the Power Mac 9600).
But why not wait for the next speed bump, and possibly the 7470s or Power4s? I have a rule I follow when buying computers: Never wait for the next “great thing” at the expense of productivity today. Just when that great thing comes along, so will some other technology (533MHz system bus! 800Mbit FireWire! Gigabit wireless!) thats expected in a soon-to-be-released model of the future.
The new Mac I just bought will meet my needs for the next 24 months at least. Beyond that … Well, its only money!
Troy S. Curtis
Director, Technology Services
University of Phoenix—Phoenix Campus
I bought a dual-867 system with DVD-R a while back. The new ones just dont make sense, especially with the bottleneck.
If Apple came out with a Power4 machine, Id sell my left arm to get one. But until then, Ill wait. The incremental increases are not enough—Apple needs to shake up their pro line like they reinvigorated the iMac and laptop lines. Its been far too long.
Power Mac G4 Letters,
Power Mac G4 Letters, Page 2
Thanks for giving me an opportunity to vent. The new CPU speed of 1.25GHz (and only a 20 percent increase in gigaflops) is again a lame upgrade, as it was last January.
I am anxious to upgrade, but I want some beef on the bun. Using old chipsets, ATA 66 and marginal speed increases is not what I want to get for my four grand, which is what my system will cost. (Thank goodness for the new audio inputs, though!)
If you look at some of the sites that have run benchmark testing, the DDR in the current architecture seems to add virtually nothing to the overall performance of the machine.
I have wanted to upgrade for nine months. I want to do video editing and burning. I was ready (really, not simply rhetoric like so many!) to purchase this August. 1.4GHz might have been enough to put me over the top, but this current round is lame. Again. And this comes on the heels of the .Mac fiasco. Lets hope Jaguar turns out to be a good thing (and my scanner and CD-RW will work!)
As a seasoned Mac user/purchaser, I know the drill: Dont buy before a Macworld Expo, and beware of rebates. And despite what Apple thinks, I depend on the “rumor” sites for news. You have to dig for it, but its better than the tea leaves Steve would have us use. On the basis of this “illegal” buzz, I was poised. Ive converted the last of my dongles to USB. Even purchased a Marathon desk mount in preparation.
I was really disappointed by the “bumps” and even more by the underwhelming benchmarks on Bare Feats. The absence of [Motorolas new PowerPC G4] 7470s on a motherboard they seem clearly designed for makes these machines feel like stop-gap hacks. So Ill wait for more benchmarks and shakedown observations. Unless something dramatic surfaces, Im leaning toward picking up a Classic dual-gig: established design, 300 bucks less, and the RAM is cheap.
Its not like I can upgrade the CPUs on a new dual-gig, so Ill wait until these motherboards have the right CPU and a new chassis … At least, thats what the rumor sites are promising!
I think all of this howling about bus speeds and processor speeds is freakin ridiculous. The G4 started out really far ahead of Intel, but Motorola wasnt able to keep up in megahertz like we all hoped they would. However, they had such a lead for a while it didnt matter.
Now I think they are still pretty much on level with the PC doing graphical things—they just dont feel as snappy in the interface because of the slower bus and Mac OS X overhead (with all the cool stuff that it does). In my experience, the G4 still is a monster at crunching media files, especially the dual models. I have a dual 867 and it is a screamer; my students who have PCs at home just love that machine when doing Photoshop, Maya and Cinema 4D.
P.S. I have ordered three of the new machines this week: one dual-867MHz and two dual-gigahertz.
Power Mac G4 Letters,
Power Mac G4 Letters, Page 3
Nice to see an all-dual-processor line, but I am disappointed in the mild bump up in speed. Im also somewhat disappointed about the lack of USB 2.0, although I understand that Apple does not want to sabotage the next generation of FireWire. This model, I think, is the Band-Aid for the next model, which I anticipate in July 2003, not January.
Ive been watching the new Power Macs for about a year, and I still havent found a release compelling enough to step up. The PPC processors dont just hinder the new systems, they cripple them. With AMD advancing HyperTransport and Wintel getting up to 3GHz (even though their instruction sets are a mile long), the masses go for raw numbers—and theyve been trained in gigahertz, not gigaflops.
If the modified Power4 was to be implemented, I would buy the day the machines were released. But right now, the Power Mac line is a little too wimpy for the money.
Am I ready to shell out for a new Mac? No; Im coasting along quite well personally with my G4 Cube at 450MHz.
That aside, I anxiously await the arrival of new Apple hardware at the first sniff that theyre due to materialize. Im glad there are some architectural improvements in the latest crop of Macs, but it seems more like a stop-gap measure than anything to get super-excited about.
The Mac rumor mill had been churning away like gangbusters since January. Partnerships with nVidia and industry alliances with AMD were spun into fabulous rumors that Macs would soon adopt an nForce-like motherboard strategy, or move to a new, high-speed architecture fueled by HyperTransport. Neither materialized. Frankly, I think Apple spun things well by not subjecting this relatively ho-hum announcement to the electron microscope of an official Apple Event, but I was disappointed anyway.
Ill start getting excited about Macs again when:
1. Apple adopts CPUs that can take advantage of high bus speeds and DDR timings (for memory and for other system processes). [PowerPC] 7470s, 7500s … Bring em on.
1b. Bandwidth must get wider and megahertz must increase, even if its only for bragging rights.
The new 64-bit PowerPC IBMs announcing seems like an excellent candidate to fulfill 1 and 1b, and HyperTransport neatly ties into the widely bandied 6.4GB/s figure. I cant imagine that it would be that hard to recompile certain Altivec operations to support the new SIMD extensions. Heck, IBM might even just put Altivec-on-steroids in there if they want Apples business.
