Anyone having an issue where past 80c on core fans all ramp up to 100% (sound-wise) and two out of three fans loos RPM output?

93f0rc3

Notchy44

cerealkeller

Notchy44

cerealkeller
What version of precisionX are you using, unless you’re using Afterburner. The voltage curve doesn’t work at all for me.

Im using Precision X 1.1.1, I had 0 issues using the V curve but it seems to lower targeted power ratings for me and increase the +core. Maybe some people that are having the low power issue are using the built in Auto VF Power Curve Tuner?

It’s weird that everything is working for you and not for some of us.
What’s your part number, it’s on the box, and the card too I would assume. Mine is
24G P5-3987-KR
Goddamn it! There has been a revision change!! The one listed on their website is 3988 now, whatever one Notchy has is the proper revision I think. The same damn thing happened with the 980 Classified cards and we all got screwed, those of us with the 2nd and 3rd rev., and EVGA never resolved the issue.

I am rocking a 24G- P5 - 3987 - KR, Same as you!

About your card @Notchy44 doing all the magic you want it to, could you maybe please check the revision of your card, like in the pic below, where the arrow points and post back?
Maybe there is an easier way to check the revision, than to take out your card, but this is the most trustworthy way I could think of :D
 

Sure my Rev is 0.1 and the model number is 24G- P5 - 3987 - KR. Please let me know if this helps at all.

Notchy44

93f0rc3

Notchy44

cerealkeller

Notchy44

cerealkeller
What version of precisionX are you using, unless you’re using Afterburner. The voltage curve doesn’t work at all for me.

Im using Precision X 1.1.1, I had 0 issues using the V curve but it seems to lower targeted power ratings for me and increase the +core. Maybe some people that are having the low power issue are using the built in Auto VF Power Curve Tuner?

It’s weird that everything is working for you and not for some of us.
What’s your part number, it’s on the box, and the card too I would assume. Mine is
24G P5-3987-KR
Goddamn it! There has been a revision change!! The one listed on their website is 3988 now, whatever one Notchy has is the proper revision I think. The same damn thing happened with the 980 Classified cards and we all got screwed, those of us with the 2nd and 3rd rev., and EVGA never resolved the issue.

I am rocking a 24G- P5 - 3987 - KR, Same as you!

About your card @Notchy44 doing all the magic you want it to, could you maybe please check the revision of your card, like in the pic below, where the arrow points and post back?
Maybe there is an easier way to check the revision, than to take out your card, but this is the most trustworthy way I could think of :D
 

Sure my Rev is 0.1 and the model number is 24G- P5 - 3987 - KR. Please let me know if this helps at all.

Same, Im running Rev 0.1 (there arent probably other revisions tbh) and Bios version I got card with is : .15

slovak_killer

Notchy44

93f0rc3

Notchy44

cerealkeller

Notchy44

cerealkeller
What version of precisionX are you using, unless you’re using Afterburner. The voltage curve doesn’t work at all for me.

Im using Precision X 1.1.1, I had 0 issues using the V curve but it seems to lower targeted power ratings for me and increase the +core. Maybe some people that are having the low power issue are using the built in Auto VF Power Curve Tuner?

It’s weird that everything is working for you and not for some of us.
What’s your part number, it’s on the box, and the card too I would assume. Mine is
24G P5-3987-KR
Goddamn it! There has been a revision change!! The one listed on their website is 3988 now, whatever one Notchy has is the proper revision I think. The same damn thing happened with the 980 Classified cards and we all got screwed, those of us with the 2nd and 3rd rev., and EVGA never resolved the issue.

I am rocking a 24G- P5 - 3987 - KR, Same as you!

About your card @Notchy44 doing all the magic you want it to, could you maybe please check the revision of your card, like in the pic below, where the arrow points and post back?
Maybe there is an easier way to check the revision, than to take out your card, but this is the most trustworthy way I could think of :D
 

Sure my Rev is 0.1 and the model number is 24G- P5 - 3987 - KR. Please let me know if this helps at all.

Same, Im running Rev 0.1 (there arent probably other revisions tbh) and Bios version I got card with is : .15

I have seen a few 3090 FTW3 pics on the net, where the revision is 0.0
Maybe someone can confirm if the revision starts @ 0.0?

