Introduction
When Intel’s Developer Forum (IDF) first started back in 1997 it was a small affair of about 850 attendees, and lasted about two days. It’s most important function was to tell the PC universe what Intel planned to do with its chipsets and boards so that everyone from graphics chip vendors to the guy who was shipping cables would know what they had to do to stay in the PC universe. It was a cozy affair. Lots of engineers and developers milling around. Not too much glitz. No flashy keynotes.
These days, IDF is a geek festival, circus, and rigorously controlled by Intel’s spin doctors. It’s more enlightening for the stuff you have to read between the lines than anything else. Judging by some of the press coverage Intel has received from this IDF, it’s not clear what we should be excited about so, we at THG are going to take a little time over the course of the next week to digest the data, do some follow-up, and try and cover some off-the-beaten-track sessions, too.
In this first report, we will focus on Pentium 4 because, it will have the biggest impact on the PC market this year, and Intel’s strategy, particularly in regards to memory has left some of us shaking our heads. In Part II we will cover Intel’s view of the computing world, and see whether the company’s strategy for handhelds, and peripheral devices is something we can feel excited about. We are also going to look at some of the directions that Intel is giving its OEM partners in terms of what systems they should be building in the coming year. Some of you can probably do a better job of building your own systems than any OEM, and get more bang for your buck than any store bought system, but it’s always nice to know what your competing against 🙂
Pentium 4 — It’s Not Personal; It’s Just Business
The first thing that we need to clear up is the issue of Pentium 4 and RDRAM support. In fact, the whole issue of Intel’s roadmap. Well, Dr. Tom’s crystal ball got it right in his review of the Intel roadmap not too long ago. What we learnt at IDF was more of a trip into the chip behemoth’s head, and it ain’t all full of Pentium 4 and RDRAM.
So, as far as we got the gist of Intel’s thinking:
- PCs ain’t PCs anymore, folks. They are rich clients. And handhelds, PDAs, MP3 players etc. they’re all clients, too. Intel’s got clients, baby!
- You can’t have clients without…? You got it, Intel’s got severs!
- Which also means… Intel’s got networking stuff!
- Which also means… Intel’s driving the digital world (so is just about everyone who is reading this on the Web, but let’s not quibble with Intel’s marketing team).
There’s a whole lot going on at Intel. It’s the Mud Theory of Management — throw enough products at enough markets and eventually enough of them will stick to make it seem like a great strategy. In the meantime, Intel was adamant that, despite the rumblings in the press, RDRAM was it.
If you don’t like it, bear this in mind, Intel will drive prices down, get volumes up, and phase out the desktop Pentium III market before the end of this year. This is not too different to the way that Intel forced the PC world to go from 286 to 386 processors. The company said as much. And again, the press was left in no doubt, «Get over it guys. It’s RDRAM, and the other stuff just isn’t going to be worth it in the Pentium 4 space unless you are just interested in running Word.» They didn’t quite use those words, but you could sense the impatience of executives whenever the specter of Athlon was raised, and the high cost of RDRAM.
The good news is that now, every performance freak, and hardware enthusiast can rest easy. They have a choice. There’s Intel and there’s AMD, and it’s really okay to just say no to Intel because, all their corporate buyers, and their OEMs, are going to make up for the fact that Intel has, to some extent, abandoned the hard core user, the early adopter, the connoisseur of performance.
Yeah. Intel’s busy building the digital world.
Now, let’s be fair to Intel. That’s the plan of action this year. Intel can afford to have us all wail about their Pentium 4 and RDRAM strategy. They are going to ramp up Pentium 4 volumes. They are going to get RDRAM prices competitive. They really do believe that RDRAM fits into their long-term performance roadmap. They have contractual obligations to Rambus that gives RDRAM a free run at Pentium 4. Pentium III is on its way to being the mobile CPU. For Intel’s product line, the company’s strategy is right. It’s not an easy transition. It’s kind of like the ham-fisted approach the company took moving the graphics industry to AGP from PCI. With AGP came the threat of Intel engulfing 3D graphics with its own integrated designs, and graphics chipsets, but heck, they didn’t, and so we said farewell to the i740.
Dr. Tom’s crystal ball got it right. Like we said. And a long time before the soft pronouncements at IDF came around.
No RDRAM — No Problem
Having made a firm commitment to RDRAM hasn’t stopped Intel from licensing Pentium 4 chipsets to third party chipset manufacturers such as SiS. So, you are going to get DDR, and SDRAM Pentium 4’s with Brookdale, but that’s not until the third quarter this year, and even then, it still might get pushed back. This is the pragmatic side of Intel. The side that knows AMD is breathing down its neck. The side that is hedging its bets. Is Intel going to make it tough on the RDRAM alternatives? Yup. Will Intel back off it once it sees that it has no other choice but to embrace DDR in 2002? Yup, but not without a fight on RDRAM.
Interestingly enough, Jon Kang, Senior Vice President of Samsung’s Memory Division summed it up nicely by saying that the choice of memory was application dependent, but that Samsung, the world’s largest RAM supplier, was poised to deliver all the RDRAM that a big Pentium 4 ramp up would need. Yet, he also outlined that Samsung was in there with DDR as well so, even on an Intel stage, the memory guys were not going to concede that RDRAM was a done deal. They just said that it was getting ready for the big time, and that if Intel wanted to push it out in volumes, they could meet the demand. Dual channel DDR200 is going to have the same bandwidth as dual channel RDRAM. It may have a slight price advantage. It’s there, too.
Samsung’s take is that memory is going to be driven by application. Intel kept saying as much for its strategy. You want office, wait for Brookdale and SDRAM on Pentium 4. You have «high bandwidth» apps, then you go RDRAM. Of course, Intel also defines high bandwidth apps as being digital photography, and digital editing, and even digital audio apps. But we’ll address the consumer high performance conundrum when we look at the Extended PC concept in Part II of our IDF round up. Let’s just say, we don’t buy that demand for Pentium 4 with RDRAM is going to be driven by digital video and photography in the consumer space, and by peer-to-peer applications in the corporate world (more on that in Part II, as well).
We’ve come a long way since EDO DRAM. If you compare EDO to SDRAM you go from a single memory block and peripheral circuitry to a memory array, peripheral circuitry AND high speed logic. Then we go from single data rates, to double data rates, to quad data rates. Samsung is going to have the first 1 GHz RDRAM part this year. It all adds up to some significant investment in chip technology, and none of the memory vendors are going to want to see dominance by a single memory type, or low volumes that impact overall pricing. We would really love to know what is going to drive volume RDRAM adoption on the apps side. I mean, we didn’t see a heck of a lot of gaming going on at this IDF. We saw the Windows XP demo, and that didn’t blow us away. We saw endless video and audio apps running on various Intel CPUs. We also saw NVIDIA, and you look at what they’re doing with their latest iteration of GeForce, think about X-Box, and you got to be wondering what’s going on to drive the high performance segment of the PC market.
To get the answers to some the questions raised by Intel’s roadmap and strategy we have to look at how OEMs are being encouraged to spec out their systems. We are also going to look at how the «rich client», the beloved PC, is going to fit into Intel’s new, digital world. We think you will be surprised, if we read between the lines of the presentations at IDF correctly.
Please follow up by reading IDF Spring 2001 — Part II.
