Question HDR problems, bad card?

TheMenace4Life

Reputable
Jan 11, 2020
19
1
4,515
I finally decided to upgrade my pc since it was about 7 years old. I was running an Nvidia 660Ti overclocked to 1306mhz with and FX-6300 cpu, overclocked as well. I swapped them out for a ryzen 7 2700x on an aorus x570 elite board with an EVGA 2070 Super gpu. nothing overclocked.

I've been unable to play games since then. I've experienced a wide range of issues with several different games.

  • not loading at all, no errors.
  • loading slowly
  • loading but then locking up forcing me to kill the task.
  • Loading then crashing.
  • loading in a window instead of fullscreen (alt+enter doesn't fix it)
  • Loading and working, but degraded video quality, stuttering, artifacts, screen tearing
  • overall buggy experience trying to play any games or even within the XBOX game pass program. Videos seem to play fine in VLC etc...
I tried a fresh install of Windows 10 to try to eliminate any potential programs that could be causing this. Reinstalled windows, drivers and went right into games but I encountered the same issues so after a day or so of stress testing everything, cpu with Prime95, GPU with heaven benchmark, checked the ram with memtest86 (everything passed on all tests), I decided to swap out the card for the 660Ti and boom, everything was working fine. No more stuttering. games loaded quickly and smoothly, no artifacts or screen tearing, but obviously this card can really only handle some of these games on low settings.

Then I ran the same set of tests on each card. Swapped it, installed latest drivers, set the screen resolution and tested gpu benchmark and a couple of games. Each time the same settings were applied in Nvidia control panel. 1080Px120HZ, limit frame rate to 118FPS, adapative v-sync: on, everything else was left at default.

Here are the benchmark and game testing results.

Benchmark Settings:
Quality: High
Resolution: 1080P
DX11
AAx2

660Ti
50FPS, (min - 30.8, Max - 105.3) score of 1263, smooth video, no stuttering, screen tearing, artifacts. 60C max temp

2070 Super
116.9 FPS (min - 41.2, Max - 118.3) score of 2944. video stuttering (not terrible but noticeable) and some screen tearing. 60C max temp

For the games I am just measuring by gameplay experience as I don't have any monitoring tools to show an OSD. I did not reinstall them after windows install in case they were the culprit.

Forza 4 Horizon

660Ti
loads quickly and plays perfectly fine on low settings. I was able to play on ultra without any issues but I did get a warning about low video memory but that seems normal.

2070 Super
slow to load, locks up in the loading screens, cant get to the game. After several cycles of swapping cards it mysteriously started working but it has stuttering issues.

Metro Exodus

This is where I really noticed something was wrong.

660Ti
loads quickly and plays perfectly fine on ULTRA settings. I was able to play on ultra without any issues and the game looks great.

2070 Super
crashes, won't load, and then after a few cycles of card swapping it started to play, really weird... This has terrible video during the intro scene, major artifacting, some stuttering etc... and oddly everything was really dark in this game. I noticed it because when I put the 660Ti in it was much brighter and more vivid colors. I realized that any dark colors, especially if you are in a sewer for example is pitch black, you can't see anything. impossible to play. With the 660Ti, it seemed fine.

Digging around in the settings for Metro I saw the setting for HDR which reminded that I have it turned on in Windows so I tried turning it on in the game but it didn't help. I turned it off in windows and went back in game and all of the colors were fixed. I could see!

Obviously I came to the thought that HDR in windows might be the culprit so I ran another round of testing. with HDR off the stress test still had some minor stuttering but it was better than before. The video quality is improved, no screen tearing, artifacts etc... but it still doesn't make sense to me.

Does anyone else experience problems like this when using HDR? Is this a windows issue or a bad card?

Shouldn't this cpu/gpu combo be able to play everything on at least high, or ultra, and especially go through a stress test without stuttering? (even with HDR off I still experienced some in the stress test and it wasn't even on extreme)

SYSTEM SPECS
Board: Gigabyte x570 Elite (F11 bios, latest chipset driver)
CPU: Ryzen 7 2700X (no overclock)
GPU: EVGA 2070 Super Black Gaming (no overclock)
RAM: Gkill RipJaws (2x8GB) DDR4, XMP profile @ 3200mhz (tried disabling XMP and that didn't help. Tried one stick as well which didn't help either)
PSU: Rosewill Fortress-750
HDD: Seagate 4TB 7200 (tested fine in benchmarks and tried my SSD too)

DISPLAY
TCL 55p607 - 4K HDR 120HZ display connected via HDMI. The TV does show status of refresh rate, resolution and HDR10 in an OSD.
 
