NVIDIA GeForce RTX 3070 Ti crashing on Diablo IV (rewrite/update bios?)

[Enrai]

Hello All,

I need some help with my NVIDIA GeForce RTX 3070 Ti, it was working fine for a few months ago, but now when I am playing (Diablo IV) my computer just restart from nothing.
Checking the Event Viewer I have this error:

Below is my GPU, I got it as:

As we can see in the pictures it is a fan blower model

and the lookup goes to these specifications: NVIDIA GeForce RTX 3070 Ti Specs | TechPowerUp GPU Database

I am using FurMark to stress the gpu, and I got no error (I tested it for 15 min)

and this is my computer specs:

Processors Information
-------------------------------------------------------------------------

Socket 1 ID = 0
Number of cores 6 (max 6)
Number of threads 12 (max 12)
Manufacturer GenuineIntel
Name Intel Core i5 11400
Codename Rocket Lake
Specification 11th Gen Intel(R) Core(TM) i5-11400 @ 2.60GHz
Package (platform ID) Socket 1200 LGA (0x1)
CPUID 6.7.1
Extended CPUID 6.A7
Core Stepping B0
Technology 14 nm
TDP Limit 65.0 Watts
Tjmax 100.0 °C
Core Speed 4289.5 MHz
Multiplier x Bus Speed 43.0 x 99.8 MHz
Base frequency (cores) 99.8 MHz
Base frequency (mem.) 99.8 MHz
Stock frequency 2600 MHz
Max frequency 4400 MHz
Instructions sets MMX, SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4.2, EM64T, VT-x, AES, AVX, AVX2, AVX512 (DQ, BW, VL, CD, IFMA, VBMI, VBMI2, VNNI, BITALG, VPOPCNTDQ), FMA3, SHA
Microcode Revision 0x58
L1 Data cache 6 x 48 KB (12-way, 64-byte line)
L1 Instruction cache 6 x 32 KB (8-way, 64-byte line)
L2 cache 6 x 512 KB (8-way, 64-byte line)
L3 cache 12 MB (16-way, 64-byte line)
Max CPUID level 0000001Bh
Max CPUID ext. level 80000008h
FID/VID Control yes

Chipset
-------------------------------------------------------------------------

Northbridge Intel Rocket Lake rev. 01
Southbridge Intel B560 rev. 11
Bus Specification PCI-Express 4.0 (16.0 GT/s)
Graphic Interface PCI-Express 4.0
PCI-E Link Width x16 (max 16x)
PCI-E Link Speed 2.5 GT/s (max 16.0 GT/s)
Memory Type DDR4
Memory Size 32 GBytes
Channels 2 x 64-bit
Memory Frequency 1795.6 MHz (3:54)
Memory Max Frequency 3200.0 MHz
CAS# latency (CL) 18.0
RAS# to CAS# delay (tRCD) 22
RAS# Precharge (tRP) 22
Cycle Time (tRAS) 42
Row Refresh Cycle Time (tRFC) 630
Command Rate (CR) 2T
Uncore Frequency 3591.2 MHz
Memory Controller Frequency 1795.6 MHz
Host Bridge 0x4C53


Memory SPD
-------------------------------------------------------------------------

DIMM # 1
SMBus address 0x50
Memory type DDR4
Module format UDIMM
Module Manufacturer(ID) Corsair (7F7F9E0000000000000000000000)
SDRAM Manufacturer (ID) Micron Technology (2C00000000000000000000000000)
Size 8192 MBytes
Max bandwidth DDR4-3596 (1798 MHz)
Max JEDEC DDR4-2133 (1066 MHz)
Part number CMK16GX4M2D3600C18
Nominal Voltage 1.20 Volts
EPP no
XMP yes, rev. 2.0
AMP no
EXPO no



Monitoring
-------------------------------------------------------------------------

Mainboard Model B560 AORUS PRO AX (0x00000451 - 0xECEAEA00)

DMI Memory Device
designation Controller0-ChannelA-DIMM0
format DIMM
type DDR4
total width 64 bits
data width 64 bits
size 8 GB
speed 3600 MHz
manufacturer Corsair
part number CMK16GX4M2D3600C18
serial number 00000000
voltage 1.200000
manufacturer id 0x9E02
product id 0x0


Storage
-------------------------------------------------------------------------

