RTX 2070 underperforming

[la_Z_boy]

Hello all,

I got a brand new bundle PC, migrated my old HDD to the new SSD, latest drivers, Windows 10 etc. Ran a couple benchmarks - my GPU is seriously underperforming, while all other components are just fine. Here are the raw UserBenchMark results:

CPU: Intel Core i7-9700K - 116.1%
GPU: Nvidia GeForce RTX 2070 - 92.8%
SSD: WDS500G3X0C-00SJG0 500GB - 131.1%
HDD: Seagate Barracuda 7200.14 500GB - 15.1%
USB: SanDisk Cruzer Blade 32GB - 8%
RAM: Unknown TEAMGROUP-UD4-3000 2x8GB - 97.5%
MBD: Gigabyte GA-Z390 UD

The results indicated the graphics card was "performing way below expectations (6th percentile). Similarly 3D Mark gave me a score of 8452 on Time Spy 1.0. Others online have posted 3D Mark results with the same GPU, CPU, and RAM combination, with a much higher score, e.g. 10088.

I've contacted the company that assembled the bundle, but I was wondering if anyone on here can help me figure out why this is happening, and how to fix it!?

P.S. Here is the PSU spec: Kolink KL-600M 600W 80 Plus Bronze Modular Power Supply

[Mira Yurizaki]

Download and run GPU-Z in the "Sensors" tab and then rerun the benchmarks. Some things to note:

• PerfCap Reason: If it's Thrm or Pwr, this could be a problem with thermals or power delivery to the video card.
• Temperatures: If they're constantly hovering at or above 80C (possibly 85C), this is could be a problem
• GPU Clock: May be an issue if the GPU is below advertised boost speeds or base clock

[Firewrath9]

seems fine for stock, UBM is shit.

[la_Z_boy]

30 minutes ago, Mira Yurizaki said:

Download and run GPU-Z in the "Sensors" tab and then rerun the benchmarks. Some things to note:

• PerfCap Reason: If it's Thrm or Pwr, this could be a problem with thermals or power delivery to the video card.
• Temperatures: If they're constantly hovering at or above 80C (possibly 85C), this is could be a problem
• GPU Clock: May be an issue if the GPU is below advertised boost speeds or base clock

 

Here is a screenshot during the render test. The temp never went above 55 (it's around 43 idle), the base and boost clock seemed in line with advertised, and the perfcap reason was VRel (I don't know if that is something to be alarmed about?).

 

I should add that I've tried playing GPU dependent games such as Rainbow Six Siege. In Siege, on Ultra, 1080p, 144hz, I'm getting 90 - 110 fps, when all the benchmarks I've seen online indicate I should be getting 150+ easily. 

[Princess Luna]

51 minutes ago, la_Z_boy said:

migrated my old HDD to the new SSD, latest drivers, Windows 10 etc.

So you're using a previus Windows image yes? did you have an older GPU on this OS?

 

You might have to still DDU in safe mode and reinstall the drivers fresh new for best performance.

[ddennis002]

11 minutes ago, la_Z_boy said:

 

 

Here is a screenshot during the render test. The temp never went above 55 (it's around 43 idle), the base and boost clock seemed in line with advertised, and the perfcap reason was VRel (I don't know if that is something to be alarmed about?).

 

I should add that I've tried playing GPU dependent games such as Rainbow Six Siege. In Siege, on Ultra, 1080p, 144hz, I'm getting 90 - 110 fps, when all the benchmarks I've seen online indicate I should be getting 150+ easily. 

What was your max GPU Clock, VDDC according to GPU-Z.

 

Is your card a Founders Edition?

[la_Z_boy]

26 minutes ago, Princess Luna said:

So you're using a previus Windows image yes? did you have an older GPU on this OS?

 

You might have to still DDU in safe mode and reinstall the drivers fresh new for best performance.

Thanks I'll try this.

 

14 minutes ago, ddennis002 said:

What was your max GPU Clock, VDDC according to GPU-Z.

 

Is your card a Founders Edition?

See screenshot attached; VDDC during render test was 1.0370.

 

Nope, the card is not a founders edition.

[ddennis002]

24 minutes ago, la_Z_boy said:

Thanks I'll try this.

 

See screenshot attached; VDDC during render test was 1.0370.

 

Nope, the card is not a founders edition.

 

Do you have Afterburner or Precision X1 installed?

[la_Z_boy]

3 minutes ago, ddennis002 said:

Do you have Afterburner or Precision X1 installed?

I have Afterburner, although I've never really used it. Why?

 

58 minutes ago, Princess Luna said:

So you're using a previus Windows image yes? did you have an older GPU on this OS?

 

You might have to still DDU in safe mode and reinstall the drivers fresh new for best performance.

I uninstalled then reinstalled the latest drivers in safe mode (rebooting each time), sadly no change to my benchmarks

[ddennis002]

Just now, la_Z_boy said:

I have Afterburner, although I've never really used it. Why?

 

I uninstalled then reinstalled the latest drivers in safe mode (rebooting each time), sadly no change to my benchmarks

Turn your power slider up to the highest it will go in Afterburner and rerun timespy. Honestly your results seem typically of a TU106-400 chip, a 400A  chip would be a lot higher, what is the exact model of GPU?

[la_Z_boy]

22 minutes ago, ddennis002 said:

Turn your power slider up to the highest it will go in Afterburner and rerun timespy. Honestly your results seem typically of a TU106-400 chip, a 400A  chip would be a lot higher, what is the exact model of GPU?

Reran Timespy with power limit increased to 114 in Afterburner; virtually identical score (8503). The GPU is a Palit GeForce RTX 2070 Dual 8192MB.

