Every video "I've made this video too long" Viewers "this was informative and enjoyable" Stop worrying about the length of videos my dude you are doing great.
yea. Basically enthusiast guy bragging to bloody disable the power limit (figuring out on an RTX PCB blueprint but never ever actually gonna get one in hand) completets watching this video enjoying then rewatching it to remember the on semi conductor and PM chip's model number.
If I had one, I'd probably send one. I have a Titan Xp Star Wars collectors edition I could probably loan him. My one request would be that he sends it to hell and back, then I'll give it back to Nvidia for RMA because they suck and won't give even a partial refund because its not usable due to Nvidia drivers. Does +900 mem on the stock cooler though, so it has some potential! That's the highest I've ever seen a Titan go on air cooling (granted with artifacts, but +840 looks fine on TimeSpy/Firestrike). I'll bench it on phase later (maybe LN2 as well)
John Phillips if Buildzoid asks, he shall receive. I’ve plopped it into a phase cooled FX build, so it’s not like I really need it (I might do slideshow testing, I mean... Raytracing tests...) I just bought an AIO GPU cooler to compare vs regular GPU waterblocks vs Titan stock cooler (Star Wars Empire collectors edition). But I do have other GPUs that I can do that testing on, the Titan is just the hottest card I have (besides the biggest Terrascale ever made, FireGl-V8650, but it’s problem is the 32 physical GDDR4 chips put out insane heat, idles at 70c while still sounding like a jet. But it didn’t cost me $1200, so I’m fine with that, it also doubles as a hand drier by the sink when running Furmark... multi-purpose GPU!). So if Buildzoid wants to tear this Titan to shreds and make sure it’s dead, better use for it than anything I’ve done so far with it...
Thanks! Your video actually helped me repair THREE RTX's that I've bought for repair/parts and gave up on repairing because they had crazy high pcie slot power draw (fake ofc) and always went into low clock mode because of perfcap. I've replaced like half of the PCBs on those and just gave up. After your video I took a microscope and inspected the shunts. ALL of them were ever so slightly partially desoldered but making a tiny contact (probably because of what made those cards deffective in the first place which was VRM failure in all cases). I've searched the net far and wide BUT YOUR VIDEO gave me the only viable clue! Thanks!
Did you know, 15% TDP could save you 15 minutes shopping for car insurance? Wait... I think the commercial goes saves you 15 minutes on rendering, I think that's it! Buildzoid's got it, so you should too, go mod those RTX cards!
As you want to keep the PCIe power connection near stock spec anyway, you can also rewire the ATX-power shunt measurements to the PCIe power shunt. So short the ATX shunt power pads and wire all shunt sensing pads to the PCIe shunt. This way you avoid the use of additional resistors.
Funny. I've been thinking the exact same thing, except with a different signature Buildzoid phrase. * Watch a Biuldzoid AHOC video. * Listen for the phrase: "The thing is: ... " * Take one drink at every "phrase interval". * Try to still be somewhat cognizant after the completion of all phrase intervals. * If you are still awake at the end of the video, you win the game. *** DISCLAIMER: Player must be over the age of 21 Earth Human Years. Play Buildzoid's "The Thing Is... " Game at your own risk. *** Cheers.
Buildzoid 2017 "Is an 8 hour stream long enough for ads? Better make it 10 hours just to be safe". Your doing good Buildzoid, making them much more information packed, and a sense of direction (plus it looks like you stopped using a mouse to draw in paint and got a writing/drawing tablet). You've improved a ton since you started! Keep it up, and crank that voltage!
Thanks for this video! Bought a 1080 ti FTW3 Hybrid and it shows a 0.0% TDP. Somehow it is running at stock MHz with older drivers (45x.x or lower) but above that it crashes as soon as the drivers load on Windows startup. I've got super limited reflow soldering skills, but just bought a new INA3221 that I may work up the courage to replace. This video has given me a ton more knowledge on what to look for to troubleshoot before I go into replacing the chip. Much appreciated.
The chips do not usually measure 12V and 11.950V. There is usually a 20x or 100x gain differential operational amplifier, and it basically sees 50mV difference directly (sure it might still be powered by the 0 and 12V), and then multiply it, to say 1V. There are 5 ways to trick the card: 1) reprogram ADC or variable amplifiers using firmware or I2C to have different gain, or assume different shunt, or report stuff to other control circuity in different way. 2) Lower the resistance in the shunt (i.e. replace it with lower value, or short it with somehow resistive element, or put another resistor on top). 3) Change differential opamp gain, 4) Put additional resistors / voltage divider between low side of the shunt and opam, to change voltage. This can be done by fixed resistors, or using potentiometer so you can change thins on the fly. 5) Remove the low side feed to opamp from shunt completely (effectively disconnecting the shunt from opamp, at which points you can even short the shunt if you want), and feed arbitrary precision voltage close to 12V (i.e. fixed 11.960V) from precision voltage source, or from a resistive divider (fixed or adjustable potentiometer) again. People are just lazy, and prefer simple hacks, that proper control of the current monitoring. Just use 100k precision potentiometer between 0 and 12V to generate any voltage you wish. The sensing current to opamp is basically zero (picoamps), so there is no need for anything more complex. For anybody wanting to research this more, looks for Kelvin probe, or Kelvin style bridge, Kelvin sensing or four terminal sensing (4T), four wire sensing / measurement.
Hey man, debauer was still able to disable the power limit on the Stix 2080 Ti by bridging one of the shunts with a wire and leaving the other shunts unmodified. This is the same thing he did when he dry-ice OCed a 2080 FE. The problem he encountered where the card goes into the 300 mhz fault mode was when he wire bridged both shunts (which he tried with the 2080 FE), or when he placed parallel 5 mohm resistors on both shunts (on the 2080 Ti Strix). Shorting or reducing by half both shunts probably put the sensed total power draw at too low a level and triggered the 300 mhz mode. Modding only one shunt, and leaving the others untouched, probably didn't pull the sensed total power draw too low so it didn't go into the fault mode.
Call it the iPower mod kit and charge 30USD :P. Eh I think this is far too niche to be viable. People with multimeters and soldering irons should have the necessary resistors.
Getting the voltage divider on a single tri-pin even with a potentiometer is less than a single dollar, but it still requires to solder one of the pins to the ground, i.e. wire.
understood. but dont underestimate the incompetence and laziness of average consumers; who have neither the dexterity or patience for diy stuff. if more of the process can be taken away from them, the better. and I'd think it'd look a lot cleaner compared to what most peoples handy work would end up looking like. perhaps you can talk to der8auer on what the process is of manufacturing functional merchandise. or maybe you can design, then license to someone else. and we should be getting you into a more capitalist mindset anyway, alleviate some inability to produce content due to budget i get that it would seem a bit ridiculous, especially to someone like yourself. but some people have more money than free time, particularly in *this* industry
Could you use a potentiometer (roughtly 10K) to tweak the value of the low resistor until the GPU is happy, if for some reason NVIDIA put a safeguard in to keep tweakers from messing with the power limit?
