Overclock.net banner

4281 - 4300 of 4348 Posts

·
Registered
Joined
·
47 Posts
[*]Random crashes to desktop
[*]Textures get fixed on the screen and go with me while I walk around the map
[*]Some bad artifact spiking happened twice
[*]Hard crash, where the screen goes black and but some textures stay on the screen in blue and red colors.
[*]Some BSODs where the reason is a file that starts with nv
[/LIST]

I was kind of thinking "damn...my card is broke" ... but it seems lots of people are having issues with the drivers.

did 2 hours of testing using 3D Mark stability tests and everything was fine, both stock or OC'led.


One of the examples of what was my screen the other day ... perfect visibility :p

This "artifact" was very dynamic, it kind of looked like the laser effects you see at the nightclubs or trance shows ... depending on where I looked it would change form and color, always with this "spiking effect" animation.




After a fresh windows install, DDU and NVClean Install ... even tried installing the default ( w/ experience app included ) because who knew there wasn't a catch ... No change.

Ok, The artifacts don't show that often, but crashes are on a daily basis. 3 days ago I actually got a piece of a rock texture that fixed on the screen and I took it everywhere I was going ... lol, just weird"]
Haven't happened to me mate (I have ReBAR enabled as well), but I mostly play Warzone, Skyrim modded and Genshin Impact, maybe is an issue with Apex.


Hey Buddy,

On the gaming OC when you opened it up was there a connector for the RGB. Reason I ask is I want to turn it off but RBG Fusion is the worst bit of software I have seen in a long time. I'm running my rig in a home cinema so trying to turn off all RGB.

Thanks
Besides my TUF 3080, I have a Gigabyte Gaming OC PRO3060Ti, which effectively has a connector for the RGB and others for the fans, if that helps (it seems the cooler is the same one, just the 3080 one is more large and width)
 

·
Registered
Joined
·
3,601 Posts
Yeah it does. I have the RGB logo removed from the shroud right now and have it double sided taped to the card lol. All the RGB has it's own plug.

Once my EVGA Hybrid FTW3 cooler shows up tomorrow or Thursday I will slap that on and see if I can keep the RGB logo somehow haha.

I am soo curious to see if the Hybrid FTW3 cooler fits properly on this Gaming OC because if it does it means it will fit any 3080 so people can effectively buy a AIO for their 3080's pretty easily. It ain't cheap tho. I spend €169 on it and will probably end up spending at least another €40 or so on VRAM and VRM heatsinks and decent thermal tape.. but in the end it will be SO worth it.
 
  • Rep+
Reactions: Jimmy2Shoes

·
Facepalm
Joined
·
9,874 Posts
Anyone having "major" issues with the latest Nvidia drivers and ReBar active? Since I updated and activated Rebar I had the following issues ( mostly on Apex since it's the only game I play daily )

( This happens on default settings, not OC )

  • Random crashes to desktop
  • Textures get fixed on the screen and go with me while I walk around the map
  • Some bad artifact spiking happened twice
  • Hard crash, where the screen goes black and but some textures stay on the screen in blue and red colors.
  • Some BSODs where the reason is a file that starts with nv

I was kind of thinking "damn...my card is broke" ... but it seems lots of people are having issues with the drivers.

did 2 hours of testing using 3D Mark stability tests and everything was fine, both stock or OC'led.


One of the examples of what was my screen the other day ... perfect visibility :p

This "artifact" was very dynamic, it kind of looked like the laser effects you see at the nightclubs or trance shows ... depending on where I looked it would change form and color, always with this "spiking effect" animation.

View attachment 2486398


After a fresh windows install, DDU and NVClean Install ... even tried installing the default ( w/ experience app included ) because who knew there wasn't a catch ... No change.

Ok, The artifacts don't show that often, but crashes are on a daily basis. 3 days ago I actually got a piece of a rock texture that fixed on the screen and I took it everywhere I was going ... lol, just weird
Have not seen this on a 3090. I did not compare with rebar on or off in this game.
Try deleting your shader cache in c:\program data\nv_cache , by first turning off shader cache in the NVCP, then browsing to that file location, deleting all the files in it, then reboot, enable shader cache in the NVCP and have it rebuild the cache.
 
