2021/08/04 11:08:30
radrokz
What kind of load did you use to reach that wattage? 
 
I own a rev 0.1 serial 2012 black lips since February and it's been working flawlessly (touching wood)
 
I've been told by many that if you deliver additional power to PCIe slots it's fine, my motherboard has a molex supply line for PCIe.
 
 
2021/08/04 11:33:31
Mienko
radrokz
What kind of load did you use to reach that wattage? 
 
I own a rev 0.1 serial 2012 black lips since February and it's been working flawlessly (touching wood)
 
I've been told by many that if you deliver additional power to PCIe slots it's fine, my motherboard has a molex supply line for PCIe.



Fair question.  I had a +100/+1000 OC and was running Kombuster to stress.  Note that this was only done specifically to check the PCIE power draw, since I wasn't sure from a comparative stance since I had 1.0 cards prior.  Getting ready to play some Cold War now, so will see what the real-world scenario looks like.
2021/08/04 11:59:55
arestavo
Mienko
G'day all,
On my 3rd card as I had 2 rev 1.0's go belly up on me.  The first RMA was a BNIB 1.0 that died after a couple days, but my 'new' card arrived today.  It's a rev 0.1 and I'm seeing PCIE power of ~80W when testing.  It's hard to find a definitive answer on whether this is acceptable or not (some say it's fine even though it's technically out of spec, while others chicken little it).  Things seem to be running okay.  Is there a consensus?  Anything I should be concerned about?  Using a Maximus XII Hero mobo, if that makes a difference.




One merely has to read the past few pages and you'll get your answer.
 
Since you haven't - here's a synopsis:
 
PCIE slot power draw max specification is 75W. GPU-Z (and I'm guessing other software programs as well) only reads 12V power draw, not 12V + 3V power draw. So a properly functioning, in spec card will show a max of 67W draw on the PCIE slot (this has been shown many times over with properly functioning cards), with an assumed ~8W added in for 3V for 75W max.
 
Some folks have come in and said that EVGA stated that 84W is fine (which would really be 92W total power draw on the PCIE slot). This may be the case for high end motherboards that are built with better traces, but 75W is the max and the PCIE standard. More than that and it is out of PCIE spec, and EVGA or anyone else saying otherwise is lying. Whether or not you find this acceptable is 100% up to you (maybe your card overclocks very well?), but it is grounds for RMA to get a card that can function within PCIE standards.
 
I really need to copy this to a text file, because typing it over and over again for people that don't bother to read the last few pages is getting tedious.
2021/08/04 12:42:06
Mienko
arestavo
 
 
One merely has to read the past few pages and you'll get your answer.
 
Since you haven't - here's a synopsis:
 
PCIE slot power draw max specification is 75W. GPU-Z (and I'm guessing other software programs as well) only reads 12V power draw, not 12V + 3V power draw. So a properly functioning, in spec card will show a max of 67W draw on the PCIE slot (this has been shown many times over with properly functioning cards), with an assumed ~8W added in for 3V for 75W max.
 
Some folks have come in and said that EVGA stated that 84W is fine (which would really be 92W total power draw on the PCIE slot). This may be the case for high end motherboards that are built with better traces, but 75W is the max and the PCIE standard. More than that and it is out of PCIE spec, and EVGA or anyone else saying otherwise is lying. Whether or not you find this acceptable is 100% up to you (maybe your card overclocks very well?), but it is grounds for RMA to get a card that can function within PCIE standards.
 
I really need to copy this to a text file, because typing it over and over again for people that don't bother to read the last few pages is getting tedious.




You know what they say about making assumptions. . .  Anyhow, I have been reading this thread (and reddit) and opinions are pretty all over the place as to the "safety" aspect of the power draw.  Nobody is arguing that it's technically out of spec.  I said it, and then you basically said the same thing (albeit, with a bit more snark than needed).  The long/short is that I've already had to replace the card twice via cross-ship and was just looking for some additional guidance on the best path forward.
2021/08/04 13:00:26
arestavo
Mienko
 
You know what they say about making assumptions. . .  Anyhow, I have been reading this thread (and reddit) and opinions are pretty all over the place as to the "safety" aspect of the power draw.  Nobody is arguing that it's technically out of spec.  I said it, and then you basically said the same thing (albeit, with a bit more snark than needed).  The long/short is that I've already had to replace the card twice via cross-ship and was just looking for some additional guidance on the best path forward.


If you keep having to replace cards, looking elsewhere in your system might be in order - the power supply, and conversely the power supply cables if they're aftermarket (someone even reported a problem that cropped up with a cablemods CPU cable with a new 3090), are a prime culprit of high end 30 series cards having problems. Having three separate PCIE power plugs, and no daisy chained ones, can also help significantly.
 
