bug bus interface gts450

[Boombastik]

asrock z97 aniverssary and gts450 nvidia gpu 391.35.
The bus interface reports idle pcie x16 1.1@x16 1.1
When i load stress test it reports: pcie x16 2.0@x16 2.0

This is false.
The good one report that need to be fixed on idle is this: Pcie x16 2.0@x16 1.1

 

[LabRat 891]

Actually... (AFAIK) That's how it's supposed to work.
Your GPU and Mobo are negotiating datarate based on powersaving/load.

If it bugs ya, or is causing an issue, you can usually force the slot in UEFI/BIOS to Gen2 (instead of AUTO).

Note: GTS450 is a PCIe Gen2 GPU

NVIDIA GeForce GTS 450 Specs

NVIDIA GF106, 783 MHz, 192 Cores, 32 TMUs, 16 ROPs, 1024 MB GDDR5, 902 MHz, 128 bit

www.techpowerup.com

 

[Assimilator]

This is false.

No, it's not. Do some research before you post nonsense.

 

[Boombastik]

what research to do? In idle the program must write this: Pcie x16 2.0@x16 1.1 (max motherboard speed@ gpu speed)
The first is the motherboard speed and the second the gpu.

In all other systems that i have when i stress the gpu only the second part change. The second part is this with bold: Pcie x16 2.0@x16 1.1
Open your gpuz and see if the first part in idle goes 1.1 speed. Only the second part lower the speed.

Anyway your answer is not polite.

 

[Toothless]

what research to do? In idle the program must write this: Pcie x16 2.0@x16 1.1 (max motherboard speed@ gpu speed)
The first is the motherboard speed and the second the gpu.

In all other systems that i have when i stress the gpu only the second part change. The second part is this with bold: Pcie x16 2.0@x16 1.1
Open your gpuz and see if the first part in idle goes 1.1 speed. Only the second part lower the speed.

Anyway your answer is not polite.

I'd check the PCI settings in the bios and make sure the card is fully seated in the slot.

 

[tabascosauz]

asrock z97 aniverssary and gts450 nvidia gpu 391.35.
The bus interface reports idle cie x16 1.1@x16 1.1
When i load stress test it reports: pcie x16 2.0@x16 2.0

This is false.
The good one report that need to be fixed on idle is this: Pcie x16 2.0@x16 1.1

All GPUs, even modern GPUs, should clock down the PCIe bus at idle when not under 3D load. As long as the width is correct (x16, because that does not change, and anything less than x16 by a x16 GPU in x16 electrical slot is a physical problem) and you are getting 2.0 x16 under 3D load (which it sounds like you are), then everything is normal. That's why the Render Test button exists in GPU-Z, to make sure applying 3D load bumps it back up to correct PCIe speed.

afaik forcing a certain PCIe bus speed in BIOS should not affect PCIe bus power saving behaviour, it just locks you out of any faster speeds if they exist. Windows power plans and NVCP Prefer Maximum Performance might, however.

The first is the motherboard speed and the second the gpu.

No, it isn't. The first is what the GPU is specced for (what it should be running at, when under 3D loads), and the second is what the GPU reports it is currently running at.

 

[Ruru]

It works just like it should.

For me it's 3.0 x8 as it shares bandwith with a M.2 slot and I'm running with a X470 mobo but still reports correctly.

 

[Toothless]

It works just like it should.

View attachment 320017


For me it's 3.0 x8 as it shares bandwith with a M.2 slot and I'm running with a X470 mobo but still reports correctly.

Yeah I don't know why your board is like that because my old x470 was 16 on GPU and 4 on NVME. Also revision won't change for the slot unless your board is actually that junk. 5800x has 20 lanes for GPU/SSD.

Also can't compare your stuff to OP with the massive year gap.

 

[Ruru]

Yeah I don't know why your board is like that because my old x470 was 16 on GPU and 4 on NVME. Also revision won't change for the slot unless your board is actually that junk. 5800x has 20 lanes for GPU/SSD.

Also can't compare your stuff to OP with the massive year gap.

Crosshair VII Hero so this should be one of the best X470 boards

 

[Toothless]

AMD Ryzen 7 5800X Specs

Vermeer, 8 Cores, 16 Threads, 3.8 GHz, 105 W

www.techpowerup.com

Crosshair VII Hero so this should be one of the best X470 boards

Yeah, probs a board issue or I've seen defective Ryzen chips have PCI issues. Either way my AsRock Taichi Ultimate x470 did full 16 lanes to GPU and was fine with the NVME boot. Your stuff is not running right.