2. Macs have got to become more price-competitive. Its amazing Macs are as inexpensive as they are, considering the company makes the OS and the hardware—but I think industrial design is kind of “over” now.
I suspect if Jonathan Ive could come up with rugged case designs made of galvanized aluminum instead of polycarbonate, use recycled IBM Selectric keyboard mechanisms (solid as a rock!) and Apple was willing to follow a dollar-per-megahertz pricing strategy (instead of almost double that), theyd shift a lot more units. Call it the Mac Front Line, a series of low-cost, indestructible Soviet Chic portables and slimtops painted in a silvery winter-camouflage motif, or even just left with the natural mottling of the galvanizing process. Oh and a big red … Apple. Theyd be heavy, sure, but revolutionary :)
2b. Apples use of low-watt processors is commendable for environmental reasons as well as practical ones—and the ever-shrinking die process of the PowerPC means they are, megahertz-for- megahertz, cooler running than Intel or AMD chips—but they could lower the price yet again (and boost their price/performance perception) by going for higher-wattage, larger-process (or alternative-process, like BiCMOS – remember Exponentials X704?), ultra-high-speed PPCs.
3. Apple lowers the price of iPods by $150 across the board.
4. Apple buys a really good PC/PS2 game studio and gets them to crank out games for OS X. (I suggest Level 5, who make the lovely RPG/Action Dark Cloud series.)
Interstar Technologies Inc.
Power Mac G4 Letters,
Power Mac G4 Letters, Page 4
There are three things that make this upgrade interesting (not looking at megahertz). They are:
1. Dual CPUs. It is true that many applications are not dual CPU-aware. However, Mac OS X is fully multiprocessor aware. The G4 architecture allows for very nice performance scaling with jobs that are inherently parallel—image processing, video editing, FFTs, Monte Carlo simulations—but lets discount all that for a minute.
At the very least, when a user is doing a computationally intensive task, one CPU takes up all the computational load, and the other CPU takes up the overhead of simply running other tasks, such as playing music, surfing internet, etc. Let us give a 25 percent performance benefit for having two CPUs. In other words, the performance megahertz is 1.25 GHz X 1.25 = 1.5625GHz (equivalent performance).
2. Mac OS X 10.2 optimization. I have not used it yet, but if you look at the info circulating on the Web, you find that 10.2 uses the graphics chips (Quartz Extreme) to do GUI jobs. These new systems ship with a very capable graphics card. In addition, if popular reports are to be believed, OS 10.2 has extensive improvements in stack management, multi-threading etc.
All these improvements, with a suitable hardware should eke out at least another 10 percent performance improvement. In other words: 1.5625 X 1.1 = 1.71875GHz.
3. Architectural improvements. There are some modest architectural improvements with DDR RAM, not having a PCI bridge (as happens in PCs), and directly connecting external FireWire devices and Gigabit Ethernet cards to the bus without going through PCI traffic.
This should eke out another, perhaps, 10% benefit depending on the task. (I think Apple really was hampered here because the system bus and the memory bus arent running at higher speeds—perhaps because of Motorola). In any event, the 10 percent benefit should result in 1.71875 X 1.1 = 1.9GHz.
If you throw in another 15 percent floating-point performance benefit associated with the PowerPC G4 (compared to a run-of-the-mill Pentium 4), you get something close to a 1.9 X 1.15 = 2.2GHz Pentium 4 performance.
If I were a power user using the best of yesterdays G4s, these upgrades represent a 80 percent to 90 percent improvement right off the bat. (These are conservative estimates.) That is nothing to sneeze at.
Of course, under more ideal conditions—such as running an Altivec-optimized, multi-CPU-aware application like Photoshop), and doing a parallel-processing task (running two batch jobs simultaneously), you will be kicking some serious performance gain with a *nix OS running it.
I think if Motorola really gets faster CPUs to Apple, they will be kicking the P4s butt. Moto keeps throwing lemons, and somehow Apple has been making the best lemonade (so far).
Perhaps someone like you should suggest Apple start advertising these as 2.0GHz Pentium 4-performance machines!
Power Mac G4 Letters,
Power Mac G4 Letters, Page 5
Ive been using a Power Mac 8600 (running Mac OS 8.6) for almost five years now. Im ready for a new Power Mac, and I thought I was going to buy one after Aug. 13. But Im not happy with the bus speed from G4 to the controller. I know the two top-end Macs have had their bus speed increased 25 percent, but its not good enough for me; I was hoping for DDR from the G4 to the controller.
What kills me is that I cant run Mac OS X, and Mac OS X 10.2 looks like a big improvement over 10.1. If the temptation of 10.2 is too much and I cant hold out any longer, Im going to buy either the low-end dual or a stripped-down new 1GHz.
I recently opened my own law firm and am using an 800MHz iMac, together with a variety of peripherals. I will need to buy a new Mac within the next few months, but Im not buying one of these new Power Macs.
Why? Frankly, the new iMac is too good of a deal. I dont think these new Power Macs offer the bang-for-the-buck difference over an eMac or iMac. Especially when I would have to spend about $1,700 on the Power Mac and then buy an LCD monitor. For that kind of money, I could buy a new iMac and a very nice FireWire scanner with autodoc feeder (once native Mac OS X scanning apps come out! %@#$%^#$$^)
If the next Power Macs sport this new IBM processor, and the prices are similar to current Power Mac prices, now youd have a compelling reason to spend $2,500-plus on a Power Mac. But right now, the difference in performance for the price simply doesnt justify it for me.
Larry J. Busch, Jr.
Attorney at Law
Busch Law Center, LLC Lees Summit, MO