Last edited:
https://www.displayspecifications.com/en/model/33acc21

Is it a 2017 model TV? You only get 120Hz through internal motion interpolation. It's a 60Hz TV. It might be able to display the 120 Hz refresh rate being fed into it, but it could be skipping frames. Try the UFO test:

 
OK, but it's a 60Hz TV. It doesn't have motion interpolation like I thought. The 120 Hz is just marketing jargon which actually refers to the backlight's PWM dimming frequency.

To display HDR properly you have to get all the GPU, TV, and Windows settings right. First, go to your TV's settings and make sure HDMI 2.0 mode is enabled and Game mode is enabled. If your TV allows changing the input icon, then change it to "Computer" or "PC" if available. Next, go to nVidia control panel and use YPbPr with 4:4:4 chroma, 10 bpc, and "Full" color range. This should work for 1080p/60Hz/HDR. For 4k/60Hz/HDR it looks like you have to change bit depth to 8bpc to get it to work - this is due to HDMI 2.0 bandwidth limitation.

See the review on rtings for your TV, particularly the "Input Lag" section which has all the info on refresh rates and resolution:

https://www.rtings.com/tv/reviews/tcl/p-series-p607-2017

EDIT: I'm going to look further into the 120Hz thing with your TV because I'm curious about it now
 

TheMenace4Life

Reputable
Jan 11, 2020
19
1
4,515
" First, go to your TV's settings and make sure HDMI 2.0 mode is enabled and Game mode is enabled. If your TV allows changing the input icon, then change it to "Computer" or "PC"

-
This had to be done before windows would allow me to enable HDR, so it was the first thing I did.

Next, go to nVidia control panel and use YPbPr with 4:4:4 chroma, 10 bpc, and "Full" color range.

- I don't have those options at 4K or 1080. Here is what are the available choices:

1920x1080:

Desktop color depth: only choice is highest only (32bit)
output color depth: 8bpc or 12bpc
output color format: RGB, YCbCr422, YCbC444
output dynamic range: limited (greyed out, cant change unless I set color format to RGB then I can select full)

3840x2160(native):

Desktop color depth: only choice is highest only (32bit)
output color depth: 8bpc (greyed out, cant change)
output color format: RGB, YCbCr422, YCbC444, YCbCr420
output dynamic range: full or limited

Having it set at 4k with rgb, full, 8bpc, my tv goes into HDR mode and seems to be working fine. I just encounter all these issues when trying to play games.
 

TheMenace4Life

Reputable
Jan 11, 2020
19
1
4,515
@mortemas I think you may have been right, setting it back to 60HZ for all resolutions seems to allow me back into games and the video seemed fine but then after a bit of gameply I had issues again. I went ahead and turned HDR off in case that was an issue and everything seems to work fine so I wanted to play metro for a while and I could play for about 10 minutes (the most I've played since I started having these issues) and it crashes. no errors. Tried again several times, same thing. I thought it might be the game, because I've seen many others complain about crashing so I tried Bioshock infinite and I got about 10-15 minutes in and it crashed. This time the screen locked up, and I was not able to get back into windows, I had to restart. I am starting to wonder if this card is just causing issues. I'm going to stick the 660Ti back in and try bioshock and metro and report back.

How can I find out what is causing games to crash? Everything seems to be working fine with the computer except when I'm trying to play games.
 
OK. I had some more info about HDR, but I'm going to hold off on that for now because I don't think it is the heart of your problem here. Your hardware is definitely underperforming. Your 2070 Super should be getting a Heaven score near or above 4000 at 1080p, as shown here:

Gigabyte2070S_27-800x611.jpg
https://proclockers.com/reviews/video-cards/gigabyte-rtx-2070-super-gaming-oc-gpu-review/page/0/5

I don't think the CPU is a bottleneck, but we'll look into it. I believe you mentioned temps were ok, but only for GPU. Can you give a full account of both temps AND clocks for CPU and GPU under load? Did you select "Load Optimized Defaults" in BIOS when you first built the PC? Can you run a Userbenchmark test so we can see how your PC compares to similar hardware? We need an apples to apples comparison. If your SSD is a Samsung, I know they have their own drivers available - is that installed? When you switch cards, are you using DDU to uninstall the drivers first?
 
