This is amazing from EVGA, but I'm having a weird issue, nothing major as everything still works fine.

Before the bios my clocks were stable between 2055-2085 and sometimes 2100, anything 2100 and over for a prolonged period crashed.

With this new bios I now have stability between 2100 and 2135 (no crashing), but my frames take a beating after 15 minutes or so of gameplay (2070-2085 zero issues, my floor increased by 20 mhz).

I have zero overclock on the memory and all sliders are set to 100%, my temps are hovering just below 69 degrees.

Anyone else with the same issue?

I'm not sure what's going on, but I did some independent tests where just my vRAM, then just my core were OCed, to see where I can get on both before combining the two and adjusting as needed... 
I plugged in +130 core +800 vRAM and got a lower score than the Core being LOWER by itself, and nearly the same score as just the vRAM OC, and GPUz is not showing a Reason for Performance Cap, other than VRel/VOp at the very beginning of Port Royal and that goes away after about 10 seconds.  Temperatures are fine, the Core clock is reading where it should be for that OC... but the result is off.  I had all background processes closed, and closed Precision after setting the clocks & setting fans to 80%.  Ran GPUz the second run to see if there was a PerfCap reason, first run was with nothing in the background.
Anyone else got any ideas?
 
vRAM +800 only: 12093 https://www.3dmark.com/pr/405082
Core +100 only: 12271 https://www.3dmark.com/pr/405357
Core +130/ vRAM +800: 12157 https://www.3dmark.com/pr/412550

All of those were done at same system settings other than the GPU OCs, to avoid anything else influencing the score.  And you can see from the FutureMark readout that my average GPU Core clock was 47MHz higher for the +130 vs +100 runs... yet the score is worse? lol  Is it possible with both OCs applied the vRAM is getting warm & ECC is kicking in?  I guess I'll have to leave Precision X1 open & on  HWMonitor to keep an eye on vRAM temps & see if they get warm warm, during the previous +800 only test they only saw about 58C during the test though.

robvoip
This is amazing from EVGA, but I'm having a weird issue, nothing major as everything still works fine.

Before the bios my clocks were stable between 2055-2085 and sometimes 2100, anything 2100 and over for a prolonged period crashed.

With this new bios I now have stability between 2100 and 2135 (no crashing), but my frames take a beating after 15 minutes or so of gameplay (2070-2085 zero issues, my floor increased by 20 mhz).

I have zero overclock on the memory and all sliders are set to 100%, my temps are hovering just below 69 degrees.

Anyone else with the same issue?

Try reducing the voltage slider to 0. I've always seen my scores drop by setting 100 with my 3090 FTW3 Ultra. Also, start testing out a memory OC after you have that other problem sorted - I bet you can get a conservative +500 on the VRAM (although it's not guaranteed to work, just likely).

jilljill2002

hayabusa09
How the heck do you guys keep your cards < 70C? Max fan speed? Sorry if noob question..im a noob :-/

I am curious too, I have to undervolt to 0.875mv to keep the card under 70C

I got a 3080 FE to hold me over until I get my FTW3. I also had to undervolt quite a bit (to as low as 0.8v/1800 MHz) to get to around 70C. This corresponds to around 240-250W TDP (as reported on GPU-Z). I have tried 0.856v/1875 MHz but the temp would slowly creep up to 76-78C (that corresponds to around 270-280W TDP). Your card’s temp is going to depend on the TDP you are pulling with the undervolt setting.

hayabusa09
How the heck do you guys keep your cards < 70C? Max fan speed? Sorry if noob question..im a noob :-/

Magic!
 
Real answer: Airflow & decent ambients.  Even under OCs my GPU core never hits 70C, highest I've seen is 68C during OC testing in Superposition with the GPU fans capped at 75% speed.
If you're seeing higher temperatures than 65C without an OC, something is going on in terms of airflow in your case that is not getting the heat out well enough to prevent the card from recycling it's own heat.
This is my first air-cooled non-Founders Edition GPU in ... 4-5 years?  So I'm re-learning what these things need to stay happy as I go.  Before this I had 2 Hybrid eVGA cards, 2 Founders Edition Blower-style cards (Titan Xp's), and before that 2 R9 295x2s which came with a liquid AIO loop on them.  There's an advantage to those blower-style coolers, they kick all the air out of the back of the case & don't really output much if any in to the case itself.  The aftermarket cooler style like the 3080 FTW3 Ultra has vents almost all of it's exhaust in to the case, and as a result, you either get recycled hot air for it if your airflow is insufficient, or you have an artificial space heater for your room if your airflow is sufficient. :-D

