Bioshock videocard performance

Julio Franco

Posts: 9,099   +2,049
Staff member
Around this time last year we got our first little taste of what could be expected from Bioshock, when 2K released an impressive 14 minute gameplay trailer. Bioshock is considered by its developers to be the "spiritual successor" of the well acclaimed System Shock 2. There is much more to this game than meets the eye, which can be discovered when playing the 1.2GB demo which was released just weeks ago.

The full version of Bioshock has also been released for both the PC and the Xbox 360, allowing us to purchase a copy and do a full analysis of how the current crop of videocards do in this next-gen FPS. The game uses a modified version of the Unreal Engine 3, also used by Epic's Gears of War and upcoming Unreal Tournament 3 game. Given that Bioshock is a single-player only game, we were a little amazed by the hype that surrounded it, yet the fact it offers breathtaking visuals and while the gameplay itself is quite good, the jaw-dropping visuals keep you entertained right till the very end.

Please leave your feedback here. Thanks!
Pretty good article. the only gripe I have is the lack of GeForce 6 series cards in the test as those are still quite commonly used (6600GT especially).
good article, but how were these fps measurements reached? tests dont mean much if the way the testing is performed is not stated.
Plus it would help people at home to perform a similar test on their current system to consider whether an upgrade to one of these cards would be worthwhile.
Thanks for the positive feedback guys! Orian here some more info for you...

Since Bioshock does not include an in game benchmark we have used FRAPS to record 60 seconds of in game action. The recording is taken from the start of the game when you are in the water surrounded by burning debris and finishes once you enter the main building, which takes you to the underwater city of Rapture.
I do not normally reply to reviews but the numbers here just seemed a bit to far off. I played thorugh the entire game with all eyecandy turned on high at 1680 x1050 on my 7900 GTO and got alot higher FPS than what is listed here. Also many sites talk about the problems with the ATI drivers not rendering correctly and leaving out some texture surfaces just leaving a black area. (I guess this was fixed in the driver version you used?)

But I have to give props to the ATI team if their 2900XT really is rendering correctly and beating out a 8800 GTS640. Good Job :D

Last how many times did you do each test run for each card? When doing manual runs with fraps it is always best to do at least 3 runs then average your scores. This will help minimize human errors. As is we have no idea if one time the player was looking at the ground the whole time and the next the ceiling.
Thanks for the feedback Athena. I think where our tests differ from other performance articles is that we used the Crash Site to test the graphics cards. Here there are a huge amount of water effects and it is a very open area. The tests were run three times and the average results were taken which explains the result trends that we did get, there are no fluctuating results. The reason our frame rates are probably lower than other sites is because I never looked at the ground or at something that would create a huge frame rate spike that would throw off the average.
Steve has LOTS of experience benchmarking this kind of games, so no doubts in that respect, you can trust our numbers.

Now, what could make a difference is what you believe are playable frame rates. Steve made it clear on the article that for him an average of 45fps (or better) is what he considers a good playable rate, while something in the 30s range is borderline and you can expect severe slowdowns depending on the conditions.
There is people that play just as well on these frame rates and don't care about a slow down here or there (especially those using laptops).
i have a 512mb x1950pro which i just bought recently and it plays all my current games very nicely. i also have an xbox 360 with bioshock and im surprised by the figures here to run the game on a pc.

very informative.
yukka thanks for the feedback we appreciate it!

Also many sites talk about the problems with the ATI drivers not rendering correctly and leaving out some texture surfaces just leaving a black area. (I guess this was fixed in the driver version you used?)

Athena, this is just a follow up post. Since I don't spend a long time in the game playing with each card we test (I played the game with the 8800 GTX) I decided to fire up the test system with our standard GeForce 7900 GT graphics card installed. At 1680x1050 using the latest Nvidia BETA drivers I found the game to be way to choppy for my liking when using the high quality settings. I imagine some gamers could play using those settings but I found it to uncomfortable.

I am not sure how many frames per second you were getting but in smaller size rooms we were getting between 24-33fps. The medium quality settings saw performance rise to about 30-40fps in the same rooms. These frame rates were seen in the Medical Pavilion level. Using the medium quality settings at this resolution I found Bioshock to be quite playable.

As for the ATI driver issues, they were never a problem on either the Intel or AMD test bed using the 7.8 drivers. The Forceware BETA drivers did generate the odd BSOD error but the official Nvidia drivers were perfect.

Also on a side not the GeForce 7900 GTO is clocked 200MHz faster than the 7900 GT and therefore does provide significantly better performance. I believe it to be around 10fps in Bioshock using medium quality settings.
Great article but I agree with Didou, the 6600GT and 6800GS/GT should have been included. They're quite commonplace. Also, is there going to be any testing done on Vista?
Fair enough points but given how the new GeForce 8600 GT/GTS and Radeon HD 2600XT performed when using low quality settings I felt there was little point including the unplayable performance of the 6600 GT.

But this is something we can look to include as we are always open to suggestions and obviously want to include what the readers want.

At this point I don't see why we cannot do a DX10 article, if the game stays as popular as it is right now it is highly likely we will do just that.