  • Rep+
Reactions: Pedros

·
Registered
Joined
·
1,278 Posts
Have not seen this on a 3090. I did not compare with rebar on or off in this game.
Try deleting your shader cache in c:\program data\nv_cache , by first turning off shader cache in the NVCP, then browsing to that file location, deleting all the files in it, then reboot, enable shader cache in the NVCP and have it rebuild the cache.
Gonna try that, thank you.

Just tried how COD:MW and it was crashing like crazy until I forced it to use DX11. No crashes afterward. Apex, on the other hand, although I've not experienced any weird artifacts lately ... now and then I get a crash ... pfff :x
 

·
Registered
Joined
·
47 Posts
Battlefield 4 1080p 150% res scale all max is one of the tests, it stays just under the power limit, and Cyberpunk 2077 1080P RT Psycho with DLSS on Performance is as well. The higher you put the DLSS (so the lower the resolution) the less power draw it has. It looks horrible but is a great stability test as it does use full ray tracing and DLSS calculations and complex scenes.
To add, today tested BF4 at 150% scale maxed, managed to get 2160-2175Mhz at 1.1V if I'm not powerlimited, at 2190Mhz it crashes, so I guess it's pretty decent for a TUF card.

Wish I wasn't power limited man haha
 

·
Registered
Joined
·
3,601 Posts
Great news! The EVGA FTW3 Hybrid cooler showed up today and it fits for the most part on the Gigabyte Gaming OC. The waterblock and the VRAM plate fit without any modifications and I can be pretty sure they will both fit any model 3080 just fine. The power plug for the pump is quite different but if you reverse it and cut the locking tab off it does fit in the fan power slot and it does work with PWM RPM and power as long as you reverse it in the socket for the fans. The pinout is the other way around.

All I gotta do now is find a way to cool the VRM's which probably ends up in sacrificing the Gaming OC stock cooler (and heat pipes unfortunately) and just using both VRM pieces as normal heatsinks without heat pipes. The VRM pads aren't connected to the heat pipes anyway. Or I have to glue on some copper Alpha cool heatsinks..

Temperature tests and pics coming later. So far the core was idling at 21c so.
 

·
Registered
Joined
·
3,601 Posts
Great news! The EVGA FTW3 Hybrid cooler showed up today and it fits for the most part on the Gigabyte Gaming OC. The waterblock and the VRAM plate fit without any modifications and I can be pretty sure they will both fit any model 3080 just fine. The power plug for the pump is quite different but if you reverse it and cut the locking tab off it does fit in the fan power slot and it does work with PWM RPM and power as long as you reverse it in the socket for the fans. The pinout is the other way around.

All I gotta do now is find a way to cool the VRM's which probably ends up in sacrificing the Gaming OC stock cooler (and heat pipes unfortunately) and just using both VRM pieces as normal heatsinks without heat pipes. The VRM pads aren't connected to the heat pipes anyway. Or I have to glue on some copper Alpha cool heatsinks..

Temperature tests and pics coming later. So far the core was idling at 21c so.
Hopefully this Akasa thermal tape is strong enough to not drop the pretty heavy heatsinks. I ended up harvesting the heatsinks from the EVGA Hybrid FTW3 shroud. They fit with a bit of trimming on a sanding block and angle grinder. Also drilled mounting holes for my Arctic backplate and trimmed it so that it will fit with a few small srews and thumbnuts and has a proper mount now. For fans it has push-pull Cooler Master MF120 ARGB's on the rad and the stock EVGA FTW3 ARGB fans are mounted to the card on the bottom for airflow over VRM / VRAM / Pump. All controlled through a SATA powered fan controller that gets it's RPM and PWM signal from the pump fan wiring so that I don't overload the GPU's fan controller.

Temps: This is with CP2077 1.12 1080P all max, RT Psycho, DLSS Quality and just running around doing some side missions. (+105 core +1400 memory)

2486589


So, all in all dropped 10c core and hotspot (even tho the difference is still quite big) and 20c memory with this mod. And, best of all, it went from a windtunnel to absolute silence. Fans are barely spinning up at 1060RPM. They go as high as 2000RPM if I have to..

