Folding Forum

I read about graphics cards folding for hours, to get only a few milli to microseconds worth of folding data.
Is this true?
How fast are current graphics cards?

Do you think hardware will ever come to fold at near to real life speeds?

"Ever" is a long time.

My slowest computer ran at 60 Hz, today they are 80,000,000 times faster. (my slowest Microcomputer ran at 895 kHz, still 5,500 times slower than now)

Is there a relation to the average PC's folding speeds vs real life?

Lets assume a GPU takes 10 hours to fold 1 millisec. This is 600 min or 36,000 sec or 36,000,000 millisec to simulate a real 1 millisec. So after one GPU gets 36 million times faster it can do folding in real time.

Perhaps the fastest currently available, graphics cards. Sans the math & kHz stuff. CPU folding is slow BTDT (been there done that), takes along time to get points. Install a GPU & speedy gonzales, points. GPU more specs more 'Shaders' more points, to add to the collective effort.

FAH was started some 15 years ago based on the "FACT" that it was impossible to solve folding on computers but Stanford U. figured out that is was possible if folks were patient enough, and if enough people helped, it actually could be done. (And yes, that was back when the Pentium CPU was considered very fast.) Computers have come a long way since then, and stream computing (via GPUS) was a huge step forward, too. Nevertheless, computers can't do anything close to real-time simulation.

Simple proteins fold much faster than ones with lots of atoms, so one cannot make any general statement about how many ms can be simulated per day, but for specific proteins, the actual numbers can be obtained for a particular CPU or GPU.

Theodore wrote:Do you think hardware will ever come to fold at near to real life speeds?

No, I don't. (Personal opinion ). The math is just too challenging, but there are frequent advances in hardware and in algorithms that I can't predict, so I'd be very happy to be wrong.

if quantum computing come true one day, maybe we could folding at real time
but it would be long way ahead, the qubit of IBM Q is just 20 qubit, for any practical calculating we need at least thousands of qubit, if not millions