Drive 0/0/-1/-1
Device Path \\?\scsi#disk&ven_&prod_st2000dm008-2fr1#4&5d48514&0&000000#{53f56307-b6bf-11d0-94f2-00a0c91efb8b}
Name ST2000DM008-2FR102
Revision 0001
Serial ZFL1RWSE
Capacity 1863.0 GB
Type Fixed
Bus Type SATA (11)
Rotation speed 7200 RPM
Features SMART
Controller Serial ATA Controller - AHCI at bus 0, device 23, function 0
Volume d:\, 1863.0 GBytes (99.9 percent available)

Drive 1/1/-1/-1
Device Path \\?\scsi#disk&ven_nvme&prod_bc711_nvme_sk_hy#5&230e705d&0&000000#{53f56307-b6bf-11d0-94f2-00a0c91efb8b}
Name BC711 NVMe SK hynix 512GB
Revision 41002131
Serial CNA4N635011304M50
Capacity 476.9 GB
Type Fixed, SSD
Bus Type NVMe (17)
Features
Controller NVM Express (NVMe) Controller at bus 2, device 0, function 0
Link Speed PCI-E 4x @ 8.0 GT/s
Volume c:\, 476.3 GBytes (69.1 percent available)


Graphics
-------------------------------------------------------------------------

Number of adapters 1

Graphic APIs
-------------------------------------------------------------------------

API D3D
API NVIDIA I/O
API NVIDIA NVML
API NVIDIA NVAPI
pwrMask 0x000C9CF3
pwrTable [F5 F6 00 00 FF FA FB E8 00 00 FE F6 F1 00 00 E0 00 00 01 01 00 00 00 00 00 00 00 00 00 00 00 00 ]

Display Adapters
-------------------------------------------------------------------------

Display adapter 0 (primary)
ID 0x10B0104
Name NVIDIA GeForce RTX 3070 Ti
Board Manufacturer NVIDIA Corporation
Revision A1
Codename GA104
Core family 0x174 (GA104-A)
Power Limit 290.0 W (current), 290.0 W (default), 290.0 W (max)
Thermal Limit 93.0 °C (max), 95.0 °C (slowdown), 98.0 °C (shutdown)
Cores 6144
ROP Units 96
TM Units 384
RayTracing Cores 48
Tensor Cores 192
Technology 8 nm
Memory type GDDR6X (Micron)
Memory size 8 GB
Memory bus width 256 bits
Current Link Width x16
Current Link Speed 2.5 GT/s
PCI device bus 1 (0x1), device 0 (0x0), function 0 (0x0)
Vendor ID 0x10DE (0x10DE)
Model ID 0x2482 (0x146A)
Revision ID 0xA1
Root device bus 0 (0x0), device 1 (0x1), function 0 (0x0)
Performance Level Base
Core clock 1575.0 MHz
Memory clock 9501.0 MHz
Performance Level Boost
Core clock 1770.0 MHz
Memory clock 9501.0 MHz
Performance Level Current
Core clock 210.0 MHz
Memory clock 405.0 MHz
Driver version 31.0.15.4601
WDDM Model 2.7

Win32_VideoController AdapterRAM = 0xFFF00000 (4293918720)
Win32_VideoController DriverVersion = 31.0.15.4601
Win32_VideoController DriverDate = 10/26/2023

Monitor 0
Model ED270R M ()
ID ACR0A91
Serial 93080557C3W01
Manufacturing Date Week 8, Year 2023
Size 27.0 inches
Max Resolution 1920 x 1080 @ 60 Hz
Horizontal Freq. Range 193-193 kHz
Vertical Freq. Range 48-165 Hz
Max Pixel Clock 410 MHz
Gamma Factor 2.2

Monitor 1
Model S27R35A (Samsung)
ID SAM7126
Serial H4UR301262
Manufacturing Date Week 12, Year 2021
Size 27.2 inches
Max Resolution 1920 x 1080 @ 60 Hz
Horizontal Freq. Range 30-70 kHz
Vertical Freq. Range 50-71 Hz
Max Pixel Clock 170 MHz
Gamma Factor 2.2


Software
-------------------------------------------------------------------------

Windows Version Microsoft Windows 10 (10.0) Professional 64-bit (Build 19045)
DirectX Version 12.0

So, I am not sure if this is a driver issue, or a hardware issue. I already changed MB and Memories, I tried this GPU in another computer, and the problem still happens.
I was wondering to reinstall or updated the GPU Bios, but I was not able to find the right bios file, I will attach the original file in case that can help.