First off, I agree with Slovak from a few pages back. All this hate towards EVGA over this issue is hardly warranted. The cards do what they say they do in advertising. The reason some are angry is because generally EVGA exceeds expectations. Now there is a sense of entitlement. This is the company that sent me a GTX Titan as a warranty replacement to my 780 Ti. Recently, after having pushed the living crap out of my PSU over my 5 years of ownership I started getting some undervoltage on my 3.3v rail and they sent me a brand new SuperNova 1600 T2 replacement.
 
RE: Frame Chaser - sad. I think he might have FAS to be honest with you.
 
RE: Card revision - my buddy's card says rev 0.1 and "evga 227" whatever that means.
 
Finally, I have a question about the shunt mod that blew the fuse from several pages back. I'm still waiting for the fuse replacements that I ordered (a long with a $300 1000W Digital Rework Hot Air Station from Amazon) but I was wondering if anyone can think of a good reason why I can't just take out the broken fuse and then solder the connection points together and go fuseless like the Asus Strix version of the card? Would this work to get the card up and running at least?

PadinnPlays
I'm still wondering if its a temperature thing. Since you are near 60c it makes sense it might be willing to draw more. I never see above 450 (usually 440), but my temperatures are in mid70s. Lowest I can get with side panel off is around 67.

Me too...
 
@ Notchy44
Dont believe GPU-Z to 100%
With the XC3-BIOS GPU-Z shows only 332 Watts "Board Power Draw"
But my energy meter draws 100Watts more power, than the 500W-BIOS in Timespy
 
 
Again (3rd try)
Is it possible to buy „LDAT + PCAT“ from NVIDIA?
And how much are these tools?
I need in particular "PCAT“ to make accurate measurements

Dwayne_Johnson

PadinnPlays
I'm still wondering if its a temperature thing. Since you are near 60c it makes sense it might be willing to draw more. I never see above 450 (usually 440), but my temperatures are in mid70s. Lowest I can get with side panel off is around 67.

Me too...
 
@ Notchy44
Dont believe GPU-Z to 100%
With the XC3-BIOS GPU-Z shows only 332 Watts "Board Power Draw"
But my energy meter draws 100Watts more power, than the 500W-BIOS in Timespy
 
 
Again (3rd try)
Is it possible to buy „LDAT + PCAT“ from NVIDIA?
And how much are these tools?
I need in particular "PCAT“ to make accurate measurements

 
Oh I dont only use Gpu-z I run a few power monitors ie HWInfo64 , Also I have a smart plug as well, but the Nvidia built in overlay one seems to be pretty solid for the 3000's series as well. With my Port royal scores and Time spy scores im 100% sure its working at 500watts regardless of all my power readings lol.

slovak_killer

Notchy44

Dwayne_Johnson

PadinnPlays
I'm still wondering if its a temperature thing. Since you are near 60c it makes sense it might be willing to draw more. I never see above 450 (usually 440), but my temperatures are in mid70s. Lowest I can get with side panel off is around 67.

Me too...
 
@ Notchy44
Dont believe GPU-Z to 100%
With the XC3-BIOS GPU-Z shows only 332 Watts "Board Power Draw"
But my energy meter draws 100Watts more power, than the 500W-BIOS in Timespy
 
 
Again (3rd try)
Is it possible to buy „LDAT + PCAT“ from NVIDIA?
And how much are these tools?
I need in particular "PCAT“ to make accurate measurements

 
Oh I dont only use Gpu-z I run a few power monitors ie HWInfo64 , Also I have a smart plug as well, but the Nvidia built in overlay one seems to be pretty solid for the 3000's series as well. With my Port royal scores and Time spy scores im 100% sure its working at 500watts regardless of all my power readings lol.

Bios is messing up all the readings, no matter which program u use if the source is glitched nothing correct will come out.

I have no reason not to believe even GPU-z as I have tested total power output from the wall with the 500 watt bios and without. This bios is rock solid at 498 avg watts on (GL) furmark - MSI on all hardware sensors. Plus once again my scores are in line with stunt modded and water cooled 3090. This bios for me is not giving me any false readings.

Notchy44

slovak_killer

Notchy44

Dwayne_Johnson

PadinnPlays
I'm still wondering if its a temperature thing. Since you are near 60c it makes sense it might be willing to draw more. I never see above 450 (usually 440), but my temperatures are in mid70s. Lowest I can get with side panel off is around 67.

Me too...
 
