We tested NVIDIA's new Max-Q technologies for laptops
animated by Series 30 RTX video cards, developed to achieve a stable framerate,
greater noise and temperature control and, of course, a more enjoyable overall
user experience. than in the past. NVIDIA's approach is based on multiple
factors. We have previously had the pleasure of traveling together the short
avenue of memories with RTX technologies, techwadia between Ray Tracing and Artificial
Intelligence . We therefore focused on the DLSS and on the model that allows
you to obtain an incredible performance boost without affecting the overall
quality. Today we will focus on other optimization techniques perfected by
thegreen team for users who prefer mobile gaming .
Resizable BAR, why right now?
One of the most stimulating points is certainly the
resizable bar or Resizable BAR , made available since the inauguration of the
Mobility line on the entire range of GPUs with Ampere architecture, unlike the
desktop line whose forerunner was the RTX 3060 launched precisely in the past
few months.
The resizable bar is no small matter. In fact, it allows to
give the CPU free access to the entire framebuffer of the video card, without
limiting access to the canonical 256MB and therefore forcing it to create stack
of commands. Most portable configurations activate this out-of-the-box mode,
therefore it will not be necessary to implement any type of manual intervention
to benefit from it. All that comes to mind when we think about the logic behind
technologies like Resizable BAR is one, simple question: why not before?
This mode of communication between the CPU and PCI-Express
devices did not arrive on the market before this generation for very specific
reasons. First of all, the intrinsic weight of the most modern titles made it
necessary to make a " call to arms " to improve as much as possible
the optimization of all the elements that contribute to the final yield. Among
these, the greater specific weight could have been given by the DLSS, with
increases sometimes close to 50% of the framerate.of departure.
In particular, what interests us most in this case are the
game assets, loaded from time to time into the memory available to the GPU to
be accessible during the pipeline . So what has changed compared to the past?
Well, if we look at modern gaming, in the space of a few years we have gone
from high-end video cards with 320MB memory cuts to a space available rarely
below 8GB, as also demonstrated by the most recent Steam polls . It is clear
that everything acquires a very precise meaning if contextualized in this way
since, if a decade ago the space available for the CPU instructions tended to
overlap the entire memory, today we are talking about tiny segments with
respect to the totality of the framebuffer. No less important is the complexity
index of the instructions that will then be sent, which will contribute to
creating the stack of commands waiting for their turn.
Currently Resizable BAR is able to provide a substantial
increase in performance in some highly selected titles. As suggested by NVIDIA
itself and also verified in the field, this technology cannot provide an
advantage without painstaking optimization work within the game engine. Some
titles, due to their conformation, will be able to obtain boost even up to 10-12%
without any external intervention, while others may even experience drops. For
this reason, it will be NVIDIA to watch over our GPUs, individually testing the
titles on the market and guaranteeing support for those who are able to fully
benefit from Resizable BAR.
Among the titles we had the opportunity to try, the impact
on Assassin's Creed: Valhalla was remarkable with an increase of 6-8 FPS on an
average base of 70, capitalizing a new ten and making the framerate even more
stable . Small but not imperceptible the impact on the minimum FPS. Another
interesting experience was Metro Exodus in both the original version and the
recent Enhanced Edition, where we were able to appreciate the extraordinary
stability of the video stream without any kind of uncertainty.
Dynamic Boost 2.0 and the TGP
We then tested the rest of NVIDIA's Max-Q luggage. Another
weapon at our disposal from this point of view is the excellent second
generation Dynamic Boost . In its first version it allowed to break down the
barriers imposed by the thermal design, overcoming the pre-established limits
in a univocal way. In this way, the video card could draw power from the one
reserved for the CPU. Today this technology allows to perform a real
intelligent power shift , in both directions. It goes without saying that now
even the most CPU-intensive tasks will benefit.
What interests us most, however, is to try to understand to
what extent a technology of this kind can benefit us in a particularly limited
situation, represented in our case by an RTX 3080 at 85W . We are facing a very
particular configuration since the limit of 85W theoretically involves a
significant bottleneck for the potential of such a powerful GPU. This limit is
more evident in traditional rasterization contexts. In the most modern titles,
the implementation of Ray Tracing calls on the other computing units present in
the GPU, namely the RT Cores and Tensor Cores, drastically limiting the
influence of the power limit.
