   |
|||
|
|||
|
:
News
: Archives
|
|
The Rising Cost of Graphics |
|
|
Yet Another Glass Ceiling? |
|||
|
|
|||
|
by Josh Walrath |
|||
|
Economies of Scale and Milestones The 3D graphics market was predicted to slow down several years ago, but so far no one in the industry has seen this. Schedules are still brutal, and the competition never sleeps. This means that within a year we will see products with the performance of the 7800 GTX 512 for half of the price. How is this so? Well, there are quite a few reasons. The 7800 GTX 512 MB uses Samsung 1.1 ns GDDR-3 for its main memory. This is very new and very expensive memory, and providing 512 MB of it makes for the most expensive component on the board. Time will pass, and Samsung and others will start making more and more of the 1.1 ns memory. Not only that, but they will introduce GDDR-4 next year, which will drive GDDR-3 prices down even further. Within a year the component that was the most expensive on the board will be cut down in price by 75%. In terms of the actual GPU, while it may be expensive now to produce a chip that will successfully run at 550 MHz, the price for producing parts on this process and achieving faster speed bins will fall as certain milestones are reached. After NVIDIA orders X amount of wafers, the cost per die will decrease. The longer a product is produced, the better the yields and speed bins will improve over the lifetime of that product, and that will further lower the cost per good die. A good example of this was the NV-2A (the X-Box GPU). By the end of its lifecycle, it is rumored to have a per unit cost ¼ of when it was first shipped to Microsoft in 2001. Also going to newer, smaller fabrication processes will lead to smaller and faster chips that will eventually cost less than the larger, slower GPU’s of today. Other component prices will fall over time as well, such as the heatsink and fan that the 7800 GTX 512 MB has. It is far more complex than the standard video card heatsink, but once these become utilized by a larger portion of cards, the less it will cost to produce each heatsink/fan individually. There is also another factor that is perhaps the strongest to decreasing the price on any single graphics product. That factor is the competition between ATI and NVIDIA, as well as the manufacturers using the chips from the big two. Companies such as MSI and Asus use chips from both fabless IC’s, and they use this to their advantage to drive prices down for the GPU’s they receive and sell. On another level we have companies like MSI, Asus, XFX, BFG, Chaintech, Leadtek, and others all competing with each other using the same chips from NVIDIA. So while NVIDIA will sell GPU’s to each manufacturer, the onus of producing the finished product is on the cardmaker themselves. They deal with the individual component manufacturers, produce the cards, and then try to find ways to make their card cheaper yet still reliable so they can undercut the competition. If it were only ATI and NVIDIA selling cards, prices would probably be even higher than they are now. It is unfortunate that this type of competition is not in place for the CPU makers. Both AMD and Intel sell directly to consumers, and so they can keep their margins high and not have their product’s value erode over time. While NVIDIA released the original 7800 GTX in June for $599, AMD released the FX-57 around the same time. In the 5 months since release the 7800 GTX has gone down in price to where users can buy them for around $450, while the FX-57 from AMD has not gone down in price at all. The same can be said for Intel and their Extreme Edition products. And when we get down to benchmark numbers, how big of a difference is a user going to see going from a $300 CPU to a $1000 CPU compared to a user going from a $300 video card to a $750 video card. The end user will certainly receive a lot more value in the extra money spent on graphics. The Final Straw One thing that many people discount is that we essentially do live in a free market economy. Sure, it is not a “true” free market economy, but when we deal with products such as this the consumer will have the final say on the success of said product. If the market cannot or will not support a consumer level $1000 graphics card, then nobody (or few people) will buy such products. This then gives the companies incentive to not offer a product at that price, or being able to find ways to achieve performance and features without the added cost. Currently NVIDIA and its partners are selling out their stock of 7800 GTX 512 MB cards, and they are all pretty much going at $750 a pop. Obviously there are good amounts of people out there that feel paying that much for a video card is not a hardship that they cannot endure. Should we complain that these people exist? We really shouldn’t, as they are providing the larger margins to these companies to help develop products for the midrange and budget sectors that will give a lot of performance and features for their respective prices. The latest 6600 DDR-2 board is a prime example of this. Here is a $119 product that give 256 MB of fast DDR-2 and provides the overall performance of a card priced at $450 two years ago. At the $320 mark we have cards that are faster and offer more features than the $499 products of one year ago. While we may eventually see the emergence of a $1000 video card, we probably shouldn’t be too terribly worried as market pressures will still exist to keep competition healthy and alive at the lower price points. While NVIDIA and its partners are making some fat margins off of the 7800 GTX 512 MB, the real bread and butter of these companies are the tens of thousands of cards being sold for under $300. This is an area they will never ignore, and the strong competition between ATI and NVIDIA, as well as the manufacturer competition one level below them, will help insure that quality products will still be offered to users at those price points. While the cost of entry into the upper enthusiast ranks has significantly increased, the majority of users have benefited from the technology and market trends we see today. Never before have we seen sub $150 video cards with so much performance and features, and much of this can be attributed to technology trickledown from the high end products reaching $499 and above. Update 2005/11/30 After writing this article something very interesting occurred to me. It seems we have seen a $1000 consumer graphics card, and it was surprisingly released this Spring. I am referring to the GeForce 6800 Ultra 512 which retailed around $999 for most manufacturers. This was a very forgettable product, as many people I had mentioned it to online had no idea what I was talking about. Before the 7800 GTX was released, the 6800 Ultra 512 was supposed to be the biggest, badest video card out there. Unfortunately, it was outfitted with slower memory to hit 512 MB and run reliably. So, not only was the user paying a lot for a card, but in 99.99% of the applications of that time, the extra memory did not help it out, but rather the slower memory speed hurt it. I think this product proved that the marketplace does have some intelligence. Here was a product that didn't offer the end user anything over what a significantly cheaper card can deliver, and so consumers balked at paying such a high price for what was definitely not a premium card. The 6800 Ultra 512 lasted only a few months, and then they were quietly taken off the market once the 7800 series came online. I believe that NVIDIA and its partners learned a good lesson here. Basically a product that offers the users nothing except a bit of flash for a whole lot more money will not sell, and it will be a waste of resources to release such a card. I think the first result of this lesson is the 7800 GTX 512, which not only offers more memory and flash, but significant performance improvements over the previous high end product.
If you have found this article interesting or a great help, please donate to this site.
Copyright 1999-2005 PenStar Systems, LLC. |
|||
