Will History Repeat Itself?
I get letters . . . .
A few months ago I added a fifth machine to my network and I switched from my normal routine of overclocked P3’s to an XP1700 overclocked. The heat and the fan noise were hardly worth the trouble. I liked the AMD machines performance but I did not unlock it. It was not worth the time or effort to me.
I sold that machine and my 4 others and built 4 new 1.6a Northwoods @ 2.5Gz they run cool no unlocking required and they are silent with the Intel boxed fans. I also build systems for many other people and purchase many CPU’S per year all Intel now again. I know competition is good for prices but AMD needs to give me something for my money at least equal to Intel before I try another one of their chips. Right now I do not see an benefit to purchasing AMD with overclocking in mind.”
I agree with you on AMD moves to make it more difficult and confusing to overclock the Athlon XP. I believe one of the main reason why he Athlons had such a strong “underground” support was for there easy ocing. Simple penciling.
Now it is becoming too much of a pain. I recently bought 2 P4 1.6a and assemble 2 new PC and I have them both running at 2.4ghz with very little work except a BIOS adjustment.
You’re getting greedy aren’t you?! An EASY 800Mhz (50%) overclock and you’re
still not happy, complaining about the problems. I wish AMD was getting
50% overclocks. If I could buy a 1333-1400Mhz XP and overclock it to 2 –
2.1Ghz thats what I would have right now instead of a P4.
I’m seeing these kinds of sentiments being expressed more and more frequently.
In the world of global CPU production, enthusiasts don’t count for much, maybe a couple percent of the total market as a whole.
However, they do tend to be a good leading indicator of which way the PC market is going. If enthusiasts are doing something today, odds are everybody will be doing it later.
Enthusiasts jumped on the AMD bandwagon well before the significant rise in AMD marketshare. If they start jumping off it . . . .
Why AMD Took Over: Good and Bad
There were reasons why AMD grabbed most of the enthusiast market. Two of them were even good reasons.
Good reason number one was that they significantly outperformed the PIIIs and the first generation of Willamettes.
Good reason number two was that they were cheap.
Now the others:
They had higher MHz numbers than the PIIIs.
If you picked well, you got a significant percentage overclock out of them.
You could unlock the multiplier easily with a pencil.
What advantages did Intel have during this?
They had a reputation for quality mobo chipsets.
Most people had Intel platforms and were reluctant to buy a whole new platform.
They used less power and ran cooler than AMD processors.
None of these kept kept most people SOI (Slaves of Intel).
So people swallowed the heat and the fan noise and the sometimes dubious stability of AMD platforms for the advantages.
The Empire Strikes Back
Let’s see what the situation is nowadays for the XPs (current and Thoroughbreds):
Price/Performance: Edge, Intel
Current XPs trail PIVs in performance by a bit, and there’s no chance AMD can pull significantly ahead with Thoroughbred them. The next window of opportunity for AMD on the performance front will be Hammer.
While Athlons are still cheap, there’s also cheap Intel processors that overclocked can match or pull ahead of them a bit in reality.
So AMD has lost its two big legitimate advantages. At best, it will be roughly even.
The “Numbers” Are On the Side of Intel
While AMD processors in reality are keeping it close in the enthusiast world, the Intel processors are throwing up the big numbers.
True, an XP at 1800MHz is little worse than a Northwood at 2400 or 2500MHz, but tell that to the cognitively challenged, which includes just about all your non-geekish friends, and enough of the geekish ones.
AMD’s PR ratings have not been truly accepted by enthusiasts. While they’ve been rejected by relatively few, you rarely if ever see an overclocker describe his AMD effort with a PR rating.
The Big Overclocks Are On the Side Of Intel
The typical AMD overclock is now relatively small: 20%-30%.
In contrast, someone buying a 1.6A today can almost certainly get a 50% overclock with just an FSB settings change and perhaps a slight voltage increase.
The AMD fan will say that you have to overclock a 1.6A 50% to do 10% better than a 20% XP overclock, and the PIV fan will say, “Yes, isn’t it wonderful?”
The Easy Overclocks Are On the Side of Intel
XPs made unlocking much more difficult than a few up-and-down pencil strokes. Unlocking is difficult enough to keep a large minority-small majority of overclockers from doing it. Pure FSB overclocking has been hampered until recently by the lack of key components that would make FSB overclocking relatively easy.