Last edited:

TheMenace4Life

Reputable
Jan 11, 2020
19
1
4,515
OK. I had some more info about HDR, but I'm going to hold off on that for now because I don't think it is the heart of your problem here. Your hardware is definitely underperforming. Your 2070 Super should be getting a Heaven score near or above 4000 at 1080p, as shown here:

Gigabyte2070S_27-800x611.jpg
https://proclockers.com/reviews/video-cards/gigabyte-rtx-2070-super-gaming-oc-gpu-review/page/0/5

I don't think the CPU is a bottleneck, but we'll look into it. I believe you mentioned temps were ok, but only for GPU. Can give a full account of both temps AND clocks for CPU and GPU under load? Did you select "Load Optimized Defaults" in BIOS when you first built the PC? Can you run a Userbenchmark test so we can see how your PC compares to similar hardware? We need an apples to apples comparison. If your SSD is a Samsung, I know they have their own drivers available - is that installed? When you switch cards, are you using DDU to uninstall the drivers first?

I will run the tests you mentioned tonight when I get home. For now, here is the userbenchmark I ran 2 days ago.

UserBenchmarks: Game 101%, Desk 84%, Work 88%
CPU: AMD Ryzen 7 2700X - 86.9%
GPU: Nvidia RTX 2070S (Super) - 114.1%
SSD: Corsair Neutron GTX 120GB - 76.2%
HDD: WD Green 1TB (2009) - 55.9%
HDD: Seagate Desktop HDD 4TB (2013) - 89.4%
HDD: WD Green 1TB (2009) - 43.8%
USB: General USB Flash Disk 8GB - 1.8%
RAM: G.SKILL F4 DDR4 3200 C16 2x8GB - 104.4%
MBD: Gigabyte X570 AORUS ELITE

When switching cards I was not running DDU, just using geforce experience to install them. Looks like they are both using the same driver as well. I did not select "use optimized defaults" in the bios. I left everything the way it is except turned on XMP profile (and tried it with it off as well). I do remember the GPU and CPU temps both being fairly consistent with each other under load. I haven't seen the cpu get over 60-65C but I'll run further testing as you requested.
 
Actually, the Userbenchmark numbers look pretty good based on what I'm seeing. A link to your Userbenchmark results would be most convenient, but I was able to search the site for each piece of hardware individually and everything is performing around where it should be. I'm just curious what the USB 8GB flash drive is for - if that's not the problem for some reason, then yeah it would seem like two main possibilities here. Something wrong with the GPU or something your system doesn't like about it. How old is your Rosewill PSU? I see a review on it from 2013. The 660 ti is a 150W card. The 2070 super is a 215W card. Not much more, but maybe enough to cause problems. Do you have the proper 8-pin and 6-pin PCIE power cables coming directly from your PSU, or are you using an adapter (or adapters)?
 

TheMenace4Life

Reputable
Jan 11, 2020
19
1
4,515
Actually, the Userbenchmark numbers look pretty good based on what I'm seeing. A link to your Userbenchmark results would be most convenient, but I was able to search the site for each piece of hardware individually and everything is performing around where it should be. I'm just curious what the USB 8GB flash drive is for - if that's not the problem for some reason, then yeah it would seem like two main possibilities here. Something wrong with the GPU or something your system doesn't like about it. How old is your Rosewill PSU? I see a review on it from 2013. The 660 ti is a 150W card. The 2070 super is a 215W card. Not much more, but maybe enough to cause problems. Do you have the proper 8-pin and 6-pin PCIE power cables coming directly from your PSU, or are you using an adapter (or adapters)?


The 8gb flash drive was to re-install windows when I did a clean install and I also used it to run memtest86. It was only installed for those reasons.

The power supply is from 2013, however I didn't use the computer for quite some time before upgrading, possibly a few years. Can I check it somehow? The psu does have an 8 pin connector and a 6 pin connector, both plugged in, no adapter.. I am going to try a different cable from the psu tonight just in case. I am getting some BSOD errors in windows reliability center.

here is the benchmark link
https://www.userbenchmark.com/UserRun/23594983
 
A quick check of PSU voltages can be found in the BIOS, usually under "PC Health" or similar, but that's just at idle. To test the PSU voltages under load use something like hwinfo to monitor 3.3v, 5v, and 12v, while stress testing the pc (both CPU and GPU). It won't tell us everything that could be wrong with it, but we can at least see if your voltages are within spec:

yiKk5.png


Beyond that, you'd have a few options regarding the PSU. At 7 years old, I'd personally replace it. It looks like it was a pretty decent PSU in its day based on the advertised specs and it seems to have gotten at least one good review which I saw on Tweaktown, although the testing was definitely less than thorough. Nowadays on several PC hardware sites, including Tom's, you'll find an exhaustive barrage of technical testing of PSUs which also includes actually opening it up and reviewing the electronic design and the individual components. So, here's your options regarding the PSU:

