A die-shrink to 32nm has given Intel enough room to pair a dual core processor and integrated graphics on the same chip package. The Clarkdale LGA1156 Core i5-661 also features Intel’s new GMA HD graphics, TurboBoost, and hyper-threading.
January 4, 2010 by Lawrence Lee
Product | Intel Core i5-661 LGA1156 Processor |
Manufacturer | Intel |
MSRP | US$196 |
To kick off 2010, Intel has released a new batch of Core i7, i5 and i3 processors,
codename Arrandale for laptops, and Clarkdale for desktops. These new CPUs are
rather interesting in that there are two chips on the same package: a small
dual core processor manufactured using their new 32mm fabrication process connected
internally with QPI to a larger 45nm chip that houses the memory controller and
an updated IGP that supports OpenGL 2.1, simply dubbed Intel GMA HD.
|
While quad core processors will continue to dominate the high-end desktop market,
there is still a strong case to be made for dual core chips. They are more affordable
(at equivalent clock speeds), use less energy and are thus easier to cool. And
though they have half the cores, higher clock speeds are more of a benefit for some
games and applications, not all of them. For the average Joe that doesn’t spend hours encoding
video or manipulating high resolution images with advanced filters, a 3+ GHz
dual core CPU is often a better choice.
Comparison Table: Dual Core Core i5/i3 Desktop Processors | |||||
Model | Clock Speed | Turbo Speed | L3 Cache | TDP | Pricing* |
Core i5-670 | 3.46 GHz | 3.73 GHz | 4MB | 73W | $284 |
Core i5-661 | 3.33 GHz | 3.60 GHz | 4MB | 87W | $196 |
Core i5-660 | 3.33 GHz | 3.60 GHz | 4MB | 73W | $196 |
Core i5-650 | 3.20 GHz | 3.46 GHz | 4MB | 73W | $176 |
Core i3-540 | 3.06 GHz | N/A | 4MB | 73W | $133 |
Core i3-530 | 2.93 GHz | N/A | 4MB | 73W | $113 |
* boxed price (1000 units) |
The first six dual core LGA1156 processors have frequencies ranging from 2.93
GHz to 3.46 GHz with all models supporting hyper-threading so the operating
system will see 4 threads instead of 2. For the budget-conscious, the Core i3’s
start at only $113 and the only major thing missing is the TurboBoost overclocking
feature. The Core i5-661 runs at 3.33GHz but with TurboBoost
enabled, overclocks to 3.60GHz when a demanding application is running that
can only use one core. The speed is bumped up to 3.46GHz when both cores are
active.
|
The Core i5-661 is an unusual processor, in name and features.
Both AMD and Intel desktop chips typically have a ‘0’ or ‘5’ at the end of their
model numbers. The “1” indicates that the part in question has a higher
speed graphics core and accompanying TDP, and lacks a few of Intel’s lesser
known technologies, the most notable being VT-d, which gives virtual machines
the ability to directly access peripherals on their host. Features like this
are usually only cut on lower-end parts, and the i5-661 is the only one of its
kind for now, so it’s a very odd duck indeed, resembling a Core i3 with an overclocked
IGP.
|
|
MOTHERBOARD
With the redesigned processor, new chipsets are required to take advantage
of GMA HD. The H55 Express is Intel’s new mainstream LGA1156 chipset.
