- Mar 10, 2005
- 14,647
- 2
- 0
Introduction
Hi, everybody. With one bold stroke, I?d like to claim several firsts. This is my first article submitted for peer review, my first hardware review, and most importantly, the world?s first English-language review of the elusive Xigmatek HDT-SD964. Oh, you lucky people!
Meet the new boss?
Xigmatek was founded in 2005. As far as I can tell, they are based in Taiwan, and employ German engineers. Recently, they have been increasing their presence in the North American market, not only with their own brand, but as OEM suppliers to Kingwin and OCZ. Their latest offering is the HDT-SD964. The nomenclature is similar to the auto industry:
Heatpipe Direct Touch ? Side Draft, 90mm fan, 6mm dia. heatpipe, 4 heatpipes
The HDT-SD964 incorporates features previously seen on other coolers: direct heatpipes from the HDT-S963 and a dual fan tower, similar to the XP-S964.
I bought the cooler from Gezekian.com for $38 plus $10 shipping. This is what comes in the box. The packaging is nice.
The obligatory manufacturer?s specs copy/paste:
Product Name ? HDT-SD964
Product Number ? CAC-S9HH4-U01
Heat Sink ? Base Material
? Copper (Heat-pipe Direct Touch)
? Fin Material ? Aluminum Alloy
Heat-pipe
? Heat-pipe SPEC F6mm
? Heat-pipe Q'ty 4 pcs.
Fan ? Dimension 92(W) x 92(H) x 25(D) mm
? Voltage Rating 12V
? Speed 1200~2800 R.P.M.
? Bearing Type Rifle Bearing
? Air Flow 39~54.6 CFM
? Air Pressure 1.7~3.3 mmH2O
? Life Expectance 50000 hrs
? Noise Level 22~30 dBA
? Connector 4 pin with PWM
Dimension ? 92(W) x 51(H) x 133(D) mm
Thermal Resistance ? 0.18?/W
Weight
? 466g (w/fan)
Application ?All Intel® Socket 775 CPU
?Core?2 Extreme / Quad / Duo
?Pentium® Extreme Edition / D
?Celeron® D
?All AMD Socket AM2 / 754 / 939 / 940 CPU
?Athlon? 64 / FX / X2 / Opteron? / Sempron?
The heatsink is composed of an aluminium base (likely extruded), 4 copper pipes, and 42 aluminium fins. There is no available information on the construction of the pipes, nor can I determine if the fins are pressed or soldered on the pipes. The fins are very thin and delicate, so once the fan is on, it?s best to handle the cooler either by the fan or the base. In fact, my cooler couldn't leave the factory without bent fins.
The heatsink has the 4 pipes and 5 aluminium fins on the bottom. The whole surface has been machined quite flat, heatpipes and all. When holding a brand new razor to the surface, at any angle and location, gaps show only where the pipes and base meet.
The supplied fan is marked as such:
M/N: AD0912HX-A7BGL P.S: (L2) DC12V 0.30A
If any fan gurus can decipher that, don?t be shy.
The fan attaches to the cooler with 4 rubber mounts. Eight are supplied for mounting 2 fans. I have no idea where to find replacements. I don?t know why, but I?m shocked at how good this fan is. At full throttle, the sound is 50% electric motor, 50% whoosh. There is no edginess to the sound at all, although SPCR might disagree. At 60%, 1600 rpm (approx. 7VDC) the fan drops below the noise level of my Noctua NF-S12. The sound at this speed is faintly electric motor, but mostly whoosh. Using a 12V battery and a Nexus fan controller, I could detect PWM ticking from 7 to 9 volts. Higher than that and the ticking is not discernable. More on this gem later?
To accommodate LGA775 installation, there are 2 screw holes for the attachment of 2 brackets fitted with twist-and-lock pushpins. For S939/AM2 installation, there is no factory bolt-through option. Instead, the loathed retention clip is used, limiting installation to either left-to-right or right-to-left. And this is where things go horribly wrong. The supplied bracket is a poor fit, with its fulcrum sitting about 3mm too low. This results in extremely high pressure needed to mount the cooler. Once I got it in on, there was more than enough tension to secure the cooler without locking the bracket into place. In fact, I needed to call a buddy to help me remove the stupid thing. Most fortunately for me, the Zalman mounting bracket is nearly a perfect fit in its place. Later, I found the clip from the stock AMD cooler will work also, but I?ll continue to use Zalman?s bolt-through bracket. Although the cooler can be installed with the motherboard in a case, the fan must be mounted to the cooler first.
