Click here to print this article.

Re-Printed From SLCentral

The Damn NDA
Author: JonnyGURU
Date Posted: October 7th, 2001

The Damn NDA

I don't quite understand NDAs.

An NDA is a non-disclosure agreement.

In the computer industry, it is not uncommon for a company to obtain product before it is available to the customer. This is important because that company can test compatibility issues, assemble potential builds to make available to their customers in the future, and in some cases "feel out" the product to see if they can find a market for it at all.

But in testing this equipment, the vendor is bound by an NDA. They are not allowed to divulge what product they have, to the public, because to do so may give them an unfair advantage over the competition, and manufacturers can't put themselves in a position to play favorites.

For the press, it is not uncommon for a website or magazine to obtain product before it is available to its readers. Websites can test the product so it can tell its readers of the pros (and hopefully not so many cons) of the product.

But name brands and/or specific models of particular products cannot be spoken before the expiration of an NDA, because to do so would be unfair to the other companies that have similar product on that platform that may also have product at another website that may not have an editor competent enough to write a glowing review.

Give me a break.

I've seen some pretty fancy finagling of this aspect of the NDA. One example is Tom's Hardware's review of the i845 chipset before it had come out. He had used a motherboard, which he was unable to disclose the make and model of, to benchmark and give impressions of the chipset. Photos were skillfully edited to show only the Northbridge chip and the new socket 478, but not any aspect of the board.

I, myself, had several boards and CPUs courtesy of Intel and Biostar, yet was bound by an NDA to discuss much of it. I could discuss so little and Tom's crew had already skillfully covered any information I wanted to share about the i845 chipset.

Having the product before the public release of it did do me some good. I like the new socket. It allows for faster speeds due to the shorter traces from the die to the pin grid array. I also like the new heatsink retention design. It allows for easier installation and greater portability of a P4 system.

I also learned that SDRAM is completely useless for the P4. It cripples the P4 by 10% in nearly every benchmark, so I tried to convince my boss to only carry the Biostar i845 board since it costs a whopping $45 less than the Intel branded socket 478 board for RAMBUS and this would be good for the cost conscious P4 user. I also tried to convince the boss to phase out the old socket 423 boards since the EOL is inevitable for that platform.

Of course, this is all old news now. What's the new news? The Athlon XP.

The release of the Athlon XP on October 9th is the most irritation from an NDA I have ever received.

First, the name is supposed to be a secret, yet there are already a number of dealers taking pre-orders of...guess what? Athlon XPs!! The PR rating of the CPU is supposed to be top secret, yet on the Register an entire article was written disclosing what actual MHz each PR rated model is. DUH!!! My employer won't even put the CPU on the website until the 9th to avoid getting in trouble with AMD. Just because everyone else is doing it, does it make it right? Not necessarily, but I shudder at the number of orders going to the competitors because they damned the NDA and choose to sell the CPUs anyway.

The whole PR thing is a rant in itself. AMD only puts the PR rating on the CPUs now and wants approved motherboards to not ever show the actual MHz of the CPU at post. Needless to say, an actual benchmark program will tell you what the speed of the CPU is, but Joe Average is just happy if Everquest works on his model 1700+ CPU with 2 GB of RAM and 75 GB hard drive.

For someone like myself, I think it's more impressive to have a 1467 MHz CPU that out benchmarks a P4 1.7 GHz CPU in most benchmarks than to have a 1700 CPU that is actually only runs at 1467 MHz! And where do the Pentium ratings actually come from? Benchmarks vary so much and are dependant on so many variables that can completely make or break a benchmark! I think AMD has dug them selves a hole with this move. What happens if Socket A technology allows for better performance overall? Does AMD change the name of all of the CPUs to reflect this? The 1700+ becomes a 1750+! What happens if technology vastly improves the performance of a P4? Does AMD have to change the name of all of their CPUs to reflect falling behind? Like I said, it's all relative. Using actual MHz to denote the speed of a CPU is not a variable! It's a given.

I think it's hilarious that I have an engineering sample Athlon XP model 1700+ that actually has 1467 engraved on the die. "Model 1700+" is just magic markered onto the package as if the idea to label Athlon XPs was an afterthought. I think I'll hide this CPU away just in case the NDA mafia comes to confiscate the chip.

At least they're using an organic package now instead of the ceramic. So maybe I'll get back fewer cracked CPUs...Oh wait...I'm I allowed to talk about that?

Re-Printed From SLCentral