We certainly couldn't avoid trying Dynamic Boost 2.0 on the
one hand to see if we could actually get real gain, on the other hand to
understand how far we could go this card. First of all, its necessary to
specify that Dynamic Boost 2.0 cannot be deactivated through a special switcher
, but by disabling the appropriate entry in the Windows Device Manager . For
obvious reasons, we recommend that you leave these values unchanged and proceed
as planned by the manufacturer.
For the first field test, we asked for a traditional title.
Forza Horizon 4 at Ultra details and with FullHD resolution allowed us to
evaluate and quantify the impact of Dynamic Boost 2.0 and we must admit that
indeed the difference is there and it is sensitive. Numerically we are faced
with a widely playable title in both cases but turning off Dynamic Boost
involves a loss of 3% in terms of framerate , with an instant passage from 200
to 193 FPS. There is a gain, but certainly less impactful, even with active Ray
Tracing. For the test, the excellent Metro Exodus came to our rescue. We wanted
to make some considerations with both the traditional title and the Enhanced
version. In the first version of the game we noticed a minimum gain of 1% with
active Ray Tracing, which becomes a more generous 4% by turning off the next
generation global lighting instead.
In Metro Exodus Enhanced Edition, Ray Tracing is, by
definition, one of the pillars. For this reason it cannot be deactivated and
the result obtained reflects the values obtained previously. Even in the
Enhanced version of the 4A Games title we have in fact obtained a gain of about
1% in the average framerate, with and without DLSS.
Numbers then confirmed on Control. Even in the title of
Remedy in fact,the use of lighting in Ray Tracing involves a general leveling
of performance, where the use of the cores dedicated to these calculations goes
to enormously balance the limits imposed on the GPU. Here, too, we travel
within 1% of net gain, with an advantage in watts which in all cases is around
15 W. In fact, from the 85 declared, the use of Dynamic Boost 2.0 has allowed
us to greatly exceed the 95W and in some cases even 99W.
Silence in the room, let's try WhisperMode 2.0
Finally, we want to devote a couple of paragraphs to another
interesting technology. If so far we have focused on methods that could allow
us to get the most out of our computer, with WhisperMode 2.0 we move on to a
different concept of Gaming on the move . In fact, this technology aims at the
acoustic management of the system to balance the noise level of the PC on the
basis of a threshold set by the user. With this tool, integrated within GeForce
Experience , we will be able to choose an efficiency threshold that allows us
to reduce noise in relation to our "goal" in terms of frames per
second.
Of course, this technology also has its limitations.The
enormous heterogeneity of notebook systems in terms of design and thermal
design results in a different impact of this mode on the overall experience.
For this reason, some manufacturers have decided to continue to rely on
proprietary software, avoiding such "interference" on temperatures
and performance. In reality, however, WhisperMode 2.0 proves to be particularly
efficient when available.
We can really say everything and the opposite of everything
about this technology. We are faced with a particularly functional optimization
tool with a high impact on the entire system. However, it is necessary to fully
understand which aspects will act at the system level.
In fact, with WhisperMode, if on the one hand we have a
predetermined noise target, on the other we have a maximum limit of the
framerate. This limit should not necessarily be seen as a decrease in
performance : a maximum threshold is set, which the system must not exceed. But
giving up a more or less generous handful of FPS has numerous positive sides,
especially if this allows us, for example, to remain stable at 60 frames per
second.
While the maximum limit could, on a numerical level, affect
the average game FPS, stabilizing this figure on a preset limit can result in a
better overall experience, limiting the oscillations and therefore that
annoying effect of changing gears of the game when there are some problems.
drastic drops in framerate. Consequently, if the goal is to limit noise, the
greatest impact will be on temperatures . With WhisperMode, these can drop as
much as 20 degrees from the peaks experienced at maximum performance and this
in turn will result in a longer life span of the machine itself, as well as
being a source of comfort during the most intense sessions.
newyorkersblog cosmopolitansblog realsimpleblog nextwebblog theinformativeblog
