JAMES GORBOLD / HARDWARE ACCELERATED
James Gorbold has been building, tweaking and overclocking PCs ever since the 1980s. He now helps Scan Computers to develop new systems.
OPINION
T
hroughoutthevariedhistoryofthePC,itsevolution
has beendrivenbymanydifferentfactors.Oneof
the mostnotable,ofcourse,wasthedevelopment
of the gaming marketinthelate1980sfroma cottageindustry
into the billion-dollarindustryoftoday.
As we saw inlastmonth’sinterviewwiththefounderof
Creative, a significantcauseofthischangewastheemergence
of sound cards, finallygivingthePCsoundcapabilitiesinline
with other homecomputersoftheday,suchastheAtariST
and Amiga. I stillhaveveryfondmemoriesofmyfirstSound
Blaster, not justbecauseitwasthefirstpieceofhardware
I bought, but becauseinconjunctionwithgamessuchas
Wing Commander,my 286 feltlikea true
gaming computer,ratherthana clunky
beige office computer.
It’s been quitea fewyearssincewe’ve
seen such a majorseismicleapforwardin
technology as thesoundcard,withother
notable leaps suchasGPUs,multi-core
CPUs and SSDs alloccurringovera decade
ago. However, there’s one factor that continues to drive the
PC forward, albeit sporadically, and that’s the continuing
game of leapfrog between the four sectors of the PC market:
desktops, workstations, servers and laptops. Just as we saw
the technology in a business machine used for games in the
late 1980s, we see that technologies developed for one of
these markets occasionally has unforeseen and unplanned
effects on the PC market.
A great example was Intel turning to laptop CPUs to help
save its desktop CPUs. The trouble came about because of
the lengthy pipeline in Intel’s NetBurst architecture, which
was expected to reach the heady heights of 10GHz, but was
sonotoriouslypower-hungrythatit simplywasn’t possible
toproducesuchCPUs.
RatherthanperseveringwithNetBurstand its successors,
IntelturnedtothePentiumM,whichhadbeen designed for
laptops,andcouldtraceitshighlyefficient short pipeline
designwaybacktothePentiumPro,Intel’s first dedicated
serverCPU.WhilePentiumMitselfwasn’tIntel’s saviour, it
ledtothedevelopmentofthefirstCoreCPUs, helping Intel
toregaincompetitivenessinthedesktopmarket.
However,arguablythebestexampleofone part of the
PCmarketchanginganotherwasthediscovery in the mid-
2000sthatGPUscouldbeusedfora lotmore than playing
games,andwerealsoincredibly potent
toolsforparallelcompute tasks, such as
videoencoding,rendering, simulations
anddeeplearning.Thisin turn led both
AMDandNvidiatodevelop billion-dollar
datacentreGPUbusinesses in addition to
thegamingGPUsreviewed in Custom PC.
ThestoryoftheGPUhas come full circle
recently too, with the design of datacentre GPUs now influencing
the design of gaming GPUs. For instance, the Tensor cores
used for DLSS in the latest Nvidia Turing GPUs (see p86) only
exist because they were first developed for the earlier Volta
architecture, which Nvidia used to build datacentre GPUs. A
true case of the tail wagging the dog.
It will be interesting to see how much future generations
of gaming and datacentre GPUs have in common, especially
with a third major player, Intel, due to enter the market in- I’ve long believed that Intel will go after the datacentre
market first, but that doesn’t mean an Intel gaming GPU isn’t
also planned for some point in the future.
The Tensor cores in the latest
Nvidia Turing GPUs only exist
because they were first
developed for datacentresCOMPLEX EVOLUTION
James Gorbold explores the ways in which mobile and datacentre
tech can influence PC gaming hardware, and vice versa