Originally Posted by DuckieHo
What do you think subsidizes the discrete GPUs? Mobile/IGPs are basically last generation or cutdown discrete GPUs. So when AMD or NVIDIA develop a new flagship design, you can expect the architecture to trickle into the entry-level market in 1-2 years. NVIDIA is losing this revenue flow. The margins are smaller on entry parts but this is made up by volume.
Current/upcoming IGPs replace the need for most entry level GPUs. Basically, the only time people will need a discrete GPU is gaming or workstations. You use to need a discrete GPU for dual monitors or watching video.
Although I agree that current IGPs are very superior to the previous concept of integrated graphics, they are still behind your a-typical discreet GPU. By your own logic the IGPs have raised their performance bar (finally), but so will up coming low-end discreet GPUs, since they will trickle down in 1-2 years time frame as you so put it. So when you look at it now, of course you are seeing the inverse as the IGP performance has really increased after years and years of "bare minimum" while the next gen of low-end discrete card manufacturing should be somewhat lagging.
Is the entry level market the only tier losing money for nVidia? Because they spell the numbers out like an omen that all that matters is the quantity of GPUs being sold right now, not the performance bracket, time frame of release, quality, architecture type, or taking into account the low-end discrete cycle or any cycle at all, which was my point you effectively pointed out
One could easily argue well AMD are still showing profitable gains, and my response to this they are still riding the new gen/release wave. Furthermore, they have much easier to swallow pricing compared to nVidia. I would ascertain that nVidia makes the most money when pushing that new GPU / generation out, it spikes and then plateaus out to nothing, where as AMD/ATi are more of a smaller spike, but the plateau sits higher. In general, this article is a bit 1-dimensional as I put it, and when you analyze the numbers a bit with some logic and understanding of the CPU/GPU tech world it sort of makes sense to anticipate it.
Although Duckie meant this for some one else about something else, this emphasizes my thoughts considerably:
Its too early to tell.Edited by RagingCain - 5/4/11 at 10:16am