I do not have any more ideas about what the problem can be, so that is why I came here looking for some help.

For now, that is all, thanks!!!!

 

[Tatty_Two]

The only thing that I can see at a glance from what you have provided is that your GDDR6X are a bit warm at 106C, do they have pads or make contact with anything?

 

[Enrai]

The only thing that I can see at a glance from what you have provided is that your GDDR6X are a bit warm at 106C, do they have pads or make contact with anything?

Hello Tatty_Two

Yeah it has pads, and it should be making contact, but I will open it again to make sure.
How much the memory temp should be under a stressing test (average)?

Thanks!!!

 

[Frick]

What power supply do you have?

 

[Enrai]

I got a brand new GIGABYTE GP-UD750GM 750 W ATX 12V v2.31 80 PLUS GOLD Certified
UD750GM Key Features | Power Supply - GIGABYTE Global

What power supply do you have?

 

[Tatty_Two]

Hello Tatty_Two

Yeah it has pads, and it should be making contact, but I will open it again to make sure.
How much the memory temp should be under a stressing test (average)?

Thanks!!!

It varies dependant on memory manufacturer and variant, card model/cooling, views on max temps can vary, some info says 105C, some 110C, my GDDR6X on my 4070 is in the 70's under stress but I don 't use Furmark.

Edit: Maybe have a play again in Diablo 4 with HWinfo64 (sensors) running in the background to check how the game affects the temps, if only to cancel out any temp issue in game.

 

[Enrai]

It varies dependant on memory manufacturer and variant, card model/cooling, views on max temps can vary, some info says 105C, some 110C, my GDDR6X on my 4070 is in the 70's under stress but I don 't use Furmark.

I see, I believe because this gpu model is a fan blower it is normal, when I am playing the memory temp stay around 80C.
If I have to guess, I will say it is a driver issue or hardware issue, but as I told before, the GPU was working fine before, then after a nvidia driver update this problem started. I already reinstalled the system and already tried different old drivers, but no success so far.

I saw that same time when we rewrite the bios file it can help, but for some dark reason I am not able to find any information about this graphic card or bios for it.

It varies dependant on memory manufacturer and variant, card model/cooling, views on max temps can vary, some info says 105C, some 110C, my GDDR6X on my 4070 is in the 70's under stress but I don 't use Furmark.

Edit: Maybe have a play again in Diablo 4 with HWinfo64 (sensors) running in the background to check how the game affects the temps, if only to cancel out any temp issue in game.

a few more information about the GPU

and this is the error that I am having when the computer does not reboot:

 

[Deleted member 182555]

are thermal pads original?
Was the gpu opened before presenting the problems?
try replacing the thermal pads with better soft ones. Check the hardness specifications first. make sure there is optimal contact with TIM.
you can also check whether vram are reliable.
run a stress test, verify that the problem is not only with that game.

 

[Chomiq]

Full system specs otherwise there's nothing that would suggest it's a GPU fault. If you're getting a full reboot with black screen and fans running at 100% then it's most likely memory issue or CPU OC (i.e. curve optimizer set too high on Ryzen).

 

[Enrai]

are thermal pads original?
Was the gpu opened before presenting the problems?
try replacing the thermal pads with better soft ones. Check the hardness specifications first. make sure there is optimal contact with TIM.
you can also check whether vram are reliable.
run a stress test, verify that the problem is not only with that game.

Hello Don,

The pad are not original, I had to replace it. I just did the thermals again, on a stress test my gpu is getting this:

Much better, but I was still getting crashs.

I just tried another stuff and it seams worked, I had to disable the xmp on the bios setting, and I was able to play for more than one hour without any problem.
But because of it I lost performance, any idea about xmp problems between ram and gpus?

I am using 4x Corsair Vengeance LPX 8GB DDR4 3600MHz

Regards!!!

Full system specs otherwise there's nothing that would suggest it's a GPU fault. If you're getting a full reboot with black screen and fans running at 100% then it's most likely memory issue or CPU OC (i.e. curve optimizer set too high on Ryzen).

Hello Chomiq,

I just tried another stuff and it seams worked, I had to disable the xmp on the bios setting, and I was able to play for more than one hour without any problem.
But because of it I lost performance, any idea about xmp problems between ram and gpus?

I am using 4x Corsair Vengeance LPX 8GB DDR4 3600MHz

Regards!!!

 

[freeagent]

That is running about 10-20c than my stock evga 3070Ti.

