Advanced Micro Devices
In Reply To:
Newisys Being Acquired?

Format for Printing

Format for printing

Request Reprints


By eachus
July 16, 2003

Posts selected for this feature rarely stand alone. They are usually a part of an ongoing thread, and are out of context when presented here. The material should be read in that light. How are these posts selected? Click here to find out and nominate a post yourself!

Newisys, the star-studded Texas server start-up that bet the farm on AMD's new Opteron chip, reportedly canned 20%-25% of its people the other day in preparation for being acquired.

[Related Story - ZDNet]

Not to be a Pollyanna, but there is another side to this story. If you look at the Newisys corporate model it should be transitioning from a company concentrating on hardware development, to much more sales, support, and software. If the layoffs are related to that square pegs and round holes issue the size of the layoff says more about the current job market for hardware engineers than the financial health of the company.

Similarly, assuming that all the major hardware designs have been finished for now, there should be no surprise that Newisys is in the red. There will be a need for ongoing software and firmware work, but the point at which the company transitions from a development company to a profit making company will depend on two things. The size of the market for its inventory of IP (Newisys hardware designs and support software) and its actual cash burn rate. The company can't do anything really to control the first, the limit is AMD Opteron production. But they can control costs.

If Newisys has 2, 4, and multiple interconnect board designs completed, when will they need to do new versions? Right now it looks like when PCI Express and DDR2 support are added to the 90 nm Opterons. PCI Express is a chipset issue, which from Newisys's viewpoint will require a new version of the AMD 8111, or a PCI Express version of the AMD 8131. DDR2 will of course require new Opterons. While I think it is likely that the San Diego will support DDR2, I can't imagine a rush on AMD's part to add DDR2 support to Athens. (PC3200 ECC? Sure, but that won't require Newisys hardware changes.)

So I see the Newisys situation right now as the need to keep a few good board designers around to deal with potential problems, and let the rest go, or move them into customer support roles.

It probably didn't help that the Newisys box is, well, over-engineered.

This one comment throws the remainder of the article completely into disrepute. What makes a board or system design "over-engineered"? The Amiga 1000 was a perfect example. It was much better than it really needed to be for its intended market. I remember after reading the power supply specs deciding that a UPS or surge protector would be superfluous. One day I was working in my study and saw a really nasty squall line approaching. I decided to stop what I was doing and check for open windows. I closed the file I was working on, and was waiting for it to write to disk, when WHAM, I saw a "bolt from the blue" hit the pole transformer about 100 feet from my window.

Needless to say, the power went out, and I was certain that in spite of the over-engineered quality of the Amiga 1000 that I was going to be replacing hardware when the power came back on. As it happened one battery-powered electronic clock was fried and a couple of other minor electronic items. As for the Amiga? Not only was it undamaged, but the file write completed successfully. I didn't even lose the I/O serial chip that was socketed because it was subject to power surges through the modem. That is over engineered. Based on previous experience with momentary power outages, the Amiga 1000 had about a 1/2 second power loss ride-through. And as you can tell from this story, the EMP protection of the case was probably better than MIL-SPEC.

So let's look at the Newisys intended market? Is such performance a potential selling feature? You bet. Especially in the 24/7 server market, the difference between four nines and five nines is a major selling feature. Does all this (over-) engineering mean that the system is too expensive? Hell no! We are probably talking around $100 in cost per motherboard. (Cost, not selling price.) And this is in a system where the customer will be paying between $1000 and $10,000 per board for the CPUs. The big cost difference is the service processor, and in the intended market, that "additional" cost will pay for itself between the time you open the crate and have the system on-line.

There may be other markets where Newisys will be unable to recoup the marginal cost of their engineering work. But that is irrelevant. The announcements we see from time to time of new Opteron systems, like the recent pair of educational systems with a total of 260 Opterons indicate that Newisys' intended market is alive and well. That is the market that Newisys engineered for, and seems to have done a wonderful job.

Will nVidia nForce3 based boards dominate the workstation market? Probably. But what has that got to do with Newisys? Newisys is aiming at dominating the market for high-end Opteron servers. I think they will do/have done that.

Become a Complete Fool
Join the best community on the web! Becoming a full member of the Fool Community is easy, takes just a minute, and is very inexpensive.