This has come up enough times in meetings with manufacturers that I think I need to address it publicly, since it's not something that I think is normally exposed.
When Intel launched Centrino, its competitors saw dollar-signs. Intel managed to brand a bunch of its own chips under a new name, called it a platform, and made a ton of money. Obviously it wasn't that simple, prior to Centrino the mobile market was honestly a mess. Wireless was far from a standard feature, and Intel's own mobile processors were severely lacking (mobile Pentium 4? Really?), not to mention issues with 3rd party chipset vendors and their drivers. Centrino put together a really good mobile CPU, with a decent chipset and wireless, it was a platform that actually did something.
Since then, everyone has tried to reproduce the success of Centrino - including Intel. Viiv, Centrino for the home, was such an attempt but quite flawed in execution. VPro, Intel's business platform, is a bit better because it actually offers tangible and useful features, but still not as widely accepted as Centrino.
Once you get outside Intel, then the attempts at cloning Centrino become even more pathetic. These days, there are folks within AMD and NVIDIA who want to turn each company into a platform company. We've actually been asked on a few occasions to review products as a part of a platform, rather than individual components to somehow showcase these magical synergies that exist between products.
The NVIDIA platform is a stretch these days; NVIDIA's motherboard + GPU combo are the primary parts of the platform, which is why snide remarks usually come from AMD/Intel whenever we talk about NVIDIA's "platform". From their view, a platform isn't a platform without a CPU.
Thus, AMD's platform is far more complete since the ATI acquisition as it now has a CPU, motherboard/chipset and GPU to offer. Furthering this logic, AMD's platform is arguably more attractive than Intel's because of superior graphics. It almost hurt to write that.
Here's the problem, while the platform story may be a good one to sell to OEMs, at the end of the day users want the best CPU, the best chipset and the best video card. Unless you can offer a bundle of three that provides an actual benefit to the end user, there is no platform to speak of. Centrino works because you can still have a Centrino branded notebook with ATI or NVIDIA graphics; you can't have a NVIDIA platform with an ATI GPU. What about chipset flexibility with Centrino? Sure, there is none, but there's not much value from alternative chipsets in the mobile space. You just need something that works, the important part is in the graphics where you do get an option. Centrino isn't the perfect platform, nor the ideal one, but attempting to recreate the platform story by simply bundling a desktop motherboard, graphics card and CPU together just isn't going to work.
Can you imagine how frustrating it would be to review AMD/ATI graphics cards only on AMD motherboards and compare them to NVIDIA graphics cards on NVIDIA motherboards? Goodbye scientific method.
Despite the obvious issues, there's constant pressure to review everything in this platform fashion and while I understand the desire for these companies to make money (selling 3 chips is better than just one), it just wouldn't fly for what we do. This post has two purposes, one to give you a glimpse of the kinds of discussions that often happen while meeting with manufacturers, and one to answer a question I'm often asked while meeting with those manufacturers.
Hope it helped.
When Intel launched Centrino, its competitors saw dollar-signs. Intel managed to brand a bunch of its own chips under a new name, called it a platform, and made a ton of money. Obviously it wasn't that simple, prior to Centrino the mobile market was honestly a mess. Wireless was far from a standard feature, and Intel's own mobile processors were severely lacking (mobile Pentium 4? Really?), not to mention issues with 3rd party chipset vendors and their drivers. Centrino put together a really good mobile CPU, with a decent chipset and wireless, it was a platform that actually did something.
Since then, everyone has tried to reproduce the success of Centrino - including Intel. Viiv, Centrino for the home, was such an attempt but quite flawed in execution. VPro, Intel's business platform, is a bit better because it actually offers tangible and useful features, but still not as widely accepted as Centrino.
Once you get outside Intel, then the attempts at cloning Centrino become even more pathetic. These days, there are folks within AMD and NVIDIA who want to turn each company into a platform company. We've actually been asked on a few occasions to review products as a part of a platform, rather than individual components to somehow showcase these magical synergies that exist between products.
The NVIDIA platform is a stretch these days; NVIDIA's motherboard + GPU combo are the primary parts of the platform, which is why snide remarks usually come from AMD/Intel whenever we talk about NVIDIA's "platform". From their view, a platform isn't a platform without a CPU.
