I think it is just the games you are playing and some of the glitches that are game specific.  I would keep trying other games out.  play crysis it is still a beast of a game on GPU's and if you do not see anything in that game I would think you card is fine.
 
you may also want to check the forums of the games you are having glitches in to see if others have the same issue.

Looks like a game glitch to me. If you're not seeing artifacts inside Unigine Valley 1.0 or Unigine Heaven 4.0 your card is fine.

RyougaLolakie
Oh ok. :) I get worrisome sometimes and I'm sorry that I've bother you. Guess I'll try out other games other than World of Warcraft and Team Fortress 2. It's weird how my card ran heaven and valley didn't and get artifacts while Team Fortress 2 did. It doesn't make any sense.

Since I underclock it for test, will it damage the gpu?

You're not bothering me. No, underclocking the card will not damage the GPU.

Didn't mean to bump this thread but I ran a 30 minute test on evga oc scanner with artifact detection on both memory and core burner. So far, no artifacts. I also ran a 3DMark demo on steam using these tests offered and I didn't see any glitches nor artifacts. So I'm guessing it was either the game nor the drivers. Although it's still strange that Team Fortress 2, an old game that's gets updated frequently shouldn't get those artifacts/glitches especially with new cards. To be honest, there hasn't been a single thread of someone who has that same issue especially with GTX 970/980 cards.

Now I'm start to feel there are some glitches in Assassin's Creed IV Black Flag. Are these screenshots artifacts or glitch? I mean I ran the OC artifact scanner both the core and burner for 30 mins and didn't find any artifacts so what the heck is going on here? I really need your guys opinion on this. These things occur rarely even though the gameplay was smooth.

Also I ran memtestg80 (recent open source version) and found no errors. Sorry that I got a bit scared I get confused about these things. Kinda need an enlightenment on how artifacts form which can vary from funky polygons to pixelated mess.

Edit: man I feel stupid. These last 2 didn't look like polygon artifacts at all compared to what I seen on Google when I searched polygon artifact. The first one, looks like the texture is missing. Guess I should've researched more. That 320.18 driver must've screwed a lot of the Gpus. LOL!

And to more testing, I've run vmt on direct x testing and it showed no errors.

Then again, I'll wait for the next driver update. I heard it'll fix some issues on the maxwell cards.

Suggestions...
 

 
Take a chill pill bro. Your cards are fine. Games have glitches. The artifacts in scene 10 inside Heaven 4.0 are normal. When you start seeing a ton of artifacts inside Heaven 4.0 & Valley 1.0 then start the RMA process. 

And btw, do these glitches caused by drivers or game specific? If it is a driver bug, then which stable version should I rollback on?

Well, I played tf2 while underclocked and I get a TDR error from the black screen. That also happen when I reset it to default and ran heaven and I got a black screen on the second loop, at least no hard locks. I'll be doing an rma on this one. As soon as I go into safe mode, I saw those little light blue horizontal lines on the top left edge of the screen. This is my first time I ever rma a faulty product. I'm really scared about the output of the replacement.

I'm also with the same black screen issue on gtx 970 sc, have tried everything nothing resolved, already opened the rma a week ago and has not been approved. evga never again.

Left4Dead 2



 
 

RyougaLolakie
@vladada011 Wow, so much pixelated / blurry shadows. I remember that my gtx 560 had those pixelated shadows in some of the games including skyrim.

Well, the replacement card came by today and I've tested Heaven benchmark for 10 loops. So far, I don't see any artifacts, crashes, black screens, or anything yet. I'll test some more in games but I'm getting sleepy. Kinda nervous on how TF2 worked out for this new replacement.

Oh yeah...During using Heaven benchmark, my gpu core boosted up to 1367ghz, same speed as my previous one. I gotta ask, do most gtx 970 sc cards shipped those kinds of turbo boost speeds that are like above product specifications? To be honest, I'm not complaining. ^^;

Noo, that's perfect normal. 
I know very well when some card present bad picture... This is absolutely normal. 1920x1080 resolution never look good on pictures.
Even I played that with last drivers, game is old but everything is clearly visible, I didn't done some mods or tweaks in graphics. 
See on someone else picture... Maybe is that possible to fix with high resolution shadows but I didn't try...
 

 
My picture is normal I know that. I immediately recognize when something is not OK.
Look RyougaLolakie's pictures, that's problematic.
I had on ASUS GTX580 similar problem, but than it's not only shadow or some thing, everything is weird like render is not finish, only 75%.
Do you know how Borderland 2 have bug example and if you open box need 0.5 sec before picture become clear, in Borderland 2 you fix that in cfg file change 4-5 numbers on 0, something like that but some cards are always give such messed picture, and you can't do nothing.

After much being stuck in the middle, I've been testing my haware, so far I've done testing...