The HDT-SD964 is an unorthodox heatsink, and that calls for some unorthodox methods. Usually, I apply a small blob of TIM in the middle of the chip and smoosh thoroughly. Doing this with the Xigmatek results in a thumbprint-looking smudge in the center that misses at least 1 of the 2 outer pipes. Next, I tried laying a bead on the bottom of the cooler and spreading it ultra-thin with a razor blade. This gave me a ?very heavy tint? and the gaps in the bottom were filled in perfectly. As awesome as it looked, the resulting temps were 5C too high. The best results came from laying a 1-2mm bead along each pipe followed by the traditional smooshing. This is about 8 times the amount of TIM I normally use, but the gaps seem to soak it up, while giving even coverage. Yes, the gaps are kind of a pain to clean.
?Same as the old boss?
The Zalman CNPS9500LED, and its family, is pretty well known, so we won?t spend too much time re-treading old ground. The construction is all copper, save the upper part of the base. It?s fitted with a non-replaceable (mostly) 3 pin, 90mm, blue LED fan. The CNPS stands for ?Computer Noise Prevention System.? Unfortunately, this is contaminated, expired, discount baloney. At 100%, the fan sounds like a hair dryer, with a pinch of sand in it, muffled by a pillow. At 60%, the fan still has a little growl to it.
Installation is excellent. The cooler can easily be mounted in any of 4 orientations while the motherboard is in a case, provided a backplate and bolt-through retention module are in place. Personally, I think the 9500 is an attractive piece of engineering. Let?s see if all that glitters is gold?
Testing Methodology:
Tests are conducted on a real-world computer, with a real-world installation of Windows. Tests conducted in the vertical orientation show no change in temperature. I am recording what Speedfan reports as ?K8? for CPU temps and Temp 2 for what I?m interpreting as northbridge temps. Ntune is used to log GPU temps. Prime95 (In-place large FFTs) is used alone and in conjunction with rthdribl (1280x720). LGA775 installation was not tested. Ambient temperature was measured to be 20° for all but 1 test (19°, normalized).
Relevant hardware:
AMD Opteron 165, overclocked to 2.5 GHz
Asus A8N32-SLI in stock form, BIOS 1303
Corsair 2x 1GB DDR400
XFX 7950GT (factory OC at 570 MHz, passive cooling)
Silverstone SST?ST60F power supply (Noctua 120mm fan, custom wiring harnesses)
Silverstone LC-20MB HTPC case (2 Nexus 80mm fans, brace and drive cage removed)
Tuniq TX-2
Relevant software:
Windows Vista Ultimate 32, all updates, some tweaks
Latest official Nvidia video and motherboard drivers
Prime95 v25.5
Rthdribl v1.2
Speedfan v4.33
Core Temp v0.96.1
Nvidia ntune 5.05.54.00
Call of Duty 4
Temps are taken at the end of each phase, with highest values being kept.
0 min: power on, no lid
5 min: idle 7V
10 min: idle 100%, start prime95 and rthdribl
30 min: kill rthdribl
50 min: put the lid on
60 min: restart rthdribl
80 min: loads removed
90 min: idle 100%
100 min: idle 7V
At the conclusion of the synthetic cycle, I remove the lid, play CoD4 Multiplayer for the remainder of the map in progress and the entirety of the following map, then repeat with the lid on.
It's getting hot in herre *Now with slightly better graphs for free!* *again*
West-East Orientation
East-West Orientation
South-North Orientation, Zalman only
HDT-SD964 Stock versus NMB-MAT
Well, well, well. It seems the Zalman, which isn?t a bad cooler, completely falls apart. I had to stop 2 of the 3 tests, before the fire department stopped them for me. Following the curve on Speedfan?s graph, there were still a few degrees left to go. Meanwhile, the Xigmatek is hardly breaking a sweat. Given TCaseMax of 65°, the HDT-SD964 enables me to sustain a 40% overclock straight through the dog days of August! And does it quietly. I?m guessing South to North installation, getting maximum draw from the case exhaust fans, would drop CPU temp by at least 5°C. Xigmatek?s great results are due to the combination of the heatsink and its superior fan.
The northbridge temps, the Zalman?s claim to fame, don?t really suffer with the Xigmatek. Or is that not really benefiting from the Zalman?
Notice how the GPU temps are positively affected by the Xigmatek? That?s again showing how well the fan deals with back pressure. For all its screaming, the Zalman just has no torque. This is particularly sad since its heatsink design is as wide open as it conceivably gets.
*UPDATE*
In order to compare the HDT-SD964's stock fan to something commonly available, I re-tested the cooler with a NMB-MAT FBA09A12H -1BX. Formerly Panasonic Panaflo, and formerly made in Japan, this example is known as the Chinaflo. With 2800 rpm, it's just as loud as the Zalman fan, but it has a more gritty, mechanical sound. At 60% it is still louder than, and stands out from, the power supply. Blech!