|
|
|
|
|
TEST METHODOLOGY
Common Test Platform:
- Corsair
XMS3 memory 2x2GB, DDR3-1600 @ 1333MHz, 9-9-9-24 (1066 MHz,
8-8-8-20 for LGA775) - Asus EN9400GT Silent Edition
graphics card – 512MB - Western Digital VelociRaptor
hard drive – 300GB, 10,000RPM, 16MB cache - Western
Digital Scorpio Blue notebook hard drive – 500GB, 5,400RPM,
8MB cache (for IGP testing) - Asus
BC-1205PT Blu-ray drive (for IGP testing) - Seasonic
SS-400ET ATX power supply - Microsoft
Windows Vista SP1 operating system – Home Premium, 32-bit
AMD AM3:
- AMD Phenom II X4 955 Black Edition
processor – 3.2 GHz, 45nm, 125W - AMD
Phenom II X3 720 Black Edition processor – 2.80 GHz, 45nm,
95W - AMD Phenom II X2 550 Black Edition
processor – 3.10 GHz, 45nm, 80W - Asus M4A78T-E motherboard
– 790GX chipset (onboard video disabled) - ZEROtherm Zen FZ120
CPU cooler
Intel LGA775:
- Intel
Core 2 Quad Q9550 processor – 2.83 GHz, 45nm, 95W - Intel
Core 2 Duo E7200 processor – 2.53 GHz, 45nm, 65W - Asus
P5Q3 motherboard – P45 chipset - ZEROtherm Zen FZ120
CPU cooler
Intel LGA1156:
- Intel Core i5-750
processor – 2.66GHz, 45nm, 95W - Intel Core i5-661 processor – 3.33GHz, 32nm, 87W
- Intel
DP55KG motherboard – P55 chipset (for Core i5-750) - Intel DH55TC motherboard – H55 chipset (for Core i5-661)
- Thermalright
MUX-120 CPU cooler
|
Measurement and Analysis Tools
- CPU-Z
to monitor CPU frequency and voltage. - CPUBurn
processor stress software. - Prime95
processor stress software. - ATITool
artifact scanner to stress the integrated GPU. - FurMark
stability test to stress the integrated GPU. - 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. - 3DMark05
as a 3D benchmark. - 3DMark06
as a 3D 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 power
and regulate the CPU fan.
Estimating DC Power
The following power efficiency figures were obtained for the
Seasonic SS-400ET used in our test system:
Seasonic SS-400ET Test Results | ||
DC Output (W) | AC Input (W) | Efficiency |
21.2 | 32.0 | 66.3% |
41.6 | 58.0 | 71.7% |
60.2 | 78.0 | 77.1% |
81.9 | 102.0 | 80.3% |
104.7 | 128.0 | 81.8% |
124.1 | 150.0 | 82.8% |
145.2 | 175.0 | 83.0% |
165.3 | 195.0 | 84.8% |
193.2 | 227.0 | 85.1% |
222.7 | 261.0 | 85.3% |
This data is enough to give us a very good estimate of DC demand in our
test system. We extrapolate the DC power output from the measured AC power
input based on this data. We won’t go through the math; it’s easy enough
to figure out for yourself if you really want to.
Testing Procedures
Our IGP testing procedure is to run the system through a video test suite featuring
a variety of high definition clips. During playback, a CPU usage graph is created
by the Windows Task Manger for analysis to determine the average CPU usage.
High CPU usage is indicative of poor video decoding ability. If the video (and/or
audio) skips or freezes, we conclude the IGP (in conjunction with the processor)
is inadequate to decompress the clip properly.
Our CPU testing procedure is designed to determine the overall system power
consumption at various states (measured using a Seasonic Power Angel). To stress
CPUs we use Prime95 (large FFTs setting) or CPUBurn, whichever produces the
higher power draw. We also perform a short series of benchmarks featuring real-world
timed tests and synthetics detailed below.
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
CPU Benchmark 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 (256kbps).
- TMPGEnc Xpress: Encoding a 1-minute long XVID AVI file to VC-1 (1280×720,
30fps, 20mbps).
Video Test Suite
H.264: Rush Hour 3 Trailer 1 is a H.264 encoded clip inside an Apple Quicktime container. |
WMV-HD: Coral Reef Adventure Trailer is encoded in VC-1 using the WMV3 codec commonly recognized by the “WMV-HD” moniker. |
VC-1: Drag Race is a recording of a scene from network television re-encoded with TMPGEnc using the WVC1 codec, a more demanding VC-1 codec. |
Blu-ray: Disturbia is a short section (chapter 4) of the Blu-ray version of Disturbia, the motion picture, played directly off the Blu-ray disc. It is encoded with H.264/AVC. |
Blu-ray: Becoming Jane is a short section (chapter 7) of the Blu-ray version of Becoming Jane, the motion picture, played directly off the Blu-ray disc. It is encoded with VC-1. |
IGP Test Results
To evaluate the performance of Intel GMA HD graphics, we setup the i5-661 in
our standard motherboard test platform featuring a notebook hard drive, Blu-ray
drive, and Seasonic power supply. Our test results for Intel’s previous flagship
IGP, GMA X4500HD were obtained using a 2.53GHz Core 2 Duo E7200. To make it
a more fair battle, we tested the i5-661 underclocked to 2.26GHz by lowering
the CPU’s multiplier to 17x in addition to stock speed.