10K is way too big for the smaller resistor. You really don't have to be too accurate a 33ohm + 120k works great. If you need more you can switch over to a 47 + 120K. You could also use a 100 or 500 ohm trimmer pot.
Maybe I'm having a brain fart or something (and after all, I didn't finish my EE degree), but isn't doing it this way, essentially the same thing as halving the shunt resistance? By halving the shunt resistor you are creating that voltage drop that you are then feeding into the monitoring system. This mod is also just lowering the voltage fed into the system. The only difference I can see is that the voltage that then feeds the card hasn't been dropped by that '50 or 25 mV'... but I don't see how that really changes anything from the sensing circuits perspective? Hopefully you or someone else in the comments can show me what I'm missing...
Only just seen this video as I picked up a cheap Watercooled 2080ti while I wait for my 3090. This circuit works by using a voltage divider to fool the current monitoring circuit into thinking it is only drawing a fixed low acceptable current. Downside is that you can't monitor the current using any tool that reads the current value from the card. Whilst this mod gives a truly unlimited power draw, I've read that if you piggyback 0.008 ohm resistors on top of the 0.005s giving almost exactly 0.003 ohms then the card won't go into safe mode and you've increased the power limit by 66% which is as far as I would like to go.
If you have issues with power balancing/turning off phases, you could make the card think it's pulling whatever is the optimal amount for the vrm, permanently
Hard to understand the circuitry well enough to do so, not everyone has access to Nvidia card’s Board View like you 😇😇. And those are very small smd components to solder onto
Indeed i am also quite surprised derbauer didn't bypass the shunts and created a voltage source for the current feedback.To me doing that looks easier(and also much safer and cheaper) than placing galinstan on them.
well, I realize I am a bit late to this video, but wouldn't it work if you just added like 20 milliohm resistor on top of the 5 miliohm resistor so the end result is 4 milliohms. (or 15 milliohm for 3.75)
Buildzoid, what about cards like the Galax 2080 RTX Ti HOF OC Lab edition that has the three 8 Pin Connectors? Would that have more power because of the extra 3rd 8 pin?
Miguel Chang The reason the power in limited on the ASUS cards isn’t because the 8 pin connectors can’t supply enough but because the amount of power draw is limited artificially. So having another 8 pin won’t let the card pull more power unless Galax has set an even high power limit that the ASUS ROG OC cards.
@@ActuallyHardcoreOverclocking Galax has a Facebook Group called "Galax OC Lab" there they have posted12 hours ago, PCB pics of their Galax OC Lab edition 2080 RTX Ti PCB and I counted 19 VRMs on two of those pictures. Looks low res so you cant tell the part numbers, but looks like that is highest VRM count on an RTX Ti card so far. I read online that the 3rd 8 Pin is exclusively for the memory. Just thought I would let you know Buildzoid.
I'm buying the 2080ti once EVGA releases their blower style cooler/ it comes back in stock, if I do so and if you still don't have a 2080ti by then and want that one to test around with, I can send it your way till mid Jan (please don't send me back a dead card if I do so, haha). Edit- Are you based in the US?
Oof, kinda rules that out then Shipping it both ways would probably wind up costing more than the card itself. :( I do wish Builazoid luck though, and really hope he gets his hand on a card.
@@CorerMaximus Does he not live in the US? I could probably lone him a Titan Xp collectors edition, as long as it passes Nvidia RMA then my 1 request is that he sends it to hell and back again. I hate Nvidia for being scum and poor drivers with even worse customer service, so my Titan is already considered a $1200 paper weight that is a useless piece of junk other than benchmarks (gaming sucks, even when it does work without crashing its worse than $330 in crossfire cards). I'm switching to 4x R9 Fury just so I have fewer driver issues, and potentially double the GPU performance, Nvidia drivers are just stuttery and crashing garbage.
The thing probably has pre-programmed current levels at idle (and maybe other use levels too), and if your mod won't show it the right draw at idle, it may well go to safety mode...
* Why does it not work to half the Ohms of the shunt resistor (soldering a resistor on top on the shunt)? Does it measure the voltage drop internal ? * What does the last shut resistor do (the one at the buttom) ? why not do the same for all 3. ? * Would it make more sense to buy a let say 0.0035 Ohm shunt resistor instead of just circumventing all ? * I guess this also works on the GTX 1080 card ?
Is the FE VRM’s current balancing and the “fancy Nvidia current balancing circuitry” redundant in any way or do they serve entirely different purposes that the other is unable too
Excellent explanation of how the mod works and how to do it, but what about the results? And besides that, how is this a better solution than just bridging it?
I'm trying to overclock my 2080 SUPER and it keeps hitting power limit (MSI afterburner wont let me go past 100%), it only reaches 50C under stress tests, think I should try this?
Pretty sure you said the additional power balancing is just for each 8 pin not for phases. VRM should work fine. The only thing I think could happen here is a fire if people aren't careful. Could short 8 pins to each other, probably not needed for this use though.
Dude I searched for this for years(since I shunt moded my 1060). Then there is real problem with the stacking of resistors. Even if you do not run into safe mode, the card wont let you go over 100% PL. There is still some kind of second PL(forgot how it was named in afterburner) that can't be exceeded. Even though GPU-Z reported 68% or something at max. the card showed PL perf cap. in the GPU-Z logfile. This clearly showed me that my mod was not a real success. Moding the pci e shunt resistor did not help either.(just tried out of frustration). Its like, the card is still perfectly knowing how much juice it sucks from the psu. Your mod seems to be much more sophisticated. I will definitely try it with my 3070 once its upgraded with a waterblock.
Secret sauce: add .008 ohm resistor to the .005 ohm, do not add .005 ohm. The minimum total resistance is .003 ohm, anything less and the card goes into emergency mode.
Really? That works? That'd be nice, cause then you can still use the power limiter in the overclock utililty, its just the tad higher that you want, no?
You have actually tried the .008 for an effective .0031 ohms? It seems this dude throws out these ideas without even testing which is a half-assed way to make TH-cam videos. Obviously if the shunt monitoring circuit only looked at the voltage drop across the shunt in a simplistic manor , his 120K and 47ohm divider would create a .0047 voltage on the sense pins, ever lower than the .003 in parallel with the card drawing 2 amps. So has this dude even tried this scheme, probably not and therefore the new scheme probably doesn't work. I'm going to look at the ICs on the board in detail and see how this all works, nevertheless I'm pretty sure the BIOS handles this PM in a more complex way that he is advertising here.
so does the 1660ti not have this issue? because i was able to 'trick' it just by soldering a wire across the shut (shorting the shunt). jw because i also have a 2070 super, assuming my 2070 super will have this safe mode issue if i straight short the shunts..
