the GTX 1063 gigabyte cards existence is
curious it's one of these two they look
the same they look the same even in the
name its gtx 1060 but it has one less SM
so in this regard the change between the
ten $6,200 card and a 10 $6,250 card is
actually somewhat significant it's more
than just a VRAM difference what we've
got is a three gigabyte card that's the
other one a three gigabyte card and this
has nine SMS it's got 128 fewer CUDA
cores so instead of 1280 it's 1152 and
on the six gigabyte card which has ten
SMS and is running the full 1280 CUDA
cores Nvidia is advertising a five
percent performance drop in the GTX 1063
gigabyte despite its 10 percent decrease
in SMS the card also lists a $50 cheaper
price than the six gigabyte GTX 1060
making it more directly competitive with
the RX 484 gigabyte and parallel in
marketing strategy though that execution
was different the curious part is why
Nvidia decided that this card should
remain with the 1060 branding but we'll
talk about that after the benchmarks
before getting to those just a quick
note to clear things up we've seen some
comments asking why this isn't a 4
gigabyte card instead of a 3 gigabyte
card because of the way the gtx 1060
build its memory subsystem because of
the 192 bit interface it's not possible
to run a 4 gigabyte solution on a gtx
1060 they would be different GPUs
generally when you see a multiple of 3
like a 6 gigabyte card that card will
only ship in other multiples of three
like a three gigabyte card the same is
true for 4 and 8 gigabyte units you'll
never see a 4 gig card with a 3 gigabyte
alternative it will always be 4
gigabytes or 2 gigabytes or 8 or
something similar
for these tests we're going to strictly
cover the GTX 10 60s so that means other
cards on the chart normally like the
1080 1070 or X 480 rx 470 none of those
will be on these charts because we're
just looking at these 2 3 vs. 6
gigabytes I'll throw that one in just to
have it there that is also GTX 1060 it's
an overclocked 1060
by MSI whereas these two are the stock
clocks so we are running a 1060 founders
Edition stock clock 17:08 megahertz
boost versus an EVGA GTX 1063 gigabyte
card so it's six versus three also 1708
megahertz boost boost 3.0 has some
variants but this is as close to
identical as we can get for the test
memory clocks are the same and we're
really just looking at the FPS here so
there won't be a thermal comparison is
not worth doing because there's no like
for like gtx 1060 s that we have to
compare that would really see a
difference in thermals from the vram
difference and from the 2-2 core
difference if you want to see the RX 4
80 1070 960 all the other cards hit the
link in the description below for the
full article which contains the review
and test methodology starting with GTA 5
at 1080p and very high and ultra
settings the gtx 1066 gigabyte card
performs at 95 FPS average with 60 6.3
fps and 60 point 7 FPS lows trailed by
the 1063 gigabyte card at eighty nine
point three FS Everage 64 1% low and 56
0.1% lows at this resolution the
differences are largely inconsequential
to gameplay but are measurable the
performance difference is about 6.3
percent between the two with the msi
1060 gaming x marginally ahead of these
a result of its higher clock rate at
1440p we're looking at 68 FPS average
for the 60 gigabyte card or about 63 FPS
average for the 3 gigabyte card the 0.1%
lows are mostly the same between the two
and that remains the case for 4k which
is obviously too taxing for the 1060
anyway but a worthwhile benchmark to see
if any hidden failures are rooted out
with a three gigabyte model shadow of
mordor produces more varied results at
1080p the gtx 1066 gigabyte card
operates at eighty seven point seven FPS
average with lows ranked at sixty eight
point seven and sixty six FPS for the
0.1% low metrics the three gigabyte card
runs at 77 FPS average a percent change
of 13.5 from old to new with the lows of
most noticeably different at 45 FPS
versus 66 fps
the six gigabyte card that's a pretty
big gap move into 1440p the trend
continues and posts the gtx 1066
gigabyte card at 63 FPS average 51 one
percent low and 49 0.1 percent lows the
three gigabyte unit sustains 58.7 FPS
average 45 point three one percent low
and thirty six point three 0.1% lows as
somewhat noticeable and significant
difference at times depending on what's
going on in the game Assassin's Creed
syndicate was revived because it has
historically shown us fairly large
performance gaps between four gigabyte
and a gigabyte or four and two gigabyte
solutions like the rx 480 which in a
four gigabyte model struggled in its
lows when compared to the eight gigabyte
rx 480 in this benchmark though we're
seeing similar performance from each
card the 1066 EQ by unit is slightly
ahead and there's no game breaking
disparity between the two stutters our
infrequent on both devices and those are
mostly identical OpenGL results with
doom are largely the same between the
GTX 1060 3 gigabytes and 6 gigabyte
cards 1080p results are just under 100
FPS average and the lows are fairly
tightly timed for each Vulcan testing
does produce some screwy results though
the 6 gigabyte card vastly outperforms
the 3 gigabyte card with Vulcan enabled
posting a 1080p performance difference
of 20 6.5 FPS that's huge or 37 percent
at 1440p we see similarly tied OpenGL
performance both cards operating around
65 fps more or less which is then tanked
for the 3 gigabyte unit with Vulcan
Vulcan positions the GTX 1063 gigabyte
card at 50 FPS with a 6 gigabyte ecard
at 63 fps that's a difference of 13 FPS
or 27% this difference only seems to
exist with Vulcan for Doom and because
the cards still perform better with
OpenGL you'd be best off taking that
setup instead call of duty black ops 3
is a title fairly sensitive to clock
rate and vram changes and we can
normally rely on it to show swings
between similar spec cards but it sort
of depends on what's going on under the
hood and if there's any major pre
overclock at 1080p the GTX 10-6
six gigabyte card is performing that one
22.7 FPS average with a three gigabyte
unit following at one 13.7 that's
performance change of about eight
percent the one percent low values are
gapped by 10 FPS with a 0.1% low is
similarly spaced at 87 and 79 FPS this
maintains a performance difference in
the low metrics of about 8 to 10% 1440p