Video Playback | ||||
Test State | Core i5-661 @ 3.33GHz (Stock) | Core i5-661 @ 2.26GHz (Underclocked) | ||
Avg. CPU | System Power (DC) | Avg. CPU | System Power (DC) | |
Rush Hour (H.264) | 3% | 31W | 17% | 31W |
Coral Reef (WMV-HD) | 8% | 31W | 10% | 31W |
Drag Race (VC-1) | 13% | 35W | 16% | 34W |
Disturbia (Blu-ray H.264) | 28% | 43W | 39% | 40W |
Becoming Jane (Blu-ray VC-1) | 13% | 39W | 16% | 38W |
TurboBoost disabled during video playback testing. |
Like most modern IGPs, Intel’s GMA HD had no problem rendering our video test
suite with low CPU utilization playing both H.264 and VC-1 content with the
CPU at stock speeds. Even underclocked by almost a third, playback was smooth,
though H.264 playback was more reliant on CPU resources than we usually see
when using Nvidia or ATI graphics. Power consumption was very low during playback.
Video Playback: GMA X4500HD vs. GMA HD | ||||
Test State | E7200 @ 2.53GHz (Intel DG45FC) | Core i5-661 @ 2.26GHz (Intel DH55TC) | ||
Avg. CPU | System Power (DC) | Avg. CPU | System Power (DC) | |
Rush Hour (H.264) | 30% | 29W | 17% | 31W |
Coral Reef (WMV-HD) | 25% | 30W | 10% | 31W |
Drag Race (VC-1) | 40% | 33W | 16% | 34W |
Despite a 266MHz disadvantage in processor clock speed, the i5-661’s GMA HD
required significantly less CPU assistance to play HD video compared to the
GMA X4500HD paired with a Core 2 Duo E7200. Either the IGP is more efficient
or the i5 processor is much faster clock for clock (or a combination of the
two). Power consumption was similar on both platforms.
3D Performance: Futuremark Comparison | |||
Motherboard | Graphics | 3DMark05 | 3DMark06 |
Asus P5Q-EM (C2D E7200) | X4500HD | 1708 | 1092 |
MSI KA760GM (X2 4850e) | HD 3000 | 2289 | 1049 |
Gigabyte MA78GM-2SH (X2 4850e) | HD 3200 | 2293 | 1116 |
Intel DH55TC (Core i5-661 @ 2.26GHz) | GMA HD | 3179 | 1676 |
Intel DH55TC (Core i5-661) | GMA HD | 3215 | 1688 |
Zotac GeForce 9300-ITX (C2D E7200) | GF9300 | 3522 | 1797 |
Asus M4A785TD-V (X3 720 BE) | HD 4200 (Sideport) | 4095 | 1789 |
Asus M4A78T-E (X3 720 BE) | HD 3300 (Sideport) | 4884 | 2205 |
System configuration: 2GB of RAM, 256MB of VRAM assigned (128MB for motherboards with 128MB of Sideport memory). |
GMA HD is also a faster IGP in 3D performance, with a hefty improvement onver
X4500HD in both 3DMark05 and 3DMark06. Its scores put it close to the GeForce
9300, but still aways off from the Sideport equipped HD 4200/3300 for AM3. For
older games at lower resolutions, it should be sufficient. Note that the i5-661’s
GPU has a 900MHz clock speed while the other Clarkdales have to suffice with
733MHz.
CPU/VRM Power Consumption
To get an idea of how much power the CPU actually requires, we measured the
current (through a 0.01 ohm shunt resistor) and voltage at the ATX12V connector in various CPU load states. This allows us to calculate the power demanded by the CPU and the board’s VRM (DC-DC converter).
DC Power Consumption | ||||||
Test State | Core i5-661 @ 3.33GHz (Stock) | Core i5-661 @ 2.26GHz (Underclocked) | ||||
System | CPU + VRM | Diff. | System | CPU + VRM | Diff. | |
Idle | 18W | 5.3W | 12.7W | 18W | 5.2W | 12.8W |
CPU Load | 60W | 46.0W | 14.0W | 45W | 32.4W | 12.6W |
CPU + GPU Load | 67W | 51.2W | 15.8W | 54W | 39.0W | 15.0W |
CPU Load (Turbo) | 63W | 48.0W | 15.0W | N/A | ||
CPU Load + GPU Load (Turbo) | 69W | 52.7W | 16.3W | N/A | ||
CPU + VRM power measured from the ATX12V connector (combined DC draw of VRMs and CPU). |
By our measurements, our i5-661 sample used 5.3W when idle and up to 46.0W
when stressing the CPU with Prime95, with an indeterminate amount lost to VRM
inefficiency. The addition of FurMark’s GPU stability test resulted in an additional
5.2W draw.
