Alpha S-PAL8952 P4 Heatsink

blog image

The Alpha 8952 is the latest in a long evolution of distinctive “blow-up” fan optimized CPU heatsinks. Can this now venerable high-end aluminum pins on copper base design keep up with the latest heavyweights with processors in the 100W range? SPCR’s review of the Alpha 8952 P4 HSF, quiet fan version.

July 19, 2004 by Ralf
Hutter

Product
Alpha
S-PAL8952M82
Heatsink
Manufacturer
Alpha
Co Ltd
MSRP
US$55.10
with fan, $41.80 w/o fan

(market price is generally lower)

The last significant development in the Alpha line of heatsinks was the PAL8942, which quickly became one of the top performing Pentium
4 heatsinks available at the time. The PAL8942 was an adaptation of Alpha’s
successful high performance Socket A heatsink, the PAL8045.

Fast forward
two years: Alpha recently released the successor to the PAL8942, which they
creatively call the S-PAL8952. The S-PAL8952 looks externally very similar to
the earlier PAL8942; the base dimensions are almost identical.

So where are the differences?

Alpha’s web site states:

“The S-PAL8952 shares many design features with the current PAL8942.
They both mount through the four holes surrounding Socket 478. We increased
the fin length by 10 mm and changed the fin pattern. The new fin pattern design
is based on the air flow within the heat sink. The shape of the outside pins
is square and inside pins are round. The intake cover is longer to prevent any
air from taking a short cut past the hottest portion of the pins.”

The original 8942 is rated up to a 3.2GHz P4 while Alpha claims that the new
8952 is…

“the ideal choice for the new Prescott, Northwood and Extreme
Edition processors from Intel, which run much hotter than the previous Northwood
chips.”


Alpha didn’t waste and money on fancy packaging or
fancy naming schemes.

ALPHA S-PAL8952 SPECIFICATIONS
HeatsinkIncluded Fan

Overall height (w/ included fan): 72.6mm
Total Weight (w/ & w/o fan): 624/524 grams
Heat Sink Material: 6063 Aluminum
Finish: Black anodized
Embedded Copper Base: 1100 or 1020
Size (W x L x H): 78.5x90x52mm

Dimension: 80mm x 15.4mm
Voltage: 12VDC
Current: 0.09A
Speed: 2000rpm
Max. Air Flow: .57mm/min
CFM: 20.13
Noise Level: 21dBA
Vendor: SANYO DENKI CO.,LTD
Model number: 109P0812M701
ALPHA fan Model #: M82


Package contents. Aluminum shroud, heatsink, fan, instructions,
mounting screws and standoff, fan mounting bolts and fan.

DESIGN

The 8952 continues the tradition we’ve come to expect from Alpha: A precision,
cold forged aluminum base; an integral, well machined copper heatspreader; and
an aluminum shroud for use with a fan oriented to suck air away from the heatsink,
instead of blowing down onto it.

It departs from
Alpha’s previous designs in that not all the fins have the same shape.
On the 8952 the fins directly over the CPU are round
or slightly hexagonal whereas the rest are square. The fins are also 10mm taller
than the fins on the 8942. Alpha claims that these changes have
improved the airflow and thermal performance.
Alpha also increased the length of the 8952’s aluminum fan shroud to force
more air past the lower warmer portions of the pins. Whatever
improved cooling the 8952 offers, it has to come from the redesigned pins and
fan shroud because the footprint of the new 8952 is the same as the earlier
8942 at 90mm x 79mm.


The cooling fins of the 8952. Note the round pins in the center section,
right over the CPU’s heatspreader.

The base of the 8952 is the same design as the earlier
Alphas, a big chunk of copper inserted into the aluminum base during
the cold forging operation for the best possible copper to aluminum contact. The base
is very flat although some slight machine marks are visible. The four corners are CNC machined so they sit down on top of the
brass risers when the heatsink is fully tightened.


The beautiful base: Copper and aluminum are essentially one piece
due to cold-forging operation.

The aluminum shroud is designed to force the air to be sucked into the heatsink
near the base of the pin where they are the hottest. This shroud is one of the
main reasons the 8952 (and all Alphas) work better in the “blow up”
mode rather than the “blow down” orientation. The shroud
also lifts the fan up a bit off the heatsink fins to act as a plenum and
prevent a “dead air spot” in the fan hub area.

