Allchin Talks Up Longhorn, 64-Bit Computing

By eweek  |  Posted 2004-05-10

Allchin Talks Up Longhorn, 64-Bit Computing

Microsoft Corp. used its recent annual WinHEC conference in Seattle to give more specifics about the roadmap for the release of Longhorn, the next version of Windows, as well as to talk up potential future adoption of 64-bit computing. Jim Allchin, Microsofts group vice president of platforms, sat down with eWEEK Senior Editor Peter Galli after his keynote address.

You talked about a new technology that allows devices and PCs to power up from a "sleeping" state in under 2 seconds called "Instant On" in your keynote. Is this technology in any shipping products yet?

This is fabulous technology, and the work around this has been quite something. We havent decided when to include it in a shipping product as yet, but it will be soon. What you saw was work that dealt with a whole lot of the peripherals, but we can get the amperage down to 50 percent if we do the rest of the work. We are almost there, and that to me is just a mindblower in terms of the impact it can have.

Microsoft has been talking a lot about the potential pervasiveness of 64-bit computing by the time Longhorn ships. Do you agree with that scenario?

I see the adoption of 64-bit computing moving pretty rapidly as I see few compatibility issues, other than the drivers, which could slow it down, and thats another reason why we have to take a hard bet on it. So the compatibility is very good in terms of applications, and the second thing is the cost: Conceptually theres no price difference. If thats true, then, wow, why dont you get it and make the 32-bit applications running on the 64-bit operating system faster, and there are other nice neat things about address spaces.

So, will Longhorn be based on a ?

We will have a 64-bit version of Longhorn, no question. Will we have a 32-bit version? The plan is yes, but now, if we learn a lot between now and then, that might change. But right now we are staying the course and it is so hard to predict how fast the run rate will be. We know where AMD [Advanced Micro Devices Inc.] will be, we know pretty much where Intel [Corp.] will be. I think we just have to wait a little bit.

So, if you decided not to have a 32-bit version of Longhorn, those customers would not get Longhorn?

Thats why were not going down that road. Theres no reason why we have to do that. Its sort of an academic discussion.

Can you tell me what the reference architecture is for Longhorn?

I dont know if thats public yet. Youre talking about memory size, processors and all that? I dont think weve made a final decision,n and I dont think we will until we hit Beta 1. We have targets inside, and we may have gotten feedback on this from some select people, but its just too early.

Next page: Syncing Longhorn releases.

Longhorn in Sync

You have shifted your strategy on the Longhorn client and server release cycles, and you are now developing them pretty much in tandem so they can ship close together. What drove the change in your thinking on that?

Internal efficiency. The alternative was to add a ton more functionality to the server so it would have taken longer. But the problem was that it meant we would have run two different source trees with all the shared components having to deal with both, with two separate runs on beta testing. The coordination effort means that while the server will not have as many features as it would have when we planned to release it two to three years after the client, but now you get all of the benefits and there will be a synchronized system until the very end, where there will be a longer test. We will take whatever time is needed to finalize the server, but it will not be years though. Basically, well keep it one milestone behind and at the end, around Release Candidate time, well let it bake a little bit longer.

Click here for more on Microsofts plan to sync Longhorn releases.

But that was the early plan with the Windows XP and client releases, and they shipped years apart. Is that same scenario avoidable here?

We think so. The amount of IQ that we have on the software engineering process is incredible at this point. The level of people working on our source and build environments and our tools now are so smart. Before we didnt have the MIT guys; we have now applied real hardcore brilliance to the problem. Were putting the dollars there, and I anticipate [what happened with XP] wont happen again. We didnt think internally that the pain with XP was going to be as bad as it turned out to be, and none of us want to do it again, we really dont.

We all know by now that you plan to cut some Longhorn features. What is currently on the chopping block, and do you have a sense for what might be cut?

The first thing you have to put in context is the magnitude of Longhorn. The things that we cut you did not even know were there and most of the people working on Longhorn did not know they were there. So, its not like the storage system, the graphics system or Indigo are not still going to be there. All the work on basics is still all there. There are things we wished that we could do that weve now decided to move to Blackcomb. There is spreadsheet after spreadsheet of little line items that when you look at them we say we just dont need those now. The one thing about Longhorn is that the quality is going to be there. This security situation is going to be addressed. We have done a very good job with Windows XP SP2, and we are redoubling, tripling, our efforts to ensure that the quality base is beyond anything weve ever done, and I would cut other features out without even thinking about it to make sure that I do a great job on security. Theres so much in Longhorn, it is so overwhelming and the specific things we cut were wish-list items really.

Next page: Finalizing the feature set.

