Overclock.net banner

221 - 240 of 382 Posts

·
Facepalm
Joined
·
8,623 Posts
yeah then it's truly 300W, not affected by paint resistance.
You should be careful with that. Do you know what wire gauge your cables are? if it's 16 aug, 300 watts should be safe.
I know the Seasonic 12 pin cable is good up to 600W.
 

·
Registered
Joined
·
93 Posts
yeah then it's truly 300W, not affected by paint resistance.
You should be careful with that. Do you know what wire gauge your cables are? if it's 16 aug, 300 watts should be safe.
I know the Seasonic 12 pin cable is good up to 600W.
not sure, its an EVGA 850G2 PSU. I had to turn off the overclock on my 10900k...and still when benching, I have seen as much as 750W power usage. I obviously need to upgrade my PSU. Can you explain to me theoretically what the max power consumption from an 8 pin PCIE power cable is? I know the safe number is 150W, but I have been pulling almost 300W benching my card all day, and it's working just fine. They cables are cold to the touch also.
 

·
High Clocker
Joined
·
3,436 Posts
150W is just the ATX spec for minimum the 8pin has to be able to support to advertise as ATX compatiable.
Then you have the EPS plug which is exactly the same plug lol but it has a 225W minimum it has to meet.
Realisticlly 300W should be fine the limiting factor is going to be the pins in the plug more then the wires. If the pins get a little lose then you have more resistance for the current to pass though.
More resistance and more current = hot.

But personally any branded PSU I woudnt worry about. Aslong as you have good connection in the plugs.
 

·
Facepalm
Joined
·
8,623 Posts
not sure, its an EVGA 850G2 PSU. I had to turn off the overclock on my 10900k...and still when benching, I have seen as much as 750W power usage. I obviously need to upgrade my PSU. Can you explain to me theoretically what the max power consumption from an 8 pin PCIE power cable is? I know the safe number is 150W, but I have been pulling almost 300W benching my card all day, and it's working just fine. They cables are cold to the touch also.
Regular 8 pin cables can draw up to 300W but that isn't supposed to be sustained power draw. Two 8 pins can pull 300 watts each for 600W but they may get warm. Someone said it may be the connector itself that heats up, rather than the actual wire.

Do you have time spy and port royal scores?
 

·
Registered
Joined
·
93 Posts

·
Facepalm
Joined
·
8,623 Posts
What pads are those? Thermalright Odyssey 12.8 w/mk or Fujipoly 11 w/mk? (they seem to be identical pads, at least I think. Or maybe the fujipoly pads are very very 'slightly' dryer, not sure..
Or are those the 17 w/mk "use once then they flake apart" fujipoly pads? :)

Are those 1.5mm pads or 2mm pads?

Also on your left side of the GPU side, you may want to trim the strip slightly. I don't think you want the pad sitting on top of those weird big square things that are just to the right of the displayport plugs. Those things seem to be higher in elevation than the VRM's themselves and may interfere with partial contact on the VRMs.

BTW do you have any idea what the original 'white' pads are? They seem to be a combination of pad and a mesh like thing. Any idea what their specs and w/mk / branding are?
 

·
Registered
Joined
·
718 Posts
these are the odyssey thermal pads from thermalright 1.5mm 12.8w/mk

yea the nvidia pads seem to be some kind of mesh + paste

i know its not surgically clean, but will do the job :p

2466237
 

·
Registered
Joined
·
103 Posts
Discussion Starter #231
welp, my bykski waterblockk doesnt fit, seems they send me the wrong version, crap
The EXACT same thing happened to me! I bought one of their "Founders 3090" blocks shortly after they came out, and ended up with a reference PCB block. Bykskis entire release of the 3000 blocks was messed the hell up, and I should've known better as I've never once had a good experience with them or their products. Oh well... Live and learn, again and again.

I ended up getting the Bitspower block, which I am really happy with currently. Also, the BP block came thermal pads that look identical to the ones in your screenshot
 
  • Rep+
Reactions: dante`afk

·
Registered
Joined
·
718 Posts
which one did you get?


was my shuntmod successful or not? i see occaisionally still power on the perfcap reason for a few seconds on the second text of timespy.

2466269



power in/power out peaks at about 850w during the test.
 

·
Facepalm
Joined
·
8,623 Posts
which one did you get?


was my shuntmod successful or not? i see occaisionally still power on the perfcap reason for a few seconds on the second text of timespy.