Cpu: Intel burn test (maximum setting 10 times) = no crashes and stable
Ram: Memtest86+ 5.01 (tested with all of the ram) = no errors

And my gpu passed all of the test in 3DMark demo with no crashes and so forth.

So lemme tell you the update on my gpu. Well, I did put the image preview in control panel to quality and it looks like I don't see any artifacts when playing team fortress 2, but right now I've experienced driver crashes which it can happen around 30 minutes to like 4 hours. Sometimes it doesnt crash during 3 to 4 hour session of tf2.This is kinda like my original problem when I have my gtx 970 where tf2 froze for a sec then the screen goes black, where as I can easily access to task manager and closed tf2. Last time I got a game crashed, I open task manager and closed tf2. Then I checked on reliability monitor where it doesn't show video hardware (ie TDR error), but in the event viewer, event 1001 reports a live kernel event happened at the time of the crash.

It lead me to wonder if those crashes could relate to the power supply or the motherboard. I guess ill go out and get a multimeter but I've never done it before so I don't know if it was worth doing it for an inexperienced person like me. Plus, I don't recall seeing local pc repair shop in area so that's a bummer, but I'll search for one.

Here's the video of the artifacts before I set the image quality to quality.

http://youtu.be/EO5DWzmT32o

If any other suggestions, I would be appreciated. :)

bsmegreg
It is likely a driver problem in which case an RMA would not help you. What you can do is you should have the memory dump file (.dmp) under c:\windows\minidump
What could help you is to use Debugging Tools for Windows to view the dump file and it should point to a particular driver file which you may be able to replace.  
Steps
1) Download Debugging Tools for Windows 
2) Open the program and select to open a dump file
3) Open the dump file and analyze which files its having issues with
4) Is it a driver file? Is it an OS file? If so, an RMA may not help the problem... You might be able to replace the file with googles help or maybe someday NVIDEA will decide to patch it.

I open the debugging tools for windows and here's i found from the live kernel events watchdog folder:
 

Microsoft (R) Windows Debugger Version 6.3.9600.17298 AMD64
Copyright (c) Microsoft Corporation. All rights reserved.

