A New Computing Performance Scale to Resolve the Disparity between Consumer Technical Knowledge and Increasing Complexity in Regard to Consumer Purchases

tangibleassistantSoftware and s/w Development

Dec 3, 2013 (3 years and 8 months ago)

57 views

A New Computing Performance Scale to
Resolve the Disparity between Consumer
Technical Knowledge and Increasing
Complexity in Regard to Consumer Purchases


Categorisations

and scales have long given order to otherwise technical and robust concepts
and aspects of discussion. From Carl Linneaus’ binomial system of classification in his
‘Species Plantarum’ through to IEEE’s very own classification systems and bands across the
radio spectrum.

Classifications and scales give interested parties a means of easily determining order and
understanding the nature of difficult concepts. In the computing age when exploration and
new goal posts for human development lie not in the distanc
e travel or lands explored, but in
the ground gained in computing power to calculate and envision fundamentally new concepts
and even conjure whole industries. It seems in users there is a growing gap between ability to
easily use and understand simpler an
d simpler software
and the

complexity of hardware and
technical facets of comput
ing/mobile technology.


O
ver 1.6 billion

people currently own a computer
according to
Computer Industry Almanac
.
With 87 million computers

sold

in the first quarter of 2012
. Th
e UN considers the growth of
computing/mobile and the internet to be such an important part of modern living that it has
declar
ed internet access a human right
.

The discrepancy lies in the consumer’s shallow understanding of more and more complicated
compo
nents associated with compu
ting. When making a purchase,
computer users often need
help consulting experts on what to buy; do they need a particularly powerful CPU for what
they are doing? Do they need a dedicated graphics card?


Is the sales rep genuinely helping them or is the sales rep trying to push an unsuitable product
upon them in order to make a sale. If two
CPU
s are particularly similar but one offers a clock
rate of 3.30GHz as opposed to the latter offering 3.20GHz. Does
this necessitate a huge
change in quality? Does it matter? The answer is yes. For example an Intel Core i7
-
3960X
with a clock rate of 3.30GHz can retail at $999.99USD, as opposed to an Intel Core i7
-
3930K
with a clock rate of 3.20GHz which can retail at $5
69.98USD. If one exam
ine
s the two CPUs
the difference is not significant. Both are exceptionally high end at this time but the price is
phenomenally different.


If one were to examine the PassMark’s benchmark test scores for both of these they would
discov
er that the former scored 12880 across 49 samples and the latter scored 12084 across
179 samples. The overall rank of the former across all CPU’s according to PassMark is 11.
The latter is 12. Thus both despite being exceptionally high end CPU’s capable of

most
conceivable modern consumer requirements


have a price difference of over $440USD.


This represents a significant loss in savings unnecessarily for a consumer who hasn’t got the
technical understanding to determine the technological difference.


The

same issue spans across many components such as video cards. Looking again at high
end graphics cards and using the GeForce GTX 680 and the GeForce GTX 690.

The former is the top rated video card according to benchmark tests, with a sampling of 286 it
sco
red 5497, the latter being particularly new scored 4880 with a sampling size of 49.


So not only is the user experience and performance better with the GTX 680 despite some of
the better technical aspects of the GTX 690. But the price difference between th
em is a
significant $560 USD


where the former costs $439.99USD and the latter costs
$999.99USD
.


The reality is both of these CPUs and both graphics cards have negligible performance
differences. The overall quality of both sets of components is extremel
y high. The only
perceptive difference is minute and the prices vary wildly.

In the interests of accuracy it has
to be said that there are some benefits to the GTX 690. It can interface with four monitors at
once, and in theory its two GK104 Kepler GPUs gi
ve it a theoretical double to any figure by
the GTX 680. However in practice each GPU experiences a slower clock speed (overclocking
not considered). Despite this theoretical doubling the actual performance whilst superior to
the 680 is still similar enoug
h that the pricing difference seems unwarranted.


It is important to note as well that the two wouldn’t compete with each other in any new scale
or system


but in fact would be clustered together at the same level.
The average consumer

cannot be expected
to know the technical ins and outs of the two cards

nor can they be
expected to thoroughly research to that technical level. But the reality is one user could
purchase a product for $1000 more than another user
and be $1000 worse off for the price
versus p
erformance value.


Which is where a new system is utterly essential if we are to progress as a society. The
proposal is to have a system where all components are giv
en a rating by out of
6
.

This could
be done easily with pre
-
existing benchmark databases an
d figures so the undertaking would
not be arduous.

This system should be watched fairly and accurately by an independent group

(Such as the IEEE)

where manufacturing and commercial interests could not unfairly taint the
scale; with all components accurately conforming to a scale where the top performing
components are given a score out of
6

and those with the weakest performances given a score
of 1

and the rest filling the void in between.


