RIPJAY Hardmod: Breaking NVIDIA’s Rules w/ Shunt Resistor Piggyback, Ft. HiCookie

by birtanpublished on September 30, 2020

we've been wanting to reply to J's high
score on the times by extreme
leaderboards for some time now J can be
he has a bit of a smug face when he's in
the lead and frankly it bothers us so we
need to end that J I'm sorry but your
reign will end soon so the RIP J n
series has always been about doing
everything short of just going – Allen –
I've got a tank over there on my set we
can done this today J did have no chance
it'd be over you'd be done skis but we
already had an Allen to battle and it
was with a different benchmark it was
with Port Royal and it was supervised by
kingpin it was for st. Jude's as a
charity battle a lot of fun today though
we want to keep furthering the RIP J and
of GN series by doing the MacGyver mods
that Jay and I have been doing where we
try to push cards as far as possible
without resorting to the easy method of
liquid nitrogen before that this video
is brought to you by the rake on
everyday e25 wireless earbuds we've
liked the rake on everyday e25 s for the
relatively long wireless range something
for which we've implemented a strict
testing methodology that we've titled
the Herculean phone das the e 25s have
six hours of playback time and include a
socketable pillow case with its own
batteries so you get packaging that
actually serves a functional purpose the
pillowcase can recharge the earbuds up
to an additional 24 hours without
needing a cable the rake on earbuds also
come in multiple colors and include
different inserts to help customize for
the best fit they're currently available
for 15% off links below or you can visit
by Raycom slash gamers Nexus so we're
going to be modifying a $2,500 Titan
r-tx video card wit well that's what it
used to cost anyway with shunts we're
gonna do a piggyback mod and attach a
shunt resistor on top of a shunt
resistor to increase the power budget of
the card without needing to resort to
ln2 to push the frequency higher while
we were in Taiwan we bought some eight
million shunt resistors at Guan hua
digital Plaza and then we took them over
to high cookie it was a professional
overclocker who works for gigabyte and
he helped us and modify in the card I'm
sure Jai will have a lot of complaints
about this and modifying the card to get
the shunt resistors on there properly
without breaking it now that we've
learned on one and we've tested it for
today's
we could do the second one so today
we'll be showing our mod with hi cookie
he's featured in this and testing with a
lot of numbers for you on liquid metal
for shorting the shunts vs a shunt
resistor piggyback mod versus a stock
cart so it should be fun to finally
learn in numbers what all this stuff
actually does okay so this is pretty
much a mod that anyone can do in a sense
we did go to a professional overclocker
as stated we went to hi cookie he's
pretty awesome he's one of the best
overclockers for CPUs he's a massive
part of the R&D team 4 gigabytes
motherboards and he's been doing this a
long time but after he taught us the
basics of how to do this mod at this
point we're fully capable of doing it
ourselves and likely many of you would
be as well it basically requires a
soldering iron you do need a high
temperature we'll talk about that and in
the clip we cut too with high cookie
later and some shunt resistors and a bit
of research beforehand now a couple of
things first
people always ask this question of why
don't you just short the shunt resistor
with the wire well that is a great idea
and in theory it would work perfectly
because you just run a wire from one end
to the other and you achieve the same
thing potentially even better because
you're looking at even less resistance
but the problem is the the way end video
works is if it sees too much power or if
it if you just short the shunt like that
what'll happen is it locks itself into
2d clock mode so you won't get the 3d
clocks that you need to actually get
scores or play games at all even it'll
lock itself down into a safety mode and
until you remove that mod the card is
functionally useless it might as well be
dead all you'll get is desktop and
that's it
so these mods same thing with the shunt
resistor same thing with liquid metal
one of the things we'll show you today
is if you do too heavy an application of
liquid metal it'll do the same thing
where it spikes really high in power
consumption then it drops to the floor
and it craters or if you do a shunt
resistor at let's say five millionths
instead of what we used which is eight
million on top of the existing 5 milli
ohm shunt resistors we do five millions
on top of five you might end up and in
this case you would end up in safety
mode totally useless so 7 we've read is
kind of functional but we only found
