Grayblade

It's mean unstable undervolt for me sadly.

So I've been trying to save my GPU undervolt to MSI Afterburner and have same settings after reboot. However after I reboot GPU clock goes either up or down by 15 MHz. Open Afterburner click on profile shows me correct settings and after I apply clock speed jumps 15 MHz. What is this madness? So far I've tryed delete profiles in afterburner and afterwards remove manually profile files left over. Installed latest beta and then tryed latest stable, same behavior with both. Am I missing something or is that software buggy?

Depends how long you planning to keep the card and your money situation. You can check performance comparisons on Youtube, just search for 3070 TI vs 3080. Just keep in mind that RTX 3080 is ~21% faster (from Techpowerup) and has 2GB more VRAM. Personally I'd pick RTX 3080 over RTX 3070 Ti any day, but I'm biased cause have RTX 3080. On my 3080 Far Cry 6 runs great on 1440p. Pretty sure RTX 3070 Ti will run Far Cry 6 great too, but without high res. textures on 1440p.

You are right on that. Went to see her PC and PSU lacks PCIe cables so can't even put decent GPU in there. So really need full upgrade as in processor, ram, motherboard, GPU and PSU. So in the end we decided to postpone GPU upgrade to next year. Thank you all for the input.

Thanks for input. I'll get miner GPU. Now I have 2 choices: ASUS TUF GTX 1660 TI Dual for 110€ and ASUS TUF RTX 2060 Dual for 175€. Which one is better value?

I'm searching used GPU for my mom. She is huge Diablo fan and got invite for D4 beta, but current iGPU A8-7650K doesn't cut it anymore. Meaning it failed to launch the game. Currently there are flood of mining GPU-s coming to market. I found ASUS TUF-GTX1660TI-6G-EVO-GAMING for 105€ that seems decent price for that card. However I'm not sure what risks are there when buying GPU that has been used for mining. Seller claims that GPU has been used for 2 months but I'm not sure about that. What are good questions to ask seller for the card conditions that are telling? What do I need to take account when buying GPU that has been used for mining?

Thanks for the input. Card does run warmer side with silent bios MAX temps are for GPU is 77C, hotspot 87C and memory about 90C. Temps should still be within spec. When running OC profile same artifacts happen and temps are bit lower. Yea, I guess best is to use card with stock setting and wait for the card to die. Edit: updated temps

Hello! I have RTX 3080 Gaming OC that has given me trouble. When playing more demanding games like Cyberpunk 2077 or Guardians of the Galaxy I get massive artifacts (attached some screenshots). In Destiny 2 and some other games I get some random artifact or flicker that does away fast. Also on desktop appeared black dots that went away after reboot and some random flicker that is rare. However running synthetic benchmarks like 3D Mark picture is fine and some games don't produce artifacts either. I have DDU reinstall drivers that did not help. When I undervolt/clock my RTX 3080 to 900mV and 1860MHz Cyberpunk 2077 does not produce artifacts. Not sure what is happening. Card is still under warranty till end of next year at least. Shop where I got it from tests with synthetic benchmark to verify GPU functionality. No point going there unless I can say for certain what is broken. I've sent ticket to Gigabyte but haven't heard from them yet. Any one with experience has idea what could be wrong? Should I just use the card until it dies or keep it undervolted?

Thanks for elaborating. I got it now. Now that I think about it makes sense that GPU creates the load.

Thank you for the explanation. If I'm understanding it right Corsair PSU provided different load to my GPU and possibly better PCIe connection. That resulted magnets in my GPU to flip out less and make less audible noise. As what exactly was the cause is more complex, since there are a lot of factors that could influence outcome. I thought it was the higher wattage and different cables. Corsair has in cable capacitors that were bit had to work with but I find them nice bonus for reliability.

Hi! Can anyone explain why did changing PSU greatly reduced my GPU coil whine? Backstory: I have R7 5800x + RTX 3080 system that was powered by good quality PSU Fractal Design ION+ 600p 660W Platinum that was about year old. Games worked great no problems except really loud coil whine. I have Define R4 silent case that is on the floor next to my computer and when I'm in Doom Eternal or Halo Infinite menu I could hear coil whine over menu music. I google around and see no good solution other than standard turn v-sync on which I've already done. So Black Friday comes around and I see Corsair RM850x 850W PSU on sale and started thinking if PSU change could reduce my GPU coil whine. It felt right thing to do so I pulled the trigger and got the Corsair PSU. To my surprise coil whine is pretty much gone. If put my ear close to GPU I can hear it, but if close the case I can't hear coil whine anymore. What a relief! Now I'm starting to wonder was it higher wattage reduced coil whine or is there something else in play? Anyone knowledgeable can tell me? Also weird thing with PCIe cables. Couldn't do double cables because I couldn't plug in 2nd cable into one of the GPU PCIe slots, but pigtail end fits right in. Both PCIe cables had that same problem. Didn't want to force it and broke my 3080 so I ended up using one cable for my GPU. It seems to work just fine.

I have NT-H1 thermal paste where they recommend 1 big dot in the middle https://noctua.at/pub/media/blfa_files/manual/noctua_nt-h1_3.5g_manual_en.pdf Maybe 1 dot doesn't work for 5800X and I need to do the 5 dots pattern. I'll wait and see what Noctua replies and then reseat the cooler.

Thanks. Out of 4 I can try adding 2nd fan and dive into BIOS when I feel like tuning. If even 360 AIO struggles with temps then I guess it's normal that NH-D15S struggles too I found AMD slide where they say it's by design and no user interaction needed. Diving into BIOS is something I wanted to avoid, but seems like best and cheapest way to get temps down. My CPU fan curve is set at 85C fan goes 100% speed that is 1500 RMP. I've thought of replacing my case fans and ordered one PHANTEKS PH-F140MP V2 to see if it's any good and how I like the sound of it. It seemed to have good price to performance with 15€. For buying more expensive fans I'll wait for new 140 Noctuas.

I rather not mess with this stuff. I don't even know where to begin. Out of curiosity ran CB R23 while Ryzen Master opened. I saw spike to 1,43V on peak core voltage and then it settled to 1,37V. Is that too much?

For thermal paste I put pea size as per Noctua instructions. I did think it's too little but instructions said too much will hurt performance. I have fresh Windows install with latest BIOS and chipset drivers. As for power plan Ryzen 5000 doesn't have one. Source this AMD post: https://community.amd.com/t5/processors/missing-power-plans-on-amd-ryzen-5000-series/m-p/421774/highlight/true#M33892 Actually had to check my fans. I knew case fans where OK but need to double check CPU fan too. I think I'll e-mail Noctua support and see what they come up with.