troubleshooting 2080 Ti underperforming a 1060 GTX

[BryonLewis]

Parts:

• Ryzen 1700
• MSI Gaming Carbon Pro x370
• 16GB DDR4 (3000) T-force Delta II
• EVGA 1060GTX (Old Card)
• ZOTAC GAMING GeForce RTX 2080 Ti AMP Extreme Core
• Corsair RM650 semi-modular power supply

Problem:

After getting the new card and installing it. It's benchmark performance and in-game performance is less than the 1060GTX. 3dMark, MSI Kombustor, CSGO and others all perform worse. Lower scores and lower frame rates.

Attempted Actions -Safe Mode DDU (driver uninstall) and reinstall multiple times -Tried alternative PCI-Ex16 slot -Replaced Modular PCI-E power cables

Additional Notes: I think I've narrowed it down to either the power supply or the card itself. After the testing I swapped back to the 1060GTX and it runs right around where a 1060 should.

Additionally during the testing the power consumption percentage for the Zotac stayed mostly in the 40% range occasionally spiking to 60% once. (The 1060 went to 100% for the benchmarks) The Core clock never really went above the base clock frequency.

This is a pretty expensive card that I just got so I want to try to eliminate all other problems. Other note is that even during benchmarks the Temp never really got above 50-52C.

The only thing I can think of beside the card at this point is the power supply. But if it couldn't provide the power I believe that it would cause other stability issues in the system instead of just not giving the performance. Meaning it would be crashing the computer because of drops in power on other parts (RAM,Mobo,Others)

 

3DMark examples (Same exact PC only changed out the GPU)

SCORE 2466 with NVIDIA GeForce RTX 2080 Ti(1x) and AMD Ryzen 7 1700

Graphics Score 2209 CPU Score 7265

SCORE 4393 with NVIDIA GeForce GTX 1060-6GB(1x) and AMD Ryzen 7 1700

Graphics Score 4106 CPU Score 7281

[lloose]

What resolution?

Are the clocks normal during benchmarking?

 

[BryonLewis]

1440p

When I first started off there were issues where the initial clock wouldn't go above 300MHz even when in a game, confirmed using MSI Afterburner OSD, as well as CPUID HWmonitor logs.  It might go to like 315MHz.  After a couple driver uninstall refreshes it was reporting that it was getting to 15XX (somewhere around the base clock).  But the performance wasn't anywhere near what it should be.  Still sub 100 FPS in CSGO and the benchmark scores that I included.

[Ankerson]

What PSU is that exactly?

 

Model number?

 

Also post the result link for 3DMArk.

 

How is the GPU pulled into the PSU?

 

How many cables?

[BryonLewis]

3dMark Link for 2080ti

 

3dMark Link for Same PC with 1060GTX

 

Corsair Certified CP-9020103-NA CX650M 650W ATX12V / EPS12V SLI Ready CrossFire

was wrong on the RM, it's actually a CX.  The power supply is the one thing I'm iffy on, but I would think there would be some other issues if it wasn't supplying enough power.

 

2 Separate Cables back to the PSU for powering.

[Ankerson]

11 minutes ago, BryonLewis said:

3dMark Link for 2080ti

 

Corsair Certified CP-9020103-NA CX650M 650W ATX12V / EPS12V SLI Ready CrossFire

was wrong on the RM, it's actually a CX.  The power supply is the one thing I'm iffy on, but I would think there would be some other issues if it wasn't supplying enough power.

 

Ugg.

 

You are running a RTX 2080Ti on that budget PSU?

 

A $1,000+ GPU on THAT?

 

Looks like the GPU is starving for power, it's not boosting even near as high as it should.

 

Get a good high quality PSU and 750W, Tier A+.

 

 

[FrostytheDopeman]

I believe RTX2080 or 2080Ti pinning itself to 300MHz is a symptom of when the on-board controller sees something wrong with power delivery. Your setup is in dire need of a new power supply anyways (a budget 650W will not do the trick for a 2080Ti), and if that doesn't work I might resort to RMAing the card. The card pinning itself to that clock can also happen when someone shunts the power delivery circuit to unlock voltage control on the card, so if you got it second-hand someone could have been digging around where they didn't know what they were doing.

 

I'd shoot for a 750W power supply at the minimum like Ankerson says, with really clean power delivery. I wouldn't gamble with my PSU when it comes to a binned 2080 Ti card...

[BryonLewis]

Thanks everyone, it was my one concern, but seeing as I live 2 hour round trip from a Bestbuy (being my closest option for picking up a higher end one) I was open to any other issues that could be causing the problem.  I would of thought that the symptom of not enough power would of been less subtle.  Just put in an order for a RM850X off that list Ankerson provided.  When it gets here Tuesday I'll update if that fixed the issue.  Thanks again everyone.

[Ankerson]

1 minute ago, BryonLewis said:

Thanks everyone, it was my one concern, but seeing as I live 2 hour round trip from a Bestbuy (being my closest option for picking up a higher end one) I was open to any other issues that could be causing the problem.  I would of thought that the symptom of not enough power would of been less subtle.  Just put in an order for a RM850X off that list Ankerson provided.  When it gets here Tuesday I'll update if that fixed the issue.  Thanks again everyone.

 

Let us know how it goes.

 

As a side note, remove the RTX 2080Ti from the machine until you get the new PSU.

[BryonLewis]

1 minute ago, Ankerson said:

 

Let us know how it goes.

 

As a side note, remove the RTX 2080Ti from the machine until you get the new PSU.

As soon as I noticed the performance Issues I swapped back to the 1060.  Been watching Linus for quite a few years and he always talks about the forums.  You folks truly are awesome.

[BryonLewis]

On 4/20/2019 at 8:46 PM, Ankerson said:

 

Let us know how it goes.

 

As a side note, remove the RTX 2080Ti from the machine until you get the new PSU.

Okay well I installed my RM850x as soon as I got it and the same exact issues.  I ran the Firestrike 3dMark on both the GTX and RTX

GTX

RTX

 

Also ran GPU-Z with it logging at the same time.  As soon as I started it up and got the driver installed I opened GPUZ and it immediately has the Perfcap Reason as Pwr, even though it's running at 300MHz and the power consumption is like 60-80W.  It never once was not Perfcap Reason Pwr.

 

Temperatures throughout the benchmark and everything never went above 56C.  It just doesn't want to ramp itself up. 

 

I'm thinking I have to escalate this to ZOTAC Support and get a RMA.  Again before I've tried different PCI-E slots, did DDU multiple times and everything else.  Still performs like crap.

GPU-Z Sensor Log.txt

[BryonLewis]

11 minutes ago, ReggieGRS said:

Seems RMA is indeed your best option. Tho i'd try to test it in another rig before proceeding. Unlikely to be it but, y'know, never hurt to be more certain.

I would if I didn't live close to 40 minutes away from anyone I would trust with a modern enough and large enough computer for it.  I've already spent like 5 hours and an additional 120 bucks trying to debug a card I paid over $1k for.