Live Performance Benchmarking Comparison for rFactor 2

I should mention that I didn't test different AA modes with otherwise identical in-sim settings. I'll add that to my 'to do' list.

 

Ah yes! That was poor of me to use performance comparison between PCI-e standards with a completely different graphics engine ultimately.

It would seem that the demand on PCI-e is not controlled by how high your fps goes in a certain graphics engine but by what graphics settings you are using in that particular graphics engine. This determines the PCI-e bandwidth demand and why with the live benchmarks max gfx settings we see PCI-e 3.0 vs 2.0 performance differences. That video i posted was comparing the different PCI-e speeds with the default settings in 3d mark. If you turn them up even more, at some point, the PCI-e speed your running at should become a factor (though it looks well optimised since even PCI-e 1.1 seems to be fine....which makes sense when you think about it because how else can you fairly compare all systems in the world to get a true comparison of all the graphics card performances....e.g. someones GTX 980 on PCI-e vs someone with a 5+ year old system still running PCI-e 1.1 with their graphics card).

 

Thank you! That makes perfect sense then. Nice overclock on the memory btw, the MSI gtx 970 i received it a bad memory overclocker. Anything past +200mhz (so 7400mhz mem effective) in afterburner starts to show artifacting whereas all the reviews i've seen say easy +500mhz (so 8000mhz mem effective) stable.

If you can add your clock speed details to your results post please that would be greatly appreciated. It helps for when i'll (or someone) collects all the results to create a graph of relative performance.

 

Yeah, i can't, lol. i5-2500k here. :S

 

Try seeing what happens when more demand is placed on the GPU by using 8xMSAA instead of 4, 12AM instead of 9AM, and rain.

 

IF 3.0 makes a difference going to 8x shouldn't be a problem with a 780 TI.

However I saw no performance drop when I did either so that doesn't make much sense either.

 

Hey rogue, TechAde pointed me to your post here: http://isiforums.net/f/showthread.p...-for-rFactor-2?p=310018&viewfull=1#post310018

And i have to apologies, you had it right but i assumed my pci-e was in 3.0 instead of actually checking to confirm it was indeed true. You had the answer! lol. My bad.

Tech also says he has a 7990 too. Maybe you ask him really nicely and he give you some scores at pci-e 3.0 x16 mode.

 

Apologies from me too, rogue22, I skipped right over your 'Could be PCIe 2.0 v 3.0' comment and thought I'd worked it out for myself. Turns out the answer was in post #7 all along!

 

However that may not be the case as others saw no difference going between 2.0 and 3.0. So I would call this 100% effective just yet.

 

Has anybody other than you reported no difference between 2.0 and 3.0 in the live benchmark?

Given that your motherboard & CPU are not capable of running at 3.0 (no matter what GPUz & AIDA say) I'd say the chances are you're actually running at 2.0 all the time.

 

Didn't we just get like 3 or 4 benchmarks that gained the missing fps after the user set the system to PCI-E 3.0 (@ 16x)? We got enough proof yesterday....TecheAde proved it, someone else proved it when they went from a 3930K (without registry fix) to a 4770K, and i believe there were another a couple other people that tested it out as well.

 

I decided to give this test a try as well, just for a laugh. Did it three times but the results aren't very consistant (probably due to the low fps).

I can't run the right resolution unfortunatly, so this is done at 1680x1050. Other than that I applied all settings.

GTX560 no overclocking, 810mhz core and 4008 mhz memory
AMD Athlon II X3 445 @ 3.1 ghz
RAM 8 GB DDR3

Take 1:
Min, Max, Average
8, 25, 17.215

Take 2:
Min, Max, Average
7, 23, 14.802

Take 3:
Min, Max, Average
6, 17, 11.873

 

I read some more of this thread. I missed those post from the other day. Which is why I didn't understand why all of a sudden PCI 2.0 vs 3.0 info came about.

So I called it, lol.

Still can't figure out why I'm getting inaccurate scores.

 

Just saw you and Ripper posted this. Sorry my confusion.

 

I wonder if this has to do with DX9 or just real bad coding on ISI's part or something because I've read review/benchmark after review/benchmark after review/benchmark and not one of them, NOT ONE, has PCI-E comparison results even remotely close to those of rF2. It just seems really odd to me.

 

That is something you'd have to ask one of the developers at ISI. I wish they would have chimed in and gave us some of there findings as it would have saved about 3 threads of all of this.

 

Yes you did lol.

I think it has to be a faulty reading and there are 3 reasons for this belief:

1) Your Bus Interface reading in gpu-z and the other program did not down regulate the bus speed when in the gpu was idle. I may be mistaken and this only happens on mine or maybe this is normal and is happening to everyones in which case according to gpu-z though, yours was not.
2) As a possible additional indicator that something is wrong with how gpu-z is reading some of your graphics cards live state values, it was not showing a value for your graphics cards memory size.
3) And the biggest one of them all, you have a 1st gen inte CORE cpu which only has PCI-e 2.0 integrated support, meaning even if you stick it in a PCI-e 3.0 capable motherboard, you wouldn't be able to get it working in 3.0 (just like with my system and i5-2500k cpu). You would need an i5/i7-3000 series or higher.

 

Massive differences holy crap!!!

Something is up with rF2 because every other review and test i've seen comparing PCI-E framerate differences - and I've seen tons from all sorts of reviews and forums - has only shown 1-3 percent differences, not 5 - 20 percent! Wow!!