Loading Dump File [C:\Windows\LiveKernelReports\WATCHDOG\WD-20150220-0031.dmp]
Mini Kernel Dump File: Only registers and stack trace are available
Symbol search path is: *** Invalid ***
****************************************************************************
* Symbol loading may be unreliable without a symbol search path. *
* Use .symfix to have the debugger choose a symbol path. *
* After setting your symbol path, use .reload to refresh symbol locations. *
****************************************************************************
Executable search path is:
*********************************************************************
* Symbols can not be loaded because symbol path is not initialized. *
* *
* The Symbol Path can be set by: *
* using the _NT_SYMBOL_PATH environment variable. *
* using the -y <symbol_path> argument when starting the debugger. *
* using .sympath and .sympath+ *
*********************************************************************
Unable to load image ntoskrnl.exe, Win32 error 0n2
*** WARNING: Unable to verify timestamp for ntoskrnl.exe
*** ERROR: Module load completed but symbols could not be loaded for ntoskrnl.exe
Windows 7 Kernel Version 7601 (Service Pack 1) MP (8 procs) Free x64
Product: WinNt, suite: TerminalServer SingleUserTS Personal
Machine Name:
Kernel base = 0xfffff800`01e55000 PsLoadedModuleList = 0xfffff800`02099890
Debug session time: Fri Feb 20 00:31:58.460 2015 (UTC - 5:00)
System Uptime: 0 days 6:39:45.960
*********************************************************************
* Symbols can not be loaded because symbol path is not initialized. *
* *
* The Symbol Path can be set by: *
* using the _NT_SYMBOL_PATH environment variable. *
* using the -y <symbol_path> argument when starting the debugger. *
* using .sympath and .sympath+ *
*********************************************************************
Unable to load image ntoskrnl.exe, Win32 error 0n2
*** WARNING: Unable to verify timestamp for ntoskrnl.exe
*** ERROR: Module load completed but symbols could not be loaded for ntoskrnl.exe
Loading Kernel Symbols
...............................................................
................................................................
.......................
Loading User Symbols
Mini Kernel Dump does not contain unloaded driver list
************* Symbol Loading Error Summary **************
Module name Error
ntoskrnl The system cannot find the file specified
You can troubleshoot most symbol related issues by turning on symbol loading diagnostics (!sym noisy) and repeating the command that caused symbols to be loaded.
You should also verify that your symbol search path (.sympath) is correct.
*** WARNING: Unable to verify timestamp for watchdog.sys
*** ERROR: Module load completed but symbols could not be loaded for watchdog.sys
*******************************************************************************
* *
* Bugcheck Analysis *
* *
*******************************************************************************
Use !analyze -v to get detailed debugging information.
BugCheck 117, {fffffa80538f2010, fffff880031a7978, 0, 0}
*** WARNING: Unable to verify timestamp for mssmbios.sys
*** ERROR: Module load completed but symbols could not be loaded for mssmbios.sys
*** WARNING: Unable to verify timestamp for dxgkrnl.sys
*** ERROR: Module load completed but symbols could not be loaded for dxgkrnl.sys
*** WARNING: Unable to verify timestamp for nvlddmkm.sys
*** ERROR: Module load completed but symbols could not be loaded for nvlddmkm.sys
***** Kernel symbols are WRONG. Please fix symbols to do analysis.
*************************************************************************
*** ***
*** ***
*** Either you specified an unqualified symbol, or your debugger ***
*** doesn't have full symbol information. Unqualified symbol ***
*** resolution is turned off by default. Please either specify a ***
*** fully qualified symbol module!symbolname, or enable resolution ***
*** of unqualified symbols by typing ".symopt- 100". Note that ***
*** enabling unqualified symbol resolution with network symbol ***
*** server shares in the symbol path may cause the debugger to ***
*** appear to hang for long periods of time when an incorrect ***
*** symbol name is typed or the network symbol server is down. ***
*** ***
*** For some commands to work properly, your symbol path ***
*** must point to .pdb files that have full type information. ***
*** ***
*** Certain .pdb files (such as the public OS symbols) do not ***
*** contain the required information. Contact the group that ***
*** provided you with these symbols if you need this command to ***
*** work. ***
*** ***
*** Type referenced: nt!_KPRCB ***
*** ***
*************************************************************************
*************************************************************************
*** ***
*** ***
*** Either you specified an unqualified symbol, or your debugger ***
*** doesn't have full symbol information. Unqualified symbol ***
*** resolution is turned off by default. Please either specify a ***
*** fully qualified symbol module!symbolname, or enable resolution ***
*** of unqualified symbols by typing ".symopt- 100". Note that ***
*** enabling unqualified symbol resolution with network symbol ***
*** server shares in the symbol path may cause the debugger to ***
*** appear to hang for long periods of time when an incorrect ***
*** symbol name is typed or the network symbol server is down. ***
*** ***
*** For some commands to work properly, your symbol path ***
*** must point to .pdb files that have full type information. ***
*** ***
*** Certain .pdb files (such as the public OS symbols) do not ***
*** contain the required information. Contact the group that ***
*** provided you with these symbols if you need this command to ***
*** work. ***
*** ***
*** Type referenced: nt!_KPRCB ***
*** ***
*************************************************************************
*************************************************************************
*** ***
*** ***
*** Either you specified an unqualified symbol, or your debugger ***
*** doesn't have full symbol information. Unqualified symbol ***
*** resolution is turned off by default. Please either specify a ***
*** fully qualified symbol module!symbolname, or enable resolution ***
*** of unqualified symbols by typing ".symopt- 100". Note that ***
*** enabling unqualified symbol resolution with network symbol ***
*** server shares in the symbol path may cause the debugger to ***
*** appear to hang for long periods of time when an incorrect ***
*** symbol name is typed or the network symbol server is down. ***
*** ***
*** For some commands to work properly, your symbol path ***
*** must point to .pdb files that have full type information. ***
*** ***
*** Certain .pdb files (such as the public OS symbols) do not ***
*** contain the required information. Contact the group that ***
*** provided you with these symbols if you need this command to ***
*** work. ***
*** ***
*** Type referenced: nt!_KPRCB ***
*** ***
*************************************************************************
Probably caused by : nvlddmkm.sys ( nvlddmkm+13e978 )
Followup: MachineOwner
---------

 
So what I'm saying is thats caused by the nvidia drivers, but somehow it also shows dxgkrnl.sys and mssmbios.sys into the picture. What does it mean?
 
 
 

If thats the case, I have similiar crash when i was using gtx 560 (which it used to be fine since 2011 until i upgraded to gtx 970 which causes the 0x117 crash). While my 970 card was in RMA, I was using a gtx 560. At that time, I was playing Osu and suddenly got the same crash as my gtx 970, but with different games. When the crash started, my gtx 560 fan's rpm was suddenly dropped to 0 but it recovered. Now, I got my rma gpu back and it does the same crash as my first gtx 970.

Now, I think I may have the conclusion that my PSU might be the cause since my PSU has survived numerous power surges and in some way it was aging despite it was 3 years old since i bought the pc from Ibuypower. I remember the time where the psu made some buzzing sound when it comes through the power plug on the psu. It goes away when I tightned the plug. Not sure if its related but i would like to add additional information.

What is the best quality PSU on the market today?

Ill check those out, Sajin. Ill reply back if the issue persisted. :)

RyougaLolakie
What is the best quality PSU on the market today?

In the 1200w category the EVGA 1200w P2 would be the best unit. http://www.jonnyguru.com/...op=Story6&reid=389