Between 13W and 16W was used by the rest of the system which consisted of the
motherboard, two sticks of memory, a notebook hard drive, idle Blu-ray drive,
keyboard and mouse.
System Power Consumption Comparison
To compare the Core i5-661 to other processors we’ve tested, we changed the
hardware to match our CPU testing platform, swapping the notebook hard drive
for a Velociraptor and using a GeForce 9400GT rather than integrated graphics.
Test Results: System Power Consumption (DC) | |||
Processor | Idle | VC-1 Playback | CPU Load |
C2Q Q9550 | 47W | 57W | 115W |
Core i5-750 | 37W | 55W | 107W |
Core i5-661 (Turbo) | 31W | 45W | 79W |
Core i5-661 | 31W | 45W | 78W |
C2D E7200 | 41W | 50W | 69W |
X4 955 BE | 56W | 79W | 171W |
X3 720 BE | 58W | 77W | 127W |
X2 550 BE | 48W | 67W | 106W |
Back in September we reviewed the Core i5-750 and i7-860 Lynnfield quad core
processors and found them to very energy efficient. The i5-661, utilizing a
smaller manufacturing process and half the cores is even more thrifty when it
comes to power draw. Power consumption was 10W less than the Core 2 Duo E7200
when idle and only 10W more on full CPU load. AMD’s fastest dual core, the X2
550 Black Edition, an 80W chip, was off the mark by 20~30W.
CPU Performance
Test Results: Benchmarks | |||||
Processor | NOD32 | WinRAR | iTunes | TMPGEnc | PCMark05 |
C2Q Q9550 | 2:31 | 2:59 | 3:33 | 3:14 | 8760 |
Core i5-750 | 2:31 | 2:38 | 3:48 | 3:21 | 8390 |
Core i5-661 (Turbo) | 2:03 | 3:30 | 3:03 | 3:32 | 9342 |
Core i5-661 | 2:08 | 3:35 | 3:10 | 3:39 | 9220 |
C2D E7200 | 2:47 | 3:34 | 4:00 | 6:10 | 7057 |
X4 955 BE | 2:27 | 3:05 | 4:34 | 2:52 | 9004 |
X2 550 BE | 2:34 | 3:09 | 4:44 | 4:54 | 7217 |
X3 720 BE | 2:47 | 3:16 | 5:13 | 5:08 | 7738 |
Our timed benchmark tests were chosen to simulate average use, thus the i5-661
performed very well compared to the competition, winning in an anti-virus scanning
race using NOD32, and an audio encoding test using iTunes. Surprisingly, the
i5-661 came close to catching its similarly priced quad core cousin, the i5-750,
when video encoding using TMPGEnc, test which favors quad cores. Even a task
using multiple threads is not a sure-fire win for a quad core over a dual core.
Benchmark Power Consumption
Average System Benchmark Power Consumption (DC) | ||||
Processor | NOD32 | WinRAR | iTunes | TMPGEnc |
C2Q Q9550 | 66W | 77W | 70W | 92W |
Core i5-750 | 57W | 61W | 57W | 88W |
Core i5-750 (Undervolted) | 54W | 53W | 56W | 76W |
Core i5-661 (Turbo) | 52W | 53W | 55W | 70W |
Core i5-661 | 49W | 50W | 53W | 69W |
X4 955 BE | 105W | 105W | 113W | 138W |
X3 720 BE | 91W | 90W | 96W | 107W |
X2 550 BE | 85W | 85W | 87W | 96W |
Total System Benchmark Power Consumption (watt-hours, DC) | |||||
Test State | NOD32 | WinRAR | iTunes | TMPGEnc | Total |
Q9550 | 2.77 | 3.83 | 4.14 | 4.96 | 15.70 |
i5-750 | 2.39 | 2.68 | 3.61 | 4.91 | 13.59 |
i5-750 (Undervolted) | 2.27 | 2.33 | 3.55 | 4.24 | 12.38 |
Core i5-661 (Turbo) | 1.78 | 3.09 | 2.80 | 4.12 | 11.79 |
Core i5-661 | 1.74 | 2.99 | 2.80 | 4.20 | 11.72 |
X4 955 BE | 4.29 | 5.40 | 8.60 | 6.59 | 24.88 |
X3 720 BE | 4.22 | 4.90 | 8.35 | 9.15 | 26.62 |
X2 550 BE | 3.64 | 4.46 | 6.86 | 7.84 | 22.80 |
When we combine the i5-661’s strong performance in our timed benchmarks with
the average system power consumption during these tests, we find that it not
only finishes tasks quickly, but does so using less total power than any processor
we’ve tested to date. In terms of total watt-hours used for our entire test
suite, the i5-661 platform was ahead of the i5-750 by 14% and 5% when the i5-750
was undervolted to its minimum stable voltage.