The included fan is an
80 x 15mm Sanyo Denki “Petit Ace”. This fan is rated at 20cfm at
21dBA. This fan is the quiet version Alpha offers with the 8952. The standard version is a 46cfm, 40dBA Delta fan.
(And that’s definitely something we don’t even want to consider here
at SPCR!) The Petit Ace has a reputation as a high quality fan,
but I’ve never seen them talked about as particularly quiet, so it will be interesting
to give this one a listen. This fan comes terminated with a standard 3-pin fan
connector. This will easily allow it to connect direct to a motherboard
fan header or something like a Zalman FanMate. For this test I plugged it into
a 4-pin Molex adapter and ran it at a fixed 12V, 7V and 5V.


Included 80mm Sanyo Denki low profile fan. Look at all those blades.

The 8952 mounts in the standard Alpha method, bolted onto brass
standoffs that the user attaches to the motherboard after removing the
stock Intel retention bracket. While not nearly as easy as using the stock Intel
heatsink mounting bracket, it’s certainly a much sturdier attachment for this
heavy heatsink. This type of mounting method requires that
the motherboard be removed from the case to install the mounting hardware. You’ll
need access to the underside of the motherboard to screw on the supplied brass
nuts. Fortunately this is a one-time procedure. Once the standoffs are attached
to the motherboard you can remove or install the heatsink while the board remains
installed in the case.

Alpha includes a complete set of hardware that includes the brass standoffs(4),
brass nuts with built-in lock washers(4), nylon flat washers(8), heatsink mounting
screws and springs(4), and fan mounting screws (2 sets, one for 15mm thick fans,
the other for standard 25mm thick fans).


Standoffs and heatsink mounting hardware.


Fan mounting bolts. Includes two different sizes for mounting 15mm and 25mm
thick fans.

ASSEMBLY

The Alpha S-PAL8952 heatsink was mounted on my standard P4 motherboard.
Assembly was not as easy as with heatsinks that utilize the stock Intel mounting
bracket, but wasn’t difficult for me because my test setup is not mounted inside
a case. I didn’t need to remove the motherboard before installing the
standoffs.

The instruction sheet is well-illustrated as easy to follow. A couple of adjustable crescent
wrenches or 9/32″ and 7/32″ nut drivers are very useful to tighten up the
standoffs and locknuts to the board.


Standoffs bolted to the board. Note the nylon washers.
Threaded end of standoffs extend through the board and are bolted on from underneath
using the included flanged locknuts and another set of nylon washers.

After mounting the standoffs, apply the thermal compound to the CPU
and insert it into the socket. Then bolt the heatsink onto the standoffs.
The four spring-loaded bolts apply the correct amount of pressure. The standoffs themselves
provide the correct height. This is a safe and foolproof
mounting method, as long as the screws are tightened in a “crisscross”
pattern to keep even pressure on the CPU. Combine this with the security of the “bolt-through”
method of attaching the heatsink to the board and you have one of the best mounting
methods available.


Completed mounting. Fan shroud channels air in through
the warmer base of the heatsink and acts as a spacer to raise the fan above
the fins.

ON THE TEST BENCH

I compared the S-PAL8952 to my sort-of-reference Zalman
7000AlCu
heatsink, as well as the stock Intel P4 heatsink. I used
the same processor that I’ve been using in all my heatsink tests, a P4-2.4C. It may not put out as much heat as a new Prescott, but using this
processor will allow me to directly compare the performance of the new 8952
against all the other high performance P4 heatsinks that I’ve tested during
the past year.

I also tested the 8952 with the SPCR Reference 80mm Panaflo L1A. This will let
us directly compare the cooling ability of this heatsink against other heatsinks
that we have tested with the Reference fan, as per the
basic SPCR heatsink testing philosophy
of keeping everything the same (including
the fan) and changing only the heatsink. The Sanyo Denki and the Panaflo L1A were
both tested blowing down on the HS as well as blowing up. In both cases the
performance with the fans in the traditional “blow down”
method ran about 2-4°C hotter than when they were used the way Alpha engineered
them. The Zalman 7000 and stock Intel coolers were tested with the fans in their
normal “blow down” orientation.


Ye Olde Test Setup

Key Components in P4 HS Test platform:

The P4 HS test platform is an open system not enclosed in a case.

Intel P4-2.4C Northwood core – Maximum power is 66.2W.
Intel 875PBZ motherboard – Intel 875P Chipset; on-die CPU
thermal diode monitoring
ATI Radeon 7500 passively cooled video card (AGP)
Mushkin PC3200 Level II – 2 x 256MB DDRAM
Seagate 80GB Barracuda IV hard drive
Seasonic SuperSilencer 400W (rev A1) PSU
Arctic Silver Ceramique Thermal Compound
Two-level metal platform with rubber damping feet. Motherboard on top; other
components below.
CPUBurn processor stress software
Motherboard Monitor 5.3.4.0 software to track CPU temperature
and fan speed

Each heatsink was cleaned and installed on the test system as per the
manufacturer’s and Arctic Silver’s instructions. Prime95 was then run for
8 hours to verify system stability and cure the Ceramique. The system was
then shut down and not restarted until the next morning when the actual
testing was done. The system was allowed to cool between tests for 30 minutes.
Each test was run for 30 minutes even though all temperatures generally
stabilized within 15 to 20 minutes.

