WITH NEW BENCHMARKS comes
a need for new zero-points and test-
beds. Not only is this a pivotal step for
our reviews, but it’s also important for
our system builds as well. Test-beds
are fairly complex things to organize,
as you need to have some flexibility in
them while also trying to keep themas closely aligned as possible to one
another to reduce inconsistency. For
instance, when testing AMD and Intel
processors, if we’re comparing them to
one from the competition, it’s ideal if we
have motherboards that cost roughly
the same, from the same manufacturer,
with a similar feature-set across the two
platforms. Then it’s all about pairing that
with a single high-end GPU (in this case
an RTX 2080 Super), two separate but
identical SSDs for the OS (one install for
AMD, and one for Intel) to avoid chipsetand program conflicts, and then the same
memory kit as well, in this case 32GB
(2x16GB) of DDR4 @ 3600MT/s.
Moving to graphics cards and SSDs,
however, complicates things. GPUs
have typically always performed better
with Intel processors due to higher IPC
and single-core performance. Ryzen’s
certainly come a long way since it first
debuted, but it’s still slightly behind its
blue competitor. But AMD has one trump
card in its arsenal, and that’s the fact it’s
the only platform right now that supports
PCIe 4.0, doubling the effective bandwidth
of any device that uses the connection
standard compared to last gen.
There are two reasons why that’s
important, and why for the first time in
over a decade we’re turning to AMD for
our graphics card testing. We expect
that within the next 12-24 months to see
the first PCIe 4.0 graphics cards emerge,
especially as RTX matures in its second
iteration. Committing time to testing
GPUs on a PCIe 3.0 platform, and then
needing to retest our entire inventory
when the new cards launch is just not
feasible. On top of that, there’s already a
plethora of PCIe 4.0 SSDs on the market.
So if we can keep a single high-spec
platform available for both SSD and GPU
testing together, it saves time and energy
stripping down parts off of the test bench.Our gaming laptop zero-point is the ............Our gaming laptop zero-point is the ............WORKSTATION SYSTEM BENCHMARKS
ZERO-
POINT
Cinebench R20 Single (Index) 00 000 (00%)
Cinebench R20 Multi (Undex) 00 000 (000%)
HWBOT X265 @ 4K (fps) 00 00 (-00%)
3DMark: Fire Strike (Index) 00 000 (00%)
AIDA64 Memory
Bandwidth (MB/s)^00 00 (0%)
CrystalDiskMark 7.0 Sequential
Read QD8 M.2 (MB/s)^0000 (-00%)
CrystalDiskMark 7.0 Sequential
Write QD8 M.2 (MB/s)^0000 (-00%)
CompuBench 2.0 N-Body
Simulation 1024K (Iterations)^00 000 (000%)
FAHBench (Index)^0000 (-00%)
Adobe Premiere Pro
Benchmark @ 4K (m:s)^00 000 (000%)
Adobe After Effects
Benchmark @ 4K (m:s)^00 000 (000%)
Power Draw Fire Strike
Combined (Load) (W)^00 00 (-00%)
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%GAMING SYSTEM BENCHMARKS
ZERO-
POINT
Cinebench R20 Single (index) 00 000 (00%)
Cinebench R20 Multi (index) 00 000 (000%)
CrystalDiskMark 7.0 Sequential
Read QD8 M.2 (MB/s)^0000 (-00%)
3DMark: Fire Strike (Index) 00 000 (00%)
3DMark: Port Royal Ray
Tracing (Index)^00 00 (0%)
Total War: Three Kingdoms (fps) 00 00 (-00%)
Assassins Creed Odyssey (fps) 00 00 (-00%)
Metro Exodus (fps) 00 000 (000%)
Red Dead Redemption II (fps) 00 00 (-00%)
Adobe Premiere Pro
Benchmark @ 4K (m:s)^00 000 (000%)
Power Draw Fire Strike
Combined (Load) (W)^00 000 (000%)
Max CPU Temp Fire Strike
Combined (Load) (C)^00 00 (-00%)
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%The Future of
Te s t B e d s
Zero-Points
ZERO-POINTS ARE fixed systems that we
compare our system builds against. When
we last overhauled the benchmarks, we
actually moved away from using a single
overly expensive zero-point, to using three
different ones instead. These systems
were aimed at three different price points,
with different gaming resolutions in mind
(1080p, 1440p, and 4K), and loosely based
on our Blueprints section.
Now we know that not everyone plays
games, but gaming PCs are typically
jack-of-all-trades when it comes to other
computational tasks. They can handle
anything you throw at them, but obviously
not to the same degree as a machine
specialized towards a particular purpose.
To that end, if we do build a very specialist
workstation rig, comparing it against a
system like this gives us a good insight as
to just how much performance is gained
for that extra investment.
To that end, we’ll be introducing a
brand new set of ZPs for build-it and
feature systems. We’re not quite sure
what they’re going to contain just yet, as
we’re still working out the details with a
new retailer to help support us.benchmarking
32 MAXIMUM PC SEP 2020 maximumpc.com