post slightly bigger scaling gaps with
the 6 gigabyte card at 78 FPS average
and the 3 gigabyte card at 71 FPS
average that's a 10 percent performance
difference or twice the 5% advertised by
Nvidia 1% lows are also fairly gaped at
60.3 and 53.7 FPS we tested a lot more
games on this - you can check the
article link the description below for
Metro last light the division and a
couple of other games so it mostly comes
down to if the difference is actually
noticeable in gameplay and if so is the
$50 saved on the 3 gigabyte unit
actually worth the performance drop in
most games from what we've tested we're
seeing about a 6 to 7 percent
performance change from the 6 gigabyte
down to the 3 gigabyte card and in some
games like doom with Vulcan the changes
are pretty massive but in that scenario
you would be better off with OpenGL
anyway and there are some sort of
existing problems with Vulcan on and
Vidia with the the new cards to begin
with so that's one of those scenarios
where there is an alternative and it
exists within doom you can go to OpenGL
but when we look at the performance
disparity between the two it's pretty
large 27% in some 1440p tests more than
30% in other tests so it really comes
down to the API at some level with doom
specifically what we can't extrapolate
is how does that impact these cards in
the future because these API is it's not
just the developer poles Vulcan off the
shelf puts it in their game and it's
done they very specifically build for an
API so doom could well be an outlier but
we just don't have a good way to know
because Vulcan is still so new to the
market but we have the x12 tests and
other tests in the article length
description below if you want to read
more specifically about the new api's
and their impact on these cards as for
the other games
the x11 games especially black ops 3
shows a difference of in some cases of
10% at the high end for the performance
change between three and six gigabyte
cards and that is of course accounting
for the one less SM as well in the three
gigabyte unit and shadow of mordor shows
a 14% change is pretty big and the
biggest for the DX 11 games the three
gigabyte gtx 1060 is fine for some use
cases and there are only a few
applications where it's frames really
dip hard but when they do it's pretty
big and can be noticeable that can be
the difference of adding noticeable
chopped to 1440p gameplay in some
instances if you're trying to use this
from 1440p because the gtx 1060 can
actually handle that resolution at least
in its six gigabyte variant with the
wrong settings combination the changes
are particularly noticeable with that
vram intensive application workload on
the three gigabyte card and going
forward we'd still recommend the six
gigabyte gtx 1060 for most use cases
it's one of the only cards this
generation available at its actual MSRP
near 250 and that makes it an agreeable
purchase this is a similar situation to
the 4 gigabyte rx 480 which we generally
would forego in favor of the 8 gigabyte
model that said we did see some larger
swings in the 4 versus 8 gigabyte our X
480 performance as for naming that's
another matter this isn't a lower vram
gtx 1060 it's a different video card an
entire SM is disabled including 1/10 of
the cards processors and it's half the
vram capacity the GTX 10 63 gigabyte
should absolutely not be called a GTX
1060 for consumers who are already faced
with seemingly endless variants of AIB
partner cards X's and Z's and gaming's
and Bessie and SSC or FTW or Strix are
classified and so on and so on this only
further obfuscates the gtx 1060 pool
further it's not a gtx 1060 it's
completely different in that one sm is
disabled so invidious choice to name a
card as such will confuse buyers into
thinking it's just a 1060 with half the
vram which is plainly false this is a
gtx 10:50 TI and nvidia decided not to
call it that it's a marketing play that
doesn't mean it's a bad thing all these
companies
make marketing plays and NVIDIA is
certainly not alone in the video card
market in that regard but this isn't one
that we can brush aside as harmless
because users will inevitably make the
incorrect assumption that the only
difference is a VRAM as for being VR
ready which I believe at least one of
these boxes says on it don't buy this
for VR the three gigabyte difference
does matter there a lot of VR
applications are recommending four
gigabytes or more of memory on the video
card which obviously three gigabyte card
does not fulfill that requirement or
recommendation and VR has got very tight
tolerances so looking at these 3d non VR
games if we're losing six to seven
percent performance kind of at a low
average that amount of performance is a
big deal in VR and obviously scannings
are different there there's dynamic
settings changes on the fly all that
stuff but a couple percent is the
difference between being able to play a
game at a slightly higher setting or
with greater fluidity and VR and not
being able to do that though we are
still working on VR benchmarking and
that will come up in the future at some
point hopefully semi soon as for the
three gigabyte GTX 1060 it has place in
a few bills so if you've got an odd
budget distribution the card does
deserve to exist I've seen some people
say that it doesn't deserve to exist I
would say there is a place for the unit
I do take issue with the name but as a
product $50 cheaper it's okay where it
is if you've got a budget distribution
that calls for a cheaper video card
maybe you want to invest the $50 in an
SSD but you're still building a budget
system or get a better power supply or
something like that
it's okay there but where possible we
would still recommend going for either a
six gigabyte GTX 1060 or in applications
where it makes sense and rx 480 at 8
gigabytes and you can check our 480 and
our 1060 reviews for more discussion on
those specific scenarios when you might
want one or the other but for three
gigabytes for six gigabytes I would mean
six three is okay in some specific
budgets so that's all for this time as
always patreon like the post well video
link the description below for more
information subscribe for more I'll see
you all next time
you
you
We are a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for us to earn fees by linking to Amazon.com and affiliated sites.