Hi congratulations on 40k sub. I’m pretty sure if you zoom all way out, on one of GN pcb pictures. You will find all power wireing diagrams. They all printed underneath, on their mod mat. You videos are not too long at all. 🔌🛠
That NCP chip triggering safe mode might just be the way the chip works rather than NVIDIA actively blocking shunt mods. I'd presume they rather like completely owning hwbot and the 3d mark hall of fame charts p.s You should rename the channel to "Unnecessarily Long Hardcore Overclocking Videos." Though they could be even longer IMO
I have questions about doing this mod on a 1080ti FE that i am hitting the power limit with. 1. Will i need any extra tools to actually use the extra power or will the card just draw as much as it needs? 2. Will the mod negatively affect the card during normal daily use when not pushing it? Also im assuming this method is better/safer if done correctly than a liquid metal mod?
Wish I had the extra money, I would send you a 2080ti and a Radeon Vii and say go at it. One day you'll get all this stuff just sent to you dude, keep at it your content is excellent.
tono man,,, think they did think about it... it seems to simple... if nvidia dont want to limit oc they would put the shunts on display as it is now...or was... its not that hard to say the value ist not changing regardless the load so theoretical power draw.
@buildzoid i have an hp prebuilt (GPU) 1050 at that overclocked really well but i being myself changed the fan considering it has two pin fan header on it had to cut an shit...anyhow shorted it out. found the issue but im not skilled in that department. if you want it just bc its not made by your normal brand and still out did most the 1050s lmk ill send it to u.
If you're good at soldering just replace the shunts entirely with 4mOhm ones. I did this on my 1080ti and it worked flawlessly, allowing 300w without touching anything, or 360w with +20% in afterburner. I guess 3mOhm shunts would work too, but more risk of triggering the safe mode. Personally I don't like the idea of completely disabling the current monitoring.
That is so low though... I pull 300w on a lightly overclocked RX 470. I was annoyed I had to BIOS flash them, but while I was changing that, I did another little change that let my cards shine... Took the #1 TimeSpy for duel RX 470 (number 2 had much faster overclocking GPU cores with a 5.2 Ghz i7 8700k and fast RAM, while I only used the stock air cooler and terrible Elpida Vram limiting me. CPU was Ryzen 1700 at 3.8 Ghz and 2666 Mhz budget RAM that can't even run its stock 2800 XMP even on my Threadripper system).
@@jakegarrett8109 Why did you tell me all that? I'm not really interested in how many watts your rx470s draw, it has no relation on nvidia pascal overclocking. 360w on a 1080ti is more than enough. I dunno why you started talking about your cpu and ram either lol
@@SuprSi I'm just saying, Nvidia has such stupid low power limits its embarrassing. RADEON makes good enough VRMs that pulling over double the TDP on air without issue (those cards are rated 120W TDP, yet my Titan Xp is gimped down to less power draw max even though its a 250w TDP). Its sad when sub $200 budget cards have a more powerful VRM than my $1200 flagship card... I'm not sure why they couldn't afford to use better components like Radeon, especially at that price level. I'm not sure why I mentioned CPU either, I guess my brain shut off around 10 last night... 360w would never be enough for me, that's leaving a lot of clocks on the table. When I pay $1200 for a GPU, I want it to do what I want, just like any other GPU. When I tell them to jump, they should jump, if I tell it to fry itself, it should give me that option. I didn't realize Nvidia owns the card and babysitting something I paid for... If I blow it up, that's fine, I don't want Nvidia holding my hand or in this case strangling my performance if I want to set records. I shouldn't be a dog held back on a choke chain, the GPU should be the dog here.
I do agree about 250W being too little for these high end cards, I replaced the shunts within a week of having my 1080ti. It'd be even more frustrating with a Titan Xp like you have. (I didn't see that until after my earlier post, apologies). But honestly, I've found 360W is enough for a WC'd 1080Ti. It hardly ever hits that limit. Maybe if I was going sub-ambient cooling, and doing physical voltmods to go beyond 1.093v then you might need more headroom. Edit - the VRM's aren't the limiting factor though, they're actually very very strong, entirely overkill for the stock limit.
Ooops, I accidentally deleted my other post with my timespy run. www.3dmark.com/compare/spy/4622155/spy/4460909 1800X @ 4.2, RAM @ 3600 CL14, 1080Ti @ 1088/6000mhz.
On the cards with only 2 8-pin connectors, is there value in using cables with higher gauge conductors once you lift the power limit to reduce voltage drop and/or stave off conductor heating?
Probably not, I use a single CPU connector to run 500w to my FX, so I really doubt it. Plus with GPUs, you can usually run duel RX 470 off a single connector, and mine bench 300w (each), so if you can run 600w through a single cable, its probably not a big deal. Just make sure the wires are physically heavy (meaning they used copper, I know my HX1200i platinum cables are the same "gauge" as another cheaper 1000w PSU I have, but you can immediately tell which set is the Corsair set because they weigh about twice as much, same length). So if you do have a crummy PSU with aluminum cables... Well... You might need massive cables, or just get a used set of higher quality Corsair cables and re-arrange the pinout (I know the - and + on mine are reversed compared to the cheaper modular cable set, so sparks WILL fly!)
hope you dont mind if i pick your brains, i have aquired a card, im told it still works but ive not tested it yet (as it has an ek waterblock (msi 1080 seahawk) and i dont have a cooler as yet), but it has 2 pins burnt on the board, ive looked at a pcie pinout and it is the 1st 2 pins (facing cpu and nearest back of pc) which i believe are prsnt#1 hot plug prescence detect and +12v i was thinking i could try and fill the hole in the plastic with some epoxy or something and glue some foil to replace the missing gold fingers, ive scratched back the solder mask to reveal the copper on the board and it is apparent that the prsnt#1 trace is microscopic and unless its a multi later board may not go anywhere and i dont fancy my chances of being able to solder to it and im just wondering if it is unused/un-nescessary? 1st q my next q, the +12v pin is common with the +12 pin next to it so i presume its not essential for the card to work but if i dont repair it will this be a problem if i want to overclock the card? ie will it draw more current from this source and possibly burn the remaining pin or is this extra current all pulled from the 12v power connectors? so basically should i attempt to repair it? or should i just leave it as is? are these pins essential? if i leave it should i avoid overclocking it or will this still be ok? any info much appreciated tia
Probably. Since I'm more into the hardware itself than the scores I don't really care about how relevant it will be benchmark score wise once it's cheap enough.
What would the repercussions of adjusting 12v rail in the psu up lets say. 0.05v or more. Most have an adjustment pot. Would that mess up rest of the system. And someone send this man a card!!!
despite premium manufacturers push for ultra stable 12V rails the actual spec is kinda loose on the precise value. i've had a card pull enough power to drag the 12V rail down to 11.3V without crashes. on the upper end i don't think even 13V would cause any issues.
