nachtfalke
-
Posts
860 -
Joined
-
Days Won
8
Content Type
Profiles
Forums
Events
Blogs
Posts posted by nachtfalke
-
-
i don't think too . i just need to find time for mod and bench it
my best matrix is dead now but i have other card no tested
he asked about Up6208 pwm witch is the gpu pwm controller , not up6205 mem controller( simply FB mod).
for matrix you can use software or hard VID mod, like i did in mine , if you want to use hard mod ask me
-
No in fact very easy [emoji3] Vid modd need
Gesendet von meinem SM-G930F mit Tapatalk
-
Gesendet von meinem SM-G930F mit Tapatalk
-
nice run man , hight memory freq make an huge difference:D
-
vga on GBT board ran x8 pci-e , x16 vs x8 in AM3 is not big difference almost nothing , but in other legacy 3D will be!
-
-
In the datasheet for the NCP5388 it says that the ILIM pin:
"Over current shutdown threshold. To program the shutdown threshold, connect this pin to the ROSC pin via a resistor divider as shown in the Applications Schematics. To disable the over current feature connect this pin directly to the ROSC pin. To guarantee correct operation, this pin should only be connected to the voltage generated by the ROSC pin – do not connect this pin to any externally generated voltages."
So if you short ILIM(pin 13) to ROSC(pin 12) it should disable OCP.
Here's the datasheet:
http://www.onsemi.com/pub_link/Collateral/NCP5388-D.PDF
On the topic of OVP it says:
"The OVP circuit monitors the output of DIFFOUT. If the DIFFOUT signal reaches 180 mV above the nominal 1.3 V offset the OVP will trip. The DIFFOUT signal is the difference between the output voltage and the DAC voltage plus the 1.3 V internal offset. This results in the OVP tracking the DAC voltage even during a dynamic change in the VID setting during operation."
The thing is I don't see how the NCP5388 could possibly know what voltage the VRM is putting out if you go and mod the VS+ instead of VFB. Admittedly VS+ goes straight to VCC but I bet you can cut the trace for it somewhere and then feed it a skewed voltage reading through a potentiometer to get the voltages that you actually want.
That's what I would try before going full zombie at least.
i tryed that few years back and only way is VID mod no other way , that way i vid moddded my all cards directly when OVP trip , that is the ONLY WAY to aviod ovp:D
-
NCP5388 is one of the "evil" pwm ,not many can manage to pass over his OVP:D
FB is pin19 , at ~ 1.3v ocp will trigger , if you lowering in half the ressistance between ILIM (pin 12)and ROSC(pin12) you will earn another ~50mv , till OVP will trigger ...and game is over!
when you see NCP on the card´s zombified them quickly
-
When I was turning the vddq trim pot I didnt see any movement in voltage, but when I turned vddr trim pot both voltages moved.
So it's ok to just use the vddr pot for both? Or does the vddq have to be moved also?
I havent really benched a ton of cards that have both rails so I guess that's why I am a little confused. On 9800 it was easy, they each had separate pots and measurement points and I just turned them the same amount.
use only vddr and see how the gpu will perform , from my experience classified aren´t very good on mem overclocking , is very hard to get one with 1500 ram , think you will bench it at 1450-1475.
on cold many it has issue when you play with hight ddr volts , so check the core vs ddr stability and stay with low volts on it ( 2.25 -2.3v)
i bench yestarday mine gtx285 matrix and very strange behavior under cold -115 had CBB , ddr don´t like volts ..max benchable was 1480-1500 at 2.25Vgddr , gpu core it´s ok , 1130mhz/2500 shader at 1.55v, over this no scalling anymore!
-
thanks , at 1.7+ vgpu ocp will trigger ...so think 1.7v is max that this vrm can do without blew up:D
my card like volts but also is an mediocre gpu , not an golden one
-
they are linked , that way i told you to be carfully when you turn the trim pot , need to be both outputs monitorised
also vddq and mvddc don´t respond very well on cold , i don´t think you will need it:D
if you wanted to measure directly from vddq or mvddc outputs , on the front side were the vrm are localised you will see one(vddq) and 2 (mvddc) tantalium cap , near inductors!
-
i don´t make units for sale! i don´t know prices , i build this cascade for bench , not just showing low temp thermometer
than many builders did.like i said this cascade use something that other don´t use ...it can hold really load , also DeltaT is very small ( evap base vs chip internal temp)
my K-Typ measure temp directly from evap base witch is have internal hole so reading are very accuraty that you see on the pic. -103`C at ~1.7vgpu , the integrated cascade thermometer measure the suction(top evap) witch is -110 load!
-
yes, R1150 (C2H4) 99,99 pure in second stage ..and some more improvment on the XV and evaporator arhitecture ( my own design)to hold much more load than regular used in cascade before , also first stage it use TEV(thermostatich exp valve) not cap tube
-
you have a nice card and very low volt , don´t think your cascade can hold 1.7V on GT200:D
-
Ofcurse is not the same like radiator in bucket of ice , is not the same even you put that radiator in dryice , because radiator arhitecture is not the same like platte HX , so heat exchange will be weak on radiator@DICE , also you will NOT have the same bench autonomy...on chiller you bench 24 hours without problem!
people should understand how an phase waterchiller it works , i´m the builder of Ramiro unit , that unit is an monster ....here on hwbot don´t exist an phase chiller more powerfull than that unit!!!!
-
waterchiller is an phase change , so it is a must to have separate category , like SS and cascade it have already! many people bench on chiller and chose custom water , so it is time to have an waterchiller category to avoid these misunderstandings!
-
-
-
Hi , i hint deep cold slow with my gtx560Ti twin frozr II/OC at ~-80*C , bios flash are this from first page , but still CS issue.
card pcb is same like gtx460hawk , but without dip switch mounted , i made manualy strap to "xtreme cool" , card ran perfectly at -68 on my phase and hint 1380 mhz gpu freq , but i can´t bench this card on my cascade ~ -115*C .
know anyone to pass over -80 CS on these card?
-
hi ,i have Z87 OCF and Z77 OCF all with original packing and accesories , mint conditions , are you intrested?
-
max mem clock was 1300 , but his gpu at 1400 durring vantage gpu tests , this card can do on ambiental water ~1200 , on chilled water 1260 and single stages 1300( speak about my units) is picked from ~15 pcs over time:D
ram clock make a huge difference when you play with gpu hight, hight mem bandwith archieve better bench fps!
-
thanks guys , my all card are binned over time prehaps 6 year of binning all legacy cards owned over 200 cards!, many legacy submission are in danger now!
-
thanks thanks thanks is that all for activating both, vmem and vgpu ?
what do you see in the pic above is only for main vrm (gpu) .
to enable the gddr vrm you need only to add +12V to gddr rail ( see first post -pic with yellow wire) , because gddr VRM was powered directly from mb pci slot , when you cut the card that rail will missing , so you need to add +12v to gddr rail manually
enjoy!
-
thanks dude i have a dead gtx580 dc2 now. but i don´t really know, if the vrm´s are functionable. today i will cut it. i have bought small modules, which can convert 12v to every lower voltage that i want. i must simply solder it to the right spots.
enjoy:D
GTX 285 Classified or reference.
in Nvidia GPU Overclocking
Posted
you also can use software ...but for me hard mod is the winner , easy to use voltage card , no voltage reset after shutdown/restart..
if you have any problem doing this VID mod , hint me with an PM!