Current iPhone vs past supercomputers

I just got a new iPhone. How far back in time would I have to travel for it to be the most powerful computer in the world?

Somewhere around the late 60s to early 70s.

As an example:

in 1976, Cray Research unveiled the Cray-1, an unbelievably fast computer designed specifically for scientific computing. It had a theoretical performance of 160 MIPS. It contained roughly 8Mbytes of main memory, weighed 5.5 tons (including its freon refrigeration system), consumed 115Kw of electricity, cost 5.5 million dollars, came with no apps (other than a FORTRAN compiler), and couldn’t make phone calls.

But could it take photographs and let you browse the Web?

The CDC-7600 was built in 1969 and a a peak performance of about 36 MFLOPS which is about the same as the iPhone 4.

That machine cost $5 million in 1969 which is $29.4 million in today’s dollars. Rather bigger too.

Does that price include the white belt and orange shirt?

That was out in California, so what do you expect?

Back in Minnesota, where those machines were designed & built, they wore plaid flannel shirts.

First you have to define what you mean by “powerful”. One way is by getting an engineer to measure how fast the internal electrons are talking to each other. But if you mean in terms of how much it can accomplish visibly to an ordinary person, I’m guessing that you’d have to go back to before Thomas Edison.

I’m being serious. Beside its usefulness as a flashlight, how much can an iPhone do when there’s no network to connect to? One of the reasons I’m staying away from the current generation of smartphones is that all their smarts vanishes when the Cloud is out of reach. Specifically, I was looking for a phone/PDA that would let me edit simple text files and save them on the device (rather than in the cloud), and I couldn’t find any. Can one even use camera, or are those saved in the cloud too?

I suppose I must admit that I’m exaggerating. I think they have calculators that can be used. I guess that brings it to the early 1970s.

I don’t know of any smartphone which can’t edit and save files locally. Depending what apps you have installed, your iphone/android/windows phone can do a heck of a lot of things with no network.

Like play media files and even edit them. You have to go to the fairly recent past to get that. I don’t think an early Cray could do much with any media file without the software to go with it which is another huge part of this question.

Pretty much anything except surf the web, get email, and make phone calls. Like any other smart phone. What gave you the impression you needed the cloud to do anything useful on the phone?

You should take another look at smart phones, the abilities you describe have been around since Windows 6.0, at least. And that well predates ‘modern’ smart phones (like Android and iOS based devices).

Music synthesis on the PDP-1, 1964 Link.

In the late 70’s I was fooling around at a recording studio and used an S-100 computer to digitize, store, and play back audio. This was with 8-bit sampling at a low (several KHz) rate. That was well before there were CDs - the basic format was probably defined at the time I did my audio work, but the availability of audio CDs was several years away.

The main limitation was the cost of storage (both main memory and disk) - in the mid to late 70’s disk drives that stored anywhere from 5MB to 200MB were the size of washing machines and very expensive. A typical hobby computer had somewhere between 8KB and 64KB of main memory and (if a high-end system) perhaps 512KB of disk (a pair of single-density 8" floppies).

I’m surprised an iPhone has so little power. I thought it was like a little computer. I know it has less power than a computer, but I thought it would at least have a tenth the power of a current desktop. Isn’t 36 MFLOPS like, a hundredth or a thousandth of a modern desktop, or am I wrong?

I googled attempting to find out the FLOPS of an iMac and ended up on this page: http://www.intel.com/support/processors/sb/cs-023143.htm that lists Intel processors in the tens of GFLOPS.

How does an engineer measure these internal electrons? And what would be a high speed?

I presume that the question is about processor power, and perhaps memory and non-volatile storage. And people are providing meaningful answers.

Floating-point operations aren’t particularly important for a hand-held device. For that matter, even desktop machines aren’t generally floating-point bound. Integer operations are going to be much more important to overall speed, and so they are what is optimized on the iPhone. Most of the calculations requiring floating point would be for screen rendering, and the iPhone has dedicated hardware for that - hardware that would blow any 30-year-old supercomputer out of the water.

I’m not sure when you looked last, but the iPhones have 8, 16, or 32 gigs of memory to store whatever you want on it locally. Take your own photos, store music, games, PDF’s, word processing documents, take your own video or watch movies, books, whatever - all to be played back at will, regardless of connection to the web. Depending on what apps are on-board, you can also edit any photos or video you have taken, along with editing documents. If the cloud is unavailable to connect, but there’s a hardwired web-enabled computer nearby, you can still connect via USB.

When you are connected to the cloud, you can stream everything and not even use any of that storage to watch youtube, netflix, or music from Pandora, Napster, whatever!

Now what year are we in? :smiley:

Per what the others said what in the world are you talking about? Where did you get this notion?

Modern smartphones have all their text editing smarts (and other programs) onboard and can save locally. This has been true ever since the first smartphones came out several years ago. The only apps that are “cloud” dependent are those they rely on an external database or data set of some kind. If you are determined to wait until a smartphone can hold the web offline I think you will be waiting a long, long time.

It is not a clear cut problem to find a historical machine or even system that relates directly to a modern computer, even more so for a smart phone. Relating to a supercomputer is even more difficult because they have really significantly different uses.

A traditional supercomputer has usually been a floating point monster. 64 bit floating point on large regular data. Your average smart phone is notihng special at that particular job, mostly because there is no good reason to provide the capability. On the other hand, pretty much all the other useful capabilities are well taken care of.

Most smart phones are ARM based, with Coretex-8 being very common. Clock rates up to 1GHz, reasonable cache, and with graphics accellerators on many phones. Support for some form of SIMD acceleration, whether proprietory or ARM’s NEON, gets things very close to the capabilities of old vector machines like the Crays, but still typically no 64 bit floating point. 32 bit, for games and graphics support yes, but the requirements of the platform just don’t need 64 bit.

The old Crays didn’t even have caches, they just had very fast memory and massive prefectch that worked well with the vector operations. This meant they didn’t slow down like more general purpose machines do when the caches start to thrash.

Even very humble phones have had significant, if very specific, computational power. Running the codecs and encryption engines requires processors with quite a bit of grunt. However the implementations are usually very targetted at the problem. So whilst they may whip an old supercomputer on that one task, they won’t beat the supercomputer on what it is good at. There was even a bespoke supercomputer devoted to QCD calculations that tried to leverage the processing power in early phone chipsets. (Lattice Gauge QCD codes require a small matrix kernel on a 8D space that only needs 32 bit float. Thus it turns out to be a nice fit. Most physics problems are not so well contained.)

So, that answer to the OP is really not well defined. You can do a rough match against various metrics on machines back in history, but a smart phone is a very well honed integrated system design, as were the old supercomputers. With very different use cases they are optimised in very different ways. So any comparison is going to find quite diverse answers depending upon the metric you use.

I’ve always heard this, but never quite understood it. Could you provide an example of something that a supercomputer’s purpose-designed chip would be better at than a general-purpose one? What exactly is it about the hardware that makes them better-suited to any one thing?