TS2012 frame rate performance charts

[PapaXpress]

Snowstorm, I hope you are reading this.... here are my post TS2012 SP1 specs and numbers:

CONFIGURATION (bold is updated)
Hardware: Intel Core 2 Duo E8400, 8 GB (2 x 4 GB) Cosair DDR2 800mhz, EVGA 460 GTX FTW, Seegate 500GB, ASUS Xonar DX.
Software: Windows 7 64-bit, latest motherboard BIOS & drivers, NVidia v301.42, disable Microsoft Security Essentials Real-Time Protection, some other Logitech junk was left on.

SETTINGS
NVidia: default 3D Application Settings
TS2012 launch window: TSX enabled, basic settings radio button set to Custom, 1680x1050 full screen, 1x1 monitor, Anisotropic x 8, SSAA 2x1, no command line except Enable EAX
TS2012 in-game options: graphical detail set to High, Depth of Field unchecked, Adaptive Bloom unchecked

Running in the cab with head light on.

Northeast Corridor 'Early Clocker' 560Ti 448 FTW (no OC):

RailWorksProc2 2012-06-19 19-23-34-47 fps.csv

RailWorksProc2 2012-06-19 19-23-34-47 frametimes.csv

[snowstorm]

Snowstorm, I hope you are reading this.... here are my post TS2012 SP1 specs and numbers:

Finally got around to it:



Same test from four months ago:



That's quite a difference. If it's due to the SP1 update, for me the first question would be how much is the result of underlying optimisations of the game engine, and how much is the tweaking of specific routes. The before/after tests I did for the previous major performance update (Feb 9) suggested mostly the latter.

I haven't updated yet, so have the benefit of being able to do a direct before/after test on any route I like, but it'll have to wait until I'm in feeling masochistic

[PapaXpress]

Thanks for crunching the numbers! Its good to see some facts behind what our MK I eye ball is picking up.

To the best of my knowledge, NEC was not touched by SP1. I have also noticed improved performance with my route, but that is more to do with draw distance and loading times as I don't have many models down on it yet. It would be nice to see if someone could do a before and after (SP1) using CoA or NSAND as they are pretty heavy with trees and assets.

[Machinist]

At the numbers side (I like stats) this is a single resume of what I got in my new rig. Even almost not impacting fps, setting Shader Quality to "low" gives much more smoothness than with the "high" setting. FPS is not all, framerate looks to be more relevant.

Stats.jpg

At the human side (subjective perception, psycological feelings and placebo facts etc.), is visible that there is an astronomic difference in quality (graphic details and colors) when playing SSAA 3x3 compared to MSAA x8. I can say I found a new world in TS2012 with higher settings, and I'm very impressed and surprised with my own lighting enhancements with SSAA 2x2 or higher (I was used to play medium MSAA x4, and also medium in-game most of time).

Therefore, and bottom line for me, I rather play with SSAA 3x3 locked at 20fps, than SSAA 2x2 at 25, or MSAA x8 at 30 (or higher).

Each scenario, even of same route, deserves different settings of system. That's the way I play, a bit more of work before play a scenario, but much more fun when driving.

Doc.

[peterhayes]

Doc
Hyperthreading ON or OFF?
cpu overclocked? @ ??GHz
Regards
pH

[Machinist]

Peter,

No idea... what are you talking about?

MB is Asus and there is a tool to turbo the pc but I didn't use yet because it warns to enhance the cooler system. So far, when in intensive use, temp is reaching 55ºC for CPU and 45ºC for GPU, so it's good.

What's the Hyperthread On or OFF and how to overclock the CPU/GPU, is worth the risk of overheating?

Doc.

[Haystack]

Peter,

No idea... what are you talking about?

MB is Asus and there is a tool to turbo the pc but I didn't use yet because it warns to enhance the cooler system. So far, when in intensive use, temp is reaching 55ºC for CPU and 45ºC for GPU, so it's good.

What's the Hyperthread On or OFF and how to overclock the CPU/GPU, is worth the risk of overheating?

Doc.

What cpu do you have? Hyperthreading is a feature on some cpus. And overclocking is when you raise the speed of your cpu. For example, say you have a cpu that runs at 3.2 GHz stock, and then you overclocked it to 4.0 GHz, that would be called overclocking. When you overclock its best if you have a after market cooler to keep temps down, The stock cooler won't cut it.

