Massive FPS gains in rf2 using PCI-e 3.0 x16 with higher end cards!

I just added more to my post:

So far i see 2 faulty readings in gpu-z, you bus interface (because it simply can't be actually running in 3.0 mode) and unable to read your memory size.

Wondering if the programs are unable to read your cards live state values properly is the problem here. Would seem much more plausible than your 1st gen intel CORE cpu allowing PCI-e 3.0.

 

Saw you added post.

I noticed that too. I'm back to stock speeds for the entire system. Still does the same thing.

AIDA64 reports it as well so I don't think its GPUz. Do they get there readings from the video card driver? Could be that causing faulty readings.

I'm downloading Heaven benchmark to test the system fully as well.

 

I honestly don't know, sorry.

I don't know of any benchmark you can use to stress the PCI-e bus and then compare your perforamnce to others who have the same card that you know are running in PCI-e 3.0 x16 mode to see if your readings are indeed 100% false.

Except for maybe the live benchmark in rf2 and finding someone who has the exact same card as you (and set same clocks) in 3.0 x16 and compare performance for confirmation.

 

Rogue22, what CPU and motherboard do you have?

 

Intel Core I7 920 and a Gigabyte X58 UD5

 

Just for you, because I'm nice ( and you are a good sort) I'll run your live bench... but my bench is run at MAX settings and at my custom res of 3780x1620..... IIRC yours is 1080p... which I don't have.... I have 4K monitor or 3K (approx.) projector not 1k (1080p) but I'll sort it and drop to the low res option of 1080p if it proves a point (that if I had a lower res I would benefit, which I don't so the point is moot)

I'll get back to ya with the results on the other thread... but how do I disable PCI 3.0 now I have run the reg fix?

 

Mr Pix, we've already had enough people in the other thread do different PCI-E tests. You don't need to bother. There are multiple tests that clearly prove the mysteriously large performance benefits that PCI-E 3.0 @ 16x brings to rF2 relative to PCI-E 3.0 @ 8x / 2.0 @ 16x.

 

@rogue22

Could be worth trying this test build version of GPU-Z as you seem to have a number of things being incorrectly read.

 

You have PCI-E 2.0, no matter what your PC tells you. GPU-Z may be reading what your card is capable of rather than what it's actually running at.

Try downloading the newest version of GPU-Z, click the little 3D/benchmark program in it and let it run for a good 10 seconds and then write down exactly what it says after 10 seconds but while it's still running.

 

No need to be like that... it doesn't benefit all of us...simple as. Nobody is wrong here, including me.

 

Mr. Pix I meant to save you time, that's all, because i misread your post.

 

He's testing it for me (i guess) as a favor and out of interest. It's interesting that if his observation is true however, it would be interesting to know why.

Looking at your post, your normal resolution might be a hint to why you didn't see a difference (but if so, it would seem counter intuitive, one would think increasing the res would increase bus bandwidth demand if any at all but i don't know, just a guess).

 

Sorry bad day at the office.. I read it again... please forgive me....

I will however perform the test... I think then the thread may have to be retitled... to Massive FPS gains in rf2 using PCI-e 3.0 x16 with higher end cards @ 1080p!

I'm convinced I am already at max at my resolutions... it sounds backwards, but maybe the extra bus width/speed is not the bottleneck past 1080p?

 

Ok I was partway through editing my original post because i realized I was in the wrong but then I saw both your replies and now that other post is partway edited, ahhh what a mess.

Basically, I'm the one who owes you an apology. I misread your first post. I thought that you believed that there was a large difference with the different PCI-E modes, but you didn't really feel like proving it with tests and such, but then decided to do so anyways (hence the "Just for you, because I'm nice ( and you are a good sort) " thing). So I was trying to save you time and effort and letting you know that you didn't really have to do all the tests if you didn't feel like it because we have quite a few test-results already. But, I see now that I completely misread your post; you were NOT convinced of the large differences between the PCI-E modes but still decided to test it as a favor for DrR1pper. Ahhh, got it.

So I am the one who completely mis-read your post, and therefore I am the one who owes you an apology.

Sorry Mr.Pix.

 

Already done this Spinelli, look back a few post you can see my screen shot here towards the bottom. Its got me stumped. I already know its 2.0 though.

http://isiforums.net/f/showthread.p...ng-PCI-e-3-0-x16-with-higher-end-cards!/page2

I'll have to check to see if its a Bios thing.

 

Ya, I'm saying to read what GPU-Z says after 10 seconds of it's 3D/benchmark running while it's running (run the GPU-Z bench program in full screen, wait for it to run for 10 seconds, do not close bench program, see what it says on the bottom of the screen, then close bench program. Try the same thing but in non-fullscreen mode as well.

 

Yeah, i'm already at character length max with the title, lol.

It's possible but without a bus interface expert here to confirm...testing is best thing we've got.

Don't feel pressured to do it though, only if you want to as well ofc.

 

Have you tried the test version I linked in post #49 as it seems hd7990 detection has been broken in the last couple of GPU-Z builds as evidenced by your screenshot which has a number of values incorrectly showing?

 

Ya, don't feel pressured. Having said that, seeing how PCI-E affects 2160p relative to how PCI-E effects rF2 in 1080p sure would be nice

 

I did exactly this, I actually discovered it on my own the reason why I asked for it to be explained. I full screened the benchmark for 20 minutes. At the bottom left it says both GPUs are running at PCI-E 3.0 @ 16x.

I'm using the latest GPUz but after reading MJPs post I've convinced that's what the problem is. However I don't think its GPUz, Because in AIDA64 it reports the correct ram, bit rate, etc........except AIDA64 claims its PCI-E 3.0.

I think it lies in the way the 7990 reports its information. You can window benchmark the car either, has to be in full screen in order for both cores to work properly. It is a rare odd acting card if you don't know how to tune it properly. I'm using the default bios on it too, the one that throttles it back when heated.

I'm not to worried, I get the performance reported at Anandtech bech in those games I play. Rfactor 2 is the only sim I can't get crossfire to work in. I wonder if this PCI thing has something to do with it.