View attachment 2466269


power in/power out peaks at about 810w during the test.
Can you do a "Heaven" benchmark run (this should not hit the power limit at all if your mod was successful) and have all of the power rails in GPU-Z set to "maximum" value?
Also, the easiest way to tell if your shunt mod was successful was to have an "no power limit" game or stress test run of something that did not hit the power limit before the mod, in GPU-Z with the maximum values shown for all rails, then run the exact same test again after the shunt mod. If all of the rails decrease in value by close to the same percentage (maybe more for MVDDC or PWR_SRC, etc) then your mod was successful.

For example here is my PR run I just did. Highest score I ever got.
I have HWinfo set to a 1.47x wattage multiplier. I was able to get this value by going to a pre-mod Timespy run (21,035 graphics) @ 112% power target (400W) and then finding the new power target that gave me the same score at the same overclock (+95 / +600 for that run back then), which was 76% power target. I then had GPU_Z open and looked at the board power draw shown and then simply calculated how much of a multiplier would put it back at 400W and came up with 1.47x.

it seemed like the PCIE slot was limiting me at first, as right after the "corrected" Mod was done, PCIE was getting up to 68W and limiting me (long story, something happened with my very first mod, which caused 8 pin #1 to start reporting 174W, and 8 pin #2 to draw 135W, with 8 pin 1 triggering the rail power limit--I noticed a massive difference in "Power target" and "power target: Normalized%", which made me come up with a new way to deal with the silver edges of the shunts being lower than the middle, causing contact and consistency issues. Basically, when this was happening, it was acting like my max power limit was only reaching like 430W because of the 8 pin limit. So I had to deal with that over a few days (I had no idea what was going on at first--I thought maybe it was ANOTHER shunt triggering that) But anyway: now PCIE seemed to "drop down" to 63-64W, causing a power flag "warning" (power limit flag can be triggered without any throttling happening, when you get "close" to the power target) which are those tiny blips you see.


2466270
 

·
Registered
Joined
·
718 Posts
honestly I did not even put the card in without mod :D

but somethings wrong, I used to get 22k gpu score in timespy and now it dropped down to 17-18k
 

·
Facepalm
Joined
·
8,623 Posts
honestly I did not even put the card in without mod :D

but somethings wrong, I used to get 22k gpu score in timespy and now it dropped down to 17-18k
Please take a screenshot of the max wattage rails in GPU-Z for me, please. I need all of them. Also put the TDP at maximum also.
At the same time, please have HWinfo64 running and have 'TDP %' and "TDP Normalized %' showing together. This is important.
 

·
Registered
Joined
·
718 Posts
Please take a screenshot of the max wattage rails in GPU-Z for me, please. I need all of them. Also put the TDP at maximum also.
At the same time, please have HWinfo64 running and have 'TDP %' and "TDP Normalized %' showing together. This is important.
2466273
 

·
Facepalm
Joined
·
8,623 Posts

·
Facepalm
Joined
·
8,623 Posts
I only put a 8mOhm on the Pcie slot shunt, should I change for 5mOhm?
Did you solder or use conductive paint? If you used conductive paint, just rip off the 8 mOhm shunt, and if there is paint contacting the edges of the original shunt, paint the entire PCIE shunt fully and completely, then let it dry without putting a shunt on top of it.

If you used conductive paint, did you scrape the edges of the shunts before applying ? it's possible that the shunt-->paint layer isn't even contacting the circuit. This is the problem I had originally when I tried (for the second mod attempt) putting a 5 mOhm shunt (yes, 5 mOhm) on the two 8 pins, with conductive paint bridging the edges. While the shunt contacted the paint, the circuit didn't seem to be completed or something on the first 8 pin very well, because when I pried the shunt off, there was barely any paint residue on it. It was all on the original shunt, even though I "pressed it down", it's possible the paint made a better connection on the onboard shunt than the top shunt. Of course all these problems happen because the edges are lower than the black housing.

When I simply removed the shunt, and painted another layer on the first layer (without shunt stacking) and applied a liberal amount of conductive paint, this fixed everything up.
I do have a 10 mOhm shunt on top of my PCIE btw, oddly enough, bridged with the conductive paint on the edges. It seems to have "broken in" as the card warmed up with time, because at 520W, it was hitting 68W from PCIE and signaling a throttling flag, but now it's gone down to 64W. So now I know at any time, if I have a problem with PCIE in the future, I can just take off the 10 mOhm shunt and paint the entire original shunt. That would probably allow more than 600W of power though.
 

·
Registered
Joined
·
718 Posts
I soldered and thoroughly cleaned up with isopropyl and scraped off the edges of each shunt

I'll have another look on the pcie shunt and probably apply more solder on the edges
 
221 - 240 of 382 Posts
Top