Test setup and application benchmark suite
Test
system specifications
The synthetic and gaming tests are done on
the MSI GX60, Razer Blade R2, and Alienware M18x. The Blade utilizes GeForce
GTX 660M mobile GPU of Nvidia, whereas the M18x deploys an Nvidia GeForce GTX
680M. The 7970M and GTX 680M stand for the fastest mobile graphics modules
currently available. The two comparison machines sport 8GB of RAM, and their
CPUs and GPUs run at its factory speeds.
We conducted our productivity-oriented
tests on the systems that are equipped with the previous-generation Core
i7-2720QM quad-core and -2640M dual-core processors. Plus, we added a 45 W Core
i7-3720QM quad-core chip to show the discrepancy that adding extra 10W creates
the GX60’s 35 W APU. While we did not have any other gaming laptops on- hand
with these specifications, we had a couple of Dell Latitude E6520 and a
Latitude E6530 computer. The Latitude lines are mainly used as business
laptops, but if you consider these tests based only on the CPU, we have no
problem using them.
All performance tests were run on an
optimized image of Windows 7 loaded on a 180 GB Intel SSD 520 drive, which we
installed ourselves. The compression tests were run on a RAM drive to minimize
the effects of storage hardware on each machine.
A
note on FPS
We recognize that frame rates (fps) that
are acceptable is something we are able to dispute on the basis of each game
and each provider, which is the reason why we only listed the average speeds.
As long as we are going to play games smoothly, we often choose higher-quality
images rather than higher frame rates. Ultimately, we wish to see all of the
work that the game designers have done to make games visually appealing.
Normally, we find that 20fps is the minimum level we are willing to call
acceptable, with that 30fps is an ideal minimum level. In multi-player games,
we prioritize frame rates rather than low latency.
Synthetic tests: 3DMark
With the latest version of 3DMark
available, we know that 3DMark Vantage and 3DMark 06 are starting to show their
age. However, the two tests still tell you the system performance, and they
allow us to compare older hardware that you may still own. The comparisons were
conducted for the Razer Blade R2 with the Nvidia GTX 660m, as well as Alienware
M18x with the high-end Nvidia GeForce GTX 680M.
In
the chart above, the scores of GPU and Physics show us a great deal of
information about the individual performance of the AMD A10-4600M APU and
Radeon HD 7970M GPU. The Radeon proved to be faster than Nvidia's best GeForce
GTX 680M.
In the chart above, the scores of GPU and
Physics show us a great deal of information about the individual performance of
the AMD A10-4600M APU and Radeon HD 7970M GPU. The Radeon proved to be faster
than Nvidia's best GeForce GTX 680M. Moreover, the Radeon is nearly 3 times as
fast as the GTX 660m, which is often present in the systems with equal, or
sometimes even higher, price levels than MSI’s GX60.
The A10-4600M APU falls short of the
Physics score compared with the two quad-core-i7 CPUs. Whether the A10-4600M is
a quad-core chip, it runs like a dual-core part as a result of the share of its
shared resources.
These results showed the major compromise
that MSI offered decrease the GX60’s price. The GPU is faster than anything
now, while the CPU gets far behind Intel’s quad-core chips. In a computer
designed to play games with the lowest possible cost, this compromise appears
to be reasonable. As long as the CPU is likely to catch up with the GPU, there
will be no loss in gaming performance.
3DMark Vantage reduces the number of
high-end rendering features that these GPUs have to demonstrate. The GTX 680M
goes ahead in GPU scoring, while differences in CPU scores prove to be much
more prominent.
3DMark
Vantage reduces the number of high-end rendering features that these GPUs have
to demonstrate.
Returning to DirectX 9, the 3DMark 06
further varies the GPU results, prioritizing Nvidia’s card. But we feel about
how each system might run in the older games. The CPU results reach close to
displaying real-life performance, with the A10-4600M running at about half the
speed of Intel’s Core i7-3632QM.
Returning
to DirectX 9, the 3DMark 06 further varies the GPU results, prioritizing
Nvidia’s card.
On the whole, 3DMark testing confirms for
what we waited: GPU performance is classy, while CPU performance is not as
impressive as high-end chips from Intel. The GX60 APU and GPU work together
like in games is what we actually want to know out there. Intel’s high-end Core
i7-3940XM is 3.5 times faster than the GX60 A10-4600M in 3DMark 11, but the
Core i7-3940XM is priced at nearly $1,100. The entire GX60 package only comes
at a $100 premium compared to that chip! Therefore, how much is the
3.5-time-faster CPU speed worth? If you are a gamer who owns the same frame
rates, are you interested in?
Productivity tests
We conduct our productivity-oriented tests
on systems that are equipped with the previous-generation Core i7-2720QM
quad-core and -2640M dual-core processors. Besides, we added a 45 W Core
i7-3720QM quad-core chip to show the difference that an extra 10W creates the
GX60’s 35 W APU.
The GX60’s AMD A10-4600M APU takes almost
more than twice as long to encode videos as the Core i7-2720QM. However, these
HandBrake results arrive with the caveat that the current version does not
support OpenCL acceleration, which reduces much of the workload for a
compatible GPU. Running the currently compatible version of HandBrake with
OpenCL support will reduce encoding times of the A10 in half, putting it very
close to the quad-core Core i7s. Once the OpenCL-supported version of HandBrake
becomes stable, the A10-4600M may rank first in this chart.
While Premiere CS6 supports OpenCL to
accelerate specific effects and playback, Adobe Media Encoder does not. This
means that Premier Pro CS6 runs faster with AMD GPU acceleration in the real
program, but once you export the video to encode, you just go back to the
CPU-only.
Once again, the AMD A10-4600M is half the
speed of the quad-core -i7 chips. If Adobe supports OpenCL acceleration in
Media Encoder, the A10 may do its best to keep up with.
The CPU-only Photoshop CS6 results keep on
showing similar performance.
CPU-only file compression benchmark also
indicates that the AMD APU falls behind the Intel CPUs.
When we consider the productivity
performance measurements from the GX60, there are a couple of things to take
notice:
·
First, the measurements are to help boost CPU
performance compared to the other CPUs. The cores of the A10-4600M x86 are
unable to catch up with Intel's chips. However, as more and more applications
fully support OpenCL, this will rapidly change because the A10 has a very
decent built-in GPU (not to mention the high-end separate 7970M of the GX60).
·
Second, you should make sure that you have the
valid priorities. The GX60 has one purpose: serious gaming with little money as
possible. If you need more performance than productivity, you should pick a
machine up with a faster CPU. However, you will also need to pay more or
sacrifice performance in GPU in the process. If all you wish to do is game, the
only thing you should be concerned about is whether the CPU is fast enough so
as not to bottleneck the GPU.
Despite the
fairly poor results, the GX60 is fast enough for daily use. And as you will
see, it is also very strong in in real-life games.
