TR Forums

So I upgraded my computer and made some benchmarks. The twist is I made a little program to draw frame time graphs.


Bad Company 2 mostly worked on the old computer with low settings. The shadows are particularly bad and there's no off setting for them. I played the multiplayer plenty even though the performance was kind of bad.


So here's a regular percentile graph. The new video card in the old motherboard is actually worse than the old video card in the old motherboard. Otherwise I guess this is what you would expect. Maybe it's crazy a secondary feature on a processor can easily beat a comparatively giant board that's a few years old.


Time spent beyond in line graph form. Maybe I should have cut the lines off at the top.


Oh that's what it's supposed to look like. I also upgraded from a 19 inch CRT that does 1280 by 960 at 85hz to one of the 1920 by 1080 at 144hz TN panels. Playing Dead Space in 4:3 is a little weird. Also something odd is going on at the top and bottom of this screenshot.


I stopped playing Hot Pursuit on the old computer because of the bad performance. It's especially bad in night races.


Again the new video card alone made frames take longer. Turns out Hot Pursuit is capped at 60 frames per second so maybe it's not the best choice for benchmarks.


Here's the first bit of the frame times graph where you can see the new video card is doing a micro stutter kind of thing but bigger. This is a complete mystery to me.


Kerbal Space Program Demo. The industry standard of benchmark tools.


I (foolishly?) decided on an upper limit of 100ms for frame times which is less than every single frame produced by my old processor in KSP. So time spent beyond is the only graph my program generates that can show my KSP benchmarks. Upgrading the processor predictably helps and adding a video card (and it doesn't matter what) helps again.


Better performance also lets KSP's simulation time run at the same rate as real time.


I also stopped playing Witcher 2 because of bad performance with my old computer. Riot Flotsam, seen here, was extra bad and where I stopped.


Too bad I didn't stop playing it with my new computer because it's a bad game!


Including deciles in the 99th percentile chart pays off here (maybe?). The new processor alone easily beats the new processor with the old video card at 99% but it's the other way around at most lower percents.


You can actually see that a lot clearer in the percentile frame times graph. They cross at about 97.3%. So which is better? Trick question they're both unplayable. Again upgrading just the graphics would have made things worse.


So there's my little program. It parses Fraps frame time files and uses .Net to make pngs. It's not clear if this download link will work: https://docs.google.com/file/d/0B5-uYmAleI21MXVlN2VzN0puMDA/edit

I was clicking around a Windows 8 in VMWare Player and was suprised by the amount of games that work mostly the way you would want them to. Naturally, benchmarking followed.



For Batman City I flew about halfway across the city in 60 seconds. I left the virtual machine running in the background with half the ram while doing the host's benchmark. I made an extra pass through the level at the start to make sure everything was loaded. Loading in the host took a lot longer than normal while in the guest was closer to what is supposed to happen. At one point the game appeared to hang with looping sound while loading for at least 10 seconds.



For Tomb Raider I ran through some of the first town for a minute. VMWare Player updates it's video output only 60 times per seconds making many games literally unplayable regardless of frametimes. Less importantly it doesn’t support DirectX 11 features. In both cases you get about half as many frames by jamming VMWare Player in the middle. Can a K Haswell get these kinds of numbers?

Thanks for sharing. How does MechWarrior Online and oh say MineCraft do?

I can't do the whole setup again but I can tell you MechWarrior Online was bad enough on my old computer I played two matches badly and left it sitting on my hard drive and now the patching process fails. Minecraft use to work fine with the old computer but at some point there was a lighting upgrade that made me pick between a new ugly mode or the good lighting that made the frames take a little longer than I wanted. Minecraft can’t make the new computer draw less than 144 frames per second without loading or calculating terrain.