Pics coming from my phone, takes a while to upload.
 
  • Rep+
Reactions: ducegt

·
Registered
Joined
·
235 Posts
To add, today tested BF4 at 150% scale maxed, managed to get 2160-2175Mhz at 1.1V if I'm not powerlimited, at 2190Mhz it crashes, so I guess it's pretty decent for a TUF card.
Btw, ASUS bioses show +30-50Mhz higher target Mhz than other manufacturers bioses.

Target? If you look clocks at @ Afterburner, GPUZ or another Manufactures OC app your GPU clocks, those clocks aren't real Mhz, them are only clocks that card try to run, and 95% any of cards cant reach it.

You, and everyone else must get an effective clockspeed, and currently HWInfo, i think, it's only software that show to us those real clocks what card is running.

What about Asus "cheat" clocks, my MSI SuprimX original bios, show example Afterburner 2100Mhz and my Effective clocks are 2080Mhz.

I use Asus Strix OC bios an couple weeks, and i "get" more speed, I get 2130Mhz on afterburner, but my effective clocks was same than with MSI bios, 2080Mhz.

You can see that all reviews since last autumn, "on all" reviews Asus cards get +30-50Mhz higher clocks on overclocking tests than another factories cards, still FPS could be same than example MSI or Evga cards.

It also depends how to overclocking , slider or V/F curve. V/F curve overclocking you get less effective clocks than basic slider overclocking /w same "request" clocks = less true power.

Example, my card V/F curve OCd 2100Mhz, Effective clocks ~2050Mhz.
But on slider OCd 2100Mhz, Effective clocks ~2090Mhz.

 
  • Rep+
Reactions: fray_bentos

·
Registered
Joined
·
3,601 Posts
I noticed the same where on my card the EVGA XC3 BIOS had consistently higher effective clocks compared to the Gigabyte Gaming OC one. I did go back to Gaming OC BIOS now but a different newer version (not the rebar one) and now it's basically the same. At 1995-2010 reported I see about 1960-1980 effective while slamming the power limit as you can see in my above HWINFO64 monitoring (1979 effective at 1995 reported).

Pics of the mod I did:

Block mounted using original EVGA FTW3 mounting hardware + VRAM plate.
2486619


Arctic Accelero IV backplate drilled and modded to mount to the original backplate holes with Arctic Accelero IV mounting screws and thumb nuts.

Stock EVGA radiator fans mounted to the card for VRAM and VRM airflow and card mounted in the case. Yes the 8 pin converter is ugly.
2486621


All done, rads in, fans in, card fully working and RGB as well. Even saved the stock Gigabyte RGB logo. Speed control on both the rad and card fans through MSI AB working.
2486622
 

Attachments

·
Registered
Joined
·
47 Posts
Btw, ASUS bioses show +30-50Mhz higher target Mhz than other manufacturers bioses.

Target? If you look clocks at @ Afterburner, GPUZ or another Manufactures OC app your GPU clocks, those clocks aren't real Mhz, them are only clocks that card try to run, and 95% any of cards cant reach it.

You, and everyone else must get an effective clockspeed, and currently HWInfo, i think, it's only software that show to us those real clocks what card is running.

What about Asus "cheat" clocks, my MSI SuprimX original bios, show example Afterburner 2100Mhz and my Effective clocks are 2080Mhz.

I use Asus Strix OC bios an couple weeks, and i "get" more speed, I get 2130Mhz on afterburner, but my effective clocks was same than with MSI bios, 2080Mhz.

You can see that all reviews since last autumn, "on all" reviews Asus cards get +30-50Mhz higher clocks on overclocking tests than another factories cards, still FPS could be same than example MSI or Evga cards.

It also depends how to overclocking , slider or V/F curve. V/F curve overclocking you get less effective clocks than basic slider overclocking /w same "request" clocks = less true power.

Example, my card V/F curve OCd 2100Mhz, Effective clocks ~2050Mhz.
But on slider OCd 2100Mhz, Effective clocks ~2090Mhz.

