Poll: In gaming, is FPS a measurement of performance of CPU & GPU, or measurement of performance of every component working together?

CPU & GPU
2 6.5%
Every component
27 87.1%
Other
2 6.5%

Status: Voting has ended

29 Total Votes

Poll Debate about definition of FPS (Frames Per Second) in gaming (Read 2025 times)

  • Avatar of Ragnar
  • Worthless Protoplasm
  • PipPipPipPipPipPipPipPipPipPip
  • Group: Premium Member
  • Joined: Jun 15, 2002
  • Posts: 6536
Man, PCs make things so much more complex.  On a NES, the amount of frames per second is always exactly 60.09982293844223, except when the CPU is unable to keep up with the instructions it needs to process, at which point it begins spreading the calculations over several frames, causing "frozen" frames.

Yes

I think the real reason I liked Metal Gear Solid 2 was because it was the first game in a decade to actually run at the right speed

even my fucking MP3 player has lag
http://djsaint-hubert.bandcamp.com/
 
  • Avatar of King Arthur
  • Heavenly King
  • Group: Premium Member
  • Joined: Jan 29, 2003
  • Posts: 20
Indirectly, since RAM does not process the frames, but the processor which does, usually needs something from the RAM. The processor might work at 100% capacity but if you suddenly need to get in new textures or something the processor has to wait for the RAM to finish the loading before it can continue processing the frames. Your FPS goes lower only because the game is loading/"paused", not because you have insufficient processing power.

CPU/GPU Performance - RAM Performance = Total Performance (Assuming CPU/GPU Performance > RAM Performance)

Using my crappy example again, if I take longer to get your your coffee I am directly affecting the time you take in getting your coffee. The same can be said here: If the RAM is slow, the CPU/GPU's performance are lowered by that amount.

Another example: You (the CPU) can sign a total of 50 contracts per minute, but if I (the RAM) only bring 10 contracts to you per minute you can only sign 10 in the end. I am directly affecting your performance.

This can also work the other way: You (the CPU) can sign a total of 50 contracts per minute, but if I (the RAM) bring 2000 contracts to you per minute you can only sign 50 in the end. In this case you are directly affecting my performance.
Last Edit: October 17, 2008, 09:50:30 pm by King Arthur
  • Avatar of Verne
  • Dwarf Giant
  • PipPipPipPip
  • Group: Premium Member
  • Joined: Nov 27, 2001
  • Posts: 492
CPU/GPU Performance - RAM Performance = Total Performance (Assuming CPU/GPU Performance > RAM Performance)

Using my crappy example again, if I take longer to get your your coffee I am directly affecting the time you take in getting your coffee. The same can be said here: If the RAM is slow, the CPU/GPU's performance are lowered by that amount.

Another example: You (the CPU) can sign a total of 50 contracts per minute, but if I (the RAM) only bring 10 contracts to you per minute you can only sign 10 in the end. I am directly affecting your performance.

This can also work the other way: You (the CPU) can sign a total of 50 contracts per minute, but if I (the RAM) bring 2000 contracts to you per minute you can only sign 50 in the end. In this case you are directly affecting my performance.

Well using your contracts example I can turn it into my favor by saying that I write each contract with the same speed so my speed of writing is always the same (processing power is the same = contracts per second or whatever) but I have to stop every now and then to wait for you to deliver the stuff to me (loading). On the second example I have far more contracts than I need so I don't have to wait at all (no loading, better pre-loading/buffering) but my processing speed is still the same.
  • BAA2U
  • PipPipPipPipPipPipPip
  • Group: Premium Member
  • Joined: Aug 7, 2007
  • Posts: 1403
You're assuming that every "contract" is the exact same size and complexity, where as instructions sent to the CPU are rarely the same of either. And not being able to "write a contract" is the ram directly affecting your performance, because without the contract you are not performing! Longer "contracts" will take more transer time to get to the cpu, again the ram having an impact at the speed the cpu can operate because of the speed the ram operates is directly effecting it.
  • Avatar of King Arthur
  • Heavenly King
  • Group: Premium Member
  • Joined: Jan 29, 2003
  • Posts: 20
Well using your contracts example I can turn it into my favor by saying that I write each contract with the same speed so my speed of writing is always the same (processing power is the same = contracts per second or whatever) but I have to stop every now and then to wait for you to deliver the stuff to me (loading). On the second example I have far more contracts than I need so I don't have to wait at all (no loading, better pre-loading/buffering) but my processing speed is still the same.

If you have to stop doing something, then that is time wasted and performance (FPS in this case) lost. You aren't operating at peak performance if you have to stop every now and then and wait.

The second example put our roles in reverse: Instead of placing the RAM as the bottleneck I made you, the CPU, as the bottleneck. You, the CPU, may be operating at maximum performance, but the system as a whole isn't because you're slower than me, the RAM.
  • Avatar of reko
  • PipPipPipPipPipPip
  • Group: Member
  • Joined: Jun 11, 2002
  • Posts: 883
I think everything that can be said has been said, no matter what example you guys come up with, Verne will counter it with same faulty logic.

big thanx to dragonslayer for sig!
Locked