[Machinist]

I have an I7-2600K @ 3.40 GHz (8CPUs).

One thing I noticed and didn't mention before is when checking Multi-Core CPU1 usage is average 90%, CPU5 40% (the others are few used). When Multi-Core is unchecked CP1 60%, CPU5 50% and CPU6 30%, ie usage is more distributed, but fps is quite the same. What is better, after all?

[Hawk]

These two little utilities may be of interest to you.

CPU-Z - http://www.cpuid.com/

GPU-Z - http://www.techpowerup.com/gpuz/

[arizonachris]

I have an I7-2600K @ 3.40 GHz (8CPUs).

One thing I noticed and didn't mention before is when checking Multi-Core CPU1 usage is average 90%, CPU5 40% (the others are few used). When Multi-Core is unchecked CP1 60%, CPU5 50% and CPU6 30%, ie usage is more distributed, but fps is quite the same. What is better, after all?

Hyperthreading is enabled. That is only a four core CPU, but with Hyperthreading, Windows "sees" 8 cores. RSC has said that the multi core check box is more for troubleshooting. I don't find any difference in performance, turned off or on. IMHO, if it's working OK, leave it alone. If you are having any issues, go ahead and mess with settings. Can't hurt anything too much, you do have several installs to fall back on.

Oh, did you ever install those two Windows Hot Fixes that I posted about for the i7 and i5?

[Machinist]

Oh, did you ever install those two Windows Hot Fixes that I posted about for the i7 and i5?

Not yet, when you posted I didn't have I7 yet, so no interest at the time. So far I'm still busy re-installing all programs (utilities) in the Win 7. I'll look into your post further.

[Chacal]

This is off topic but when I got my computer I noticed it gets abnormaly hot, what I did was take the large cooling fan and flip it arounnd to blow air out rather than take air in. Is this alright? The computer seems cooler but I don't know about the CPU its self. I can post my PC specs if needed. Also I run everything 100% maxed...not shure if this is important to the situation

Assuming you have a tower case - this is correct. The large fan (especially if on top of the case) is for extracting hot air from the PC. Looks like yours was assembled backwards.
Cooler air normally flows in from the front, goes over the disks, the CPU and the cards, and flows out from the top and back.
Some cases allow the addition of intake fans at front and exhaust fans at rear.
Good cases have air filters at the front intakes, these should be cleaned regularly otherwise they will restrict air flow.

[snowstorm]

Anyone have any TS2012 vs TS2013 frametime logs to contribute? I've done a bunch of new comparative benchmarks with the intention of kicking off another thread with TS2012 vs TS2013 charts, but with the mixed results I'm getting, I'd really like to see what's happening on other systems first.

On the one hand, the overall frame rates, as others have reported, have spectactularly improved across the board, suggesting TS2013 should appear much more smooth. On the other, that wasn't my subjective experience - if anything it actually felt more stuttery than TS2012. This wasn't reflected in the results I compiled - the charts and the >50-400ms frametime counts told me I was deluding myself - TS2013 is a huge improvement.

This got me thinking about one of the major shortcomings in my previous test results - they're largely blind to frame rate consistency as frame rates increase. Frame times can vary by more than 50% from one frame to the next, but as long as it stays above my >50-400ms thresholds for frametime counts, no black mark will be recorded.

So I redid my spreadsheets to mark each instance a frametime varied by more than 25% or 50% from the last, then calculated those instances as a percentage of total frames. The results were jaw dropping and suddenly made sense of my subjective experience - TS2013, despite its higher frame rate, was way more jumpy in its delivery of frames. For example, in my NEC Early Clocker test, >25% frametime jumpiness was 24.05% for TS2012 vs 91.78% for TS2013, despite the average framerate almost doubling in TS2013.

Anyway, I'm planning to do additional tests to measure the extent to which FPSLimit mitigates the problem, but in the meantime I'd really appreciate some results from others, as there's little point posting a bunch of definitive-looking benchmarks if all they're reflecting is some quirk of my particular system.

[PapaXpress]

I will see if I can get you my benchmark this weekend.

[up_8677]

I'm downloading SH now, but I tried some of my own scenarios on various routes. From what I see--on my system--the performance improvements are pretty darn impressive. Also, the SD75 cablight no longer brings my computer to it's knees.