I am somewhat confused about how NVIDIA distributes its graphics chips. I have a PNY graphics card which has an NVIDIA graphics chip. It seems that several companies market this chip on their graphics card; not just PNY.
Now when the Fermi is released who will market the chip, NVIDIA? Will it be some third party vendor that simply has the Fermi in its retail product. I know that NVIDIA made graphics chips on several retail items not marketed by them.
Who sells the Tesla? I know the chip is an NVIDIA manufacture, but who sells the Tesla?
Finally, I read that the Fermi will be released it late March 2010. It is now late March in 2010, so where is it?
[i]NVIDIA’s engineers manufacture the chip technology, separate parties such as EVGA, Zotac, PNY, ASUS ect. are the distributors of their variation of the unit itself. NVIDIA specific boards are known as “reference” units, these are available in early production dispatch, and or launch partner distribution. As for Tesla that is a special workstation “supercomputer” board manufactured in house, and is not to be considered a GPU by any means as it operates more like a massive computational unit or Central Processing Unit dependant on users necessity.
Finally concerning “Fermi’s” launch, limited quantities were available for testing purposes and sparse pre-order at this time(all of which are filled), regulated by one board per person on the limited stock. As for full public availability or “mass availability” they are slated for within the next few weeks dependant on supplier. I would look for purchase dates between the 12th and end of April.[/i]
NVIDIA doesn’t make or sell complete retail graphics cards, they sell “chipset kits” (for want of a better expression) to add-in board manufacturers who build retail cards and sell them into channel. Professional products (like Telsa and Quadro) seem to be OEMed from PNY and resold by a number of big name manufacturers (like HP and Fujistsu Siemens), as well directly into channel by PNY.
Fermi based retail boards have just been announced (see here). I understand Fermi based Tesla products are not due until the northern hemisphere summer.
Are you sure, because I read everywhere, that all maufacturers are doing is put some stickers and put it in their own box (at least for Fermi). The whole card design is the reference design from NVIDIA at this time, and the manufacturers get the boards from NVIDIA. I thought it was like you say above, but from what I have read recently, the current Fermi cards are all manufactured by NVIDIA.
Yeah I am pretty sure. Reference design kits are just that, reference boards + components, and the AIB manufacturers stuff them, assemble them, test them and sell them into channel.
The thing that made me wonder a bit is quotes like this:
“few people have told us that Nvidia plans to ship more finished Geforce GTX 480 cards in the first week of May. Furthermore, some lucky ones will get chips to design their own GTX 470 cards and strap them with some custom designed coolers.” (News)
Which seems to suggest that it is currently NVIDIA who is shipping cards to their partners who just put personalized stickers on and boxes around it
I think there is a terminology usage problem here. “Partner” is not necessarily the same thing as AIB manufacturer. There are NVIDIA cards retailed under a lot of brands but most of those come from a relatively small number of AIB manufacturers who get kits from NVIDIA (gpus aren’t unique in this regard, motherboards, laptops and the like are the same). There are plenty of “put sticker on reference card” products going into retail, but those come from AIB manufacturers through OEM channel, not directly from NVIDIA. For the Fermi launch, the rumour has it that either one or two AIB manufacturers (possibly PNY and/or Zotac) are building the first batches of GTX 470 and GTX 480 reference designs that everyone will be seeing in retail. Later, some of the “top tier” partners (so EVGA, ASUS etc) will get to do their own customised versions which deviate from the reference design (coolers, clocks). They may or may not actually be manufactured in-house, but still made by the same AIB manufacturers.
enlightenment followed ;) Thanks, have some better grasp on the whole thing now.
Now you are on a roll, is it the same for MXM modules? Is there any way to buy a bunch of MXM modules? Who actually manufactures these things, so where to go as a company to buy them.
I would assume MXM modules are going to be similar: NVIDIA supplies kits, AIB manufacturers put them together and push them into channel, laptop builders and whichever other OEMs that use them (industrial and embedded manufacturers maybe?) put them into product that goes to resellers or retail. I would guess that PNY make them (they seem to make a lot of NVIDIA AIB product), but I really don’t know enough about the machinations of Taiwan Inc to say anything authoritative. A reseller is probably going to be the only place your will get them from.