Why would a vendor choose GS over GT 8400m or 8600m?

Just curious why there is even a GS offering of the 8400m and 8600m notebook GPU's. Does the GS really offer anything at the lower pricepoint like power savings? Aren't the GT and GS really more or less the same just with performance artifically reduced?

Why not just offer the GT?

 

Same architecture with half the shaders, maybe it costs a lot less manufacturing gimped cards to sell to consumers.

 

Also heat issues, more shaders and pixel pipelines means more heat, which in a small cramped-for-space notebook like my M1330, more heat is deadly.

 

But the issue is that a lot of times they sell the GT for even the smaller laptops as an upgarde. I don't know that it costs them less to manufacture as they are usually made on the same die, just disable features to sell it as a lesser card.

 

There are two ways this can be profitable.
First, the obvious, it allows them to use otherwise defective cards. What if a couple of shader units don't work? They obviously can't sell it as a normal card,because it crashes when that shader unit is used.

Scrapping the card isn't exactly profitable either.
Disabling some shader units, though... It lowers performance, so they have to lower the price as well. But they're still getting *some* money, which is better than the alternative, scrapping the chip entirely.

Second, a more subtle reason is that having a GS allows them to sell the GT at a higher price point.
There are people who are willing to pay X dollars for a GT (call them the GT crowd), and there are some who don't want to pay that much. (The GS crowd)

So they have two immediate options. Either sell the card at X dollars, which means only the "GT" group will buy. The rest may go to ATI instead.

Alternatively, they could sell the card, running at normal GT speed, at a GS pricepoint.
Now they'll be able to sell to both the GS and GT crowds, which is nice. Unfortunately, they're getting less money out of the GT crowd. (because they're only paying GS price rather than the GT price they were otherwise willing to pay)

So the smart thing to do is take some of the cards, sell them as GT's at a GT price, and then take the remaining ones, artificially throttle performance, and then sell them at a GS pricepoint to the GS crowd.
Obviously that earns them much more money than either of the other options.

So yes, part if it can be manufacturing costs (allowing them to use otherwise defective chips), but part of it is also just plain pricing strategy. You need to have a card available to sell to the cheapskates, but you also want to be able to wrestle more money out of those who are willing to pay.

 

Exactly.

For example, my 7900gs is basically a 7900gtx that fell at a low point on the "normal distribution" scale. So they just cut a few pipelines and lowered the clock speeds, and sold the card at a lower price.

The same principle applies to the Core 2 Duo CPU's. The best cores will end up as core 2 duos... but if one core is faulty, they disable the defective core and sell it as a core-solo.

This will conceivably apply to all types of technology. Designing a new core is a huge waste of money, when a single design will let you sell 3 products (GS, GT, GTX) and reduce the number of chips you have to throw away.

Jalf is also right about the marketing/pricing aspect. Although I think that vendors are partially cheating the consumer by listing turbocache memory as the actual amount of memory on a card. For example, a 128mb integrated card will not play most games that require a card with 128mb (dedicated) memory. They need to sort out their advertising methods.

One good thing, GS and GT cards should therefore be able to overclock very well, since the chip is technically a GTX with lower clock speeds and less pipelines.

 

Some GS models are not of the same core as the GT. Take for exapmple the 8600GS is a 8400GT overclocked, the 8700GT is a overclocked version of a 8600GT.