IT IS THAT TIME of year again, Nvidia’s quarterly conference call for the financial set. Since the company gets pouty every time we try to ask it questions, we decided to publish an open list of questions for you, the people they have far less disdain for than SemiAccurate, to ask directly.
These are all open questions that have been spun, half answered, or selectively responded to. Nvidia has a habit of answering tough questions verbally in phone calls and other private forums. Some would call this a Reg FD violation, others just obfuscation, but since the SEC has better things to do, we won’t try to pigeonhole it.
In any case, if you do ask, try to get a written answer, and compare that to what it said in the last quarter’s conference call. Save a copy for the next quarter, too. Things change a lot when no one holds a company accountable for its past statements.S|A
How many GTX480s cards are going to be shipped by Nvidia before the end of April 2010?
After the initial allotment of GTX470 and GTX480 cards, will there be any more produced? If so, how many? Are these made from new wafers run, or simply parts not sold and trickled out over a few months?
What is the cost of a fully functional Fermi chip? What is the yield? When do you expect that to improve, and why?
Can the current Fermi/GF100 chips be sold to the consumer market at a profit without subsidies from compute or workstation variants?
At GTC, Nvidia promised 512 shader chips to several analysts. Now Nvidia is telling AIBs not to expect 512 shader versions at launch, if at all. Why?
There are four derivatives of Fermi on the roadmaps. When do expect them to get to market in quantity? Have they taped out yet? If not, why not?
If the derivative parts are unlikely to be out in the market before the end of 2010, what do you plan to sell in between the sub-$100 GT240 and the high end Fermi?
Are there any other 40nm GPUs coming from Nvidia, or is Fermi GF100 the only one not based on the older architecture?
Nvidia has no DX11 parts on the market. Fermi GF100 is going to be the first one, and it is exclusively high end. How is Nvidia planning on filling that gap? When?
Last quarter, Nvidia blamed the lack of G200b based parts, GTX260, 275, 285 and 295, on shortages of wafers at TSMC and bad planning. TSMC does report shortages on 40nm wafer starts, but it has not reported 55nm shortages, then or now. There has been more than ample time to raise production of the 55nm parts since the problem was discovered last fall. As of February 17th, Newegg has one GTX260-216, zero GTX275s, four GTX285s and one GTX295 in stock, all at prices far higher than any competition. Why is the supply of GTX200 series cards still non-existent? When was the last time G200b wafers were started? In what quantity? How many have been run in Q4/2009 and how many will be run in Q1/2010?
Can G200b based cards be manufactured and sold at a profit?
Has Nvidia abandoned the gaming market? If not, why, and what does it plan to sell to that market?
Nvidia has no DX11 laptop parts, and only low end (96 shaders max) DX10.1 parts. When do you expect to have DX11 laptop GPUs on the market? What do you plan to do until then?
ATI has a smaller die size and higher performance at every price point on the market. ATI also has notably higher yields on 40nm chips. It can make all of its parts cheaper than the Nvidia alternative. How do you plan to combat this? Will margins hold if you have to compete on price?
This is the initial list. We will try and update it, below here, with any curiosities from the conference call. There are several discussions on this topic in the forums, so feel free to join in there too. Enjoy. Ed.
Latest posts by Charlie Demerjian (see all)
- What does Qualcomm’s server SoC look like - Apr 15, 2015
- How does Qualcomm’s SenseID fingerprint scanner work? - Apr 9, 2015
- How fast is Qualcomm’s 64-bit Kryo server core? - Apr 7, 2015
- Amazon is spending lavishly on game development - Apr 6, 2015
- What is the name of Intel’s Cannonlake +1 server platform - Apr 6, 2015