Finalizing the Feature Set

When will the feature set be more final?

Around Beta 1, where well hand it out and see what people think. But we are very clear on the vision and the core feature set, and well make the final decision once we do the beta. But you will see a general push to quality beyond what we have achieved to date, and that will be across the board.

For a look at the latest Longhorn build, click here.

I am sure that you have seen the recent concern expressed by members of the open-source community about specific Longhorn features like XAML and Avalon and how they plan to address these, claiming that these moves could be the "final takeover for the Web." How do you respond to that?

I take their response as flattering, but how they jump to see this as a plan to take over the Web, that is ludicrous. Do people want to have a safe browsing experience? I think so. The new application model in terms of XAML and the like is good. But we are not doing this to make them concerned.

Some may say that the fact that they want to duplicate these features is a sign that the open-source community is again playing catch-up.

Absolutely. They are cloners.

There was discussion within Microsoft about a possible interim client release between Windows XP and Longhorn. Why did you decide in the end not to do this?

We looked at what we should do. No six-month period goes by where we do not ask ourselves what is happening in the market and what do customers need. We didnt plan on doing all the things we are currently doing in XP SP2. So, no, there is no plan to do an intermediate client release because we can get Longhorn done.

But this is the longest time release between Windows client releases in Microsofts history.

Do I wish we could ship it earlier? Of course, but it is what it is, and what it is is the most comprehensive client release we have ever tried to do. Windows XP is a great and solid base, and it is doing very well. Weve been able to stair-step it with a couple of versions of the Media Center, multiple versions of the Tablet, and now a 64-bit version. So we have been able to come out with a whole lot of different releases all based on this core, so the way I view it is that every long period of time we come out with a new architectural shift. Even XP, although it was a new technology over Windows 98, was not an architectural shift. Longhorn is an architectural shift in terms of the capabilities There are truly deep architectural changes in it. So I would expect it to come out and then there will be offshoots of that. The cycle is architectural [followed by] intermediate, architectural [followed by an] intermediate [release].

Did you move a lot of the Longhorn team to work on XP SP2?

I took a lot of people off [Longhorn] to go back and work on that, and that makes me even more determined that when Longhorn ships we are not going to have to go back and do that. So it makes me even harder core about the foundation and the basics, whether it be deployment, security, reliability or compatibility.

Next page: The upgrade issue.

The Upgrade Issue

Many customers used to complain that you had too many upgrades too often. But Software Assurance did change that to an extent in that many users thought they would get a client upgrade within those three years, and now it is clear that many wont. What are you doing within Microsoft to make sure those customers are kept happy as this has the potential to be a very big issue?

This is a very important issue for us. As you probably know we are already providing our Software Assurance accounts and there will be even more software that we will be announcing later this year that will be available to those accounts. This is all part of a strategy to assure that there is value that came with that, but we already have some software that certain accounts have been beta testing for some time, and most of it is related to service-level capabilities. So I would say that we are acutely aware of it and will continue to work on delivering some software for those customers.

Who do you see as your biggest competitor for Windows, and has the competitive landscape changed at all recently?

On the client there are so many to chose from. But the landscape has not really changed. Linux coming from IBM and Red Hat [Inc.] and the other distribution shipping that, with , and then theres Apple [Computer Inc.]. For us its also the installed base, but I think a few of us have always seen IBM leading the Linux charge and is a huge competitor for us on that front and probably leads our top of mind. But the other somewhat pureplays that exist like the Red Hats of the world and the Lindows, all have targeted areas where they may be strong in an emerging market here and a particular strength there. They all fit into that.

Red Hat last week announced a new desktop offering that is directly and aggressively targeting those Microsoft customers facing the end-of-life support for their Microsoft products.

Click here for more on Red Hats Linux desktop.

The first thing is that we are very cognizant of the end-of-life aspects, and I dont have anything else to say other than we are cognizant of that and we will do the right thing. We also believe that people are going to stay where they are today because they are happy and the big thing is that if someone is really looking ahead they are going to be thinking about the next generation of applications and they are going to want to move to Longhorn. It will take them [competitors] a long time to clone the core technologies we have in Longhorn. They might do a surface-level thing, but we didnt spend all these billions [of dollars] indiscriminately. This is hard stuff and is not like cloning the moonscape we had in bitmap or the Bliss screen we had in XP. The bottom line is that it is competition and I hope that customers see that we are offering great value for the price we are charging.

Check out eWEEK.coms Windows Center at for Microsoft and Windows news, views and analysis.
Be sure to add our Windows news feed to your RSS newsreader or My Yahoo page:  

Rocket Fuel