You are here

Intel Q9550S: A Greener Quad Core?

Intel's Q9550S: A Greener Quad Core?

June 22, 2009 by Lawrence Lee

Product Intel Core 2 Quad Q9550S

LGA775 Processor
Manufacturer Intel
Street Price US$335~$360

The high-end desktop processor market is saturated with products
that use a lot of energy. The fastest, Intel's Core
i7s, all have thermal envelopes of 130W. AMD's highest-clocked Phenom II quad
cores are rated for 125W, while most Intel
Core 2 Quads are rated for 95W. Intel is in a enviable position with a sizable
lead in performance at the very top, and energy efficient alternatives to rival
the best AMD has to offer.

Enter Intel's Q9550S, Q9400S, Q8400S, and Q8200S, 65W variants of some of the
existing Core 2 Quad line. Rather than ramp up the performance of their Core
2 Quads (which would eat into their Core i7 and future Core i5 profits), Intel
has taken a page from AMD's book, releasing more energy efficient chips.
AMD brands their energy efficient chips with an "E" at the end of
their model numbers, while Intel has decided to use "S."

Comparison Table: Current Intel Core 2 Quad Lineup
Model
Clock Speed
Total L2 Cache
TDP
Mfg.

Price
Best Online Price
Q9650
3.00 GHz
12MB
95W
$316
$320
Q9550S
2.83 GHz
12MB
65W
$320
$335
Q9550
2.83 GHz
12MB
95W
$266
$220
Q9400S
2.66 GHz
6MB
65W
$277
$287
Q9400
2.66 GHz
6MB
95W
$213
$214
Q8400S
2.66 GHz
4MB
65W
$245
N/A
Q8400
2.66 GHz
4MB
95W
$183
$174
Q8300
2.50 GHz
4MB
95W
$183
$190
Q8200S
2.33 GHz
4MB
65W
$183
$236
Q8200
2.33 GHz
4MB
95W
$163
$160
Pricing according to Intel's Processor
Price List
and SPCR/Pricegrabber as of June 18

A 30W drop in power is nothing to sneeze at, especially for a high performance
chip like the Q9550 that's likely to be put to good use on demanding applications.
Lower power consumption has obvious benefits, namely a reduction in operating
cost, but from silent computing perspective, a lower power processor puts out
less heat, making it easier to cool quietly. A 65W Core 2 Quad would be a better
choice for a small form factor system where performance
cannot be sacrificed yet cooling is not optimal. The "S" versions do come with a heavy price
premium, though. Currently, the slower model S processors
retail for $75 more than their 95W counterparts. For the Q9550S, the difference
is $115.



Our Q9550S sample.

Operating Voltage

Rather than tweaking processor design to yield a more energy efficient chip,
it is likely the "S" line are simply standard Core 2 Quads capable
of operating at lower voltages. They could be easily identified, plucked off
the assembly line and binned separately. Binning is an age-old practise that long precedes CPU manufacturing. Apples from the same orchard, for examples, have been binned into different grades, going back many centuries to when they were first grown for sale or trade.



CPU-Z screenshot: Q9550S on load.



CPU-Z screenshot: Q9550 on load.

The Q9550S sample from Intel will be compared
primarily to a recent retail sample Q9550. Our Q9550S sample runs
at 1.144V according to CPU-Z, 64 mV less than our Q9550 sample.



CPU-Z screenshot: Q9550S at idle.



CPU-Z screenshot: Q9550 at idle.

With SpeedStep in effect, both processors run at the same voltage when idle.

TEST METHODOLOGY

Common Test Platform:

Intel:

AMD:

Measurement and Analysis Tools

  • CPU-Z
    to monitor CPU frequency and voltage.
  • CPUBurn
    K7

    processor stress software.
  • Prime95
    processor stress software.
  • Cyberlink
    PowerDVD
    to play H.264/VC-1/Blu-ray video.
  • Eset NOD32 as
    an anti-virus benchmark.
  • WinRAR as an
    archiving benchmark.
  • iTunes
    an audio encoding benchmark.
  • TMPGEnc
    Xpress
    as a video encoding benchmark.
  • PCMark05
    as a general system benchmark.
  • Seasonic
    Power Angel
    AC power meter, used to measure the power consumption
    of the system.
  • Custom-built, four-channel variable DC power supply, used to regulate
    the CPU fan speed.