Each heatsink was tested three times on consecutive mornings to check
to the consistency of the results. All results were within 1-2°C of
each other and the average readings are included in the charts.

Ambient temperature was measured at 71-72?F (22°C) over the entire
series of tests. No tests were run unless the ambient temperature was at
that reference level.

* All temperatures in degrees Celsius.
* Diode: Reading from P4-2.4C CPU diode via Motherboard Monitor.
* °C Rise refers to the difference between ambient temperature and the
diode reading. .
* °C/W refers to the °C of temperature rise per watt of heat dissipated
by the CPU.

RESULTS

I tested both the included Sanyo Denki fan and the Panaflo L1A at a fixed 12V,
as well as at the quieter settings of 7V and 5V. The 12V results are shown below:

A) 12V
Fan Results
Heatsink
idle
load
°C
rise
°C/W
MP*
°C/W
TDP*
Alpha 8952 w/ stock fan
25°C
41°C
19
0.25
0.29
Alpha 8952 w/ Panaflo L1A
27°C
42°C
20
0.27
0.30
Zalman 7000
24°C
40°C
18
0.26
0.27
Stock Intel HSF
25°C
46°C
24
0.34
0.36
*For °C/W – TDP calculations, Intel’s
TDP of 66.2W was used.
*For °C/W – MP calculations, CPUHeat
& CPUMSR Projects’
estimate of 75W was used

At 12V the 8952 with the stock Sanyo Denki fan performed well but was also
fairly noisy. It had a fairly noticeable whine and a moderate amount of rushing
air noise. It’s definitely too loud for a quiet system but it cools surprisingly
well for being such a low-airflow fan.

Next I swapped out the Sanyo Denki fan for the 80mm Panaflo L1A, running at
12V. It cooled the CPU virtually as well as the stock fan with less air
turbulence noise; the electrical/mechanical noise was less objectionable
as well

At full speed the Intel HSF sounded worse than the stock Sanyo Denki fan and
it didn’t cool as well. It’s as expected, running in such high class company.

The Zalman at 12V was certainly audible, but not as
objectionable as the stock fan or the Panaflo at 12V. The Zalman
cools as well as the 8952 with noise lower in level and pitch, but still
not acceptably quiet by SPCR standards.

B) 7V
Fan Results
Heatsink
idle
load
°C
rise
°C/W
MP
°C/W
TDP
Alpha 8952 w/ stock fan
27°C
50°C
28
0.37
0.42
Alpha 8952 w/ Panaflo L1A
28°C
52°C
30
0.40
0.45
Zalman 7000
24°C
41°C
19
0.27
0.29
Stock Intel HSF
28°C
55°C
33
0.47
0.50

At 7V the
Sanyo Denki fan was much quieter than at full speed, with considerably less whine. The air turbulence noise was also noticeably quieter. I heard just the slightest bit of clicking, but only at within about 6″ of the fan. At
7V, the Sanyo Denki sounded a lot like the Panaflo L1A at 7V, but with less
clicking than the Panaflo.

The 8952 was a decent performer with the Panaflo
L1A at 7V. The noise was pretty low and probably wouldn’t be objectionable
for most quite PC enthusiasts, as long as they aren’t real sensitive to the
quiet clicking noise. Chances are it would be muted by the case and noises from other components. The 8952 would certainly be suitable for cooling a 70-80W
P4 with either its included fan or the L1A Panaflo running at 7V, and both
fans are very quiet at those settings.

The Zalman still worked better at 7V and its
noise was pretty unobtrusive, consisting of a bit of air noise and
a quiet clicking.

The Intel HSF was already starting to breath pretty hard
at this voltage and still had a bit of whine and air noise.