Now if you do three voltage mods in the morning, then wake up, you might find that hard to believe and you’ll probably not think much of that group. Well, you’re hardcore, and they’re softcore, but then again, more people buy Playboy than . . . never mind. 🙂
The AMD fan will say that his 30% overclock was a far greater accomplishment than the PIV owner’s 50%, and the PIV owner will say “What’s bigger, 50% or 30%.”
Now the AMD fan is right on both points, but look deeper, and it’s really an argument for masochism.
The PIV user can say, “if I can at least match you with next to no effort, what good is your effort?”
Ease-Of-Mind Is On the Side of Intel
Toleration of faults and foibles for a greater good ends when the greater good ends.
You’re in a group of people. I show up and show everyone what looks to be two lumps of clay and ask you to carve a little face in one of them. I tell you that one lump really is clay, and the other one is plastic explosive. I give you instructions, and tell you that carving the plastic explosive is no big deal provided you follow the instructions and take the proper precautions, and don’t do anything really stupid.
I offer each person $1 to carve the clay and $100 to carve the explosives. Since all of you were together because you were on the unemployment line, you all really need the money.
So most of you go for the $100, and only a relatively few blow themselves up.
Next week, I come back with the same offer, but this time, carving the plastic explosives will only get you $1.
What do you do? What do you do if the week before left you stressed out?
People swallowed the risks of burning up or crushing Athlons when it got them something. That didn’t mean most liked it. They accepted it because they got better performance out of it.
The same applied to the noise of Delta 38s or the weight of copper bricks, or the quirkiness of Via chipsets.
But when you have a solution that you can’t burn up or crush that’s quiet (at least at the moment) and can’t be used as a lethal weapon, and has at least the reputation of being reliable, and all that does as well or better than your old solution, again, we get back to the love of masochism.
The AMD Advantages: Inertia and Ideology
Like Intel a couple years ago, AMD does have the advantage of an established base of users who aren’t willing to incur the additional expenses associated with changing platform.
This will be a bigger advantage to AMD than it was to Intel because unlike a few years ago when the shoe was on the other foot, overclocked Intel platforms will not have a marked performance advantage over AMD platforms.
AMD holds another advantage in that there’s not one but two lights at the end of the tunnel. Hammer is less than six months away for those willing to jump away from Socket A, and Barton isn’t much further away for those who aren’t.
It’s easy to make the case for the PIV for a new system. It’s not easy at all to make the case to change from a socket A to socket 478 if the socket A owner isn’t already dissatisfied.
For many enthusiasts, the XP in a Throughbred XP should stand for Xtra Procrastination. 🙂
Some people find buying AMD an ideological statement: rooting for the underdog, fear of monopoly, hatred of the big and powerful. However, the ideologically committed are a relatively small minority within the AMD tent. If Intel offers a significantly better deal, most will jump ship, just as they did the Intel ship a few years back.
Dilberts vs. Dodos
There’s a key philosophical difference between AMD and Intel. AMD thinks it’s Dilbert selling to Dilberts. Whatever Intel is, they realize that they’re selling to Dilbert’s boss.
AMD’s marketing seems to be done by engineers. “If we build it, they will come, and if they don’t, it’s their fault.” If they build a good product at a reasonable price, their job is finished.
Intel, on the other hand, takes the more modern marketing approach of “give the people what they really want; ego gratification. Pander, pander, pander.”
See how the two advertise to average people. First, AMD hardly does it, while Intel does it all the time, and the gap is wider than the company’s relative sizes. People like to be asked, “Please buy our products?”
The typical AMD ad will be product-oriented, “This is good, it’s cheap, so buy it.” What else could people want?
Apparently, some ass-kissing, which Intel ads provide them. Intel ads are pander-oriented, “This will make you feel good, important, powerful, yet safe.”
This difference in philosophy continues when it comes to the care and feeding of their products. While not apparent to the average computer owner, it is to us.
AMD thinks you should know what you’re doing when you do it yourself with their products; Intel doesn’t.