 

[Mussels]

WHEA errors usually arent from the GPU - you're likely looking in the wrong place for the cause of your issues

Can you fill out your system specs properly here
TechPowerUp Forums
and post a photo of the PC in this thread - simple things that you think are okay may not be, your software copy paste cant tell us what case its in, what powersupply is used, extension cables etc etc

I just tried another stuff and it seams worked, I had to disable the xmp on the bios setting, and I was able to play for more than one hour without any problem.
But because of it I lost performance, any idea about xmp problems between ram and gpus?

I am using 4x Corsair Vengeance LPX 8GB DDR4 3600MHz

That's exactly what was needed in the first post.

You've got an unstable memory controller. Your 11400 does not support that RAM speed.

Intel Core i511400 Processor 12M Cache up to 4.40 GHz Product Specifications

It can overclock to 3600, but with four sticks of LPX RAM (Which is known to be problematic since 'revisions' are entirely different RAM with the same product name) is not going to just work with XMP.
Without a K series CPU the required voltages are locked, you should enable XMP and then lower the RAM speed to 3200, and lower it one step each time the system crashes until its long term stable.


If you can alter the SA voltage (System agent) in the BIOS, report back with what voltage it's running at and what it can be adjusted to

 

[Enrai]

WHEA errors usually arent from the GPU - you're likely looking in the wrong place for the cause of your issues

Can you fill out your system specs properly here
TechPowerUp Forums
and post a photo of the PC in this thread - simple things that you think are okay may not be, your software copy paste cant tell us what case its in, what powersupply is used, extension cables etc etc


That's exactly what was needed in the first post.

You've got an unstable memory controller. Your 11400 does not support that RAM speed.

Intel Core i511400 Processor 12M Cache up to 4.40 GHz Product Specifications
View attachment 321492

It can overclock to 3600, but with four sticks of LPX RAM (Which is known to be problematic since 'revisions' are entirely different RAM with the same product name) is not going to just work with XMP.
Without a K series CPU the required voltages are locked, you should enable XMP and then lower the RAM speed to 3200, and lower it one step each time the system crashes until its long term stable.


If you can alter the SA voltage (System agent) in the BIOS, report back with what voltage it's running at and what it can be adjusted to

Hi Mussles,

Thanks for the directions, so far I am running 2x 8gb at 3600, and so for it is going good, I will try the 4x 8gb at 3200 as the cpu recommendation.

I will keep posting as soon I have more test done.

Here is my little one:

Thanks so for for the help all!!!

 

[Solaris17]

Thanks so for for the help all!!!

Glad you narrowed it down!

As a quick tip for you and everyone else reading.

First; big thanks for actually looking in your error logs, many do not.

When you do look at your logs though, dont focus a ton on the actual error, read it and note the time. You actually generally want to look at events BEFORE the error happens, this can give you a better picture into what went wrong.

 

[Tatty_Two]

If by disabling XMP it improves the situation but you have a noticeable performance drop then maybe dial in the primary timings and speed manually that XMP gave you but add a little extra voltage that may stabilise as you would not be the first with these kind of issues, especially with 4 sticks of ram, so if XMP gave you 1.35v then maybe dial in 1.37 or 1.38v to see if that works for you, also what are your VCCIO and VCCSA (term can differ slightly) settings? Are they set to auto or have you manually set them.

 

[Keullo-e]

VBIOS isn't a driver which needs updating..

 

[Mussels]

Hi Mussles,

Thanks for the directions, so far I am running 2x 8gb at 3600, and so for it is going good, I will try the 4x 8gb at 3200 as the cpu recommendation.

I will keep posting as soon I have more test done.

Here is my little one:

View attachment 321493View attachment 321494


Thanks so for for the help all!!!

That's a nice setup, nothing i can see physically that isn't done right.


LPX ram has revisions on the sticker - check to see what yours are, and make sure you match the pairs up.
So if you had two rev 1.1 and two 1.2 you'd want

1.1|1.2|1.1|1.2 when they go in at the same time

Then CPU-Z screenshots of the RAM, one for a stick from each kit so we can compare their timings - they may be the same, they may not. The idea is the manually match the slowest sticks, lowering the speed every crash. If you can manually set your SA voltage, you'll have a lot better luck with this - ryzen has the exact same issues with these RAM speeds as well, it's not unique to your setup.