Thus, AMD's platform is far more complete since the ATI acquisition as it now has a CPU, motherboard/chipset and GPU to offer. Furthering this logic, AMD's platform is arguably more attractive than Intel's because of superior graphics. It almost hurt to write that.
Here's the problem, while the platform story may be a good one to sell to OEMs, at the end of the day users want the best CPU, the best chipset and the best video card. Unless you can offer a bundle of three that provides an actual benefit to the end user, there is no platform to speak of. Centrino works because you can still have a Centrino branded notebook with ATI or NVIDIA graphics; you can't have a NVIDIA platform with an ATI GPU. What about chipset flexibility with Centrino? Sure, there is none, but there's not much value from alternative chipsets in the mobile space. You just need something that works, the important part is in the graphics where you do get an option. Centrino isn't the perfect platform, nor the ideal one, but attempting to recreate the platform story by simply bundling a desktop motherboard, graphics card and CPU together just isn't going to work.
Can you imagine how frustrating it would be to review AMD/ATI graphics cards only on AMD motherboards and compare them to NVIDIA graphics cards on NVIDIA motherboards? Goodbye scientific method.
Despite the obvious issues, there's constant pressure to review everything in this platform fashion and while I understand the desire for these companies to make money (selling 3 chips is better than just one), it just wouldn't fly for what we do. This post has two purposes, one to give you a glimpse of the kinds of discussions that often happen while meeting with manufacturers, and one to answer a question I'm often asked while meeting with those manufacturers.
Hope it helped.
7 Comments
View All Comments
Screammit - Wednesday, October 24, 2007 - link
With this topic in mind, I am reminded of NVIDIA's decision to restrict SLI to their own motherboards. Their strategy makes sense once you think that they are desperately trying to hold on to the platform line of thinking. Never mind how much money they could make in one fell swoop if they even ALLOWED SLI to work on other boards. They wouldn't even need to bother supporting it. Their current model just seems bad for business.Sunrise089 - Friday, September 28, 2007 - link
I was surprised to see new entries on the side bar - I had given up hope. There entries are really different than anything else posted on the site (other than maybe Gary's occasional posts over in the forums), so please keep it up.Calin - Wednesday, October 3, 2007 - link
I remember when NVidia made automatic overclocking of bus frequency (by 25%) when one of its graphic cards was paired with one of its mainboards (well, a mainboard with NVidia chipset). However, the performance bonus was not there.I would like such a platform - Intel processors and NVidia chipset with integrated video. As it is now, I will go for the AMD processor to pair with the NVidia "platform"
crimson117 - Friday, September 28, 2007 - link
As Anand hints at, Centrino's success was because it entered a fractured market and made wireless (with good driver support) a standard feature, right when 802.11b/g was taking off.Intel predicted that people would want wireless laptops with good battery life and decent performance, the OEM's agreed, and together they made a ton of money. Much like the iPod, platform competitors won't be able to catch up any time soon because the market is no longer thirsting for this particular solution.
The next winner will be the one who manages to predict the next significant trend in computing.
Insomniac - Friday, September 28, 2007 - link
For the most part, laptops just aren't something the enthusiast takes apart and swaps parts in like their desktop PC. As you said, the platform may make more sense to an OEM who sells pre-built systems. Those same volume (and rebate) discounts won't be available to the enthusiast even if they wanted to buy the bundle.It would only make sense for AnandTech to review those platforms if you are comparing OEM systems in a round-up.
FlameDeer - Friday, September 28, 2007 - link
This is really an issue of bundled platform. Hopefully manufacturers will provide more flexibility to their platform. Platform better to get more flexibility. By doing so should also help them to achieve better sales. Finally, keep up the good work Anand!thwzr - Friday, September 28, 2007 - link
Makes me glad to know you're willing to stand your ground and just do what you designed AnandTech to do. You guys rock! Excellent job!! I started reading AnandTech's articles I guess 6 or 7 years ago now -- and you guys are just as good now as you were then, probably better .. seeing as now you blog too! :-) heh. Wish you all the best!!