report of that and we saw multiple
reports of eight Milian was working so
that's what we bought this is something
that you could damage your card in the
process first of all and after it you
could also damage your card because it's
a hell of a lot of power going into this
thing now quick recap on the Nvidia
boosting algorithm the biggest
limitations first is power and video
limits to power because it's trying to
maintain efficiency that's the strongest
point and after you get past the power
limitation your next limitation is going
to be temperature frequency boosts based
on how much power is available and then
after that or depending it'll boost
based on how much thermal head rooms
available and both of those can be
solved with chilled water for rip J and
with the shunt resistor mod again for
rip J so we've done some of this stuff
with liquid metal before but nothing as
good as this another one final note is
that liquid metal we've never
recommended anyway because like the shut
mod it can be different they can be
dangerous but for different reasons if
you apply it incorrectly like at the
base of the shot you get some liquid
metal down there at the base where
soldered on it could end up wiggling
itself off and killing the card
eventually we've never had that problem
we've left liquid metal on cards for
over a year and they've been fine but is
very much your mileage may vary don't
get it on the legs of the shunt and
separately if you do too much it's
useless anyway and then it's also a risk
of dripping off and shorting something
so not really it's fine for temporaries
for competitive overclocking not
something we'd recommend on average this
however is much safer in every other
sense except for the power consumption
skyrockets so you might kill other stuff
like MOSFETs but let's uh let's jump
over to Taipei where we worked with hi
cookie at gigabyte HQ and then we'll cut
back to the test results yeah so now we
just pinky pick yeah okay yeah yeah easy
so you already showed your soldering
iron that you like using in the other
video yeah but the reason you like this
one is because it heats up in eight
seconds to two hundred degrees yeah it
Phung to two hands you're only a second
so just just gonna meet two that we want
to short this this one of them one on
the back
how often do you get the soldering iron
out for projects these days not many
yeah that's just cuz your engineers do
it all for you yeah yeah typical holiday
pushing pushing okay yeah what is it
oh by through the sea by way sharing
flights yeah and you need a more
temperature hmm
definitely a more temperature and this
this one I think it up to 500 degree mmm
no this one's for hungry is it just
because it's too hopper in the PCB I
guess so its heat sink in it
yeah I guess it's down
it's not perfect just needs to be on
there okay it's try anyone okay
do you use any specific type of solder
like LED or for this one yeah no no just
just find a good quality yeah 3:05 five
hundred degrees four hundred agrees 450
now for the Jerry is liquid metal too
short stuff out before no just just the
pen yeah yeah you can use that for CPUs
right yeah like the track the pins okay
that's pretty good as you bypassed the
PCI Express
OCP right yeah you trick it yeah it
tricks I think the way I understand it
is by changing the resistance for the
because these are both for the PCI yeah
it's one for it it should adjust for the
one connected right
cool well that from my perspective was
easy cuz I didn't have to do anything
thank you for doing the shunt mod this
will help me get a bit more power to it
for the upcoming ripped Jay series
revival that were hoping to do with 3970
x3 990 X so thanks for doing the mod
maybe before we get into the
results charge for the hard mod and
stock results we wanted to illustrate
what the power consumption looks like
when too much liquid metal is used so
this would be the standard way to try
and bypass power limitations but it's
not an easy way to do it because you
basically have to go by experience for
how much liquid metal you should put on
the shunts this is an amount where it's
not as bad as shorting the shunt with a
wire which would lock us into 2d clocks
but instead where we get mixed power
consumption consistency depending on the
application this chart shows DC current
to the PCIe 12-volt cables against the
constant 100% work load from firm arc
this load does not fluctuate and should
never dip we see a massive spike beyond
the piggyback shunt mod which we'll put
on the screen in a second up to 45 amps
or a staggering 540 watts for PCIe
12-volt alone but we also see power
consumption crater unexpectedly that's
because the liquid metal application is
too heavy and although it allows more
power through it also start stripping
protective circuits or just exhibiting
instability increasing the power budget
