| ||||||||||
>Just to be clear, I'm actually saying I think the whole "image processing on the cloud" thing actually will happen at some point, but I see the most obvious bottleneck being how to get massive source material from point of acquisition up to the cloud servers in short order People have been saying this future of dumb terminals and central processing – now called the cloud, has been coming for decades now. But those that argue for it ignore three very basic fundamental facts. 1. Physics: The speed of light means when you have two equivalent CPUs, the closest one will render faster, thus with less total time, energy and cost. Thus, to create distant processing requires more powerful remote CPUs. Distance means time, energy and therefore costs must be absorbed to process remotely as opposed to locally. Moore’s Law has applied to CPUs but not to data transmission, thus meaning unless this flips, data transmission will NEVER catch up to increases in processing. 2. Trend is computing is distributed, local cores, not central powerful cores. Smartphones, tablets have not changed the computing paradigm of the last 20 years at all. They contain the same parts and computing model as workstation – an increasing amount of multiple cores for powerful local computing (Intel sees 100 core smartphones), increasing core count GPUs and larger local storage – with slow increase in data transmission. If neither smartphones, tablets nor workstations are trending towards cloud computing, then how can you argue that is where everything is going? Just because Google Maps and Siri are popular, don’t forget Angry Birds needs a fast CPU and GPU and storage running locally. 3. The smaller, more mobile and lighter you make something, assuming processing power and technology is the same, the slower, more expensive and hotter it will be. Just look at Ivy Bridge vs. Sandy Bridge. While incremental technology improvements have helped, we have yet to make the breakthrough to completely flip this around. If you have breakthrough that makes a smartphone sized processor faster, it will also make a desktop faster – but much faster, cooler and cheaper in a larger form factor. stephen van vuuren 336.202.4777 http://www.insaturnsrings.com/ A film is – or should be – more like music than like fiction. It should be a progression of moods and feelings. The theme, what’s behind the emotion, the meaning, all that comes later. –Stanley Kubrick From: After Effects Mail List [mailto:AE-List@media-motion.tv] On Behalf Of Carey Dissmore Hey Teddy, Just to be clear, I'm actually saying I think the whole "image processing on the cloud" thing actually will happen at some point, but I see the most obvious bottleneck being how to get massive source material from point of acquisition up to the cloud servers in short order. Once it's "up there" a lot of things can become possible with thin clients and streaming a single video output stream back to the user. Look at the OnLive gaming platform. The tech is good even if the business model, in that case, suffered. But at the same time I'm also not saying we are all dumping desktop computing in short order either. Carey On Nov 6, 2012, at 6:01 PM, Teddy Gage <teddygage@gmail.com> wrote: I'm with Carey. I was in the middle of an email about this when I read all the replies, so I scrapped it. He said everything I was saying. Basically, I don't think the "desktop" will ever just go away. Once you start getting into physical simulations, like realflow or fumeFX, even a brand new hex core workstation feels like working on a cellphone from 1993. Not even to mention rendering that stuff at 4K with GI and raytracing etc. I don't even think we'll hit some kind of limit where "ok, this is fast enough", because it can always look better or render faster. |