@ Notchy44
Dont believe GPU-Z to 100%
With the XC3-BIOS GPU-Z shows only 332 Watts "Board Power Draw"
But my energy meter draws 100Watts more power, than the 500W-BIOS in Timespy
 
 
Again (3rd try)
Is it possible to buy „LDAT + PCAT“ from NVIDIA?
And how much are these tools?
I need in particular "PCAT“ to make accurate measurements

 
Oh I dont only use Gpu-z I run a few power monitors ie HWInfo64 , Also I have a smart plug as well, but the Nvidia built in overlay one seems to be pretty solid for the 3000's series as well. With my Port royal scores and Time spy scores im 100% sure its working at 500watts regardless of all my power readings lol.

Bios is messing up all the readings, no matter which program u use if the source is glitched nothing correct will come out.

I have no reason not to believe even GPU-z as I have tested total power output from the wall with the 500 watt bios and without. This bios is rock solid at 498 avg watts on (GL) furmark - MSI on all hardware sensors. Plus once again my scores are in line with stunt modded and water cooled 3090. This bios for me is not giving me any false readings.

Notchy44

slovak_killer

Notchy44

Dwayne_Johnson

PadinnPlays
I'm still wondering if its a temperature thing. Since you are near 60c it makes sense it might be willing to draw more. I never see above 450 (usually 440), but my temperatures are in mid70s. Lowest I can get with side panel off is around 67.

Me too...
 
@ Notchy44
Dont believe GPU-Z to 100%
With the XC3-BIOS GPU-Z shows only 332 Watts "Board Power Draw"
But my energy meter draws 100Watts more power, than the 500W-BIOS in Timespy
 
 
Again (3rd try)
Is it possible to buy „LDAT + PCAT“ from NVIDIA?
And how much are these tools?
I need in particular "PCAT“ to make accurate measurements

 
Oh I dont only use Gpu-z I run a few power monitors ie HWInfo64 , Also I have a smart plug as well, but the Nvidia built in overlay one seems to be pretty solid for the 3000's series as well. With my Port royal scores and Time spy scores im 100% sure its working at 500watts regardless of all my power readings lol.

Bios is messing up all the readings, no matter which program u use if the source is glitched nothing correct will come out.

I have no reason not to believe even GPU-z as I have tested total power output from the wall with the 500 watt bios and without. This bios is rock solid at 498 avg watts on (GL) furmark - MSI on all hardware sensors. Plus once again my scores are in line with stunt modded and water cooled 3090. This bios for me is not giving me any false readings.

Idk if we are talking about same thing, all im saying is with XC3 bios flashed im sure all power related readings are false.

cerealkeller

Tripod1990
One thing I noticed is that you can increase power consumption by setting the core voltage slider to 100%. This is not surprising but still interesting. With default voltage slider (at 0%) I peak around 460W, with the slider at 100% I can go up to 490W. Still not 500W but higher. I also get a 15mhz core clock bump even with identical core offset (from 2085mhz to 2100mhz). To me this looks like the cards, for some reason, does not go into higher points, on voltage/clock curve. You can force it to some extend by pushing the voltage slider, which basically manually unlocks some higher points on the curve. But the question is why the cards doesn't use hit those points on their own (and even with 100% I am still short of 500W).
 
Maybe it is because the card hits max voltage limit before hitting max power limit. Since some people were able to hit 500W with better cooling, it might be related to this, since you can push more current (and watts) at the same voltage if temperature is lower. Voltage = current x resistance, where resistance increase by temperature, so higher the temp gets, card will reduce current to maintain core voltage.  

To be fair, I haven’t been touching that because it appeared to do nothing, I’ll have to check that...

Can confirm that setting voltage offset to 100% will give u better boost because of slightly higher vcore, we are talking about 30-40mV cca though.

cerealkeller

slovak_killer

cerealkeller

Tripod1990
One thing I noticed is that you can increase power consumption by setting the core voltage slider to 100%. This is not surprising but still interesting. With default voltage slider (at 0%) I peak around 460W, with the slider at 100% I can go up to 490W. Still not 500W but higher. I also get a 15mhz core clock bump even with identical core offset (from 2085mhz to 2100mhz). To me this looks like the cards, for some reason, does not go into higher points, on voltage/clock curve. You can force it to some extend by pushing the voltage slider, which basically manually unlocks some higher points on the curve. But the question is why the cards doesn't use hit those points on their own (and even with 100% I am still short of 500W).