also did nothing for this version of the
mod but we saw similar fall-off in
performance the liquid metal mod ended
up crashing out of the test the
piggyback shunt mod meanwhile held about
31 to 34 amps constantly and is reliable
in its stability the stock card held
about 19 to 23 amps constantly also
stable but a lot lower this illustrates
how the liquid metal mod isn't so easy
since going to light on the application
makes it a waste of time but too heavy
makes it unstable it's hard to control
and the only way to really nail it is
with experience our first bar chart is
for power consumption at the PCIe 12
volt cables tested with a physical
current clamp we're not measuring PCIe
slot power draw for this test as it's
irrelevant and we didn't mod the PCIe
slot shunt resistor so it's going to be
the same this could also blow off
slot or melt the 24 pin connector if you
mod that one so don't do it
we only mod the PCIe cable shunt
resistors power consumption is the first
step toward understanding if this mod
worked fir mark puts the hardest load on
the GPU and ensures a fixed 100% TDP
remember that GPU power limitations are
calculated against whether it's bumping
against 100 % TDP limit including if
that TDP has been raised through OC
software and so we need something that
doesn't have a variable load for easier
testing fir mark has the unmodified card
at two hundred thirty eight point eight
Watts with the piggyback shut mod from
high cookie doing four hundred Watts
that's a massive increase in power of
161 watts or about sixty eight percent
more power consumption from the mod
that's significantly more Headroom for
boosting – but Nvidia's efficiency
normally plummets at this point in the
curve so it won't be anything close to a
linear increase doing a reasonable
amount of liquid metal and amout where
it's not nearly dripping off the shunt
we get about a 3% increase in power
budget overstocked it's not much and not
particularly helpful for pushing clocks
for the heavy liquid metal allotment we
end up all over the place as we showed
earlier sometimes it pulls more than the
shunt resistor piggyback MOT but then it
randomly craters to 12 to 24 watts from
earlier 500 watts it's not very useful
if it's going to crash and drop clocks
but we still wanted to show the Peaks
note that this is the marked in the
legend as inconsistent so don't consider
these heavy liquid metal results as
actually useful supplying 114 percent
power to the GPU the maximum allowed via
the v bios the unmodified card moves to
271 watts or about a 14% increase as
you'd expect the modded card goes to 455
watts both are about a 14% increase has
the V bios expects but one is 55 watts
and one is 32 watts this is extremely
simple math but a lot of people get this
wrong constantly in forums and on
Twitter this is something that people
screw up when they compare things like
AMD and NVIDIA cards where you might
have a 180 watt baseline and 50% power
offset and in video cards where you
might have a 250 watt baseline and a
lower power percent offset ultimately a
percentage is an offset from the base
number and a smaller base number means a
lower absolute value for wattage in this
instance the modded card is pushed
extremely hard in 500 watts for just the
12-volt cables that's insane and it's
going to take serious cool and keep the
vrm a lot
– bye power consumption was about 236 to
240 watts on the unmodified car door 362
382 on the modified one hi cookies mod
also increased our power consumption to
322 watts in gaming where it was running
about 240 watts previously the light
liquid metal mod has us around 3 percent
improved once again with the heavy
liquid metal closer to the piggyback
shunt mod but still cratering and
consistently gaming results had us about
320 – watch for the piggyback shunt mod
240 watts or stock now a 3% for the
light liquid metal application if you
take a look at hell-blade though you'll
see where the heavier liquid metal
application didn't pull much beyond the
lighter liquid metal application that's
because again it's inconsistent
unreliable this next chart will look at
frequency in times by extreme useful
because it's functionally a perfectly
repeatable result every rot for this one
we're looking at the unmodified card
versus the piggyback modded Kart the
cards were maintained at the same
temperature of 68 to 69 degrees Celsius
GPU this is without any overclock
implied so we're looking at only these
stock behavior which is going to
basically follow the power limitations
on the Kart the stock card dips to about
1830 megahertz and gt1 and about 1740 to
1800 bagger hertz in gt2 checking the
logs this card was power limited it
would have held higher clocks and
boosted higher but Nvidia's boost
behavior limits the card on power first
to maintain efficiency so we see this if