You have a delicate CPU? AMD says, “Be careful.” Intel comes up with protective heat spreaders and idiot-proof clamps and cages. CPUs overheat due to faulty heatsink placement? AMD builds relative furnances, but automatic shutdown is a low priority. Intel builds cooler CPUs but makes automatic shutdown work in its CPUs. Intel essentially says, “You bumbling little darlings, don’t worry, Uncle Intel will make this nice and easy for you.” AMD, in contrast, seems to take the attitude, “if you don’t know enough to do it right, learn.”
It’s the difference between taking care of Dilberts vs. taking care of dodos for whom ease and convenience are important.
While neither AMD nor Intel like the idea of overclocking, the two companies take much different approaches to it.
AMD lets you change multiplier and increases FSBs, but hassles you about it, whether it’s trenches that need to be filled, or PCI divisors that are miserly doled out, one at a time.
AMD’s effective attitude is, “You must prove yourself worthy of this.”
Intel either stops you cold with multiplier locks, or does nothing about it (FSB, outside of their own products).
Intel has to know what people are doing with 1.6As, and they obviously don’t care enough to do anything about it. Contrast that to what AMD did to make unlocking a challenge.
What AMD Needs To Learn From Intel
AMD needs to realize that they’re selling to people, not engineers. Even among the enthusiasts who think themselves quasi-engineers. On the whole, most are really are more like Dilbert’s boss than Dilbert.
Hammer is a product designed and produced by and made for Dilberts. It’s not designed to rack up huge GHz numbers. It would take the average newcomer about ten minutes of benchmark comparisons to figure out how good it is. It probably won’t be terribly overclockable. Its major strength is reduced memory latency, a rallying cry for Dilberts if there ever were one.
The PIV is a product made for Dilbert’s boss (or his kid). It is designed to rack up big numbers, which means it takes one second for the average person to figure out it’s “better.” It can be very overclockable.
In short, it’s designed to appeal to the average, shallow person. Intel understands KISS “Keep It Simple, Stupid.” It also understand kissing ass pays off more than kicking it.
To illustrate all this in action, let’s put together a Dilbert cartoon:
(Dilbert’s boss, Dilbert, and co-workers at a meeting)
Dilbert’s Boss: I’ve decided what new computers to buy.
Dilbert (thinking): The man can’t count past two.
Dilbert’s Boss: Our engineers recommended AND Hammer machines over Intel.
Dilbert (thinking): Count? He can’t even say three letters right.
Dilbert’s Boss: I have learned that the Intel machines run at 3GHz, while the AND machines run at 2GHz.
Dilbert (thinking): Oh no!! The Intel salesman taught him “three.” We’re doomed!!!
Dilbert’s Boss: It seems like our brilliant engineers don’t know that three is more than two.
Dilbert (thinking): Education is futile.
Dilbert’s Boss: Does anyone disagree that three is more than two?
Dilbert(thinking): Violence is the only answer.
(Dilbert leaps upon table, throttling boss)
Dilbert: You fool! Hammer has reduced memory latency!!!
(Dilbert being led away in straight-jacket mumbling, “It’s the old 20-stage pipeline trick.”)
Dilbert’s Boss: Why did he do it?
Wally: He thinks he’s Thor.
This is marketing genius. Intel has created and positioned a product that the average person “knows” has to be better. Once the average person knows a Hammer is really a 2GHz machine, most times, the complete thought process will be “3 has to be more than 2.” and “reduced memory latency” is no answer to the average person.
Before you scoff, don’t you “know” a 50% overclock is better than a 20% overclock? Doesn’t that make you feel better, or more fulfilled? If it does, you’re no better.
Actually, the great overclockability of the 1.6A for the average person has nothing to do with the skills of the overclocker, and everything to do with Intel selling a very low rated speed version of the CPU.
You could get 200% overclocks of Athlon XPs tomorrow; provided AMD started making 600MHz XPs. See what I mean?
The true indicator is the level of performance for the price. It’s the results, not the ritual, that count. I like the 1.6A doing 2.67MHz because I get about 10% better performance than AMD’s current best for $130; not because I got a 67% overclock out of it.
If AMD came out with a 2.5GHz XP for $150 tomorrow that couldn’t overclock 1Mhz; I’d tell all of you to buy it in a heartbeat.
But I bet a lot of you would prefer something less capable, but more overclockable. That way you’d get “more” for your money.
AMD really needs to start worrying less about making products that are good, and be more worried about making products the average person “knows” are good and then making them feel good about buying their product.
There’s just so many Dilberts out there.