GPU comparison

[johnwhelan]

https://www.videocardbenchmark.net/gpu_list.php as mentioned in the survey.

It seems a more comprehensive list than Tom's and looking at the Passmark G3D score makes comparing cards fairly easy. You can also sort the cards by videocard value.

The interesting scores are Intel 4000 - 456 which is enough to run Middleton for laptops at 23 fps 1600 by 900. It is fairly easy to see why it doesn't have enough horsepower for speedtrees. The 5000 series Intel integrated graphics come in at 567 which means Middleton runs at 30 fps by the way.

The GTX 1060 6 gig is normally considered to be a reasonable card for TANE. 9093 but the 3 gig version comes in at a very respectable 8969

The old GTX 980 is 9639 by the way.

The GTX 1050 TI comes in at 6000 which is about as low as I'd suggest for normal TANE. Interestingly enough running the same route on a GTX 980 and a GTX 1050 TI same stock etc if the route and stock are not too demanding the results on the screen look much the same.

So yes a GeForce RTX 2080 Ti is wonderful but you may not notice any difference to a GeForce RTX 2070.

Cheerio John

 

[clam1952]

I've always used Passmark and rebranded Futuremark https://benchmarks.ul.com/compare/best-gpus

Slightly different order to Passmark at UL, however they are usually a few weeks behind with newer cards.

 

[wilts747]

I think the theory of 'How long is a piece of string' very much comes into play when benchmarking anything to do with Trainz and there is a lot to be said for acceptability is in the eyes of the user. I can't recall using any other software that utilises computing resources as Trainz does hence the difficulties in choosing an appropriate benchmark standard to provide a meaningful gauge of any computer to manage running Trainz. Does it matter, perhaps not assuming we are satisfied with the result we are achieving, but equally it would be comforting to know that currently we had spent our money wisely financing our interest or that in the future we were about to. Peter

 

[johnwhelan]

I think the theory of 'How long is a piece of string' very much comes into play when benchmarking anything to do with Trainz and there is a lot to be said for acceptability is in the eyes of the user. I can't recall using any other software that utilises computing resources as Trainz does hence the difficulties in choosing an appropriate benchmark standard to provide a meaningful gauge of any computer to manage running Trainz. Does it matter, perhaps not assuming we are satisfied with the result we are achieving, but equally it would be comforting to know that currently we had spent our money wisely financing our interest or that in the future we were about to. Peter

There are so many variables which is what makes it so difficult. Undoubtedly the content used and how it is used has a very big impact.

Cheerio John