Something I don't get: der8auer said the card went into safety mode, if he fed it too much power with a traditional shunt mod. the mod he did - as i understand it - basicially doubles the power, the card can consume in relation to what the card THINKS its pulling. so the card might think its taking 250W but in reality its taking 500W (numbers might be different). but 250w would still be under the power limit. So safety mode should'nt be triggered. Your mod makes the card think its always pulling a small amount of power, no matter how much it is actually pulling. My point is: both mods make the card think its pulling less power, then power limit. So i would assume the safety mod would still be triggered, because the card needs to have some other way of understanding how much it is pulling. otherwise der8auer's mod should also have worked. What am I missing?
addition: from what i understand, even though your mod gives much more control of how to set the "fake power consumption", the WAY the card gets its information is the still the same in both mods. therefore i think the safety mode could still be triggered.
its prob just thinking its pulling to little power when booting up and goes into safety mode. that wont happen with this method unless you let it think its using only like 2% or 1% or something all the time.
1:56 "I'm not sure what the safety mode looks like for a 20-series [card]" If you skip to about 7:33 in der8auer's video titled "RTX 2080 meets DRY ICE" you can see his card running at 300MHz in that situation. He explains how he tripped that just before that section in the video.
maybe making the card thing it's using 90% of tdp makes it use all VRM's ? or maybe the vrm is monitored by the controller so no need :) as you said you need to test one to know feelsbadman
If 50% of viewers donated 1 euro / dollar / pound... whatever, he would have RTX on hand now. NVIDIA is trying so hard to lock their cards down. It is good to see someone standing up and saying NO. If i pay for that damn thing i should be able to do whatever i want to do with it even overclock the crap out of it, or kill it trying...
Could you not just drop a blob of solder or some braid heavily reinforced with solder to all 4 lines? with the 2 high-side scene lines tied positive it will report no power draw and let you pump whatever in to it or does it error out if you?
yes that gives you a 300mhz clock. it did on the old cards to. if that braided wire was 4mOhm it would probably work but 2.5 is to low and 5 gets split in half unless you remove the 5 all ready fitted
If you're on a ~$600 budget for CPU and Board, going for a 9900K and a more mid-range board should be within reach, and would be considerably faster all-around. Alternatively, you could go with the X470 Gaming Pro Carbon Buildzoid reviewed the other day and put the $100 savings towards a better GPU or faster/more RAM.
I would honestly recommend the 1700/1800x, and whatever board you please. I am a 4790k user aswell, and plan to get the r7 1800 and the asus rog strix b450-f GAMING
yes its twice the cores + slightly higher clocks. but only your wallet can answer (also not every editing/virtua' (add software title here) software cares about cores so check that first) my 1800x is 4ghz 8core at 1.4v of beast and is smashing its way through h.265 encodes all day every day and was worth the money last year (still going strong this year)(and a few years more to come) the 2700x is better even if its just because you can run the memory faster (but you'll probably get more clocks to)
Hey man I was planning to do the R003 resistor on top on my shunts on a 1080ti, this will work right? Would it be wise to do the shunt for the 8pin test the card and then do the 6pin or should I just do both and hope I don't trigger the debug mode?
its hard to find a r003. i was thought about making one out of vape coil but measuring the difference between .4mm and .39mm (can remember actual lengths) and actually having that amount showing after soldering is never going to happen.
He should be allowed, as long as he says "this EVGA" or other brand. He may not be allowed to say for Asus products, or he may actually have a very limiting proprietary technology contract that stems up to all of Nvidia from his association with Nvidia. It may involve non-disclosure of new info, but if it becomes "common knowledge", then that's not really disclosing anything if everyone already knows. It just depends where he got the information from, and if it is protected under any guideline or contract.
Every video "I've made this video too long"
Viewers "this was informative and enjoyable"
Stop worrying about the length of videos my dude you are doing great.
i watch at 2x speed anyway :P
yea. Basically enthusiast guy bragging to bloody disable the power limit (figuring out on an RTX PCB blueprint but never ever actually gonna get one in hand) completets watching this video enjoying then rewatching it to remember the on semi conductor and PM chip's model number.
Somebody get this man a 2080 STAT #BuildZoidForPresident
#BuildZoidForPresident
My first thought too. Must be some rich bugger watching these out there, buy this man a 2080ti!
If I had one, I'd probably send one. I have a Titan Xp Star Wars collectors edition I could probably loan him. My one request would be that he sends it to hell and back, then I'll give it back to Nvidia for RMA because they suck and won't give even a partial refund because its not usable due to Nvidia drivers. Does +900 mem on the stock cooler though, so it has some potential! That's the highest I've ever seen a Titan go on air cooling (granted with artifacts, but +840 looks fine on TimeSpy/Firestrike). I'll bench it on phase later (maybe LN2 as well)
John Phillips if Buildzoid asks, he shall receive.
I’ve plopped it into a phase cooled FX build, so it’s not like I really need it (I might do slideshow testing, I mean... Raytracing tests...)
I just bought an AIO GPU cooler to compare vs regular GPU waterblocks vs Titan stock cooler (Star Wars Empire collectors edition). But I do have other GPUs that I can do that testing on, the Titan is just the hottest card I have (besides the biggest Terrascale ever made, FireGl-V8650, but it’s problem is the 32 physical GDDR4 chips put out insane heat, idles at 70c while still sounding like a jet. But it didn’t cost me $1200, so I’m fine with that, it also doubles as a hand drier by the sink when running Furmark... multi-purpose GPU!). So if Buildzoid wants to tear this Titan to shreds and make sure it’s dead, better use for it than anything I’ve done so far with it...
@@jakegarrett8109 collector Star Wars
GamersNexus really needs to give this man a modmat so he doesn't have to rely on wikipedia for pin outs
This is the only sponsorship Buildzoid really needs 😂
Steve offered BZ a modmat when they first came in, but apparently it wouldn't fit on his workspace
@@KaitlynFedrick minimodmat?
He seems to be using one in this video
no, he's using images supplied by gamers nexus of the pcb sitting on a modmat. He doesn't have one.
Thanks! Your video actually helped me repair THREE RTX's that I've bought for repair/parts and gave up on repairing because they had crazy high pcie slot power draw (fake ofc) and always went into low clock mode because of perfcap. I've replaced like half of the PCBs on those and just gave up. After your video I took a microscope and inspected the shunts. ALL of them were ever so slightly partially desoldered but making a tiny contact (probably because of what made those cards deffective in the first place which was VRM failure in all cases). I've searched the net far and wide BUT YOUR VIDEO gave me the only viable clue! Thanks!
Did you know, 15% TDP could save you 15 minutes shopping for car insurance? Wait... I think the commercial goes saves you 15 minutes on rendering, I think that's it! Buildzoid's got it, so you should too, go mod those RTX cards!
Awesome video, hopefully der8aur does this in a video soon.
Hopefully Buildzoid does this in video soon.
As you want to keep the PCIe power connection near stock spec anyway, you can also rewire the ATX-power shunt measurements to the PCIe power shunt. So short the ATX shunt power pads and wire all shunt sensing pads to the PCIe shunt. This way you avoid the use of additional resistors.
New drinking game: take a shot every time he messes up ohm/milliohm
Funny. I've been thinking the exact same thing, except with a different signature Buildzoid phrase.
* Watch a Biuldzoid AHOC video.
