Stop Fanning the Flames: Seeing Beyond CPU Zealotry
I know what you're thinking.
"Here comes Jonny with his opinions again. What's wrong with that guy? Why can't he leave things alone?"
Well, I'll tell you what. I don't have any intention on changing your mind, but I do like to make you think.
This month, Dave asked me what I thought about the latest in the Pentium 4 versus the AMD Athlon debate.
All I can do is think, "When will this stop? Will it stop? When will we all live in harmony?"
We aren't living in harmony any time soon.
As I look at the benchmarks for a Pentium 4 1500 and 1700 MHz CPUs versus an Athlon 1400 MHz, I can't help but to notice that the Pentium 4 falls short on every benchmark except for Quake 3 Arena. Of course, benchmarks are not very real world, and the only time I ever use them personally is to make sure that a customer's machine is running within spec when compared to a reference machine.
Earlier this week, upon hearing that the 1.7 GHz Pentium 4 had landed, a customer asked about a potential upgrade from an Athlon 1.4 to the P4 1.7.
I couldn't help but to have a funny, poke me in the ass with a stick, look on my face when my coworker Stan said, "you'll LOVE the P4 1.7 GHz if you're a gamer."
I asked him what rock he grabbed that information out of and he responded, "Well, it's not really any FASTER, but at least it works."
I need to get that tiny bit of insight out of Stan that he must have had that made him feel that all of those Athlon machines are apparently NOT working, yet all of the P4s are.
Of course, Stan gave me the song and dance about the limited selection of 1.4 GHz approved boards, and the difficulty of finding quality RAM, and problems with multiple sticks of RAM in Athlons and cooling issues and problems with power supplies having enough capabilities on the 3.3V and 5V combined power rail...
Our current PC XT standard has many limitations... NOT JUST the CPU or the "level of quality" you have to maintain to pacify the CPU's needs.
CPU's speeds are tripling per year, while memory designs fumble around by increasing bus speeds while decreasing latency or seeing how many times we can read and write data on a single clock cycle's fall and rise on the same old bus speed. The basic principle of memory bottleneck is to STILL add more RAM, but the penalty for this is the increase of latency to keep the PC stable because of the greater capacitance load the multiple sticks of RAM introduces. Sigh. Are we back on that trip?
The focus to develop the personal computer seems to be a bit misdirected. Certainly PCs are getting much faster over all at an amazing rate, but if Intel and AMD quit making faster CPUs for a minute and could actually help to make the PC more efficient overall with an improved bus, faster RAM and maybe even 64 bit architecture, I could have my buddy HAL writing this as Arthur C. Clarke had promised.
Then again, scratch that. I can see myself playing Quake on HAL, getting pissed off at being fragged three times in a row by the same camper and being told " . . . you ought to sit down calmly, take a stress pill, and think things over." as I'm disconnected from the server. Not good.
But how much of this performance increase are we really taking advantage of? I'm willing to guess that about 30% of the people reading this have upgraded their PC in the last year. I'm also willing to guess that those people did an upgrade that increased the speed of their CPU by at least 30-40%. I'm going to go out on a limb and guess that at least 10% of that 30% have overclocked their CPU by at least 20%. I'm also willing to guess that 90% of them can't tell the difference between their 750 MHz Athlon and their 1.2 GHz or their PIII 733 and their Pentium 4 1.3 GHz without running a benchmark program.
Of course, if AMD or Intel couldn't convince the customer to senselessly upgrade their PCs, they wouldn't be getting much of the customer's money, now would they? And the PC business IS a business and making money is what business is all about.
But it seems that the industry has been testing the waters a bit lately. In an attempt to perhaps spread the wealth about the industry, they seem to be making an attempt to convince the computer customer that they NEED to buy even more parts than just a faster CPU... and it's working! Maybe because CPU cost is now no longer the most expensive component of the computer, it seems to make perfect sense to be willing to spend more money on more parts.
Think about it. Let's say that three years ago, you had a socket 7 motherboard. Your next upgrade would be a Pentium II or K6. The K6 might work on your existing board, maybe not. The PIII uses a Slot 1 board, so you'll need a new board. It's quite likely, however, that in either case you already have SDRAM and an ATX case, so you're cool there. Next upgrade is a Pentium III or Athlon. With the Athlon, you may need a better power supply, but it's still ATX. Your existing one might actually work, so you give it a try. The PIII CPU may work in your board, but worst-case scenario you have to buy another board. All and all, the upgrade path has already been fairly easy for you for the past three or four years. One or two major parts move over to the next machine, but over all, no major investment. $200 or $300 at a shot? Even if I wanted a dual processor Pentium III machine, I could just swap out the motherboard in my case and buy two matching CPUs and call it a day for about $300-$400.
>> Next Page