A question to 4870 users

[samplaire]

Hi,
For many reasons I work on a Mac and its MacOSX. But I found it convinient to have a Windows partition to be able to play TS. The machine specs are as follows: a quadcore Nehalem Xeon 2.66GHz CPU, 7GB of 1066MHz RAM, SATA2 HDD and nVidia GT120 512MB graphics card, Win 7 64bit. Recently I got the 4870 512MB ATI Radeon card (Mac edition) because I've read benchmarks it is almost 3 times faster than the GT120 (there are not too many options to choose from in the Mac market, this was the 'cheapest' one, next was the 2 times more expensive). I've ran 3DMark06 to see it's indeed almost 3 times faster - 16000 points (ATI) to 6600 points (nVidia). I've ran those tests to ensure everything is alright because I don't see literaly ANY improvement in performance both in TS2010 and 2009 using OpenGL! I'm completely disappointed with it and additionaly Mac cards are really expensive... Now I put the GT120 back in the machine but I realized I forgot to test it in DirectX mode... Can anybody tell me if there will be a signifficant performance boost when in DX mode? I don't want to swap the cards one more time without any knowledge in the DX case - these are fragile devices. Any comment appreciated!

Wojtek

[edit]
I forgot to mention I intentionaly 'test' TS2009/10 on my route's slowest point where windows shows 7-10fps when I alt+tab to desktop. Perhaps this is not a professional measure tool but the numbers are the same both with ATI and nVidia, no difference. When running the 3DMArk tests, the 3D movies were visibly different and the numbers the same - nVidia 18FPS, ATI 50-60FPS so I expected 7-10fps to 17-20fps increase with ATI in TS...

 

[whitepass]

Know idea on a Mac but for me DX is about 10x faster that OGL according to Fraps.

 

[samplaire]

Thanks! +1 point to reconsider testing the card

 

[JCitron]

Wojtek,

Dropping to the desktop will impede the performance and give you incorrect readings. Try using Fraps, which will run in the background while Trainz is running. This will give you more realistic results.

Over the years, when a manufacterer says that something is 6x faster than a previous model, or one by a competitor, I've never found that to be true. In the real world, the results are more like 2.5x to 3x the results, which is what you seem to be seeing.

John

 

[samplaire]

Thanks for your input! As I said, the going-to-desktop method isn't a valuable tool but the results IMHO can be comparable when you test 2 things in the same conditions. And while the 7-10fps can be missleading or just not true one could expect dramatic difference between cards from 2 worlds. As a sound engineer working in the commercial business I'm aware of the '6x', 'thousands more better' etc blah blah so I found the info regarding performance differences made by independent testers (http://www.barefeats.com/nehal05.html). And as I said I also tested the 2 cards with 3dmark06 to see the enormous difference between them both. TS2009/10 seem to behave exactly the same. Not even 10 percent... I hope you see now

 

[JCitron]

Trainz does put a lot of stress on a lot of components so the video performance is probably going to suffer a lot more than it would with other programs. I'm sure if you were playing with PhotoShop or some other regular graphics program you'd see the difference in performance.

A professional sound engineer, very interesting. My university offers a top notch SRT program, which is very difficult to get into and remain in once your there. There are so many people that apply to it because they think it's easy, and end up dropping out. I'm not in the SRT program, but instead as a music performance major. Hopefully I can remain in school next year because my living exepenses are very high.

Take care,

John

 

[samplaire]

Hi John, thank you for your assistance I'm an attempted economist (4 years at a university, but dropped it while working at a local radio station) but to be honest I'd like to be a train driver So I became a sound engineer. Weird, isn't it?

Back to the topic: I couldn't resist myself and put the card back in the machine to discover the DX mode is my solution, really!!!! Now the desktop test shows 23fps in the problematic place! So the increase is enormous )) I'm so glad I didn't return the card The only thing is some rolling stock has white bars instead of writings but I'm sure I saw people resolving the problem at our Polish Trainz Team site.