* Listen for the phrase: "The thing is: ... "
* Take one drink at every "phrase interval".
* Try to still be somewhat cognizant after the completion of all phrase intervals.
* If you are still awake at the end of the video, you win the game.
*** DISCLAIMER: Player must be over the age of 21 Earth Human Years.
Play Buildzoid's "The Thing Is... " Game at your own risk. ***
Cheers.
"Oh, no I've made it unnecessarily long" -BuildZoid 2018
Buildzoid 2017 "Is an 8 hour stream long enough for ads? Better make it 10 hours just to be safe". Your doing good Buildzoid, making them much more information packed, and a sense of direction (plus it looks like you stopped using a mouse to draw in paint and got a writing/drawing tablet). You've improved a ton since you started! Keep it up, and crank that voltage!
Adding fuel to the #RIPEVERYONE flames
This video can be summarized to 2 words. "Voltage. Divider."
Thanks for this video! Bought a 1080 ti FTW3 Hybrid and it shows a 0.0% TDP. Somehow it is running at stock MHz with older drivers (45x.x or lower) but above that it crashes as soon as the drivers load on Windows startup. I've got super limited reflow soldering skills, but just bought a new INA3221 that I may work up the courage to replace. This video has given me a ton more knowledge on what to look for to troubleshoot before I go into replacing the chip. Much appreciated.
The chips do not usually measure 12V and 11.950V. There is usually a 20x or 100x gain differential operational amplifier, and it basically sees 50mV difference directly (sure it might still be powered by the 0 and 12V), and then multiply it, to say 1V.
There are 5 ways to trick the card: 1) reprogram ADC or variable amplifiers using firmware or I2C to have different gain, or assume different shunt, or report stuff to other control circuity in different way. 2) Lower the resistance in the shunt (i.e. replace it with lower value, or short it with somehow resistive element, or put another resistor on top). 3) Change differential opamp gain, 4) Put additional resistors / voltage divider between low side of the shunt and opam, to change voltage. This can be done by fixed resistors, or using potentiometer so you can change thins on the fly. 5) Remove the low side feed to opamp from shunt completely (effectively disconnecting the shunt from opamp, at which points you can even short the shunt if you want), and feed arbitrary precision voltage close to 12V (i.e. fixed 11.960V) from precision voltage source, or from a resistive divider (fixed or adjustable potentiometer) again.
People are just lazy, and prefer simple hacks, that proper control of the current monitoring.
Just use 100k precision potentiometer between 0 and 12V to generate any voltage you wish. The sensing current to opamp is basically zero (picoamps), so there is no need for anything more complex.
For anybody wanting to research this more, looks for Kelvin probe, or Kelvin style bridge, Kelvin sensing or four terminal sensing (4T), four wire sensing / measurement.
I felt dumb when you whipped out the voltage divider. :P
Hey man, debauer was still able to disable the power limit on the Stix 2080 Ti by bridging one of the shunts with a wire and leaving the other shunts unmodified. This is the same thing he did when he dry-ice OCed a 2080 FE.
The problem he encountered where the card goes into the 300 mhz fault mode was when he wire bridged both shunts (which he tried with the 2080 FE), or when he placed parallel 5 mohm resistors on both shunts (on the 2080 Ti Strix). Shorting or reducing by half both shunts probably put the sensed total power draw at too low a level and triggered the 300 mhz mode. Modding only one shunt, and leaving the others untouched, probably didn't pull the sensed total power draw too low so it didn't go into the fault mode.
perhaps there is a market for prefabricated shunt replacements
which you can just solder over the empty spot left after you remove the shunt
this mod is not even 5USD in parts to do.
It could be 15 when sold as a package with handy instructions.
Call it the iPower mod kit and charge 30USD :P. Eh I think this is far too niche to be viable. People with multimeters and soldering irons should have the necessary resistors.
Getting the voltage divider on a single tri-pin even with a potentiometer is less than a single dollar, but it still requires to solder one of the pins to the ground, i.e. wire.
understood. but dont underestimate the incompetence and laziness of average consumers; who have neither the dexterity or patience for diy stuff. if more of the process can be taken away from them, the better. and I'd think it'd look a lot cleaner compared to what most peoples handy work would end up looking like. perhaps you can talk to der8auer on what the process is of manufacturing functional merchandise. or maybe you can design, then license to someone else. and we should be getting you into a more capitalist mindset anyway, alleviate some inability to produce content due to budget
i get that it would seem a bit ridiculous, especially to someone like yourself. but some people have more money than free time, particularly in *this* industry
Should have broken everyone's record, and then release this video on how you did it XD
Did you just shuntsplain der8auer? -TRIGGERED!!!1!
My gtx 680 soc can go up to 208% power in MSI afterburner... fan speed maxes out at 75, almost room temp under full power draw
Love your scientific approach while others use the empirical approach.
Could you use a potentiometer (roughtly 10K) to tweak the value of the low resistor until the GPU is happy, if for some reason NVIDIA put a safeguard in to keep tweakers from messing with the power limit?
10K is way too big for the smaller resistor. You really don't have to be too accurate a 33ohm + 120k works great. If you need more you can switch over to a 47 + 120K. You could also use a 100 or 500 ohm trimmer pot.
Maybe I'm having a brain fart or something (and after all, I didn't finish my EE degree), but isn't doing it this way, essentially the same thing as halving the shunt resistance? By halving the shunt resistor you are creating that voltage drop that you are then feeding into the monitoring system. This mod is also just lowering the voltage fed into the system. The only difference I can see is that the voltage that then feeds the card hasn't been dropped by that '50 or 25 mV'... but I don't see how that really changes anything from the sensing circuits perspective? Hopefully you or someone else in the comments can show me what I'm missing...
Only just seen this video as I picked up a cheap Watercooled 2080ti while I wait for my 3090. This circuit works by using a voltage divider to fool the current monitoring circuit into thinking it is only drawing a fixed low acceptable current. Downside is that you can't monitor the current using any tool that reads the current value from the card. Whilst this mod gives a truly unlimited power draw, I've read that if you piggyback 0.008 ohm resistors on top of the 0.005s giving almost exactly 0.003 ohms then the card won't go into safe mode and you've increased the power limit by 66% which is as far as I would like to go.
40K subs! Congrats! You deserve it!
If you have issues with power balancing/turning off phases, you could make the card think it's pulling whatever is the optimal amount for the vrm, permanently
Why so difficult way? Just add one single resistor in parallel with sense input of the power monitoring IC....
Hard to understand the circuitry well enough to do so, not everyone has access to Nvidia card’s Board View like you 😇😇. And those are very small smd components to solder onto
reported
Nice video. Buildzoid solving the world's problems, one at a time!!
Indeed i am also quite surprised derbauer didn't bypass the shunts and created a voltage source for the current feedback.To me doing that looks easier(and also much safer and cheaper) than placing galinstan on them.