  1. Keep trying other things (BIOS updates, mobo driver updates, software updates. underclock, breadboard, etc)
  2. Test 2070 Super in a friend's PC that hopefully has equal or better hardware
  3. Keep PSU, exchange the 2070 Super (if you get same problem with new GPU, PSU is again suspect)
  4. Borrow a known good PSU or bring PC to repair shop so they can test with a known good PSU
  5. Buy a new PSU
Ah, I forgot to mention it but good to see you have already been to Windows Reliability Monitor. Any details on the BSODs? You can also check the Event Viewer for Critical, Error, and Warning events and report back here.

Wondering if maybe your Hard Drive(s) is(are) causing stuttering? I'm curious to see a retest of Heaven, since the article I found puts the 2070 super's score at around 4000. Maybe try to duplicate their test settings exactly.

Here's a test for 120Hz:
-Make sure you are at 1080p /120Hz in nVidia control panel
-Ensure that there is no scaling by typing Make Everything Bigger in the Windows start bar, pressing Enter, and selecting "100%" if Make Everything Bigger is not set to that already
-Open Chrome and ensure there is no scaling there either (zoom set to 100% in browser menu)
-Go to https://www.testufo.com/
-Change UFO test Speed to 1920 Pixels Per Second
-Use your phone to take a still picture of your screen like below (I used Burst mode on my iPhone to capture 5 or 10 frames to ensure I got the UFO at the right moment):
49376571853_2215117f94_k.jpg


You should also try 1080p/60Hz for comparison (change your setting to 60Hz in nVidia Control Panel, again set UFO test speed to 1920 Pixels Per Second):
49377026291_3db69accd2_k.jpg


Note that at 120Hz/120FPS the UFO images overlap tightly, with spacing of about 1/2 inch between each image (look at the left leg of the very first UFO image and compare it to the left leg of the next UFO image). If it's 60Hz/60FPS, then the overlap distance between each image is double at about 1 inch. These measurements are specifically for a 55" TV. Actually, I believe they are a bit off and I need to check with the folks at BlurBusters but it is close enough for identifying 60Hz vs 120Hz (I expected 0.4" and 0.8" spacing). You don't need to use a ruler. You can clearly see the difference in the above images. You'll note that at 60 Hz the dome of each UFO just touches up against the next one, whereas at 120Hz the edge of the next dome cuts almost down the middle of the previous dome. Identifying 60/120 this way visually (not with a ruler) is independent of screen size as long as you are at 1080p and the UFO test speed is set to 1920 pixels per second (assuming 1080p is a supported mode for your display).
 
Last edited:

TheMenace4Life

Reputable
Jan 11, 2020
19
1
4,515
@mortemas ok so oddly enough... I went home and tried another cable on the psu and then started testing games and everything seems to be fixed :-/ The overall feel is much better, it has been quite slow and buggy any time I have been running any graphics. I went ahead and tested several games, Bioshock, Gears 5, Forza, Warframe, all played perfectly on ultra settings once I reduced resolution down to 1440p, it struggled at 4K. I even tried with HDR turned on and it seemed to work fine (aside from some weird windows issues, like the screen flickers every time I open hdr settings). Interesting enough all these games had a steady 60FPS, which is something I hadn't seen before. In previous test FPS have been jumping all over the place. The only game that continued to crash was Metro, I really think that's a game issue however. The devs gave me a list of things to try to get it to stop crashing but none of them worked so I'll keep working with them. I ran the heaven benchmark again and on Ultra settings I got a just over 1550 score and on high I got over 4500. I ran a realbench stress test (15 minutes) and GPU max temp has 65C while the cpu max temp was 82C. I am using the stock prism cooler. I did get a handbrake instability error but tried again and it worked fine. After several hours in and out of several games, I didn't experience any glitches, crashes, bsod's nothing (aside from Metro). Reliability center has nothing except metro and oddly coretemp keeps crashing but compared to yesterday getting a list of events. I will report back if experience more issues on this but for today it seems resolved. Could it really be something wrong with the one cable from the PSU? It certainly was not a connection issue as I reinstalled that card 10X times and ensured both plugs were in completely until they clicked. Voltages are all within spec at this time.
 
Last edited:
Wow, that’s great that the problems are gone! When you say you tried a different cable on the PSU, do you mean the 120VAC mains cable? It’s definitely possible that a bad connection on any of the 3 wires of the power cable could cause problems.
 

