[IDEA] Binning: Best parts go to the top model


#1

As you may already know, no component is made equal. In silicon chips like the CPU and GPU, its called “silicon lottery”, where two identical chips can perform quite differently. But the same applies to other components like the display or battery. The manufacturer will give their “guaranteed” value, in which Eve as a customer is allowed to return the product as defective if it doesnt meet the guaranteed value. On the opposite end, for components that go beyond the guaranteed value, it counts as a bonus.

Enter the term “binning”. It is quite a common practice in the industry, but the best example is probably the GPU AIB industry. Part of the reason why one manufacturer offers half a dozen of different GTX 1080s is often for that reason: the more expensive one is more likely to overclock better, if not overclocked from the factory already. The impact could go as large that it may come close to the GPU above it in price level. Thats why Nvidia doesnt allow partners to bin the GTX 1070 Ti, as it may eat into GTX 1080 markets. But its an isolated case.

Anyway, for us it seems that we have a lot of variety in the display (brightness can be anything from 380 to 520 nits), and possibly also the case for other components. So my idea is to have the top 5% display, for example, for the range-topping $2000 model. We could set the threshold with something like, 450+ nits brightness, no backlight bleed, etc.

That could go for other components (e.g. battery with 50 Wh or larger capacity) , basically anything where some parameters can vary by quite a noticeable amount.

In my opinion, people who pay $2000 for the range-topping model deserves better experience than the people who pay less than half of that on the Core m3 model. (and Im speaking here as an m3 user myself)

What do you think?


#2

i understood most that you said.
But what do you want to do?
I agree that “people who pay $2000 for the range-topping model deserves better experience than the people who pay less than half of that on the Core m3 model” and the notebookcheck review scaried me a little bit.


#3

It’s a good idea but is it feasible from a production point of view? Won’t the process of testing and selecting the components be too costly?


#4

Well, it has to be done by Eve, and as BrightonBreezy mentioned, it might not be possible.


#5

What about the other i7 SKU ?


#6

What I wrote above is just an example. maybe we could put the threshold for the i7 models, or the 16 GB RAM models.

Or do the opposite and make sure the bottom 5% (e.g. under 400 nits, has backlight bleed) goes to the m3, so other models would get less chance of getting those units.


#7

i kind of understand that, and i agree, although it doesn´t seem very feasable.
It will certainly be terrible for someone who paid the i7 and get an V worse than the m3.


#8

This would require so much manual work that it wouldn’t work in real production line.

Every display should be species and checked individually every time. Just wouldn’t work


#9

Well, we do calibrate each display manually already. Not saying youre wrong, it will certainly require more work and time


#10

It is not something to do on Eve’s side, but on the semiconductor side. Eve is only to integrate the sorted devices (and eventually check the spec time to time for QA)
Semiconductors do leakage sorting on regular basis as long as there is enough volume to ensure that all devices will find a buyer. It is then very usual to find same component with, for example, different class of temperature range for a very variable price.

Main question is to identify the key components and verify whether the vendor actually have implemented such kind of market strategy for this component.


#11

But those re done after assembly. What would happen if after assemby u find out th the screen is nit good for the top spec one

Who would pay the extra manual work done?


#12


:stuck_out_tongue:


#13

I think it is ludicrous to base this on a bunch of outcomes carried out by different people in different places on non-consumer units.


#14

Maybe one could sent those reviewers who got an somewhere faulty unit (shown by beyond normal test results) an fully qc-ed sample?


#15

This definitely is an interesting idea, but imho it doesn’t sound like something for Eve (currently) though.

Firstly, Eve’s business model is based on quality. The same checks should be performed for all devices leaving the facility, all customers should feel equally honored.

Secondly, someone investing $800 might spend more from the stash of available money than someone who easily can take the highest price out of the bank account.

Thirdly, as a newcomer, Eve should be thankful for everyone purchasing their product. I have not recognized the team as a group that pursues the biggest margin as primary goal.

So this might be something down the road (introducing a special edition or so), but for now the effort that has to be put into it takes away valuable resources from areas needed to stabilize business and puts focus on the wrong aspects.


#16

Totally agree, I dont mean we should leave the m3 as a buggy low-quality product.

The idea is that the m3 would give guaranteed performance, but not much more than that, while the i7 1TB would have more chance in getting much-better-than-guaranteed performance.

As I said above, the m3 should still give all the guaranteed performance, but regardless of the margin, a device that costs $2000, in general should offer better experience than a device that costs less than half of that.

On your greater point (and others), I agree though, it might not be easy to do in the manufacturing, and our “guaranteed” levels are already very good.


#17

Didn’t the guys say that it’s pretty difficult to take a V apart again once built? Eg high risk of damaging the (expensive) screen? Though they did show us that screen taking off machine in the factory…

End of the day, this seems pretty much like a non starter as the costs outweigh the benefits IMHO

If a V is not performing as it should, it should be picked up by QC, or failing that, returned by the customer for inspection.


#18

Actually if I get the idea, it would not be to tear appart existing Vs but rather to select the components from the start in order to get only “best corner” type of components in a premium V (at least for the key elements).

This would constitute a kind of premium device as it could offer better power consumption and performances.

Thinking about this twice, regardless of the availability of those sorted components on the semiconductor vendor side, i kinda wonder whether this would really be marketable.
A nVidia can do because they use their own components.
A bose will do because they build their own amplifiers.
In case of a PC, as a while system, I wonder whether it could really be valuated. Of course need to see the whole system, but in the end it will always be the same Intel MCU(for example) that you can find on any other system. Unless Intel value it differently(name/spec), it would be difficult to justify performance/power(and thus price) i think.

Edit : just read @Patrick_Hermawan last post on this, and this is basically the same point i think


#19

I thought however that you wouldn’t know which components were the best till you actually plugged them in and tested them. ie for some bizarre reason we have a super m3 right now that is outperforming it’s intended performance


#20

That super m3 could even be an disguised i7 what slipped through qc!
(Info from Sherlock Holmes himself).