FRAPS performance hit

Luck100

FRAPS claims to have minimal overhead. I was using it for BF4 since precision X doesn't work (due to 64 bit binaries?).

Anyway, I was curious how much impact it was having so I ran Unigine Valley with and without fraps, everything else the same.

Without fraps: 52.5 FPS
With fraps: 48.3 FPS

That's nearly a 10% drop, just to display the framerate! I'm not doing any video capture or fraps benchmarking, just displaying FPS. Those numbers are with Unigine Valley at 1440p, using 2-way SLI GTX 670.

Surprised I haven't seen any mention of these numbers anywhere else, it was a lot more than I was expecting.

Jaafa

FRAPS's FPS hits are well known. It's simply the price you pay. Try another program like Dxtory maybe.

marco_polo

Not sure it is a valid test tbh, of you are running frame counting software on top of a program with inbuilt counter as well, they could easily be conflicting with each other.

Luck100

marco_polo wrote: »

Not sure it is a valid test tbh, of you are running frame counting software on top of a program with inbuilt counter as well, they could easily be conflicting with each other.

You might be right, but how else can I measure the performance impact?

Luck100

Jaafa wrote: »

FRAPS's FPS hits are well known. It's simply the price you pay. Try another program like Dxtory maybe.

I'll give DXtory a try, I was just looking at the website now.

As for FRAPS well known hit for FPS - after a few minutes of googling and reading I found exactly 0 measurements of fraps impact for FPS display. The only mention I saw at all was verbatim quotes from the fraps web site, which claimed "minimal" impact (no examples or numbers).

There are lots of discussion of fraps impact for capture, but I'm not doing capture. Just FPS display.

Jaafa

Luck100 wrote: »

I'll give DXtory a try, I was just looking at the website now.

As for FRAPS well known hit for FPS - after a few minutes of googling and reading I found exactly 0 measurements of fraps impact for FPS display. The only mention I saw at all was verbatim quotes from the fraps web site, which claimed "minimal" impact (no examples or numbers).

There are lots of discussion of fraps impact for capture, but I'm not doing capture. Just FPS display.

I see what you mean now. What's your CPU out of interest? And ram?

marco_polo

Luck100 wrote: »

You might be right, but how else can I measure the performance impact?

I don't know tbh. But all it is doing is counting Direct3d present() method calls so you would be talking milliseconds worth of impact especially in benchmarking mode.

Luck100

Jaafa wrote: »

I see what you mean now. What's your CPU out of interest? And ram?

System is in my sig - 3770k and 2 x 8GB DDR3 running at 1333 MHz. I bought 1600 MHz memory sticks but they sent me 1333 and I didn't notice for ages, so I couldn't exchange them.

Luck100

marco_polo wrote: »

I don't know tbh. But all it is doing is counting Direct3d present() method calls so you would be talking milliseconds worth of impact especially in benchmarking mode.

Ok, that's good to know. I really didn't have a clue how fraps actually works. I wonder if precision X has similar impact if FPS stats are enabled.

Luck100

I tried running Unigine Valley with precision X running and the FPS monitor active on screen. It had exactly zero impact on the framerate reported by Valley (to the tenth of a frame per second!).

Gumbi

Luck100 wrote: »

System is in my sig - 3770k and 2 x 8GB DDR3 running at 1333 MHz. I bought 1600 MHz memory sticks but they sent me 1333 and I didn't notice for ages, so I couldn't exchange them.

Enable your XMP Profile for 1600mhz...

Luck100

Gumbi wrote: »

Enable your XMP Profile for 1600mhz...

They're not XMP sticks, they don't have an XMP profile. I though I was losing my mind because I couldn't find the XMP settings in the BIOS, and finally I pulled the sticks and looked at the product stickers. The product number is for an inferior 1333 MHz non-XMP RAM.

marco_polo

Just gave it a test there, got pretty much the same results with and without fraps bar .1-2 of a FPS. (44.9)

However I did have to start the benchmark, stop it and start it again because I was seeing some wierd framerate locking @30fps in the opening scenes the first time it ran and fraps was on. On the runs where that happens the score takes a 10 ish percent nosedive like you are seeing. (41.7)

It just looks like they don't play nicely together as there are definately glitches.

cherryghost

Bandicam is much easier on the system

Luck100

marco_polo wrote: »

Just gave it a test there, got pretty much the same results with and without fraps bar .1-2 of a FPS. (44.9)

However I did have to start the benchmark, stop it and start it again because I was seeing some wierd framerate locking @30fps in the first few opening scenes the first time it runs and frap was on. On the runs where that happens the score takes a 10 ish percent nosedive like you are seeing. (41.7)

It just looks like they don't play nicely together as there are definately glitches.

That's interesting, I wasn't paying close attention while I was running the bench. It almost sounds like it was trying to do vsync. I double checked and I definitely have v-sync disabled for valley (from Nvidia control panel).

marco_polo

Luck100 wrote: »

That's interesting, I wasn't paying close attention while I was running the bench. It almost sounds like it was trying to do vsync. I double checked and I definitely have v-sync disabled for valley (from Nvidia control panel).

Definately had it off as well, can't explain what is going on tbh but it seem to happen more often than not. It seems to start behaving normally a few scenes in but by then the average number has already been considerably lowered