Question GPU artifacts at lower clocks ?

[_D_D]

With memory clock, artifacts usually happen when going too high, but this happens on a small range in a low region.

GPU: Palit GamingPro RTX 5070 12GB
Dual VBIOS (250W TGP)
VBIOS 98.05.36.00.4C
VBIOS 98.05.36.00.4D
Memory clock adjustment range is limited to -2000MT/s to 6000MT/s or -125MHz to +375MHz
Asrock Taichi X99 PCIe 3.0
W10 Drivers 572.83, 576.52 and 576.80
Linux Driver 575.75

Error happens in between memory clock 1649.125MHz and 1657.125MHz (DDR 13193MHz and 13257MHz, MT/s 26514MT/s and 26386MT/s). IOW applying a negative memory offset of say -97MHz for PState 0.

Usually causes Event ID 14 followed by TDR attempt which seldom works and many ID 153 Errors per second until freezes or power cycled.

So curious if this is just my card or happens on other 50 series but be careful if you try as hard reset by power cycle can cause file corruption.

 

[helpstar]

are you using more than one monitor?
which resolution and refresh rate?

should happen on all vga cards

 

[_D_D]

Single monitor, typically 120Hz @ 2k using DP however as part of the testing I had added a second graphics card and ran the RTX 5070 headless (no monitor attached). While obviously I couldn't see artifacts the system still locked up when the RTX 5070 was set in between that small band of memory clocks. The fault is 100% producible. Not sure if it's driver or HW cause but being HW would seem odd. Then again I was hoping to have a discussion to include some other things related to this card (RTX 5070) but seems an Administrator has edited the thread into a question, oh well.

 

[Teknoman2]

This seems like a classic case of faulty memory that slipped by QA so RMA is in order.

 

[_D_D]

Could be just that but I would like to have some confidence it's not the driver or VBIOS before taking that path.

 

[helpstar]

set to 60hz and try again,

with which tool are you underclocking it?
did you accidentally set a power limit or rduced voltage as well? tra withsi afterburner

are the latest nvidia.com drivers installed?

which other hardware is involved?
motherboard plus BIOS version
cpu
psu
ram

which vbios are you using? is it happening on both vBIOS versions?

 

[_D_D]

set to 60hz and try again,

Same

with which tool are you underclocking it?

Usually my own SW but have used nvidia-settings too

did you accidentally set a power limit or rduced voltage as well?

No

tra withsi afterburner

Sorry, prefer to give that a miss thanks

are the latest nvidia.com drivers installed?

Yes, 576.88 for W10, not sure about Fedora off hand. Other versions tried are listed in first post.

plus BIOS version

P1.80 but a fair bit of modification by me from some years ago while playing with Xeon turbos

cpu E5-2696-v3
psu RM850x shift
ram 32GB GSKILL DDR4 2133MT/s quad channel

is it happening on both vBIOS versions?

Yes.

Some questions answered in first post.

 

[helpstar]

reset the BIOS to default

try BIOS 1.82 [Beta]

 

[_D_D]

Just a small update. Contacted Palit support and because the fault doesn't occur at default settings it isn't covered by warranty so no RMA.