this frequency drawback as power budget
becomes too strained to support the
higher frequencies with stability the
piggyback modded card is drawing
significantly more power and is horribly
inefficient you should never do this if
you want it to actually be efficient but
it's holding about 1900 guards stable in
both tests you want a completely flat
line here and we're pretty close to that
at this point since we're no longer
significantly power limited our next
best method to push frequency will be by
lowering the temperature that's doable
pretty easily with liquid nitrogen less
easily with chilled water and a pretty
standard at least with normal water
cooling we also need to do some
overclocking for overclocking testing we
maintained an above ambient temperature
of 68 degrees Celsius GPU to ensure the
boosting algorithm behaved as normally
we later did some sub-zero overclocking
with liquid nitrogen this chart will
show
by extreme scores it's our last one and
it's the most important this helps us
understand if there was actually any
point to doing all of this remember our
goal is to chase points and scores not
to chase efficiency this mod again is
terrible for every other reason we
didn't sort this by points but instead
we kept repeated order the first three
rows are all the unmodified card tests
one is full stock test two is full stock
except with 14% power increase of 114
percent power no overclock applied and
test three is the maximum overclock for
a stable result and 114 percent power
the order repeats for the piggyback mod
below it the stock card scored sixty
nine seventy six points aggregate for
gt1 and gt2 fps results while the
piggyback mod at 100% power scored 75 27
points an increase of 8% remember that
100% on the piggyback is actually
drawing significantly more since it was
tricked into thinking powers lower than
it is pushing power on the stock card
increased the results to 70 102 points
while the piggyback increased to 7826
the gap is widening part of that's
because 14% is a again a percentage so
total power consumptions higher 14% of
33 amps is a lot more than 14% of 19
amps the overclock the numbers on
ambience were 80 87 points for the
unmodified card while the modified card
scored a massive 91 63 points that's an
improvement of 13% which is huge when
considering we're at 68 to 69 degrees
Celsius for all of these tests the vrm
would overheat in this configuration if
we ran it sustained for longer than just
times by so we'll have to pull the card
down to at least zero degrees Celsius to
keep it relatively alive for competitive
overclocking liquid nitrogen will allow
us to push the score higher still of
course but we wanted to keep it simpler
now note that we were using Allen 2 to
control temperature for these results
locks to what the air cooler could do at
68 C but the modded Titan RT acts
required significantly more cooling
power than the unmodified card so we
wanted to control temperature for
frequency accuracy and for a direct
points comparison but it is not a direct
power or thermal comparison it's not
really fair to the unmodified card in
this instance if you took the same
amount of cooling power boost a bit
higher from the frequency increase but
then it'd be at a lower temperature as
well so depends on what you want to
compare the stock card managed at 2010
the 2040 megahertz core frequency with
one 90 microts offset it held 2040 to
2050 five megahertz in the final scene
of GT Wanda and then the modded card
when we started really pushing it was
approaching twenty three to twenty four
hundred megahertz wrapping all of this
up then
first of all quick issue we noticed with
the ax 1600 I the coarser power supply
it's a smart power supply and this is
something we had to reach out to build
Zoid about for some help from actually
hardware overclocking problem with the
ax 6100 I is even though you can squish
it into a single rail mode through
software it's it doesn't always work so
it would end up tripping over current
protection when the card started pulling
30-something abs through the PCIe cables
which is somewhat an intended behavior
except when you switches a single rail
mode it's not supposed to do that
so it's buggy and we thought I was
genuinely worried for a second that the
card had died in transit to and from
Taiwan because I mean it was stuffed in
a camera bag in a pouch for for like 48
hours of flying so we were worried it
died but in fact switching the power
supply to EVGA supernova t2 fixed it
that power supply as built so I
described it as a dumb power supply in a
good way and that is single rail and
there's no intelligent detection of
current to try and protect the thing so
we switched to the PSU and it was all
fine after that but yeah this it works
very well the frequency is higher the
overclocking capabilities are
significantly higher and we don't have