Dabadger84
I'm not sure what's going on, but I did some independent tests where just my vRAM, then just my core were OCed, to see where I can get on both before combining the two and adjusting as needed... 
I plugged in +130 core +800 vRAM and got a lower score than the Core being LOWER by itself, and nearly the same score as just the vRAM OC, and GPUz is not showing a Reason for Performance Cap, other than VRel/VOp at the very beginning of Port Royal and that goes away after about 10 seconds.  Temperatures are fine, the Core clock is reading where it should be for that OC... but the result is off.  I had all background processes closed, and closed Precision after setting the clocks & setting fans to 80%.  Ran GPUz the second run to see if there was a PerfCap reason, first run was with nothing in the background.
Anyone else got any ideas?
 
vRAM +800 only: 12093 https://www.3dmark.com/pr/405082
Core +100 only: 12271 https://www.3dmark.com/pr/405357
Core +130/ vRAM +800: 12157 https://www.3dmark.com/pr/412550

All of those were done at same system settings other than the GPU OCs, to avoid anything else influencing the score.  And you can see from the FutureMark readout that my average GPU Core clock was 47MHz higher for the +130 vs +100 runs... yet the score is worse? lol  Is it possible with both OCs applied the vRAM is getting warm & ECC is kicking in?  I guess I'll have to leave Precision X1 open & on  HWMonitor to keep an eye on vRAM temps & see if they get warm warm, during the previous +800 only test they only saw about 58C during the test though.

yes e cc is kicking in. I can push my memory to 900 but I get sho **** FPS. Anything higher that 600 I lose score points

TheGuz4L

Dabadger84
I'm not sure what's going on, but I did some independent tests where just my vRAM, then just my core were OCed, to see where I can get on both before combining the two and adjusting as needed... 
I plugged in +130 core +800 vRAM and got a lower score than the Core being LOWER by itself, and nearly the same score as just the vRAM OC, and GPUz is not showing a Reason for Performance Cap, other than VRel/VOp at the very beginning of Port Royal and that goes away after about 10 seconds.  Temperatures are fine, the Core clock is reading where it should be for that OC... but the result is off.  I had all background processes closed, and closed Precision after setting the clocks & setting fans to 80%.  Ran GPUz the second run to see if there was a PerfCap reason, first run was with nothing in the background.
Anyone else got any ideas?
 
vRAM +800 only: 12093 https://www.3dmark.com/pr/405082
Core +100 only: 12271 https://www.3dmark.com/pr/405357
Core +130/ vRAM +800: 12157 https://www.3dmark.com/pr/412550

All of those were done at same system settings other than the GPU OCs, to avoid anything else influencing the score.  And you can see from the FutureMark readout that my average GPU Core clock was 47MHz higher for the +130 vs +100 runs... yet the score is worse? lol  Is it possible with both OCs applied the vRAM is getting warm & ECC is kicking in?  I guess I'll have to leave Precision X1 open & on  HWMonitor to keep an eye on vRAM temps & see if they get warm warm, during the previous +800 only test they only saw about 58C during the test though.

yes e cc is kicking in. I can push my memory to 900 but I get sho **** FPS. Anything higher that 600 I lose score points

Depends on the benchmark etc, they are all different.
 
If you work up slowly and stop once the score drops and drop the clock back to the prior setting.
 
I do the core 1st, max it out working up slowly.
 
Then do the memory until I see the score drop.
 
 

SoldierRBT
Does the new 450W BIOS let you control all 3 fans properly in MSI Afterburner?

No it does not.

I score more with mem at 200 than 500. Think its the ECC as said above. It still runs, but performance suffers, people see its not artifacting and think its ok, when really ECC is just correcting it I think.

Thanks, this kinda worked. The frames don't tank anymore, but anything over 2085 on the seems to run worse.

I'll start testing memory and see what gains I can get there.