 

[Ruru]

AMD Ryzen 7 5800X Specs

Vermeer, 8 Cores, 16 Threads, 3.8 GHz, 105 W

www.techpowerup.com

Yeah, probs a board issue or I've seen defective Ryzen chips have PCI issues. Either way my AsRock Taichi Ultimate x470 did full 16 lanes to GPU and was fine with the NVME boot. Your stuff is not running right.

Not a CPU problem. My old 3600 worked with this card at 4.0 x16 with my B550 board but 3.0 x8 with this Crosshair. It's because of the M.2 slot(s) being used and Asus has screwed up how those should be aligned.

Crosshair VII: M.2 drives steal PCIE bandwidth

If the user installs an M.2 drive in the upper slot closest to the CPU, the graphics card runs at 8x instead of 16x. This seems like poor design, given that the heatsink is installed by default on that M.2 slot. It would lead the average person to assume it is the primary M.2 slot, but it seems...

rog-forum.asus.com

 

[Toothless]

Not a CPU problem. My old 3600 worked with this card at 4.0 x16 with my B550 board but 3.0 x8 with this Crosshair. It's because of the M.2 slot(s) being used and Asus has screwed up how those should be aligned.

Crosshair VII: M.2 drives steal PCIE bandwidth

If the user installs an M.2 drive in the upper slot closest to the CPU, the graphics card runs at 8x instead of 16x. This seems like poor design, given that the heatsink is installed by default on that M.2 slot. It would lead the average person to assume it is the primary M.2 slot, but it seems...

rog-forum.asus.com

Thread shows bottom M.2 and forces x16 mode will fix the issue.

 

[Boombastik]

It works just like it should.

View attachment 320017


For me it's 3.0 x8 as it shares bandwith with a M.2 slot and I'm running with a X470 mobo but still reports correctly.

Kissamies as you see in idle the @x8 3.0 goes in x8 1.1
The pcie x16 4.0 stays always like that.

In my motherboard the card is pcie 2. When i am in idle it goes to 1.1 but it goes 1.1 also the first part , it should says always pcie x16 2.0. I am wrong here?

see the photo i attached your card is always x16 4.0.

....
No, it isn't. The first is what the GPU is specced for (what it should be running at, when under 3D loads), and the second is what the GPU reports it is currently running at.

We said the same thing my card is specked for pcie 2.0, so it should said Pcie x16 2.0@x16 1.1 when i was idle but instead it says Pcie x16 1.1@x16 1.1

In load all are ok : Pcie x16 2.0@x16 2.0

 

[LabRat 891]

Kissamies as you see in idle the @x8 3.0 goes in x8 1.1
The pcie x16 4.0 stays always like that.

In my motherboard the card is pcie 2. When i am in idle it goes to 1.1 but it goes 1.1 also the first part , it should says always pcie x16 2.0. I am wrong here?

see the photo i attached your card is always x16 4.0.


We said the same thing my card is specked for pcie 2.0, so it should said Pcie x16 2.0@x16 1.1 when i was idle but instead it says Pcie x16 1.1@x16 1.1

In load all are ok : Pcie x16 2.0@x16 2.0

Either an error/bug in GPU-z, or the GTS450 itself is 'gearing down' to PCIe Gen 1.1 when in idle.
(At idle, the card requests/reports Gen1.1; at work, requests/reports Gen2)
Since it is switching 'up' to Gen2 under load, it is a little strange but, expected behavior.

 

[OneMoar]

its a GTS450 it litterally doesn't matter it could be x4 and not matter
but there is nothing wrong here its just pcie power saving doing power saving things

 

[:D:D]

@Boombastik it's an old GPU and may not have support to show maximum supported link speed and so maybe shows current value instead? @W1zzard would now for sure and may have already explained it many times before. A quick google shows other older cards doing the same with ver 2.29 of GPU-Z.

The value on my nvidia card is one that's read on driver start up and depends on what the mainboard root port is set to as well as what the card can handle so isn't necessarily the maximum spec of the card as stated by someone earlier.

 

[Boombastik]

Thanks for the answer! At least i have explained it better and you understand what I wanted to say.

 

[W1zzard]

This is very interesting and shouldn't happen. Per PCI specification the first field displayed in GPU-Z is the maximum capability reported by the card, and it should be static. However, given how old this device is, it's not unreasonable that NVIDIA implemented a hack to allow speed switching to work on older OS's/BIOSes

 

[Boombastik]

The os is windows 11 and the bios is the last for motherboard. The driver is the last for the Fermi card.

 

[natr0n]

I thought it said bang bus in her face gts450

I also should be sleeping

 

[Naki]

I thought it said bang bus in her face gts450

I also should be sleeping

You are clearly watching too many screens at once, and/or need some sleep.