NVIDIA Tegra 4 Benchmark, Power Saving, and Web Speed Demonstration
NVIDIA Tegra 4 Benchmark, Power Saving, and Web Speed Demonstration
2013-02-26
okay so we're here with nvidia and
they're about to show us some benchmarks
were very excited the moment of truth is
upon us I'm neeraj shridhar he's with
technical marketing with nvidia and he's
going to show us a few things that
demonstrate the power i'm taking for
platform so i'll let him take it away
sure hey guys are all of you must have
seen the benchmark results of Tegra 4
that was recently published but many of
you may be wondering whether Tegra 4 is
really a low part ship and here I have a
demo to prove that degra forked consumes
really low power and can fit in phones
and tablets what I have here is our
Tegra 4 phone development platform
obviously this board is really big it's
not going to fit in a phone and the
reason for for that is that it's got all
these debug ports that are software
engineers and our OEM partner engineers
used to debug the phone to optimize it
for performance and part and I hooked up
this development go to a power meter
through these wires and this parameter
is basically measuring the power
consumption of the phone at the battery
terminal right now the phone is in
standby mode and it's consuming less
than 10 milli watts apart the
sensitivity of this meter is like 10
nemeth 10 milli watt range that's why
it's showing zero but the moment I turn
on the display you'll see it will spike
up a little bit but eventually it will
settle down to around point seven words
so basically when the phone is idle with
the display on it's consuming only
around Oh point 7 watts and most of that
power consumption is due to the you know
the display backlight power let me
unlock the screen here and you guys can
take a look at the meter
there you go the phone is a sitting idle
the power is settling down it comes
comes down to 0 point 7 watts in idle
mode so let me launch the gallery and
play a 1080p video clip play start over
and tilt this board a little bit so that
the video comes to landscape mode and
let's let the power settle down and what
you notice is this is a 1080p screen the
phone is decoding a 1080p high profile
clip and you see it's consuming only
around point 88.9 wats when you compare
this to a computing system a 1080p phone
like the droid DNA for example we found
that the droid DNA consumes around 1.2
wats so Tegra 4 is actually considering
twenty-five percent lower power than
competing systems out there and it gives
you approximately two to two and half
hours more battery life so you get both
high performance as you've seen in the
benchmarks and low power from Tegra 4
excellently in terms of you know the
relative performance a lot of the
industry analyst from what I've read at
least they say that Tegra 4i is not take
her for can you speak to that a little
bit well let us it's not Tegra 4 but it
as performance comes close to photograph
for work and user experience is close to
Tegra 4 and the reason for that is you
can play all the Tegra 4 optimized games
on Tegra 4i because it's got a 60 core
geforce GPU it has the same i 500 modem
architecture in there it supports the
same computational photography features
the Tegra 4 supports and it's going to
deliver great web browsing great gaming
experience and reading long battery life
so in all these aspects I I believe that
you note 84 is much closer to Tygra for
performance but and clearly it's aimed
at the mainstream smartphone market and
Tegra 4 is for the no superphone
high-end hero smartphone and tablet
markets a person with ridiculously high
resolution right I was reading that
apparently we're on the dawn of them
having 4k 4k resolution display
tennis is one factor you know it's not
too far off there are they getting ready
for it yeah but I mean take her for can
actually support you know that
resolution almost yeah the Tegra 4i
think they have a demo downstair that
shows Tegra 4 driving a 4k TVs can
decode a 4k video flawless area you're
very confident in saying that we can
continue to increase performance but at
the same time we can get more efficient
devices right and this is technically
possible you know yeah definitely
because the process technologies you
know continuously improving like you
know a few years ago we were on 65
nanometer 40 nanometer 28 and they're
going down and these processor
technology improvements gives us you
know more power efficiency and
architecture is also getting more
efficient like we went from single quote
of dual core and sub well-known fact
more course running in parallel you know
can run at lower frequency and voltage
and gives you better power efficiency so
I think that trend will continue and as
we provide more and more performance in
these chips people will start you know
using up this performance I mean if you
recall a few years ago when nvidia
introduced the first dual-core mobile
processor people who are wondering do
you really need a dual-core in a smart
phones which three but they're pretty
soon everyone moved to quad core because
they realize you need this small
performance and as they use these
devices more often they want to use it
like that own personal computing device
and you know and that's what we're going
at run speaking with our last night
actually they were saying that within
two years we'll have a proven
performance to avoid used to
experiencing a new pc could be yes in
terms of android and whatnot i mean
obviously I know anybody is a big
supporter of Android but is it true
really that Android can take full
advantage of this mini-course is it you
know
the app developers have some flagging in
a single core instruction set mentality
not really I know Android as you know
was based on linux and unix add support
for multi-core and android today
supports multi-core architectures like
the operating system inherently is
capable of scheduling threads across
multiple cores through the web browser
takes advantage of multiple cores
exactly especially the chrome browser
then supports multiple tabs can like
scheduled threads on for each tab on
different cores and demonstration that
can perhaps show something to that
effect this is a web browsing demo what
we have set up here is this a Tegra 4
development platform and this is a nexus
10 that is powered by a dual-core a 15
and both these tablets are connected to
a web server through you know wired
Ethernet okay and the reason we are
going wired and reason we are going to a
local web server that is under this desk
is to eliminate all the network latency
the server load Layton sees and just
purely benchmark the capability of the
you know these two tablets so I'm going
to launch this web page load test and
what this test is doing is it's
basically loading up 25 websites and
these 25 websites are the top 25 sites
that are most visited in the US and we
rip those sites store them on this local
server and I'm going to launch this test
on the Tegra 4 and on the Nexus 10 and
as you can see Tegra 4 is really like
you knows loading this fast and the cash
has been cleared the cache has been
cleared everything is being loaded from
the server and strike or a 15 the
services quarry floating rate of things
and so Tegra 4 will load these 25 pages
in approximately 25 to 27 seconds and
you'll see that soon so I gadget there
yeah so there you go 27 seconds to load
up 25 pages that's approximately a
second and a half well the Nexus 10
takes anywhere between 45 to 50 seconds
you know delivering a sixty to seventy
percent performance benefit on the
quad-core tegra for substantial program
so really when do you think we're going
to be able to see Tegra 4 devices in the
wild earliest q2 of this year q2 yes
yeah ziti and Toshiba already unknowns
that you're working with us to bring the
tablets and phones to the market and
you'll see more devices show up early in
57 seconds for the nexus 10 the a-15
Joker versus the 27 seconds for the book
almost two times faster and this is sort
of the the fifth sort of ninja chorus
yeah definitely yes so yeah tigra Flores
has this 4+1 architecture with the fifth
battery saver course right so the
battery saver core itself is a you know
cortex a15 goal but it has been
optimized for ultra low power
consumption so for example let's say
right now the sitting idle doing nothing
so we turn off all the four main course
and just run it off the battery saver
code and if the display is turned off
and you're just doing you know social
networks things in the background or
playing musics all of that runs on the
battery saver co nice and the fact that
on Tegra 4 that battery saver code is
cortex a15 it's much more powerful so it
can handle more Orange yeah it can
handle a lot more task in this low-power
State San and are you know the four plus
one technology is the second generation
technology on tagged report which means
we get even more power saving because we
have optimized the algorithms further
that's from nice thorough rundown I'm
completely sold how much you guys are
bringing to market thank you really one
of this to say thank you very much for
your time that's nice not
congratulations thank you nice talk to
you too all rights android authority
we're going to sign off for today we
pulled an all-nighter last night to try
and get you some videos online we'll
have lots more moves news on nvidia and
lots more to come here from Mobile World
Congress in Barcelona
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.