Oh I know, the effective clocks were 2150-2160Mhz depending of the scene in the game (actually played a little, BF4 story isn't bad at all to my surprise lol)

I just set an offset (+255) and then just moved the 2175Mhz dot from 1.093V to 2160Mhz, since I didn't want to go below 1.1V if I weren't power limited.

For example in my 3060Ti (Gigabyte) my max OC was 2145Mhz but effective clocks were 2120-2110Mhz in average or something like that

Here is a pic of HWiNFO64 and BF4 in-game (RTX 3060 = 3060Ti, have to change the name lol)
2486634
2486636
 

·
Registered
Joined
·
202 Posts
Does somebody have a Resizable BAR Suprim X or Strixx OC BIOS to cross flash on a Trio?
The Techpowerup database only seems to have the ReBAR version for the FE.
 

·
Registered
Joined
·
235 Posts
Search it from TPU unofficial bios database.
 
  • Rep+
Reactions: BluePaint

·
Registered
Joined
·
139 Posts
@BluePaint here is my, uploaded via GPU-Z

Strange is that when I download my file from techpowerup it is binary different than the one that I save to my pc with gpu-z.
So below I putting additional downloads:
Saved with gpu-z: 391.5 KB file on MEGA
Update file coming from MSI live update: 391.4 KB file on MEGA
 
  • Rep+
Reactions: BluePaint

·
Registered
Joined
·
2,783 Posts
It also depends how to overclocking , slider or V/F curve. V/F curve overclocking you get less effective clocks than basic slider overclocking /w same "request" clocks = less true power.

Example, my card V/F curve OCd 2100Mhz, Effective clocks ~2050Mhz.
But on slider OCd 2100Mhz, Effective clocks ~2090Mhz.

I think MSI did great with the Suprim X bios and how it handles overclocking with undervolting.

I get the same effective clocks regardless of my overclocking method, Offset slider, 1 Point VF Curve, 8-10 point VF Curve with 15Mhz increments or even the recommended 3 voltage points for 1 frequency. All will result in the same effective clock for me. This is probably what made it hard for me to understand which is the better as all 3 methods give me the same effective clock is is around 10-20Mhz less than my target clock
 

·
Registered
Joined
·
830 Posts
Who cares about a gap in the effective clocks? It's performance that matters (unless you want to balance energy use I guess.) For my card, I use a VF curve for more performance than offsetting the entire curve.
 

·
Registered
Joined
·
2,783 Posts
Who cares about a gap in the effective clocks? It's performance that matters (unless you want to balance energy use I guess.) For my card, I use a VF curve for more performance than offsetting the entire curve.

Because that is where the performance comes from, a core clock of 2100 vs 2000 if both have the same 1980Mhz effective will perform at the same speed.
 

·
Registered
Joined
·
830 Posts
Because that is where the performance comes from, a core clock of 2100 vs 2000 if both have the same 1980Mhz effective will perform at the same speed.
While we're just making up numbers, how the cards boost, and theorizing... let's just say OC A has a 120mhz gap and OC B has only a 20mhz gap like in your example. If OC A can be pushed to 2120 (2000 effective) and OC B is maxed at 2000 (1980 effective), then it's easy to understand which one is faster. All I'm saying is the gap arguably has nothing to do with "how people should be overclocking." Use a benchmark to determine which method nets the best performance because you can't use a single clock speed at specific period of time to tell what's faster during dynamic situations.

To spell it out, I have my card set to max at 2175 1.1v with custom VF curve because 1) it's stable and I push as much as I can, 2) it actually does boost that high momentarily in benches and games, and 3) there's even some games it'll hold ~2160 like Half Life Alyx. However, I can't offset the whole curve to reach 2175 1.1v because it's unstable and I could careless if I could run the whole curved reduced at 2100 so the effective gap would be lessened because it nets less performance. So basically, whatever nets higher effective clocks is indeed important, but the gap is not.
 

·
Registered
Joined
·
2,783 Posts
So basically, whatever nets higher effective clocks is indeed important, but the gap is not.
That is exactly my point, the effective clock drives the performance not the gap hence the example of 2100Mhz core vs 2000Mhz core when both are at 1980Mhz effective clock. You just wanted a more complex explanation, but you do you.
 
4281 - 4300 of 4348 Posts
Top