RICHARD SWINBURNE / VIEW FROM TAIWAN
OPINION
A
fter launching its RTX 3000-series of GPUs for laptops,
Nvidia decided to remove the Max-P branding,
allowing laptop makers to set the chips’ power draw
and clock speeds in practically any way to fit their designs.
The Max-Q branding stays, but you could be buying any of
28 variations of the RTX 3060, 3070 or 3080.
This means an RTX 3080 with baseline 80W TDP would
have a fraction of the performance compared with one with a
full-fat 150W TDP, yet it still costs the same amount of money.
Buying a mobile RTX 3080 only guarantees that you’ll get
6,144 CUDA cores – otherwise, the base clock ranges from
780MHz to 1350MHz, while boosts vary from
1245MHz to 1710MHz. Nvidia says you can
find this information in the Control Panel,
which is pretty useless if you’re still deciding
which laptop to buy.
After some outcry, Nvidia recently changed
its mind, saying it ‘strongly encourages’
(but doesn’t require) partners to list the power use and clock
speeds on the product pages of each laptop, which it says
‘communicates the expected GPU performance in that system’.
No, it doesn’t. Due to the way transistors work, increasing
power use doesn’t increase performance linearly, and the
cooling factor of any laptop isn’t quantified, so how do you
know how any system will actually perform?
If you’re shopping for a new gaming laptop, you’re better
off creating a spreadsheet of data, cross-referencing the
chip spec with power and boost clocks. There’s an 88 per
cent difference in power draw between the top and bottom
options of each chip, so it’s exceptionally difficult to make a
reasonably accurate guess. The Max-Q line-up is slightly less
confusing, with only three varying options for each chip, but
there’s still a 92 per cent difference in power draw between
the lowest and highest versions (60-115W)!
Product names used to signify performance on a branded
scale of low to high, but they now merely convey the silicon
being used. As this latest naming rule doesn’t require
disclosure on sales pages such as Amazon, you’ll have to dig
into manufacturer websites to find it. What’s more, laptops
can have the same name but different specs in different
regions, so you have to be careful which URL you’re checking- how many times have you ended up reading the ‘global’ US
website instead of the local UK one?
What’s particularly frustrating is that
Nvidia is keen to protect its branding in every
other product – G-Sync monitors have a long
list of very strict quality and performance
requirements to which manufacturers
have to adhere before they can use the logo.
Nvidia does this because it claims it cares about the gaming
experience, so why are laptop chips any different?
Nvidia could require every laptop to issue a performance
quotient for easy comparison – an RTX 3060 at 80W could
be 1, then an 80W RTX 3070 might be 1.3, but an RTX 3060 at
150W might be 1.4. This gives buyers a clearer understanding
of what they’re getting for their money without having to
dig into every laptop maker’s product page, while having to
guess how much power equals how much more performance.
Reviews can ultimately provide this insight, but manufacturers
rarely sample every spec, so there are inevitable gaps. If Nvidia
really cares about the quality of gaming experience, it will
do more to articulate these details up front.
WHAT’S IN A NAME?
Buying a laptop with a GeForce RTX 3060, 3070, 3080? You might
be getting one of 28 versions, explains Richard SwinburneRichard has worked in tech for over a decade, as a UK journalist, on Asus’ ROG team and now as an industry analyst based in Taiwan @ricswi
Increasing power use
doesn’t increase
performance linearly