 

[Dr. Dro]

I got a brand new GIGABYTE GP-UD750GM 750 W ATX 12V v2.31 80 PLUS GOLD Certified
UD750GM Key Features | Power Supply - GIGABYTE Global

(edit: just read thread. Follow the advice pertaining to RAM first.)

I'm fairly sure that is "new and improved" version of the IED that GN had at their office a while back. Yep, TPU review by Aris:

Gigabyte UD750GM 750 W Review - No Lightning This Time

The Gigabyte UD750GM uses a better version of the platform used in the P750GM, with vastly improved protection features. Overall performance is also higher, but still needs more work to be competitive. A lower price might help since the Corsair RM750x—the king of this category—only costs $10 more.

www.techpowerup.com

This is a poorer quality power supply than its equivalently priced and positioned power supplies, it also suffers from high ripple and uses more bottom shelf components than it has any right to.

Your event log entries do not contain NVIDIA driver errors. Furmark causes power limiters to kick in so it's unreliable for power testing. It sounds like your power supply is unable to manage the power transients demanded by the graphics card. Every GPU does this and power supplies must be able to accommodate the transient spikes in power delivery, or this will occur.

106 degrees C is well within acceptable parameters for GDDR6X on Ampere. Try lowering your GPU core clock and voltage by using the afterburner curve editor. Try like 1500 MHz at 850 mV. If the crashing stops replace PSU.

 

[Gica]

Your 11400 does not support that RAM speed.

Intel Core i511400 Processor 12M Cache up to 4.40 GHz Product Specifications


Without a K series CPU the required voltages are locked, you should enable XMP and then lower the RAM speed to 3200, and lower it one step each time the system crashes until its long term stable.

10th and 11th have unlocked SA and CPU VCCIO2.
VCCIO2 (memory controller voltage) is the equivalent of VDDQ TX on series 12, 13 and 14, also unlocked. It has all the tools for overclocking to 3600MHz. If there are problems, they come from somewhere else, not from the voltage.

As a 10500 and 11600K owner, I also recommend 3200MHz. Higher frequencies require voltages towards the maximum allowed limit and bring "increase" in performance only in AIDA.

 

[Enrai]

10th and 11th have unlocked SA and CPU VCCIO2.
VCCIO2 (memory controller voltage) is the equivalent of VDDQ TX on series 12, 13 and 14, also unlocked. It has all the tools for overclocking to 3600MHz. If there are problems, they come from somewhere else, not from the voltage.

As a 10500 and 11600K owner, I also recommend 3200MHz. Higher frequencies require voltages towards the maximum allowed limit and bring "increase" in performance only in AIDA.

Hello All!!!

Thanks once again for helping me.
I got new memories 4x 8gb XPG Spectrix D50 3200Mhz, I am running my pc with the xmp on, and so far no crash or any other error

That is my gpu and cpu performance during Diablo IV, running at 235 fps average.

 

[Enrai]

Good Day all!!!

Yeah, it looks like the problem is solved, I was afraid that it could be my GPU, but in the and it was only the memory frequency between CPU and RAM. When I got the rams, I just checked the frequency that was accepted from the motherboard, and completely forgot about the CPU.

Once again, thanks everyone for your time to help me with this problem!

Regards,
Enrai

 

[Mussels]

10th and 11th have unlocked SA and CPU VCCIO2.
VCCIO2 (memory controller voltage) is the equivalent of VDDQ TX on series 12, 13 and 14, also unlocked. It has all the tools for overclocking to 3600MHz. If there are problems, they come from somewhere else, not from the voltage.

As a 10500 and 11600K owner, I also recommend 3200MHz. Higher frequencies require voltages towards the maximum allowed limit and bring "increase" in performance only in AIDA.

From what i've been told by ir_cow, 10 and 11 had more options than 12 onward - but it's a complicated mess that's poorly documented so it's a real PITA without them in front of me to check.
It's his advice that SA voltage is likely the issue, you yourself just said you'd run 3200.


Looks like you moved to 4x single rank sticks which is a lot easier to run, but without the information from the corsair modules it's impossible to say if they were as well

 

[ShiBDiB]

Glad you narrowed it down!

As a quick tip for you and everyone else reading.

First; big thanks for actually looking in your error logs, many do not.

When you do look at your logs though, dont focus a ton on the actual error, read it and note the time. You actually generally want to look at events BEFORE the error happens, this can give you a better picture into what went wrong.

Also the solution is very very very rarely anything bios related with your gpu.. so stop fricking messing with them