for AIO cooling, it isn’t going to
offer you better performance than
the cheaper none OC-LAB, GALAX
HOF 1080 Ti.
Where this card stands out from
the others is when there’s an LN2
container mounted on it. All the fancy
components from which the VRM
is built and power phases matter
here, where you’ll be pushing the
card past its limits, adding anything
up to 550MHz to the core clock. An
impressive feat given that NVIDIA
has essentially extracted all the
overclocking headroom from the
silicon already, hence the ability to
run 2GHZ+ on air cooling.
That said, running such GPUs on
LN2 is important and worthwhile,
because even though the 1080 Ti
ships at such high clock speeds,
there is performance to be had if
you crank up the core and memory
clock frequencies. With all things
working as they should, it scales the
performance accordingly.
My own experience with the HOF
was a peculiar one. Mixed in with my
own inexperience (this was literally
the first GTX 1080 Ti I had the
opportunity to run under LN2), I still
managed to hit a GPU core clock of
2,500MHz. It is possible to go higher
I suspect on this very same sample,
however by the time I figured out how
to go about overclocking this GPU I
was literally out of time. So for all
intents and purposes this is the first
round and what will follow in another
issue is the 2nd round.
As you can tell as well, the
results show low efficiency which
could be BIOS related or just me
being a n00b. The point is that what
GALAX promised, turned out to be
true that overclocking this graphics
“... IT IS A SERIOUSLY IMPRESSIVE CARD,
BOTH VISUALLY AND OF COURSE IN ITS
LN2 OVERCLOCKING POTENTIAL.”
28 The OverClocker Issue 42 | 2017
card simply needs an LN2 container
and a fan for the VRM and you’re
good to go.
One of the things that those with no
LN2 experience on graphics cards
may not be familiar with is that. Each
GPU behaves differently when cooled
with LN2. In fact, a good way to start
overclocking or testing LN2 limits,
(be it you have an LN2 BIOS or not) is
to just cool the GPU down and keep
load temps at around 20’C. Which is
exactly what I did with this graphics
card initially. There’s nothing
advanced here and you’re literally
figuring out the limits of the GPU as
is by eliminating any scaling limits
you might have due to temperature
(AIO cooling will not give you 20’C
loads unless it’s chilled and even
then on a 1080 Ti it may prove too
much). This sample didn’t yield much
over the 2050MHz limits, but I was