I would NOT mess with the third shunt, no point in burning your bus up.
well, I realize I am a bit late to this video, but wouldn't it work if you just added like 20 milliohm resistor on top of the 5 miliohm resistor so the end result is 4 milliohms. (or 15 milliohm for 3.75)
The GPU power connector diagram is printed on the modmat that the card is on 😉
Love the vid man, thanks for sharing your thoughts.
Buildzoid, what about cards like the Galax 2080 RTX Ti HOF OC Lab edition that has the three 8 Pin Connectors? Would that have more power because of the extra 3rd 8 pin?
Miguel Chang The reason the power in limited on the ASUS cards isn’t because the 8 pin connectors can’t supply enough but because the amount of power draw is limited artificially. So having another 8 pin won’t let the card pull more power unless Galax has set an even high power limit that the ASUS ROG OC cards.
Galax has special permission from Nvidia for more XOC geared BIOSs for the HOF cards.
@@ActuallyHardcoreOverclocking Galax has a Facebook Group called "Galax OC Lab" there they have posted12 hours ago, PCB pics of their Galax OC Lab edition 2080 RTX Ti PCB and I counted 19 VRMs on two of those pictures. Looks low res so you cant tell the part numbers, but looks like that is highest VRM count on an RTX Ti card so far. I read online that the 3rd 8 Pin is exclusively for the memory. Just thought I would let you know Buildzoid.
seen it. I'll try ask them for better pictures
I'm buying the 2080ti once EVGA releases their blower style cooler/ it comes back in stock, if I do so and if you still don't have a 2080ti by then and want that one to test around with, I can send it your way till mid Jan (please don't send me back a dead card if I do so, haha).
Edit- Are you based in the US?
UK. Shipping is a nightmare.
Pro tip: When you loan stuff out, always make sure you can live with not getting it back 100%
Oof, kinda rules that out then Shipping it both ways would probably wind up costing more than the card itself. :( I do wish Builazoid luck though, and really hope he gets his hand on a card.
@@CorerMaximus Does he not live in the US? I could probably lone him a Titan Xp collectors edition, as long as it passes Nvidia RMA then my 1 request is that he sends it to hell and back again. I hate Nvidia for being scum and poor drivers with even worse customer service, so my Titan is already considered a $1200 paper weight that is a useless piece of junk other than benchmarks (gaming sucks, even when it does work without crashing its worse than $330 in crossfire cards). I'm switching to 4x R9 Fury just so I have fewer driver issues, and potentially double the GPU performance, Nvidia drivers are just stuttery and crashing garbage.
He doesn't, from what I checked on his website after the initial reply on my comment, Builazoid is somewhere in the EU most likely.
The thing probably has pre-programmed current levels at idle (and maybe other use levels too), and if your mod won't show it the right draw at idle, it may well go to safety mode...
well you could tweak the resistor values until it doesn't go into safety.
* Why does it not work to half the Ohms of the shunt resistor (soldering a resistor on top on the shunt)?
Does it measure the voltage drop internal ?
* What does the last shut resistor do (the one at the buttom) ? why not do the same for all 3. ?
* Would it make more sense to buy a let say 0.0035 Ohm shunt resistor instead of just circumventing all ?
* I guess this also works on the GTX 1080 card ?
You did good Mr. Buildzoid.
Is the FE VRM’s current balancing and the “fancy Nvidia current balancing circuitry” redundant in any way or do they serve entirely different purposes that the other is unable too
it's for balancing the power going into the card. It doesn't control how the VRM is current balanced.
Excellent explanation of how the mod works and how to do it, but what about the results? And besides that, how is this a better solution than just bridging it?
I'm trying to overclock my 2080 SUPER and it keeps hitting power limit (MSI afterburner wont let me go past 100%), it only reaches 50C under stress tests, think I should try this?
Elmor Labs sells adjustable voltage regulators with knobs... They're grrrrrreat!
Pretty sure you said the additional power balancing is just for each 8 pin not for phases. VRM should work fine. The only thing I think could happen here is a fire if people aren't careful. Could short 8 pins to each other, probably not needed for this use though.
Dude I searched for this for years(since I shunt moded my 1060). Then there is real problem with the stacking of resistors. Even if you do not run into safe mode, the card wont let you go over 100% PL. There is still some kind of second PL(forgot how it was named in afterburner) that can't be exceeded. Even though GPU-Z reported 68% or something at max. the card showed PL perf cap. in the GPU-Z logfile. This clearly showed me that my mod was not a real success. Moding the pci e shunt resistor did not help either.(just tried out of frustration). Its like, the card is still perfectly knowing how much juice it sucks from the psu.
Your mod seems to be much more sophisticated. I will definitely try it with my 3070 once its upgraded with a waterblock.
how would you go about doing this mod on a 1050ti asus strix? could you send some advice my way. thanks.
But how much ore performance without EXOTIC cooling can u expect to get from this mod? maybe 15 % more? than standard boost?!
somebody send this to DerB8uer, please
I also cringed on his implementation(he also only shorted one shunt, right?)
Secret sauce: add .008 ohm resistor to the .005 ohm, do not add .005 ohm. The minimum total resistance is .003 ohm, anything less and the card goes into emergency mode.
Really? That works? That'd be nice, cause then you can still use the power limiter in the overclock utililty, its just the tad higher that you want, no?
You have actually tried the .008 for an effective .0031 ohms? It seems this dude throws out these ideas without even testing which is a half-assed way to make TH-cam videos.
Obviously if the shunt monitoring circuit only looked at the voltage drop across the shunt in a simplistic manor , his 120K and 47ohm divider would create a .0047 voltage on the sense pins, ever lower than the .003 in parallel with the card drawing 2 amps. So has this dude even tried this scheme, probably not and therefore the new scheme probably doesn't work. I'm going to look at the ICs on the board in detail and see how this all works, nevertheless I'm pretty sure the BIOS handles this PM in a more complex way that he is advertising here.
so does the 1660ti not have this issue? because i was able to 'trick' it just by soldering a wire across the shut (shorting the shunt). jw because i also have a 2070 super, assuming my 2070 super will have this safe mode issue if i straight short the shunts..
Hi congratulations on 40k sub.
I’m pretty sure if you zoom all way out, on one of GN pcb pictures. You will find all power wireing diagrams. They all printed underneath, on their mod mat.
You videos are not too long at all.
🔌🛠
That NCP chip triggering safe mode might just be the way the chip works rather than NVIDIA actively blocking shunt mods. I'd presume they rather like completely owning hwbot and the 3d mark hall of fame charts
p.s You should rename the channel to "Unnecessarily Long Hardcore Overclocking Videos." Though they could be even longer IMO
Simple and right! Very nice.
I have questions about doing this mod on a 1080ti FE that i am hitting the power limit with. 1. Will i need any extra tools to actually use the extra power or will the card just draw as much as it needs? 2. Will the mod negatively affect the card during normal daily use when not pushing it? Also im assuming this method is better/safer if done correctly than a liquid metal mod?
Probably 70% would be better since they said the they are disabling phases based on load in the gamers nexus interview
the voltage controller does that based on the per phase current. Not based on the external readings.
