By using this site, you agree to our Privacy Policy and our Terms of Use. Close
alephnull on 17 February 2010
Edit Thread
Squilliam said:
Icyedge said:
Squilliam said:
Of course. It could be argued that thousands of people died because the people who folded on the PS3 could have bought the discovery faster with GPU folding.

You joking right? The PS3 is 10 X faster than an average computer at folding.

 

http://folding.stanford.edu/English/FAQ-PS3#ntoc6

Thats an average computer.

There are fewer than half the number of active Nvidia GPUs but they deliver four times the performance that the PS3 gives.

http://fah-web.stanford.edu/cgi-bin/main.py?qtype=osstats

From the same website (see paragraph 2)...

How are the FLOPS calculated?

 

People often use the number of floating point operations per second (FLOPS) as a metric for the speed of a computer. One question that arises is how to compare machines with radically different architectures. In particular, what requires only a few operations (or even just a single operation) on one machine could require many operations on another. Classic examples are evaluations of functions like the exp(x) or sin(x). On GPU and Cell hardware, functions like this can often be calculated very quickly, say in one cycle, while this is often counted as 10-20 operations for other machines.

We take a conservative approach to FLOP calculation, rendering quantities such as exp(x) or sqrt(x) as a single FLOP, if the hardware supports it. This can significantly underestimate the FLOP count (as others would count an exp(x) as 10 or 20 FLOPS, for example). Others take a much less conservative approach and we are considering giving two counts, adding a more traditional (less conservative) count as well.

The ideal comparison would be to run Folding@home on the supercomputer itself to test its speed. In this sort of comparison, FAH would likely do very well, and we are investigating the best way to perform this benchmark, as we expect people would be very interested.

 

How should the FLOPS per client be interpreted?

 

We stress that one should not divide "current TFLOPS" by "active clients" to estimate the performance of that hardware running without interruption. Note that if donors suspend the FAH client (e.g. to play a game, watch a movie, etc) they enlarge the time between getting the WU and delivering the result. This in turn reduces the FLOPS value, as more time was needed to deliver the result.

 

It seems that the PS3 is more than 10X as powerful as an average PC. Why doesn't it get 10X the credit as well?

 

We balance the points based on both speed and the flexibility of the client. The GPU client is still the fastest, but it is the least flexible and can only run a very, very limited set of WUs. Thus, its points are not linearly proportional to the speed increase. The PS3 takes the middle ground between GPUs (extreme speed, but at limited types of WU's) and CPU's (less speed, but more flexibility in types of WUs). We have picked the PS3 as the natural benchmark machine for PS3 calculations and set its points per day to 900 to reflect this middle ground between speed (faster than CPU, but slower than GPU) and flexibility (more flexible than GPU, less than CPU).