Vertically mounted cards with PCIE risers have also been an issue, with some reporting that the PCIE 4.0 risers cause an issue and the PCIE 3.0 risers are better / installing the card normally helped.
 
However, without knowing all of your computer specs, all I/we can do to help is regurgitate what we've seen in the hopes that one of them applies.
2021/08/04 13:51:46
Mienko
arestavo
Mienko
 
You know what they say about making assumptions. . .  Anyhow, I have been reading this thread (and reddit) and opinions are pretty all over the place as to the "safety" aspect of the power draw.  Nobody is arguing that it's technically out of spec.  I said it, and then you basically said the same thing (albeit, with a bit more snark than needed).  The long/short is that I've already had to replace the card twice via cross-ship and was just looking for some additional guidance on the best path forward.


If you keep having to replace cards, looking elsewhere in your system might be in order - the power supply, and conversely the power supply cables if they're aftermarket (someone even reported a problem that cropped up with a cablemods CPU cable with a new 3090), are a prime culprit of high end 30 series cards having problems. Having three separate PCIE power plugs, and no daisy chained ones, can also help significantly.
 
Vertically mounted cards with PCIE risers have also been an issue, with some reporting that the PCIE 4.0 risers cause an issue and the PCIE 3.0 risers are better / installing the card normally helped.
 
However, without knowing all of your computer specs, all I/we can do to help is regurgitate what we've seen in the hopes that one of them applies.


Thanks man.  Appreciate it.  For reference Asus Maximus XII Hero, CableMod cables, AX1600i PSU (no daisy chaining), no riser at the moment (but I have both a Phanteks 3.0 & a LinkUp 4.0 and planned on using one of these).  2080Ti works completely fine without any issues whatsoever.  Have the rev 0.1 3090 in right now and it's been chugging along for the last 4-5 hours or so with no issues thus far.
 
That said, I went ahead and emailed EVGA via the 3090OC address to get their take on the power draw for this new card.  If they say to RMA it again, I will.  If they say it's good and then things catch fire or melt as a result, at least I'll have the paper trail for legal considerations.  lol
 
Thanks again!
 
Quick notes on the 2 failures:
  • First one happened a few days after installing the EK block.  Additionally, this was on the 3.0 Phanteks riser.  Not saying either of those are related, just full disclosure.  Pretty sure the failure happened with a bad PX1 bios flash.  The card was working fine, but then upon shutting down after a gaming session the fans ramped to 100% like it was going into flashing mode and then never came back.  No LEDs or anything.  Removing the riser and cable swapping made no difference.
  • Second one started giving a red LED over the middle power connector.  No riser, no bios flash.  Swapped cables and whatnot, but that middle red LED just kept staring at me.
2021/08/04 15:13:20
LegallyBlind85
3x ft s3
2021/08/04 18:02:05
Mienko
arestavo
 
If you keep having to replace cards, looking elsewhere in your system might be in order - the power supply, and conversely the power supply cables if they're aftermarket (someone even reported a problem that cropped up with a cablemods CPU cable with a new 3090), are a prime culprit of high end 30 series cards having problems. Having three separate PCIE power plugs, and no daisy chained ones, can also help significantly.
 
Vertically mounted cards with PCIE risers have also been an issue, with some reporting that the PCIE 4.0 risers cause an issue and the PCIE 3.0 risers are better / installing the card normally helped.
 
However, without knowing all of your computer specs, all I/we can do to help is regurgitate what we've seen in the hopes that one of them applies.

Shot that note over to the 3090OC email and they're getting it situated with another swap.  Guess that answers that.  :-)  Thanks for the assist!
2021/08/05 00:37:26
Sam1987
This load balancing issue just doesn't get resolved it's 10 months ... Personally first and last time i'm buying from them the whole influencer promo suggesting 900 watt bios validated the whole situation further in my opinion. Don't care if lower power bios is the solution to stay in safe margin. Could have bought the cheapest 3090 card instead. If you get the card back and have the same issue and want the max power from bios. And have a decent motherboard you may look into shunting the pcie and double shunting 8 pins (maybe with powerboost) etc.   
2021/08/05 06:57:47
arestavo
Sam1987
This load balancing issue just doesn't get resolved it's 10 months ... Personally first and last time i'm buying from them the whole influencer promo suggesting 900 watt bios validated the whole situation further in my opinion. Don't care if lower power bios is the solution to stay in safe margin. Could have bought the cheapest 3090 card instead. If you get the card back and have the same issue and want the max power from bios. And have a decent motherboard you may look into shunting the pcie and double shunting 8 pins (maybe with powerboost) etc.   


See post #2 of this thread for EVGA Jacob's consolidated responses for the OC 3090 special RMA program for the load balancing issue (the rev 1.0 cards have a new voltage controller to help resolve the issue).

Use My Existing Forum Account

Use My Social Media Account