This would then allow for an overall score per laptop/mobile device that would better inform
the purchase of computers and mobiles. It would be useful also to determine where various
uses for computers correlate
to the scale. For instance a high end gamer would require a
computer or laptop with a level
3
-
6

score. Where
6

is compatible at an optimal level with the
most advanced games currently available


and
a 3

conforms to being able to play such
games at the rec
ommended level, with

some performance sacrifice.
The new system would
inform the decision making process better for users who want maximum value for their
budget. In the case of the graphics card example mention
ed

previously


where both the GTX
690 and GT
X 680 would score a
6
/
6

on the scale, it would then be down to the consumer to
isolate the difference between the two cards.

However the important difference is they know they are looking at an overall quality of
6



thus they can pinpoint the differences
rather than hoping to interpret the many idiosyncrasies
and in doing so this opens up what is quite a technical concept to the layman and further
familiarises

the average consumer with more complex concepts in an effort to streamline
understanding and begi
n to narrow the gap in hardware and software comprehension.


There
are 8
discernible

use
s for computers


not taking mobile technology into consideration
:

1) Gaming

2) Graphics/Design/Media Manipulation (Audio/Video)

3) Programming

4)
Spreadsheets/Databases

5) Word

processing/Desktop Publishing

6) Internet/E
-
mail

7) Education

8) Communication/Skype
etc.

These

different tasks
have
different graphics,
processing

and networking requirements.

As
such each objective would rank differently on

each level with the most care being given to
the first three listed uses taking the most processing and graphics power and thus the most
consideration at a higher level, ranked all the way down to simplistic internet and e
-
mail
browsing.

CPU Component Sca
ling

This section makes use of Pass
M
ark’s

Common CPUs

table

which
is

ranked by benchmark
score and thus performance.

It lists the most common CPUs by benchmark testing. The
following gives an idea for how such CPUs would be scaled and the criteria for scaling to be
based upon.


The obvious aspect for heavy CPU use is gaming and graphics/animation rendering.

As such the
top CPU category


that of level
6

should be determined by the most up to
date/
resource intensive game. At the time of writing one of these games is

Far Cry 3.


The optimum suggested CPU is “Intel Core i7
-
2600K/AMD Bulldozer FX4150 or equivalent
quad
-
core”
.

The recommended CPU is “
Intel Core i3
-
530/AMD Phenom II X2 565 or equivalent quad
-
core”.

The minimum CPU required is “Intel Core2 Duo E6700/AMD Athlon64 X2 6000+”


It is accepted

however that the minimum required CPU will give a terrible performance with

a low frame rate. Whilst the game will play the user experience will be very poo
r. Th
e
recommended CPU will provide for

middle

of the range gaming experience, with a greater
frame rate and an enjoyable experience but not at the level required to achieve a mark of
6
.


Thus if it is accepted that gaming is one of the most intense drains a CPU can be subjected to.
We can safely say
that at this current time anything that scores above or at the level of an
Intel Core i7
-
2600k or equivalent quad
-
core CPU


is
a level
6

CPU as of December 2012.

Thus from our list, CPUS 1 through to 11
would be Level
6
s. The question
presents itself



ho
w does one then determine the next levels and where the CPUs fit?

The cut off for Level
6

was a benchmark score of 8297. The lowest scoring 16 items scored under 2000 in the same
tests. It is important however to
recognise

that there is a massive list of CPUs and thus
relative scoring would not be realistic to determine the levels of the scale


except in the case
of relative scoring with level
6

as an indicator of optimu
m performance. If the

list of the high
end CPUs is ex
amined, it
can be determined that at a score of 8000 with the same Ivy Bridge
and quad core technology despite a clock rate of 2.70GHz, the Intel Core i7
-
3740QM still
operates at Level
6

quality. To determine a level
4
CPU one has to discover that is infer
ior in
comparison to the least high scoring Level
6
.


However we are aware of the recommended settings for Far Cry 3


this is “
Intel Core i3
-
530/AMD Phenom II X2 565 or equivalent quad
-
core”. Even for minimum requirements a
quad core is suggested. So
we c
an therefore determine that to score at level 3 a CPU has to be
a quad core.

PassMark suggests that the poorest scoring CPU on the high end CPU list is the AMD
Opteron 4122 with a score of 2942 (equal to the Intel Core2 Quad Q6600)
. This is a
significant
quality difference especially when considering that Intel’s I3, I5 and I7 branding is
aimed to make Intel CPU qualities partially easier to understand for the consumer. There are
I3’s on the list that are scoring 4406


and I7’s that are scoring 4095.