Thermals
Temperatures | |||
Sensor | Avg. Core | Chipset | VRMs |
Core i5-661 | 41°C | 39°C | 39~44°C |
Core i5-750 | 51°C | N/A | |
Core i7-870 | 53°C | ||
Thermalright MUX-120 with stock fan @ 5.6V (900 RPM). Ambient temperature: 22°C. |
The average core temperature reported by SpeedFan stabilized at 41°C after
10~15 minutes on full CPU load, about 10°C lower than its quad core cousins.
The Thermalright MUX-120 heatsink used during testing was cool to the touch
even when the processor was stressed to the limit. Board temperatures were also
excellent with the VRMs around the CPU socket staying below 45°C and the
single chipset heatsink never passing 40°C at its hottest point, as measured
by an IR thermometer.
Motherboard Impressions
Budget Intel boards typically have very restrictive controls in the BIOS and
the DH55TC is no exception. The BIOS lacked controls for memory speed and did
not allow any voltage modification. In addition, TurboBoost could not be customized;
only an on/off toggle was provided.
As for compatibility, neither the Core i5-750 or i7-870 would POST in the DH55TC,
even with the latest BIOS from Intel. The DH55TC claims support for i7, i5,
and i3 on the box, so in theory it should work, if not now, then in the future.
The i5-661 would not work in our older DP55KG LGA1156 motherboard either.
FINAL THOUGHTS
The Core i5-661 is more than just a dual core version of the Lynnfield core
processor. A die-shrink from 45nm to 32nm gave Intel
the space to fit both a dual core processor and their latest onboard graphics
chip, GMA HD, onto the same package. This combination is very energy efficient,
rivaling the power consumption of a low speed Core 2 Duo, despite the Core i5-661’s
much higher clock speed and overall performance.
GMA HD is a marked improvement over GMA X4500HD, rendering HD video using significantly
fewer CPU cycles and delivering much better 3D performance. Regarding the latter,
it should be noted that the GPU core of the i5-661 has a higher operating frequency
than the version of GMA HD found on the rest of the Clarkdale lineup. Even with
this advantage, it still trails behind AMD’s HD 3300/4200 with Sideport memory. Still,
with either CPU/chipset, if you play even 3~4 year old games at the typical native resolution
of a modern widescreen LCD monitor, you will probably need a ‘real’ video card
to get decent framerates and image quality.
The Core i5-661’s <$200 price tag brings up the question
of whether the equivalently priced quad core Core i5-750 is superior. If you’re a heavy multi-tasker or often use demanding applications that can take advantage of four cores like some video encoders, professional image editing software, and games, then the i5-750 is the better
candidate. However, if a program doesn’t scale that well with multiple cores,
the higher clock speed of the i5-661 can give the i5-750 a run for its money.
For example, in our TMPGEnc video encoding test, the i5-661 finished only 5%
slower than the i5-750.
If you’re a more casual user but you still value snappy performance, a dual core like the i5-661 is an excellent choice, particularly if
you have no need for discrete graphics. In many instances the higher clock speed
is actually more useful. In addition, the i5-661 is very energy efficient and
runs fairly cool, so you don’t need an enormous, expensive heatsink/fan
to make it a quiet computing experience. The 6×0 series should run even cooler
as they utilize a slightly slower version of GMA HD; Intel lists their TDPs
at 73W vs. 87W for the i5-661. In summary, the Core i5-661 is a significant landmark in Intel’s long stream of processors.
* * *
Articles of Related Interest
Intel
D510M0 Motherboard: Atom 2.0
Athlon II X4 630 & 620: Affordable
Quad Cores
Intel’s LGA1156 and Lynnfield
core
Intel Q9550S: A Greener Quad Core?
AMD Phenom II X2 550 BE & Athlon
II X2 250
AMD Phenom II X4 955 Black Edition for
AM3
* * *