danbenbb

Mathieas

ComicBookAssassin
What are people hitting without the bios as I don’t think this bios is available for the black ultra

I have a thread of my OC test record. All results were achieved using stock bios and stock hybrid cooler. Some runs have 100% fan speed
and are noted, otherwise stock fan profile.


https://forums.evga.com/m..aspx?m=3163459&fp=3


TLDR:
Max Port Royal:
Memory +1100/Core+175/voltage+45: 14643 Maxtemp: 55C (fans at max and high performance filters)
https://www.3dmark.com/3dm/54235724


Game stable: +750mem/+130core/+30voltage/maxtemp:58C@20C ambient: 14405 (fan curve stock, filter quality high performance) https://www.3dmark.com/3dm/54351435

Wow, how could you get such a low maxtemp ? Mine were over 67C with average 60. All 8 Fans were on full speed and aggressive fan curve for gpu. (https://www.3dmark.com/pr/596573).

hybrid card!

My bad, missed that part.

For anyone using Waterblocks I suggest trying out the XC3 bios, fan control is busted in that bios(won't matter for us), but I got higher overall core clocks. Hopefully EVGA releases a future bios that incorporates XC3 boost clock table/performance with higher power limits. So we don't have to use a XC3 one :)

Mathieas

ComicBookAssassin
Are people really gaining over the 1950 without crashing

I am working on my daily stable. Current over 1hr stable settings are generating a Port Royal of 14405.

+750mem/+130core/+30voltage/maxtemp:58C@20C ambient: 14405 (fan curve stock, filter quality high performance)
https://www.3dmark.com/3dm/54351435

slovak_killer
Did the shunt mod today, first we tried only 8pins shunt mod, still got PWR perfcap even when reading very low values, after that we tried shunt modding also PCIE, this fixed the card and never saw PWR perfcap again, unfortunately this is only possible fix. 
On the screenshot u can see full load timespy run 2100mhz, PCIE readings maxed out at 17W and some rails even bugged out at 0W. Can confirm now card can full boost, no downclocking or undervolting, EVER !
 

I had two cards and wish I tested the one I sent back (2014) more before I sent it back. I kept the 2012 that seemed to have no issues OCing for me (well, except the power limits themselves...).
 
I'm guessing it won't do much for my card to mod it in this way if it is already hitting above 2100MHz.

QueueCumber
[
I had two cards and wish I tested the one I sent back (2014) more before I sent it back. I kept the 2012 that seemed to have no issues OCing for me (well, except the power limits themselves...).
 
I'm guessing it won't do much for my card to mod it in this way if it is already hitting above 2100MHz.

Which bios are you using to hit 2100mhz? Do you happen to have a Port Royal run link?

I also notice that with all my cards, I am around 150-200 points lower on my 5950x setup versus my 10900k setup, even though I don't have the memory fine tuned on the 10900k setup and it is colder in the basement with the 5950x setup and I have all the same stuff turned on and off between both setups (Actually I went further trying to shut stuff off on the 5950x setup to get it to the same scores I received on the 10900k...). 

Oh, just noticed that I had my 10900k OCed at 5.3GHZ in one score (the first one) and 5.2GHz in the other score (5.3GHZ on the lower score... I guess PR really isn't accounting for the CPU!).
 
I can bench at 5.3GHz pretty well, but it will crash every so often if using it for gaming, so I back it down to 5.2GHz and forget about it...

Before I changed the thermal paste and thermal pads my card would run 440w's in time spy with small peaks in the 470w's.... now I changed all that out... the card in time spy runs in the 470w's and peaks in the 490w's no OC! But the card doesn't OC for crap now..... so that's telling me something got better cooling and something else lost cooling lol! Guess once my waterblock shows up with the proper thermal pads and cooling I will see what happens.... fingers crossed optimus ships this week

pellaaa

slovak_killer
Did the shunt mod today, first we tried only 8pins shunt mod, still got PWR perfcap even when reading very low values, after that we tried shunt modding also PCIE, this fixed the card and never saw PWR perfcap again, unfortunately this is only possible fix. 
On the screenshot u can see full load timespy run 2100mhz, PCIE readings maxed out at 17W and some rails even bugged out at 0W. Can confirm now card can full boost, no downclocking or undervolting, EVER !
 

whats the point of this mod??

To VOID your warranty ... but allow more power to flow ...

Cool GTX

pellaaa

slovak_killer
Did the shunt mod today, first we tried only 8pins shunt mod, still got PWR perfcap even when reading very low values, after that we tried shunt modding also PCIE, this fixed the card and never saw PWR perfcap again, unfortunately this is only possible fix. 
On the screenshot u can see full load timespy run 2100mhz, PCIE readings maxed out at 17W and some rails even bugged out at 0W. Can confirm now card can full boost, no downclocking or undervolting, EVER !
 

whats the point of this mod??

To VOID your warranty ... but allow more power to flow ...

Its total shunt?!

vlcapata
Before I changed the thermal paste and thermal pads my card would run 440w's in time spy with small peaks in the 470w's.... now I changed all that out... the card in time spy runs in the 470w's and peaks in the 490w's no OC! But the card doesn't OC for crap now..... so that's telling me something got better cooling and something else lost cooling lol! Guess once my waterblock shows up with the proper thermal pads and cooling I will see what happens.... fingers crossed optimus ships this week

You need to bench (or have a bench saved) for both of those configurations you did, because the board drawing more power could also mean the boost clocks are holding better.
 
BTW someone somewhere said you get better overclocking results on the memory if you overclock the *MEMORY* first, hit apply, THEN overclock the core, and don't change the memory after the core is overclocked.  Reset clocks to default in X1 or Afterburner first.
 
This won't help increase your max memory clock (if your memory fails at +800 with +0 core overclock well, that's blatantly obvious) but it will apparently help prevent your core from handicapping the memory clock (e.g. +150 core first, causing anything higher than +250 Vram to crash), but absolute max core clocks may be reduced a little bit.

Btw the shunt in the mid front assuming src, does also show continuity to the pcie..

Once on water i dont know if this really do a difference but i think this can broke ur mobo or the card itself, coz that cable right at pcie dont look really good to me. Everyone said to not shunt the pcie and this is the worst you can do at the pcie resistor.

vlcapata
Before I did the thermal paste and pads.... I was getting 200mhz and 1400mhz... now I struggle with 170mhz and 1000mhz.... I've got some exploring to do lol

It is not the absolute offset that determines the quality of your OC. Instead it is the average clock achieved in a given bench. I had this same misconception and was helped by the awesome folks on this forum. In Port Royal for example how did your average freq change with the before and after? Did your score increase or decrease? My apologies if you mentioned this prior, so many posts here :-P.

vlcapata
Ya sooo many to sift through for sure! And after the thermal change out my score when down because my fps were all over the place! I mean during the beginning of port I'm usually 64-72fps and now I was 60-50fps.... it was strange! But my none OC 119% and 91F in time spy was a little lower as well but the power was 470w's during the first scene ND the second 460-492w which tells me something good happened along with something bad.... now to fix the bad and I should be good lol

 
Does it still jump around on stock settings as well?