User

data shows that despite the 2942 score on the Core2 Quad Q6600


Far Cry 3 can still
be played with high settings and thus user experience is not sacrificed greatly


however the
game cannot be played on Ultra
-

indicating that the CPU cannot hope to achi
eve a Level
6

score.

However we do know that the recommended CPU has a score of 2603. Thus if the
benchmark for a recommended gaming experience is the Core i3
-
530 at 2603 and the Quad
Q6600 is scoring in that area


then it naturally falls in the Level 4 v
icinity. Far Cry 3 runs at
an average 35+ fps with this CPU according to user data.

In gaming the 60+ fps margin is
generally considered the desirable
fps
.

Taking 35+ fps into consideration at Level 4; Level 5 needs

to run at a better

quality


but
also manage the
High
settings

to be superior to

Level
4
. There is a significant testing gap
between the highest quality CPUs as over 8000, and the recommended CPUs scoring over
2603. An assessment of the AMD Phenom II X4
15

paired with vario
us graphics cards shows
that the card does not achieve the 35+ fps required consistently. It scores 4000


thus it would
be unwise to place the benchmark standard here. (One has to consider that graphics designers
will also need to evaluate Level 5 CPUs an
d thus there needs to be an assessed difference
between the very best and those that are very good but below the 8000 level).

After
examining various CPUs with various graphics cards


one particular CPU that managed very
high settings with a 50
-
60 fps was

the i7 920 which conveniently scored 5000 in benchmark
assessments.


As such: 8000
-

((8000


2603) / 2) =
5301.5

is our midway point.

Thus 5000 is a safe place
to put the Level 5 threshold.




That leaves a calculation for Level 3. The minimum requiremen
t for Far Cry 3 is the Core2
Duo E6700.

It scores at 1583.

Thus for December 2012:

To achieve a Level 3 score a CPU must have a benchmark between 1583 and 260
2
.

To achieve a Level 4 score a CPU must have a benchmark between

2603

and

5205
.

To achieve a
Level 5 score a CPU must have a benchmark between

5206

and 7
807
.

To achieve a Level
6

score a CPU must have a benchmark of over

7807
.

Anything under 1583 belongs to Level 1 and 2 to be discussed later.


Following this logic our table of common CPUs
(reference)

list 1 to 11 as Level 6’s. Test data
shows these to comfortably play Far Cry 3 on Ultra settings with over 60 fps.

List items 12 to
38 are Level 5’s. 39 to 76 are level 4s and level 3 is comprised of list items 77 to 89. 90 to

100 are
sub
-
standard
.

Gaming is very resource intensive


switching over to graphics design, rendering and
multimedia use


GPU and RAM comes into play (admittedly it does as well for gaming
obviously). Thus these same level standards can apply to graphics design
and other
computing tasks.


Following the same logic, metro 2033


which has a lower recommended requirement of
“any quad core or 3.0+ dual core CPU”.
Following that logic the E5700 CPU which was
launched in 2010 and is now end of life fits the description
. It scores 1856 in benchmarking
tests. Suggesting a top level requirement of 5568 (where 1856 * 3), assessing a CPU with this
score we can see the i7 k 875 with a benchmark score of 5578. A test of shows the game runs
at an optimum level with that CPU (as
suming GPU standardisation).


However at 5578 that CPU ranked with the latest technology (assessed with the most
resource heavy far cry 3) would score a Level 5. Accurate consideri
ng the difference between
CPUs


and
to validate,
the
i7 k 875

was tasked with running far cry 3 (standardising GPU
again).

We find then that this CPU can play far cry 3, but not quite to the stan
dard that a level
6 CPU could;
proving the effectiveness of the system.

Environmental impact

Adherence to the level syste
m could raise awareness for pc recyclability. For instance one of
the most advanced games of 2010 was Battlefield 3. If someone owned a level 4 CPU in 2010
this would equate to a benchmark score (using the poorest scoring Intel quad core the Q9000)
of 2486
. Which would mean to run battlefield 3 at a level 6 capacity would require a
benchmark score of 4972.

Such a CPU in 2012


2 years later would still reach a level 4
score. The level 4 score of 2010 however (2486) would now score a level 3 in 2012. What
th
is means is that using the scale provided users can track their PC’s real value quality


a
level 3 or 4 is still usable and thus upgrade is not entirely necessary, so parts upgrades are
still a valid method of improvement for a PC. More so such high level

components could
come with labels to encourage their resell or trade off. They would be less likely to appear
worthless to individuals when their scores can be re
-
factored for their current year.



By David Gladwin


gladwinput.com