GPUZ reports wrong PCI-E version on X58 Mainboard

[Aerpoweron]

You should be able to check that with HWinfo. Just uncheck the "sensors only" box.
But there is a lot of info, have to find the right one.

I think i found it.

 

[W1zzard]

Naki, what do you think the number in front of @ means? What about the number after @?

 

[Naki]

You are the low-level programmer and PCI expert, you tell me. If I knew this, I would not ask you.
I have done lots of Windows GUI programming, but not low-level stuff, nor any drivers programming. The lowest level I have used is some serial port read/write, nothing more.

I am pretty sure a 2.0 video card cannot run at 2.0 speeds on a 1.1 mobo, and likewise a 3.0 card cannot run at 3.0 speeds on a mobo that only has 2.0.
Why would ASRock and other mobo manufacturers need to make any Gen3 mobos, if Gen1 (with 2.0 support on them) can run 3.0 cards at 3.0 speed?

Such backward compatibility may work on those new AM4 socket boards, but that is besides the point.

 

[Final_Fighter]

You are the low-level programmer and PCI expert, you tell me. If I knew this, I would not ask you.
I have done lots of Windows GUI programming, but not low-level stuff, nor any drivers programming. The lowest level I have used is some serial port read/write, nothing more.

I am pretty sure a 2.0 video card cannot run at 2.0 speeds on a 1.1 mobo, and likewise a 3.0 card cannot run at 3.0 speeds on a mobo that only has 2.0.
Why would ASRock and other mobo manufacturers need to make any Gen3 mobos, if Gen1 (with 2.0 support on them) can run 3.0 cards at 3.0 speed?

Such backward compatibility may work on those new AM4 socket boards, but that is besides the point.

he actually answered your first question already, he is asking what YOU think.

the first field is the reported maximum link speed the card can run at as reported from the card itself, irrespective of the motherboards capable max speed. the second reports what the current link speed is.

Naki your Screenshots show 1.1, which is expected for idle. Any chance you're misinterpreting the field? It's "max supported @(at) currently active"

The issue OP is having is specific to Vega

For the second part, this applies to intel cpus only:

boards that support pcie 2.0 should be able to run 3.0 and asrock may have updated the board for other reasons. many manufactures dont update the user manuals and sense ivy bridge was not out yet it may not have gotten updated. btw, Every h61-z77 that i dropped an ivy bridge cpu in supported 3.0 and reported running at said speed. the reason is that the first pcie x16 port is supposed to be wired directly to the cpu and unless the bios tells it to specifically run at less than what the cpu is designed to run at it will always default to the max capabilities of the cpu. some manufactures may have used less than adequate components and wiring for the connection between the cpu and pcie slot but that would not stop the cpu from wanting to communicate at its maximum capable speed. it would result in other problems. so, its also possible that the board manufacturer limited the speed using software because they knew the board might experience problems. the only way for them to limit the max speed is threw the bios and sense that is software, its possible windows and the drivers are ignoring that and running at max possible speed. (cheep programing on the bios).


why would asrock make a gen3 mobos?

the change may have been for marketing and to clear up confusion. just come out with a revision and dont update the old models Manuel so as to make people buy the updated model with a higher price tag. people want to make sure they are getting the best and selling a board that says pcie 2.0 when one of the marketing points of ivy was pcie 3.0 is not good marketing and will probably cause customers to look at other options.

hope this helps clear things up for you.

 

[revin]

Yep ^^ Intel did the same with the DZ68, put out a <failed imho> BIOS patch to run IvyBridge on them, thus they run 3.0 but "Tech Spec's" only showed 2.0

 

[W1zzard]

the first field is the reported maximum link speed the card can run at as reported from the card itself, irrespective of the motherboards capable max speed. the second reports what the current link speed is.

Exactly that, and I suspect Naki is misinterpreting it

 

[Naki]

Okay, I will rephrase - "I am pretty sure & think that..." -- rest of it what I said.

Anyway, no, this did not help any. Thanks for trying. But let's leave it this way. Yes, I am no engineer - my university degree was in Chinese language, so math & technology was sadly absent during my years at uni. (Actually, we had a few hours of basic computer use - Windows, MS Word, MS Excel and a few other common programs but I already knew all of that and much more anyway. )

Unless of course somebody from the PCI spec folks or one working in mobo design comes along to explain this, which I doubt can happen here in these forums.

 

[W1zzard]

That's what's displayed in front of the @, you are simply expecting it to show something it's not showing.

http://www.lttconn.com/res/lttconn/pdres/201402/20140218105502619.pdf

 

[Aerpoweron]

Naki, you can try to run 3dmark. Especially the PCI-E test. and you can see the real speed you get

You should get around the speed as shown in the graph.

Source: https://en.wikipedia.org/wiki/PCI_Express#History_and_revisions

 

[Aerpoweron]

I have run the tests myself. Here you have PCI-E 3.0 16X vs PCI-E 3.0 8X.

I get 12.6 Gigabyte a second compared to 6.7 Gigabyte a second.

You should get around 6.7GB/s when you run at PCI-E 2.0 16X

https://www.3dmark.com/compare/pcie/50014/pcie/50011

 

[Naki]

Sorry, what? I know in theory PCI Express 2.0 16x is the same as 3.0 x8, but this is not a valid test.
I do not have the time to test this now, but surely will when I get a chance.

 

[Aerpoweron]

I would say if you run the 3d mark pcie-test, and you get above 8GB/s then your board is truely running at PCI-E 3.0 16x

And how is this not a valid test? It is made to be comparable when you run it on different hardware

 

[Naki]

No, 3DMark test is fine! What I meant your method of testing in 8x mode on a PCI Express 3.0 mobo. (which I do not have here - only 2.0)
In theory, using a 3.0 slot in x8 mode should be the same as 2.0 in x16 mode, but I doubt this is true in practice.

Actually, I think my parents may have 1-2 PCs here with PCI Express 3.0 on them but all of them have no discrete cards, just Intel CPU built-in GPUs.
(Those PCs are used for work + 1 is used as HTPC, so no gaming involved and thus no need for discrete cards)