[la_Z_boy]

Just to add; others I've asked seem to focus on the PSU not being up to snuff (Kolink 600w 80+ bronze), and are adamant that since I migrated my old OS, to a new motherboard, I should really do a clean install of Windows.

 

Any opinions on these two points?

[Spotty]

19 minutes ago, la_Z_boy said:

others I've asked seem to focus on the PSU not being up to snuff (Kolink 600w 80+ bronze)

Few different Kolink 600 80+ Bronze models. From what little information I could find on them, none of them seem to be any good. Despite the models being called "600", it's really a 500W unit, and appears to be a group regulated design with only ~450W available on the 12V.
While it's not a good PSU, it won't cause performance issues.
 

20 minutes ago, la_Z_boy said:

are adamant that since I migrated my old OS, to a new motherboard, I should really do a clean install of Windows.

Run DDU (Display Driver Uninstaller) to clean up and old driver files that might have been left over that might be causing issues. Then install the latest display drivers from the Nvidia website.

Looking at the scores you posted above, it seems to be normal. Those 10,000+ 3DMark scores might be the top overclocked results? As per the graph posted by @Firewrath9 a score of 8500 seems normal for TimeSpy for a RTX2070.

[la_Z_boy]

6 minutes ago, Spotty said:

Few different Kolink 600 80+ Bronze models. From what little information I could find on them, none of them seem to be any good. Despite the models being called "600", it's really a 500W unit, and appears to be a group regulated design with only ~450W available on the 12V.
While it's not a good PSU, it won't cause performance issues.
 

Run DDU (Display Driver Uninstaller) to clean up and old driver files that might have been left over that might be causing issues. Then install the latest display drivers from the Nvidia website.

Looking at the scores you posted above, it seems to be normal. Those 10,000+ 3DMark scores might be the top overclocked results? As per the graph posted by @Firewrath9 a score of 8500 seems normal for TimeSpy for a RTX2070.

Ok, that's good to know regarding the PSU.

 

I ran DDU previously in safe mode and reinstalled latest drivers.

 

I'm aware the TimeSpy 3DMark scores are in the normal range, but playing GPU-dependent games such as Rainbow Six Siege, I'm barely breaking 90FPS at 1080p, 60hz, ultra settings.

Other benchmarks posted online with similar rig and not overclocked are 150+FPS, and that's at 1080p, ultra settings, but higher refresh rate of 144hz!

[la_Z_boy]

Incidentally I ran Valley Benchmark if that makes any difference to others viewing this thread (I don't know why results say Win 8 as I have Win 10). Similar rigs seem to get higher scores, don't appear to be overclocked!

 

[Firewrath9]

5 hours ago, la_Z_boy said:

Incidentally I ran Valley Benchmark if that makes any difference to others viewing this thread (I don't know why results say Win 8 as I have Win 10). Similar rigs seem to get higher scores, don't appear to be overclocked!

 

seems alright

red score is for 1080p XHD 8xAA, they have a 6850K (8c16t) and thats a fac OC

[ddennis002]

5 hours ago, la_Z_boy said:

Incidentally I ran Valley Benchmark if that makes any difference to others viewing this thread (I don't know why results say Win 8 as I have Win 10). Similar rigs seem to get higher scores, don't appear to be overclocked!

 

 

Looks good to me! This is my Valley Score with my EVGA 2070 at stock.

[la_Z_boy]

Ok, maybe I'm crying over spilt milk, in so much as other non-overclocked rigs seem to get 10-20FPS more. 

2 hours ago, Firewrath9 said:

seems alright

red score is for 1080p XHD 8xAA, they have a 6850K (8c16t) and thats a fac OC

Ok, maybe I'm just making a mountain out of a molehill, as my score isn't bad, as has been pointed out. Interesting that the non OC 2070 Armor in that graph appears to get an average 20FPS higher than mine though! 

[ddennis002]

3 hours ago, la_Z_boy said:

Ok, maybe I'm crying over spilt milk, in so much as other non-overclocked rigs seem to get 10-20FPS more. 

Ok, maybe I'm just making a mountain out of a molehill, as my score isn't bad, as has been pointed out. Interesting that the non OC 2070 Armor in that graph appears to get an average 20FPS higher than mine though! 

So after looking into your card specs, that card sports the 400 chip not the 400a and has a power limit of 200W which is on the very low side for a RTX 2070.

Based on this your boost clock is probably only going to hit 1900Mhz max. Compared to a 400a and cards with power limits of around 240W. Your card is going to run significantly slower.

My card has a power limit of 240W and with just gpu boost hits 2085Mhz without any overclocking, and I can hit almost 2200Mhz when OCd.

[Firewrath9]

14 hours ago, la_Z_boy said:

Ok, maybe I'm crying over spilt milk, in so much as other non-overclocked rigs seem to get 10-20FPS more. 

Ok, maybe I'm just making a mountain out of a molehill, as my score isn't bad, as has been pointed out. Interesting that the non OC 2070 Armor in that graph appears to get an average 20FPS higher than mine though! 

they have better cpu, and quad channel ram in their test bench, and the armor has a fac oc, and a better cooler (gpu boost 3.0) and its open air thereore better temps therefore higher clockspeeds

[la_Z_boy]

Hi, as a follow-up, I decided to do a full reinstall of Windows wiping all personal files etc. and that fixed the issue (previously I'd reinstalled but keeping files).

 

I'm now getting much higher FPS across the board. I don't know if this meant the issue was driver conflict or just that my old Windows had become bloated in some way, but I've created a system restore point since everything is running well now! 

 

Thank you all for your help!