Rumors of RDRAMs Death Are Premature

Add Your Comments

A few weeks ago, we saw a number of . . . errrr . . . rather pro-RDRAM articles. I had a few things to say about that,
and a lot more to say in general.

We are now seeing some “Bury RDRAM NOW” articles, most notably over at Tom’s Hardware.

There are two issues here:

  • Does RDRAM stink now?
  • Will it always stink?

The answer to the first doesn’t answer the second.

The “Bury RDRAM NOW” folks cite a whole lot of benchmarks that show that RDRAM is not a cost-effective solution for Coppermine systems.

I agree. When I wrote my articles, I thought that was such a dead issue I barely mentioned it. To me, that was like whipping a long-dead horse; hitting it more won’t make it any deader; it just stirs up the flies.

I can understand why others might wish to write to counteract what essentially was Rambus propaganda.

What I do have a problem with is people saying “because RDRAM stinks now, it has to stink forever, so bury it now.”

The “Bury RDRAM NOW” folks don’t talk about Willamette and RDRAM, and there is a big potential difference between that and current implementations.

Current Intel chipsets, even the dual-channel 840, only support 133Mhz FSB, only a fraction of the total bandwidth a dual-channel RDRAM system theoretically has.

In contrast, the Tehama chipset is going to support a 400Mhz FSB. If RDRAM is any good at all, that should make some difference. None less than Bert McComas mentioned it as a potential factor in his recent Radio Wall Street interview (to be sure, Mr. McComas still prefers DDR, but if he isn’t burying RDRAM quite yet, why should you?).

Do I know for sure it will? No, of course not. How sure can anyone be of any product that won’t come out in another four months? I do know, though, there’s a good chance it will make some difference, and it will be worth another look then.

Maybe RDRAM will stink even using Tehama. If it does stink, bury it then, not now. You lose nothing by it.

Saying that RDRAM might not stink with Tehama doesn’t mean “Buy it now.” That would be absolutely foolish given the current price, even if I knew (and I certainly don’t) it would be wonderful.

If it proves to be that wonderful, we will surely see mass demand and production and a dramatic drop in prices. You want to consider RDRAM guilty until proven innocent, that’s very understandable, but why execute and entomb now?

You might ask, “Why isn’t Intel and RAMBUS trumpeting this?” and the answer is quite simple. If they too clearly point out the need for increased FSB to take full advantage of RDRAM; that leaves current systems crippled, now doesn’t it?

Intel’s language on its website for the 840 chipset is very cagey, they don’t say “133Mhz FSB;” they say “uses 100Mhz and 133Mhz processors.”

If you have or plan to get a Coppermine system, RDRAM does you no good because of the FSB. DDR probably will at some point; probably not enough for most people to justify upgrades, but more than RDRAM.

If you plan to get a Thunderbird/Duron system, I doubt you’ll have a choice between RDRAM and DDR. You’ll get DDR, at least in 2000.

Willamette’s still out there, though. Why bury RDRAM for good until we see what happens there?

RDRAM may still cost too much, even if it does provide some benefit on Willamette systems. It doesn’t have to become affordable. But it could. If it doesn’t, don’t buy it. But if it is, then you got somebody to dig up.:)

There are reasons why RDRAM might not look too good four or six or eight months from now, but there are also some reasons why it might.

Yogi Berra once said, “It ain’t over ’til the fat lady sings.” Tehama is the fat lady in this opera. If she’s lousy, leave, but it doesn’t cost you a dime to find out if she’s any good, so why walk out now?

It’s Not Judgment Day Yet

Many believe in a Judgment Day, but some people act as if God gave them a deputy’s badge and told them to start work right away.

That’s the feeling I get with these funeral articles. Why the rush?

You want to say don’t buy it now, fine. But what’s the harm in waiting a few months and see what Tehama does before rendering final judgment? Why the burial now?

RDRAM may seem like a corpse to some, and corpses usually don’t get better. In the pre-embalming days, though, every once in a while, the “corpse” did just that. Unlike real corpses, though, you can just put this one away for a few months, then look for signs of life. So why not do that?

DDR systems will probably be out towards the end of the 3Q. By then, we should have a preliminary idea if Willamette with Tehama is any good (and much more important, where prices are going). You’re going to have to wait around that long for DDR, anyway. Would it not make more sense to take another look, then decide? Why not?

If choice is important, why try to bury the only competition?

Let’s Be Completely Honest With The Honest Hatred

Many despise Rambus and RDRAM. There’s even some good reasons for it. If you don’t like how this company and/or Intel conduct themselves, that’s a perfectly legitimate reason for not doing business with them. I don’t doubt
that’s the motivation over at Tom’s Hardware.

However, you don’t have the option of buying chipsets and memory from Mother Teresa’s Sisters of Charity. No one in this business is pristine clear, so let’s not pretend that that’s the choice.

Sure, you can say “the alternatives are better,” and there’s nothing wrong with basing a judgment on moral grounds. If Intel had to sacrifice virgins to get Willamettes to run at 2Ghz, would it matter how well they worked?

However, if that is your real reason, then you should state, “These people are bad, and you should never buy from them.”

What you don’t do is justify your moral stance by stretching the technical evidence. If you want to condemn Rambus on moral grounds now and forever, fine. You can do that on cost/benefit now, too. What you don’t have is technical justification to condemn them “forever”, and you can’t until Tehama comes out. Then you may have good reason, but not until then.

There’s No Need To Pick Sides for Next Season Yet

Few if any of us were inclined to buy RDRAM systems anyway; we looked at the price of RDRAM, went “Ugggh,” and that was the end of that. If we looked at benchmarks, very few of us saw anything that made us change our minds. I know I didn’t.

For the Coppermine generation, this game was over a long time ago; we all went “Ugggh” and didn’t buy it.

All I’m saying is that Willamette starts a new game, and Tehama is like a newly-signed free agent. It doesn’t mean RDRAM is bound to or even is likely to win, but at least watch them in practice before deciding they’re no good.

We may all go “Ugggh” again, maybe even for the same reasons, and the end result will be the same, but there’s no rush. Saying RDRAM is dead now is almost as bad as the RDRAM folks calling DDR “vaporware.” Wait and see.

Email Ed


Leave a Reply

Your email address will not be published. Required fields are marked *