Interesting observation regarding PCIe lanes

[Abyssic]

guys... i've got a strange anomaly to talk about...

i wanted to test the performance impact when switching from PCIe3.0x16 to x8 on my R9 290X.
to do so, i placed the card in my lower slot, wich is only x8. i only changed that, clocks, settings even psu connection stayed the same.
well, it turned out that it really had quite the impact... but here it comes: POSITIVE impact!

i used valley to benchmark the results and got around 54 fps with my settings at x16. when using x8 i got 68 fps!

i also tested 3D Mark 11 wich gave me 500 gpu-points more on x8

my mind is blown and i have no idea what is reality anymore. i tested both multiple times, with switching the slot between each run, watching temps and volts. nothing changed, except more fps...

any opinions on that? similar experiences?

 

my specs can be found below or in my profile

[theKumba]

Are you... like..... sure about that?

[Interfectorem]

I'm sorry, but I highly doubt it.

[Padunkadunk]

Are you sure you have your lanes identified correctly?

[A Bubble]

Pics or it didn't happen, because I sense something wrong here.

[Monarch]

I have an odd issue. My 290x is running at x8 3.0 (fully stressed) in a 16x 3.0 slot according to GPU-Z. This is on Z87-HD3. Any ideas why?

Anyways in Valley I get 61 fps at stock 1040/1300

[Abyssic]

so, what would you like me to provide to prove this? pictures of what?

btw: i'm very disappointed about you guys deeming me a liar and i'm not highly keen on doing those swaps all over again just to silence you.

what about "if this is true..." responses?

[holy-sock]

I dont believe you, it must be magic.

[WoodenMarker]

The lower slot might be better for airflow which could lead to less throttling and more stable performance. The difference in bandwidth shouldn't affect performance much at all. 

[Queek]

How many times did you run the test?  

You should run it at least three times to get a measure of the variance.  Interesting result, I wonder what is causing it.  

[-rascal-]

what about "if this is true..." responses?

How are we suppose to believe something that directly opposes our very own logic reasoning as humans.

Less is somehow more??

[-MOBIUS-]

Less is somehow more??

 

Thermal paste

[-rascal-]

Thermal paste

That is one exception but....

If you have no money, then you are rich?

If you have less bandwidth, then you have higher data transfer?

If your CPU is slow, then you have more FPS?

[Abyssic]

okay, i ask again, how can i prove it for you guys? i mean, i could take screenshots and photos of the card in the slot but i guess that still won't satisfy some of you.

it has nothing to do with the cooling. the card won't throttle in the upper slot and also not in the lower.

[zuboici]

You need a screenie of details on screen like this:

 

I just benched my SSD in VBox and it out performs the same test in the host machine.

 

[zuboici]

Pics or it didn't happen, because I sense something wrong here.

 

See above.

[A Bubble]

See above.

What does that have to do with GPU speeds in a different person's PC?

[Queek]

okay, i ask again, how can i prove it for you guys? i mean, i could take screenshots and photos of the card in the slot but i guess that still won't satisfy some of you.

it has nothing to do with the cooling. the card won't throttle in the upper slot and also not in the lower.

 

How many times did you run the test?  

You should run it at least three times to get a measure of the variance.  Interesting result, I wonder what is causing it.  

^ This

Tell us the number of samples you took, and the details of those samples, with screen shots.  

You really can't expect a forum of tech enthusiasts to believe something that directly opposes their understanding of how something works.  You need to provide proof, which on the internet, is usually in the form of "pics or it didn't happen".  

Show us some screen shots, we want to know more about what happened.  

[Faa]

Your i5 3450 thing is OC'ed with the BCLK, go get it stable first.

[Abyssic]

i'll deliver in the next few days. i'm gonna do all the swapping again for you guys.

but please, when i go through the hassle to do it all over and document it, don't go "fake!" on me. as i said before, i don't really know how i could prove it fake-proof. i guess screenshots and maybe a few photos have to do the job.

btw: my cpu isn't oc'd by BCLK anymore, information in my profile was old. i updated it. i have it oc'd just by mult to 3.7GHz since a few months.

and yeah, i understand why you can't believe me, i couldn't believe it myself at first.

[zuboici]

What does that have to do with GPU speeds in a different person's PC?

 

Yeah idk why i quoted you. ignore it. I was giving an example for the OP.

 

 

i'll deliver in the next few days. i'm gonna do all the swapping again for you guys.

but please, when i go through the hassle to do it all over and document it, don't go "fake!" on me. as i said before, i don't really know how i could prove it fake-proof. i guess screenshots and maybe a few photos have to do the job.

btw: my cpu isn't oc'd by BCLK anymore, information in my profile was old. i updated it. i have it oc'd just by mult to 3.7GHz since a few months.

and yeah, i understand why you can't believe me, i couldn't believe it myself at first.

 

It's not about proving anything not-fake, just do the screenies of the results so we can see the same thing you're seeing - maybe you missed something obvious, we all do it.

 

"Pics or it didn't happen" is just a turn of phrase, nothing offensive expressed or implied. Surprised you haven't seen it before.