TheMenace4Life

Reputable
Jan 11, 2020
19
1
4,515
Wow, that’s great that the problems are gone! When you say you tried a different cable on the PSU, do you mean the 120VAC mains cable? It’s definitely possible that a bad connection on any of the 3 wires of the power cable could cause problems.

Wow, that’s great that the problems are gone! When you say you tried a different cable on the PSU, do you mean the 120VAC mains cable? It’s definitely possible that a bad connection on any of the 3 wires of the power cable could cause problems.

Yeah it's really odd but I'll keep an eye on it. The cable I am referring to is the cable inside the pc, with the 6 and 8 pin plug for the gpu. There is 2 of them, coming from the psu and one wasn't being used so I tried it.
 
Ooooh, yeah, definitely could be a bad pin or something. An intermittent connection can absolutely cause problems. Arcing during high loads can cause loss of available power and also noise, and therefore all sorts of system errors. The GPU has the male connector. If you were using the same cable on both the 660 and 2070 and only had a problem with the 2070, it is quite possible that the thickness of the pins in the 2070 are just a hair smaller such that the female connector on the power cable might have poor contact only with the 2070. Just one possible explanation. Could also be a bad crimp of one of the female contacts, but that wouldn't explain why it was only a problem with the 2070 Super. At least you had an alternative. Both cards have 8-pin and 6-pin power connections, right?
 

TheMenace4Life

Reputable
Jan 11, 2020
19
1
4,515
Ooooh, yeah, definitely could be a bad pin or something. An intermittent connection can absolutely cause problems. Arcing during high loads can cause loss of available power and also noise, and therefore all sorts of system errors. The GPU has the male connector. If you were using the same cable on both the 660 and 2070 and only had a problem with the 2070, it is quite possible that the thickness of the pins in the 2070 are just a hair smaller such that the female connector on the power cable might have poor contact only with the 2070. Just one possible explanation. Could also be a bad crimp of one of the female contacts, but that wouldn't explain why it was only a problem with the 2070 Super. At least you had an alternative. Both cards have 8-pin and 6-pin power connections, right?


Whatever it is, I don't really want to troubleshoot the other cable, just going to cross my fingers it keeps working haha but no the 660 used two 6 pin connectors and the 2070 used a 6 and an 8.
 
That's very interesting. It is quite likely a problem with the sense pin on the "+2" part of the 6+2 connector. An intermittent connection there would make the system think there was only a 6-pin plug plugged into the 8-pin connector of the 2070. It would be irrelevant with the 660 ti because it didn't need the +2 plug, therefore no problem. This could also explain the 2070 Super's lower than expected score in Heaven since the system might have been power limiting the card because the bad connection at the sense pin would make it look like there were only 2 x 6-pin connectors plugged in (75W + 75W + (75W from the mobo slot) = 225W) instead of an 8-pin and a 6-pin (150W + 75W + (75W from the slot) = 300W total). This wouldn't cause a real deficit in power, it just makes the system think there's less power available.
 
Last edited:

TheMenace4Life

Reputable
Jan 11, 2020
19
1
4,515
I wanted to come back here and give an update in case anyone else has issues. After I thought all the issues were gone, games kept crashing. I rma'd the card. So I finally got a replacement card from EVGA. Stuck it in and I should've tested games first but I ran a stress test first and this time using realbench I decided to do a 15 min first, which it passed just like before but after doing more research I found that I really should run at least a 2hr so I did. System instability at 58 minutes. I never ran more than a 15 min test with the other card so it's likely I never would of found this issue. I ran it again and the same thing. I decided to pop out the cpu and figure out if there was some issue there because I remembered when I initially installed it, I got an error about the cpu fan (don't know why, it was working) so I took the cooler off and it popped out the cpu with it, they were stuck together because I think I had too much thermal paste on there and when I tried to get it off I think I got a little bit of paste on the pins of the cpu. I tried cleaning it but apparently not good enough. I couldn't see any to the naked eye but I took it to work and under the microscope cleaned it with alcohol 99% and a brush, there was definitely some on the pins. all of the sudden all problems are completely gone. I played gears 5 and metro for hours yesterday with no issues and after running 2hr stress test it passed no issues. I noted that the cpu is 3-4 degrees cooler at idle and sitting around 10-12 degrees cooler at 100% load. It was getting up to 82 degrees and now after 2 hour stress test max temp was 71. I am wondering if the first card was fine and I screwed up the cpu installation. but whatever it's working now just wanted to put this hear in case anyone else has issues to make sure they didn't do the same thing as me.