Sunday, July 7, 2013

The Hypothetical Nikon D400: EXPEED Processor Analysis


As long suffering Nikon D300 and Nikon D300s users know, the D400 has been something of a mirage on the horizon; we've seen it from a distance, but the closer we've come, the further away it gets. Last year, (2012) the rumors were all over the place, but the most credible rumors were uncannily close to what the D7100 eventually turned out to be. If you match specs to rumors, the D7100 is almost what people expected, minus appropriate buffer size, full metal body and the traditional Nikon "pro" control layout. Fast forward to the middle of 2013, and the D400 rumors are on again. So, if the D7100 is already a great camera, all we need to do is add back the full metal frame and larger buffer, and then people will be happy, right?

No; and here's why:

In product marketing, you want to shoot for where the market will be going, not where it is already. A better version of the D7100 is what the semi-pro crow wants now, but it won't remain competitive over the 2-3 year model lifespan. As I said last year, it was almost certain that the D400/D7100 (we didn't know which back then) would be a 24mp sensor, only that it wouldn't be the same as the one used in the D3100 at the time. It would be better. Same principle applies today; the D400 can't be the same as the D7100, it must be better.

But how? It used to be that Nikon and Canon would first introduce a sensor family in their semi-pro lines, and then roll out a cheaper version with their enthusiast cameras. That worked for a number of years; you would always get some D90 level shooters who couldn't wait and who instead grabbed the D300 because it came out first and was the latest and greatest at the time. This strategy worked because it preserved the higher average selling price of the Dxxx cameras, and made sure that Nikon could maximize their profit potential before rolling out the lower margin enthusiast level DSLRs. In recent years this strategy got turned on its head. The flooding in Thailand and the earthquake in Japan altered a number of shipping dates, the world economy has not lived up to forecasted expectations, and both Nikon and Canon disrupted their normal product rollout with the introduction of affordable full frame cameras. Now what we have is the lower end being introduced first; we have yet to see the higher end. We've had the D7100 for a number of months, and new we have the Canon 70D announcement. Still, there is no concrete information about the D400 or the 7DmII.

The D200 and D300 both had larger buffers and faster frame rates than the D80 and D90. The D7100 has the same frame rate as the D300, and can almost pass for a semi-pro camera were it not for the buffer. I would argue that the D400 would have to be capable of 8fps in order to differentiate it from the D7100; 7fps won't cut it because it's too close to 6fps. Also, 8fps slots in nicely under the D4 in the 9-11 fps range. This is assuming that the D400 uses the same or a variant of the Toshiba 24mp sensor.

Here is a table showing the processor throughput of various Nikon DSLR's since the D3/D300 generation, as expressed in megabytes per second. (Throughput = resolution * bit depth * fps). Despite how the marketing simplifies the messaging, there is no such thing as a common Expeed processor; within each generation, each camera uses a different variation. In general, the full-frame professional cameras tend to get higher specifications than the DX cameras, with the overall throughput champ being the D4.


Camera Launch Date Expeed Processor Generation x y fps Mb/s







D300 23-Aug-07 1 4288 2848 6 122
D3 23-Aug-07 1 4256 2832 9 181
D3x 01-Dec-08 1 6048 4032 5 203
D3s 30-Jul-09 2 4256 2832 9 181
D7000 15-Sep-10 2 4928 3264 6 161
D4 06-Jan-12 3 4928 3280 11 297
D800 27-Feb-12 3 7360 4912 4 241
D7100 21-Feb-13 3 6000 4000 6 240
D400 01-Sep-13 4? 6000 4000 8 320

Below is a chart depicting the progress in Nikon DSLR processing power over time. Note that while there is a definite progression, we are not seeing a Moore's Law type of exponential increase over time. (Classically defined as a doubling of power every 18 months, though current estimates indicate that the actual doubling time is between two and three years within the overall computing industry). To give you an idea of how much further there is to go, Expeed 3 is manufactured on a 65nm process; Intel's Ivy Bridge series is produced on 22nm fabs.




What this shows is that Nikon currently doesn't have the processing power (at least in current models) to propel a hypothetical D400 to 8fps as a means of differentiating it from the D7100. To get to 8fps from 6fps is a 33% increase in processing power. Dropping in the D4 specification of Expeed 3 would bring the D400 fairly close to that benchmark, but would likely be an expensive way to do it. With it's high average selling price, the D4 can afford to use a brute force solution to processing, but because the D300/D400 line sells in relatively high volume, it would be bettered served by moving to a new generation of processor rather than overclocking the current one.

Dropping the resolution would be another way to achieve 8 frames per second, but this would be a retrograde step, as the resolution would come back to 18mp, not much more than the D7000. So based on expectations of where the D400 will have to be in terms of future performance, I think we will be looking at another generation of Expeed processors soon. Also, don't forget that having a lot of frames per second is useless unless the camera's autofocus can keep up. We're also talking about an upgrade in predictive AF to almost D3 performance levels. If and when the D400 arrives, it will almost certainly be an exciting camera, and not just "better." After all, it has the legacy of the D200 and D300 to live up to.



3 comments:

  1. I have a D7000 since 2010; I like to shoot several themes but mainly: Landscapes, macro, sport\ public events and street.

    The only thing I don´t like in my camera, and I really need is a bigger buffer. A higher Dynamic range and a better ISO performance are also welcome.

    I don´t need more resolution!

    So the D7100 is almost a downgrade . I do not want (and can…) move to the low full frame range, that also do not solved for the better all this questions (D4 is not a option).

    If I had a D300s I would be very upset the Nikon…

    I think the “D400” have a nice market waiting for a really upgrade. Do it Nikon!

    ReplyDelete
  2. i hope next d400 aviable 50 iso ....

    ReplyDelete
  3. D400
    20MP (used in new SonyA)
    8fps (9fps with Battery Grip)
    Improved Buffer size
    Expeed 4
    Wi-Fi
    GPS
    51point AF tweaked with more cross type & Better LV/Video AF
    ISO from 50
    No AA filter display

    Lens - AS-F 16-85 f/4 VR

    New SU900R + SB1000R both radio triggered

    Awaited D5300 will have a touch screen. Or touch AF



    ReplyDelete