C) 5V
Fan Results
Heatsink
idle
load
°C
rise
°C/W
MP
°C/W
TDP
Alpha 8952 w/ stock fan
28°C
62°C
40
0.53
0.60
Alpha 8952 w/ Panaflo L1A
30°C
63°C
41
0.55
0.62
Zalman 7000
26°C
45°C
23
0.33
0.35
Stock Intel HSF
28°C
64°C
42
0.60
0.63

At 5V the stock 8952 fan just barely cooled enough but was very quiet. There
was just the slightest bit of air noise and no discernible clicking at all.
No noise could be heard at all from over one meter away. Switching from the
Sanyo Denki fan to the Panaflo at 5V was a wash. Both fans cooled equally and were basically silent. The Panaflo did have a bit more clicking
noise than the 8952 fan but you had to get within 6-12″
before you could begin to notice it.

Of course the good ‘ol Zalman was still working just fine down at 5V. Its cooling
performance was still very good and it’s noise level was very quiet. I could
barely hear any wind noise and just the slightest bit of very quiet clicking
from about .5-1 meter away.

The Intel HSF is marginal at 5V. Load temps
were starting to get high and its whine and hum were louder than any of the
other fans in this test.

WITH A HIGHER POWER CPU?

Alpha claims that their new S-PAL8952 will cool the Intel Prescott,
Northwood and Extreme Edition processors but I only
tested it with the (relatively) moderate output of the 2.4C processor. We
can however, use the °C/W numbers to extrapolate the theoretical
performance of the 8952 with any other P4 processor, including the 100+ watt
P4 3.4GHz Prescott. According to Intel’s “TDP”
specs, the 3.4 Prescott puts out 103W. The higher “MP”
power rating for the 3.4Ghz Prescott is 115 watts so we’ll use those numbers
to calculate the estimated maximum CPU temperature for a 3.4E running 2xCPUBurn.
The fan used for these calculations is the included Sanyo Denki :

Extrapolated S-PAL8952M82 temps for a 103W, 3.4GHz
Prescott using the “TDP” power rating

Fan Voltage
C/W
°C rise
load
12 volts
0.29
30
52°C
7 volts
0.42
44
66°C
5 volts
0.60
62
84°C

Extrapolated S-PAL8952M82 temps for a 115W, 3.4GHz
Prescott using the “MP” power rating

Fan Voltage
C/W
°C rise
load
12 volts
0.25
29
51°C
7 volts
0.37
43
65C
5 volts
0.53
61
83°C

Depending on how much faith you have in these calculations, it does look like
the S-PAL8952 with its quiet fan can sufficiently cool the hottest running Prescott, at least at 12V. At 7V I would call the performance barely acceptable. Inside a case with minimum >10°C higher ambient temps, it probably would not do.

Would this Prescott cooling performance be acceptable
to quiet PC enthusiasts? In my opinion, even at the fan at 12V, barely. The results wouldn’t change much if we
switched to the Panaflo fan because it’s performance is virtually identical
to the Sanyo Denki fan.

FINAL THOUGHTS AND CONCLUSIONS

Alpha’s latest P4 heatsink is a minor revision of its previous top-of-the-line
model, the PAL8942. It follows the Alpha design philosophy of cooling
the CPU by sucking the hot air off the processor instead of blowing cool air
down on top of it. This seems to be just as effective in cooling ? plus it
leaves open the tantalizing possibility of ducting the CPU heat right out
of the case, instead of just recirculating most of it.

The CPU is generally the
hottest component in a modern PC (some of the latest high-end video cards excepted)
so evacuating its heat directly out of the case can put much less load of the
rest of the cooling system. This means that the user will need less airflow
to cool the rest of the system, which of course is a nice bonus for someone
looking to run their system as quietly as possible.

So how effective is the 8952 at quietly cooling a P4 CPU? Well, it’s
sufficient to run nearly silent and cool a 70-80W processor to a perfectly decent
level. Once you get up around 100 watts, things start getting marginal. You’d
need to run the included fan, or something similar like the L1A Panaflo at 10~12V to get adequate cooling and that’s above the limit that I could
tolerate, noise-wise.

The S-PAL8952 is a good heatsink, the best performing of this type
that I’ve tested. At low fan speeds it’s a better performer
than the recently reviewed Scythe
Samurai
HSF, another “sucking” type of heatsink. It may be
marginal with a low airflow fan and the hot Prescott CPUs, but silent PC enthusiasts should
give those 100W+ CPUs a pass anyway.

PROS

* Fan is quiet at lower speeds
* Good performance, even at slow speed
* Very strong, stable mounting system
* Will fit any Socket 478 motherboard with zero issues
* Good choice for using in a ducted system

CONS

* Moderately expensive
* Motherboard needs to be removed from case to mount
* Marginal performance with hot Prescott CPUs

Many thanks to Alpha
for the opportunity to review this HSF.

* * *

Discuss this this article in the SPCR Forums.

Leave a Comment

Your email address will not be published. Required fields are marked *