Canon R5 Mark II - speculated in Q2/2023, specs

Tazz93 wrote:

Bigger wrote:

Tazz93 wrote:

The point on the dual Digic setup I partially agree with you. Personally, I don't expect it to have one, like you mentioned, but for different reasons. In the past, Canon has used dual Digic setups in the 5DS and 5DSr, but I don't think it would be necessary in this application. I just don't see a need to increase the processing. Dual Digic would be an aid to oversampling 60+MP down to the various 8K and 4K resolutions, but that's it. 8K 60 is very likely possible on the current hardware (as on the R5C), if not for the overheating.

And that's the reason for dual processors. Power consumption would vary linearly with clock speed at the same voltage, but the need for increased voltage to get reliable operation at higher clock speed makes it more quadratic. So two digic processors doing the same work at half the clock speed each would only use about half the power, and would spread that power dissipation out over twice as much contact area.

Canon obviously hit the power dissipation wall with the R5, and only made incremental improvement in the R5c. The dual processor arrangement would make the most sense to get added computing power for and/or for 8k120 raw and/or down sampled 8k to allow focus breathing correction.

I understand your point, but still disagree. A dual setup could help with thermals, but that isn't a guarantee. Aside from the additional heat soak in the camera, (basically 2 heaters at 70C instead of 1 at 110C), they would have more head room to do other tasks, so its more likely they would use some of that headroom and keep pushing the processors to a higher duty. Also note, I think the biggest factor in temps come from the speed of the processor (more power is needed with faster clock speeds), so if they were to reap the benefits of lower temps, it would require a significantly slower clock speeds. Theoretically, you could have two cores running at 1Ghz instead of 1 at 2Ghz. That would definitely hurt performance in a number of areas. IMO, they would be better off creating a faster more efficient premium processor.

However, that isn't my biggest contention. Looking at the board layout there simply isn't a ton of room to add a second processor without significantly changing the architecture, layout, and potentially the form factor. Significant changes costs money, which means the consumers pay for it or they strip the camera down. But ultimately, IMO, it sounds very counter productive to add more expensive hardware to mitigate heat when a properly designed set of heat sinks would do the trick.

If the R5ii is going to step up the game by adding 8K60 raw and/or 8K oversampled, more processing power will be needed. Assuming they are re-using the digic-x (and not a digic xi) using two chips would be the logical way to get the extra horsepower. Yes this may require board and body tweaks to cram another pound of manure in the bag, but this is a mark ii update after all. And Canon already upgraded the heatsinks in the R5c, so I don't see much room for improvement there.

The single digic x architecture obviously pushed the practical limits at 8k30 raw without oversampling. The temps are just a effect of the power dissipation, not a cause per se. Trying to get one processor to do twice as much work will approximately quadruple the power dissipation. Even a small increase in computing will have a big impact because of the quadratic (x^2) increase in power. So there would be no room for extra workload with a single digic x (assuming the processor geometry is similar).

What they could do is have the one processor do most of the same work as the R5, and just have the second processor active for video on the R5ii. This would make maximum re-use of the existing development, since the extra processor would only do new tricks. As for the expense, a 2023Q3 R5ii release date would be three years after the original R5, so two cycles of Moore's Law to mitigate the cost.