This shunt mod on one or two RTX Titans, and an AMD 3950X all of it watercooled... I wonder what performance you would get on that :o
Wish I had the extra money, I would send you a 2080ti and a Radeon Vii and say go at it. One day you'll get all this stuff just sent to you dude, keep at it your content is excellent.
BZ has the longest, and we like it
i thought people had issues with not being able to increase core voltage, not the actual power limit?
I love this stuff. Thanks!
tono man,,, think they did think about it... it seems to simple... if nvidia dont want to limit oc they would put the shunts on display as it is now...or was... its not that hard to say the value ist not changing regardless the load so theoretical power draw.
@buildzoid i have an hp prebuilt (GPU) 1050 at that overclocked really well but i being myself changed the fan considering it has two pin fan header on it had to cut an shit...anyhow shorted it out. found the issue but im not skilled in that department. if you want it just bc its not made by your normal brand and still out did most the 1050s lmk ill send it to u.
Can you comment on the chinese chip that was secretly added onto server boards? How does it work, the consequences , etc
If you're good at soldering just replace the shunts entirely with 4mOhm ones. I did this on my 1080ti and it worked flawlessly, allowing 300w without touching anything, or 360w with +20% in afterburner. I guess 3mOhm shunts would work too, but more risk of triggering the safe mode. Personally I don't like the idea of completely disabling the current monitoring.
That is so low though... I pull 300w on a lightly overclocked RX 470. I was annoyed I had to BIOS flash them, but while I was changing that, I did another little change that let my cards shine... Took the #1 TimeSpy for duel RX 470 (number 2 had much faster overclocking GPU cores with a 5.2 Ghz i7 8700k and fast RAM, while I only used the stock air cooler and terrible Elpida Vram limiting me. CPU was Ryzen 1700 at 3.8 Ghz and 2666 Mhz budget RAM that can't even run its stock 2800 XMP even on my Threadripper system).
@@jakegarrett8109 Why did you tell me all that? I'm not really interested in how many watts your rx470s draw, it has no relation on nvidia pascal overclocking. 360w on a 1080ti is more than enough. I dunno why you started talking about your cpu and ram either lol
@@SuprSi I'm just saying, Nvidia has such stupid low power limits its embarrassing. RADEON makes good enough VRMs that pulling over double the TDP on air without issue (those cards are rated 120W TDP, yet my Titan Xp is gimped down to less power draw max even though its a 250w TDP). Its sad when sub $200 budget cards have a more powerful VRM than my $1200 flagship card... I'm not sure why they couldn't afford to use better components like Radeon, especially at that price level.
I'm not sure why I mentioned CPU either, I guess my brain shut off around 10 last night...
360w would never be enough for me, that's leaving a lot of clocks on the table. When I pay $1200 for a GPU, I want it to do what I want, just like any other GPU. When I tell them to jump, they should jump, if I tell it to fry itself, it should give me that option. I didn't realize Nvidia owns the card and babysitting something I paid for... If I blow it up, that's fine, I don't want Nvidia holding my hand or in this case strangling my performance if I want to set records. I shouldn't be a dog held back on a choke chain, the GPU should be the dog here.
I do agree about 250W being too little for these high end cards, I replaced the shunts within a week of having my 1080ti. It'd be even more frustrating with a Titan Xp like you have. (I didn't see that until after my earlier post, apologies).
But honestly, I've found 360W is enough for a WC'd 1080Ti. It hardly ever hits that limit. Maybe if I was going sub-ambient cooling, and doing physical voltmods to go beyond 1.093v then you might need more headroom. Edit - the VRM's aren't the limiting factor though, they're actually very very strong, entirely overkill for the stock limit.
Ooops, I accidentally deleted my other post with my timespy run. www.3dmark.com/compare/spy/4622155/spy/4460909
1800X @ 4.2, RAM @ 3600 CL14, 1080Ti @ 1088/6000mhz.
Please, someone send this man a 2080Ti !
On the cards with only 2 8-pin connectors, is there value in using cables with higher gauge conductors once you lift the power limit to reduce voltage drop and/or stave off conductor heating?
Probably not, I use a single CPU connector to run 500w to my FX, so I really doubt it. Plus with GPUs, you can usually run duel RX 470 off a single connector, and mine bench 300w (each), so if you can run 600w through a single cable, its probably not a big deal. Just make sure the wires are physically heavy (meaning they used copper, I know my HX1200i platinum cables are the same "gauge" as another cheaper 1000w PSU I have, but you can immediately tell which set is the Corsair set because they weigh about twice as much, same length). So if you do have a crummy PSU with aluminum cables... Well... You might need massive cables, or just get a used set of higher quality Corsair cables and re-arrange the pinout (I know the - and + on mine are reversed compared to the cheaper modular cable set, so sparks WILL fly!)
@ActuallyHardwareOverclocking - Apologize For YT Nerf`n My Computing My Day PlayList - But I Have You Now !!!
So why doesn't lowering the shunt work, does it have some kind of lower limit violation?
hope you dont mind if i pick your brains, i have aquired a card, im told it still works but ive not tested it yet (as it has an ek waterblock (msi 1080 seahawk) and i dont have a cooler as yet), but it has 2 pins burnt on the board, ive looked at a pcie pinout and it is the 1st 2 pins (facing cpu and nearest back of pc) which i believe are prsnt#1 hot plug prescence detect and +12v
i was thinking i could try and fill the hole in the plastic with some epoxy or something and glue some foil to replace the missing gold fingers,
ive scratched back the solder mask to reveal the copper on the board and it is apparent that the prsnt#1 trace is microscopic and unless its a multi later board may not go anywhere and i dont fancy my chances of being able to solder to it and im just wondering if it is unused/un-nescessary? 1st q
my next q, the +12v pin is common with the +12 pin next to it so i presume its not essential for the card to work but if i dont repair it will this be a problem if i want to overclock the card? ie will it draw more current from this source and possibly burn the remaining pin or is this extra current all pulled from the 12v power connectors?
so basically should i attempt to repair it? or should i just leave it as is? are these pins essential? if i leave it should i avoid overclocking it or will this still be ok?
any info much appreciated tia
You think in the future you might get one of your hands on one of these to Frankenstein like the 1080ti AHOC edition?
(once it's not ridiculously fucking expensive)
Probably. Since I'm more into the hardware itself than the scores I don't really care about how relevant it will be benchmark score wise once it's cheap enough.
What would the repercussions of adjusting 12v rail in the psu up lets say. 0.05v or more. Most have an adjustment pot. Would that mess up rest of the system. And someone send this man a card!!!
despite premium manufacturers push for ultra stable 12V rails the actual spec is kinda loose on the precise value. i've had a card pull enough power to drag the 12V rail down to 11.3V without crashes. on the upper end i don't think even 13V would cause any issues.