the huge pain in the ass of liquid metal
which is always a massive pain because
you have to be careful how you put the
card down you can't put it down flat on
this back it ends up like dripping off
or you end up actually bumping it with
your hand when you're attaching the
water block or whatever it's really
sucks to work with so this is a more
permanent solution it is not a solution
we recommend for everybody but it can be
a lot of fun if you're trying to really
get into the hard modding scene of
enthusiast overclocking without going
too deep into it we're gonna do this mod
ourselves to the other card now and then
we'll start doing some streams again for
the RIP J series we'll start with
probably water or chilled water because
that's what Jay and I were doing we try
not to go into L&T oh it's kind of the
gentlemen's agreement for this series so
we're just trying to
stuff as much as we can without going on
to and that's what makes it fun check
back for that will either do a streamer
and upload not sure yet the the world is
weird right now but that's it for this
one subscribe for more huge thanks to
high cookie for helping us out educating
us and the audience on how this stuff
works it's our first venture into hard
modding and it was a lot of fun I'm
looking forward to doing more of it I'm
sure many of you are experienced in this
already but I can appreciate that people
like builds away to have a lot of
knowledge on this and we can learn from
them as well so subscribe for more could
a store document sex it's not net to
helps out directly or patreon.com slash
gamers Nexus to get some
behind-the-scenes videos featuring other
team members as well we'll see you all
next time

Related Videos

because the Atari VCS is definitely something that's coming out we need to do a teardown on one to get some perspective and Atari if you don't know it&...
Hey how's it going guys Jack I'm out here with the toasty rose and today we're going to be doing a four hundred and fifty dollar PC that any of you ...
The biggest rule in testing coolers is to never trust anything don't trust the numbers don't trust the software don't trust the firmware and definit...
Hey what's up guys Jack and Matt here with the toaster brothers and today we're gonna be finding out if this 350 dollar laptop can game this is gonna be...
Everyone welcome back to another episode of ask GN sorry I'm not doing the intro with snowflake for this one we're flying to Taiwan and we need to get t...
Everyone welcome back to another hardware news recap for the week we are in Taipei right now working on a factory tour series and we actually have a point for y...
Hey how's it going guys Jack and Matt here with the toaster bros and today we're gonna be doing our first 20 20 thousand dollar gaming PC but guess what...
Coursers a 500 has been an embarrassing show of performance for the company's first revisit to air cooler isn't a long time but the upside is that it ma...
When I received the thread Ripper 3990 X on loan from a yet unnamed youtuber we quickly threw together an extreme overclocking stream and got the CPU to about f...
Every one so we're in Taiwan right now we're doing our factory tour series and while we were out here we stopped by Lian Lee's offices and they want...
Hey what is up guys Jack and Matt here with the wait what do you do it on sex you back your desk back to your desk sorry about that okay hi Jack and Matt here w...
We thought we'd found the limits of viewer interest when he posted our tour of a screw factory last year as poor Freight it was more of a screw shop than a ...
Everyone welcome back to another hardware news recap for the week we're in Taiwan where we venture outside to film things that's always a nice change of...
Hey how's it going guys Jack and Matt here with the toasty rose and today we're gonna be doing a 350 dollar gaming PC build it's gonna be really awe...
out of all the factory tours we've done until today we'd never gotten an opportunity to see how water blocks and open-loop cooling components like fitti...
Paints is one of the chief processes to dial in for product manufacturing and today's tour will bring us to a taiwanese factory replete with a mix of automa...
hey what's up guys Jack I'm Matt here with the tasty bros and today we're gonna be showing you our 2020 CES recap live from our CES Airbnb but befor...
except you can't wipe your ass with it well I mean we're going to take a gamble on the YouTube D monetization system and talk about the impact of corona...
everyone we're closing out our trip in Taipei for our factory tourists we have a lot of factory footage live already and we have more coming to the the chan...
hey what's up guys Jack and Matt here with the Thai stereos and today we're gonna be finding out is a six hundred dollar laptop worth it in 2020 it'...