Dealing with the Tesla non-availability issue

If you haven’t heard, Tesla’s are hard to come by. We have several Pegasus systems that customers have purchased, that we can’t get the units for. All of the distributors and resellers we have spoken to indicate that they are getting a small fraction of their orders filled. We have had units on order over a month. Several more orders, and a hard deadline to get units filled.

We have heard rumors of fabrication problems and part recalls (given the past history with other chipsets, some believe this to be the case, though I am not sure … more likely a low yield coupled with something else).
We have heard of extreme demand due to a sale. We have spoken with a few folks that indicate they have backlogs and a trickle of parts fulfilling them. But the backlog numbers aren’t huge.
I have a different theory. I think this is the “Dell” effect, coupled with low yields. Dell signed on to sell Teslas. In fact, if you go to one of their systems configuration web sites, they tell you what will delay your shipment. The Dual ATI Firepro? Yeah that will delay your shipment. The Tesla C1060? Nope.
Looks like a simple case of resource allocation. This is speculation. The reason the rest of the market can’t get Tesla’s is due to NVidia not having sufficient yield in its production runs to satisfy Dell and the rest of the market. So if you were in NVidia’s shoes, who would you prefer to service? I don’t blame them, even though it is negatively impacting our customers.
We are working around the shortage by getting a box full of GTX260’s with 1.7GB ram. Once the Tesla’s come in, we will swap the GTX260s out.
I’ve never bought a box full-o-graphics cards before.
Here’s hoping my analysis is right about NVidia, and the issue is simply one of capacity planning. This is something that Dell has to worry about. They did worry about it with AMD. I seem to remember similar problems with Opteron supply when Dell started selling AMD.

Dell had other concerns about AMD’s ability to supply Dell with a sufficient number of chips, said Marty Seyer, senior vice president of AMD’s commercial segment.
“The message to us (from Dell) two to three years ago was loud and clear: AMD needed to put in place the capacity to meet the expected demand,” he said.

Dell can move awesome volumes of product. They have to worry about scalability of supply. I think, again, pure speculation, that NVidia wasn’t ready for this spigot to be turned on. Their supply would have been fine had Dell not decided to sell Tesla.
This is more plausible to me than the other rumors we had heard. Unfortunately, NVidia has been mum on this, so we are left guessing.
I should point out that Dell’s raw firepower in shipping is one of the reasons why I (continuously) say it is a bad idea to try to out-Dell Dell. HP is pretty much the only other group I would even think could try. HP can also ship massive firepower.
Both are reasons why supply of various things disappear from the market every now and then.
Like Tesla.

5 thoughts on “Dealing with the Tesla non-availability issue”

  1. @Jeff
    Darn it. There goes my theory.
    Could be high demand. Or low yield. I am hearing backlogs of well under 500 in total from a variety of vendors. This leads me to believe that the demand is less of an issue than the supply. I had hoped it was simply a matter of Dell grabbing every last bit.
    Our box of GTX260’s arrived, put one in a machine going out to a customer soon (machine has been delayed due to the Tesla). Loading FC11 on it. I feel a post on the futility of anaconda based installs bubbling up 🙁
    “Hey, I’ve encountered a correctable problem. I’ll let you ‘debug’ anaconda, or exit the installation. No, I won’t let you return to a previous state, to try again, because THAT WOULD BE TOO BLOODY OBVIOUS, and all the other distros let you do that. Have a nice day.”

  2. Maybe they are delaying so they can lump all sales into a single quarter to make it look like an astronomically growing market.
    Low yield? They really aught to be able to just disable the parallel cores if it has a bug like the Cell for PS3. Or are the chips with broken cores the ones that go in the GTX260’s

  3. @Rohit
    Yeah, NVIDIA bins their parts. So do most vendors.
    This allows you to do bios/driver level hacks that “convert” one card into the other.
    I haven’t looked into the real difference between the platforms.

Comments are closed.