Lets get this guy a card!
so make a voltage divider on the input for the voltage controller, EZ
Something I don't get: der8auer said the card went into safety mode, if he fed it too much power with a traditional shunt mod. the mod he did - as i understand it - basicially doubles the power, the card can consume in relation to what the card THINKS its pulling. so the card might think its taking 250W but in reality its taking 500W (numbers might be different). but 250w would still be under the power limit. So safety mode should'nt be triggered. Your mod makes the card think its always pulling a small amount of power, no matter how much it is actually pulling. My point is: both mods make the card think its pulling less power, then power limit. So i would assume the safety mod would still be triggered, because the card needs to have some other way of understanding how much it is pulling. otherwise der8auer's mod should also have worked. What am I missing?
addition: from what i understand, even though your mod gives much more control of how to set the "fake power consumption", the WAY the card gets its information is the still the same in both mods. therefore i think the safety mode could still be triggered.
Same thing I was wondering. Maybe they do some initial resistance sensing?
That would be really interesting sadly i can't try without a card in his hands.
We route the power around the sensors and fake a power flow over the testing pads u nder the shunts
its prob just thinking its pulling to little power when booting up and goes into safety mode. that wont happen with this method unless you let it think its using only like 2% or 1% or something all the time.
What does putting liquid metal on top of the shunt even do?
1:56 "I'm not sure what the safety mode looks like for a 20-series [card]" If you skip to about 7:33 in der8auer's video titled "RTX 2080 meets DRY ICE" you can see his card running at 300MHz in that situation. He explains how he tripped that just before that section in the video.
because I don't want to be stuck using 1 specific BIOS.
Will add to patreon today. I hope you will get a 2080 ti.
Can't afford a tshirt. Keep it up!
Love your videos keep it up!
maybe making the card thing it's using 90% of tdp makes it use all VRM's ? or maybe the vrm is monitored by the controller so no need :) as you said you need to test one to know feelsbadman
bro, do an complete power limit removal video for amd gpu's
This video got RUINED because no one on earth donated Buildzoid his 2080Ti's!
Is there any reason you couldn't use a 20 turn 100ohm pot instead of the 33 or 47 ohm resistor in the voltage divider?
No
But you will need range limiting resistors
bruh I had no idea this shit was so complex
if derbauer added another resistor on top of the shunt wouldnt that actually double the resistance ?
Nah if same value parallel halves it and series doubles it.
Just found your channel bro! Nice! I'll have to collab with you one day soon when I hit 40k.
Do you have to also mod the gpu bios alongside the shunt mod in order for it to work(properly)?
No, the whole point of this mod is that bios mods are not required
Oi, this is lit. Just checked my sub box and seen this.
If 50% of viewers donated 1 euro / dollar / pound... whatever, he would have RTX on hand now. NVIDIA is trying so hard to lock their cards down. It is good to see someone standing up and saying NO. If i pay for that damn thing i should be able to do whatever i want to do with it even overclock the crap out of it, or kill it trying...
Actually i already started. I send 5 pound to Buildzoid. Let's get him the card. I mean only a little is enough, if enough of us will do it.
You remove R005, add fusible car 10 A red
RTX 2080 ti: Buildzoid Edition
I just love your Video's so much good info thats hard to find if you didnt do this ;)
plz do a breakdown on asus strix rog b450 f gaming...love the channel
Not good for the price
@@tanishqbhaiji103 ...not now...it really is over priced now...it was a decent deal when I bought it two years ago
Could you not just drop a blob of solder or some braid heavily reinforced with solder to all 4 lines? with the 2 high-side scene lines tied positive it will report no power draw and let you pump whatever in to it or does it error out if you?
Nevermind i finished watching the video, should have done that before xP
yes that gives you a 300mhz clock. it did on the old cards to. if that braided wire was 4mOhm it would probably work but 2.5 is to low and 5 gets split in half unless you remove the 5 all ready fitted
Is it worth getting a 2700x and x470 crosshair VII right now? coming from a 4790k I usually do gaming/light editing/virtualization
If you're on a ~$600 budget for CPU and Board, going for a 9900K and a more mid-range board should be within reach, and would be considerably faster all-around.
Alternatively, you could go with the X470 Gaming Pro Carbon Buildzoid reviewed the other day and put the $100 savings towards a better GPU or faster/more RAM.
I would honestly recommend the 1700/1800x, and whatever board you please. I am a 4790k user aswell, and plan to get the r7 1800 and the asus rog strix b450-f GAMING
I'm selling both of those things used in a combo on eBay rn. It's the unorganized one with alphacool loop
yes its twice the cores + slightly higher clocks.
but only your wallet can answer (also not every editing/virtua' (add software title here) software cares about cores so check that first)
my 1800x is 4ghz 8core at 1.4v of beast and is smashing its way through h.265 encodes all day every day and was worth the money last year (still going strong this year)(and a few years more to come)
the 2700x is better even if its just because you can run the memory faster (but you'll probably get more clocks to)
Why cant you increase the 12 volt to 13 volt? Wouldn't that increase the power output also?
no it would go into safety mode.
Hey man I was planning to do the R003 resistor on top on my shunts on a 1080ti, this will work right? Would it be wise to do the shunt for the 8pin test the card and then do the 6pin or should I just do both and hope I don't trigger the debug mode?
1080Tis shouldn't get trigger quite as easily as 20 series cards do. I recently picked up a fresh 1080Ti to work on so I'll check that for you.
Thanks for replying and the great content.
i would also like to see some testing on this. Just plain solder did not go well and liquid metal is stupid
its hard to find a r003. i was thought about making one out of vape coil but measuring the difference between .4mm and .39mm (can remember actual lengths) and actually having that amount showing after soldering is never going to happen.
Not sure where your located but buildzoid linked a place in the UK that stocks them in one of his posts. I got mine from element14 in Australia.
Do you think I can use 100kΩ to ground instead of 120kΩ ?
Yes
a sligthly easier? sligTHly?
could it be possible roman CAN'T disclose how to actually hardcore mod the card?
He should be allowed, as long as he says "this EVGA" or other brand. He may not be allowed to say for Asus products, or he may actually have a very limiting proprietary technology contract that stems up to all of Nvidia from his association with Nvidia. It may involve non-disclosure of new info, but if it becomes "common knowledge", then that's not really disclosing anything if everyone already knows. It just depends where he got the information from, and if it is protected under any guideline or contract.
do the shunt resistors need to be taken off first for this method?
yes they do watch my video about the 1080Ti Palit Gamerock I modded to see how the finished mod is supposed to look IRL
Hey Buildzoid, I heard you like dual GPU cards, i have a gtx 590 from an old Alienware pc and wouldn't mind sending it to you.
I already have 2 590s. Though I guess I could use it to demonstrate a VRM failure so sure. Send me an email to buildzoid@gmail.com about it.
I wonder if this will work for 1650 Super lmao
P S very interesting video. Couldn't stop watching
Yes it will work for all modern Nvidia cards
I just did this to my 1070. Now i thinks it is pulling 200-300% power under load. WTH? :)
Can you overclock further with that modification?