Maybe it is because the card hits max voltage limit before hitting max power limit. Since some people were able to hit 500W with better cooling, it might be related to this, since you can push more current (and watts) at the same voltage if temperature is lower. Voltage = current x resistance, where resistance increase by temperature, so higher the temp gets, card will reduce current to maintain core voltage.  

To be fair, I haven’t been touching that because it appeared to do nothing, I’ll have to check that...

Can confirm that setting voltage offset to 100% will give u better boost because of slightly higher vcore, we are talking about 30-40mV cca though.

How high is your voltage going, mine hits 1.081 without being boosted, which based on the last two generations would have been over-volt

1063mV at 0% voltage slider, 1094mV at 100% voltage slider

Notchy, have you been setting your voltage slider to max? And did you flash manually using nvflash or did you use the BIOS linked on the first page of this thread? Also, what driver are you using? I’m trying to eliminate anything I think may be affecting the rest of us by replicating your setup. Did you do a fresh install of windows when you installed your card, or just slapped it in and went from there?

slovak_killer

cerealkeller

Tripod1990
One thing I noticed is that you can increase power consumption by setting the core voltage slider to 100%. This is not surprising but still interesting. With default voltage slider (at 0%) I peak around 460W, with the slider at 100% I can go up to 490W. Still not 500W but higher. I also get a 15mhz core clock bump even with identical core offset (from 2085mhz to 2100mhz). To me this looks like the cards, for some reason, does not go into higher points, on voltage/clock curve. You can force it to some extend by pushing the voltage slider, which basically manually unlocks some higher points on the curve. But the question is why the cards doesn't use hit those points on their own (and even with 100% I am still short of 500W).

Maybe it is because the card hits max voltage limit before hitting max power limit. Since some people were able to hit 500W with better cooling, it might be related to this, since you can push more current (and watts) at the same voltage if temperature is lower. Voltage = current x resistance, where resistance increase by temperature, so higher the temp gets, card will reduce current to maintain core voltage.  

To be fair, I haven’t been touching that because it appeared to do nothing, I’ll have to check that...

Can confirm that setting voltage offset to 100% will give u better boost because of slightly higher vcore, we are talking about 30-40mV cca though.

slovak_killer

cerealkeller

Tripod1990
One thing I noticed is that you can increase power consumption by setting the core voltage slider to 100%. This is not surprising but still interesting. With default voltage slider (at 0%) I peak around 460W, with the slider at 100% I can go up to 490W. Still not 500W but higher. I also get a 15mhz core clock bump even with identical core offset (from 2085mhz to 2100mhz). To me this looks like the cards, for some reason, does not go into higher points, on voltage/clock curve. You can force it to some extend by pushing the voltage slider, which basically manually unlocks some higher points on the curve. But the question is why the cards doesn't use hit those points on their own (and even with 100% I am still short of 500W).

Maybe it is because the card hits max voltage limit before hitting max power limit. Since some people were able to hit 500W with better cooling, it might be related to this, since you can push more current (and watts) at the same voltage if temperature is lower. Voltage = current x resistance, where resistance increase by temperature, so higher the temp gets, card will reduce current to maintain core voltage.  

To be fair, I haven’t been touching that because it appeared to do nothing, I’ll have to check that...

Can confirm that setting voltage offset to 100% will give u better boost because of slightly higher vcore, we are talking about 30-40mV cca though.

Yep it doest help much, especially in terms of real world performance due to higher temps (it doesn't make it worse, but the improvement is negligible). But in my case it, it made some higher clock offsets stable. With voltage slider at 0%, max core OC I can achieve is +110mhz, with voltage at 100% I can get to +135mhz. This wont help with anything for real world applications, but for benchmark purposes it still gives a noticeable bump. It is interesting that ampere (or at least 3090) could be one of the few never gen cards that benefit from higher core voltage.  In pascal or Turing, you would hit power limit before coming close to core voltage limit, in ampere it looks like you hit core voltage limit before power limit, hence overvolting helps a bit. 
 

Is there a reason why the Strix OC performs better than FTW3 Ultra in timespy, despite the same GPU/V-RAM clock and CPU? (400-500 more Graphics-Score)
 

arestavo

slovak_killer
 
 
REV 0.0 might be/are probably pre-production versions, review samples etc. But yea lets see if someone has one.

0.1 and EVGA227 here, and my card is one that doesn't like to pull more than 440W average ever.

That's what mine is as well