Benchmark Test Details

  • Eset NOD32: In-depth virus scan of a folder containing 32 files of
    varying size with many of them being file RAR and ZIP archives.
  • WinRAR: Archive creation with a folder containing 68 files of varying
    size (less than 50MB).
  • iTunes: Conversion of an MP3 file to AAC.
  • TMPGEnc Xpress: Encoding a 1-minute long XVID AVI file to VC-1 (1280x720,
    30fps, 20mbps).

Our testing procedure is designed to determine the overall system power consumption
at various states. To stress CPUs we
used Prime95 (large FFTs setting) or CPUBurn (which produced the higher power
draw). We also performed a short series of benchmarks featuring real-world timed
tests and synthetics.

Cool'n'Quiet and/or Intel SpeedStep were enabled (unless otherwise noted).
The following features/services were disabled during testing to prevent spikes
in CPU/HDD usage that are typical of fresh Vista installations:

  • Windows Sidebar
  • Indexing
  • Superfetch

TEST RESULTS

Both our Intel and AMD systems consist of a moderately priced motherboard with
DDR3 memory support, 2x2GB of DDR3 memory in dual channel set to the 1333 MHz
with timings of 9-9-9-24. A GeForce 9400GT graphics card, WD VelociRaptor and
OEM 400W 80 Plus Seasonic power supply round out the test configuration.

Test Results: General System Power Consumption
Test State
X4 955 BE
QX9650
Q9550
Q9550S
Idle
73W
68W
64W
62W
VC-1

Playback
99W
86W
75W
73W
CPU Load
(2 cores)
157W
122W
120W
105W
CPU Load
(4 cores)
201W
142W
140W
125W

The Q9550S used slightly less power than the Q9550 during general power
consumption testing. System power was 2W less when idle and during VC-1 video
playback. Placing a load on the CPU increased its lead by 15W. It is undoubtedly
more energy efficient, but the difference is hardly what one would expect considering
the 65W rating. Our Core 2 Extreme QX9650 used only slightly more power than
the Q9550, which suggests its 130W TDP is quite conservative.

Benchmark Comparison
Test
X4 955 BE
QX9650
Q9550
Q9550S
NOD32
2:27
2:23
2:31
2:31
WinRAR
3:05
2:51
2:59
2:57
iTunes
4:34
3:20
3:33
3:33
TMPGEnc
2:52
3:07
3:14
3:15
PCMark2005
9004
9061
8760
8766

As expected, the Q9550 and Q9550S performed almost identically in our
benchmarks, with the subtle differences well within margins of error. Both chips
are only slightly behind AMD's fastest, the Phenom
II X4 955 Black Edition
.

Average Benchmark Power Consumption
Test
X4 955 BE
QX9650
Q9550
Q9550S
NOD32
128W
88W
84W
78W
WinRAR
128W
98W
96W
87W
iTunes
137W
92W
89W
81W
TMPGEnc
167W
118W
113W
102W
Est. Total Benchmark Power Consumption (Watt-hours)
NOD32
5.23
3.50
3.52
3.27
WinRAR
6.58
4.66
4.77
4.28
iTunes
10.43
5.11
5.27
4.79
TMPGEnc
7.98
6.13
6.09
5.53

The Q9550S also used less power during the time it took to run each
timed benchmark, between 6W and 11W less depending on the test. The total difference
for all the benchmarks combined was 9%. If we operated the Q9550 system continuously
for a year running just TMPGEnc, it would draw 78.84 kWh more than the same
system with a Q9550S. According to the EIA
the average cost of electricity in the United States in March of 2009 was 11.38
cents per kilowatt-hour. Average savings for one year: $8.97. Average time to
recoup a $115 price difference: 12.8 years.

Undervolting

Perhaps the most important value factor is whether a Q9550 can meet or exceed the lower power consumption
of a Q9550S through undervolting. After some fiddling, we managed to
get our Q9550 sample stable at 1.024V according to CPU-Z (1.03125V setting in
the BIOS). The Q9550S was stable at a minimum of 1.064V (1.075V in the
BIOS), which coincidentally is the same voltage applied when idling with SpeedStep
enabled.



CPU-Z screenshot: Q9550, best undervolt.



CPU-Z screenshot: Q9550S, best undervolt.

Undervolted Test Results

Test Results: General System Power Consumption
Test State
Q9550
Q9550S
Stock
UV
Stock
UV
Idle
64W
64W
62W
63W
VC-1

Playback
75W
75W
73W
74W
CPU Load

(2 cores)
120W
99W
105W
99W
CPU Load

(4 cores)
140W
117W
125W
118W

Undervolted, our Q9550 produced a 21W improvement when 2 of its cores were
fully stressed and 23W when all 4 were put to to pace. These results place it
ahead by 6-8W of the Q9550S at stock voltage. Undervolting the Q9550S
resulted in only a 6-7W improvement. The difference between the two processors
when undervolted to their respective minimum stable voltage was negligible.

Average Benchmark Power Consumption
Test
Q9550
Q9550S
Stock
UV
Stock
UV
NOD32
84W
75W
78W
74W
WinRAR
96W
83W
87W
83W
iTunes
89W
77W
81W
76W
TMPGEnc
113W
95W
102W
95W
Total Benchmark Power Consumption (Watt-hours)
NOD32
3.52
3.15
3.27
3.10
WinRAR
4.77

4.13

4.28
4.08
iTunes
5.27
4.56
4.79
4.50
TMPGEnc
6.09
5.12
5.53
5.15

Undervolted, each processor used approximately the same amount of power
during our timed benchmarks.

FINAL THOUGHTS

While the Q9550S does indeed use less power than the Q9550, it is much
less than the 30W suggested by their TDP ratings. Thermal Design Power is an "up
to" specification indicating maximum power dissipation under the most extreme loads, so caution should be used when looking at these numbers.
We found that the difference in power consumption was much too small to justify
the $125 higher price-tag. To add insult to injury, our retail Q9550 sample could be undervolted
to the point where it became more energy efficient than the Q9550S at
stock voltage. Our Q9550S didn't undervolt any better — when both
processors were set to their respective minimum operating voltages, the numbers
were indistinguishable. There were no significant performance differences between
the two.

The performance of our retail Q9550 sample suggests that Intel's
45nm manufacturing process has matured. A Q9550 purchased today and
properly undervolted should come close to the energy efficiency of a
Q9550S. All that is required is a capable motherboard, a little
time, effort, and know-how.

For a user looking for energy efficiency both to reduce their system's overall
operating cost and the amount of heat that needs to be dissipated, the Q9550S
may be a viable choice depending on the circumstances. If, for
instance, the system in question is mini-ITX, the Q9550S merits
some consideration: A SFF PC is difficult to cool and most mini-ITX
motherboards lack the ability to undervolt.

For corporations and OEMs which simply don't have time to undervolt and stress
test each processor individually to ensure system stability, the Q9550S
is a questionable choice, given its current market price. The lifespan of the
processor is probably not long enough to offset the $100+ cost premium unless
the electricity rate is very high. The slower "S" models currently
cost about $75 more than their 95W counterparts — still high, but a far
more palatable price to absorb.

Our thanks to Intel
for Q9550S sample.

* * *

Articles of Related Interest

AMD Phenom
II X2 550 BE & Athlon II X2 250


AMD Phenom II X4 955 Black Edition for
AM3


Phenom II: AMD pulls closer

Intel Core i7: Nehalem Launched

* * *

Discuss this article in the
SPCR forums.

Google

www SPCR