The Chinaflo was installed by tying it to the tower with 1mm round black StretchCord. Without any source for replacing the stock rubber mounts, I didn't want to risk losing any to damage. One pass of cord around the frame decoupled (padded, really) the fan from the tower. This stuff is great for improvising fan installation, and can be used to suspend hard drives. $1 to $1.50 gets you all you'll need.
The test was performed at 20C. The cooler was mounted once for the two fans, in the East-West orientation.
HDT-SD964 Stock versus NMB-MAT
I cannot explain how the Chinaflow dropped the CPU temp by 4 while raising the GPU by 6. The only variable was the fan. Either way, the noise is intolerable. Open air testing showed almost identical results for both fans.
And in the end?.
If the HDT-SD964 looks like this, then the CNPS9500 looks like this.
I hope you?ve enjoyed and benefited from this little article as much as I have.
Hi, everybody. With one bold stroke, I?d like to claim several firsts. This is my first article submitted for peer review, my first hardware review, and most importantly, the world?s first English-language review of the elusive Xigmatek HDT-SD964. Oh, you lucky people!
Meet the new boss?
Xigmatek was founded in 2005. As far as I can tell, they are based in Taiwan, and employ German engineers. Recently, they have been increasing their presence in the North American market, not only with their own brand, but as OEM suppliers to Kingwin and OCZ. Their latest offering is the HDT-SD964. The nomenclature is similar to the auto industry:
Heatpipe Direct Touch ? Side Draft, 90mm fan, 6mm dia. heatpipe, 4 heatpipes
The HDT-SD964 incorporates features previously seen on other coolers: direct heatpipes from the HDT-S963 and a dual fan tower, similar to the XP-S964.
I bought the cooler from Gezekian.com for $38 plus $10 shipping. This is what comes in the box. The packaging is nice.
The obligatory manufacturer?s specs copy/paste:
Product Name ? HDT-SD964
Product Number ? CAC-S9HH4-U01
Heat Sink ? Base Material
? Copper (Heat-pipe Direct Touch)
? Fin Material ? Aluminum Alloy
Heat-pipe
? Heat-pipe SPEC F6mm
? Heat-pipe Q'ty 4 pcs.
Fan ? Dimension 92(W) x 92(H) x 25(D) mm
? Voltage Rating 12V
? Speed 1200~2800 R.P.M.
? Bearing Type Rifle Bearing
? Air Flow 39~54.6 CFM
? Air Pressure 1.7~3.3 mmH2O
? Life Expectance 50000 hrs
? Noise Level 22~30 dBA
? Connector 4 pin with PWM
Dimension ? 92(W) x 51(H) x 133(D) mm
Thermal Resistance ? 0.18?/W
Weight
? 466g (w/fan)
Application ?All Intel® Socket 775 CPU
?Core?2 Extreme / Quad / Duo
?Pentium® Extreme Edition / D
?Celeron® D
?All AMD Socket AM2 / 754 / 939 / 940 CPU
?Athlon? 64 / FX / X2 / Opteron? / Sempron?
The heatsink is composed of an aluminium base (likely extruded), 4 copper pipes, and 42 aluminium fins. There is no available information on the construction of the pipes, nor can I determine if the fins are pressed or soldered on the pipes. The fins are very thin and delicate, so once the fan is on, it?s best to handle the cooler either by the fan or the base. In fact, my cooler couldn't leave the factory without bent fins.
The heatsink has the 4 pipes and 5 aluminium fins on the bottom. The whole surface has been machined quite flat, heatpipes and all. When holding a brand new razor to the surface, at any angle and location, gaps show only where the pipes and base meet.
The supplied fan is marked as such:
M/N: AD0912HX-A7BGL P.S: (L2) DC12V 0.30A
If any fan gurus can decipher that, don?t be shy.
The fan attaches to the cooler with 4 rubber mounts. Eight are supplied for mounting 2 fans. I have no idea where to find replacements. I don?t know why, but I?m shocked at how good this fan is. At full throttle, the sound is 50% electric motor, 50% whoosh. There is no edginess to the sound at all, although SPCR might disagree. At 60%, 1600 rpm (approx. 7VDC) the fan drops below the noise level of my Noctua NF-S12. The sound at this speed is faintly electric motor, but mostly whoosh. Using a 12V battery and a Nexus fan controller, I could detect PWM ticking from 7 to 9 volts. Higher than that and the ticking is not discernable. More on this gem later?
To accommodate LGA775 installation, there are 2 screw holes for the attachment of 2 brackets fitted with twist-and-lock pushpins. For S939/AM2 installation, there is no factory bolt-through option. Instead, the loathed retention clip is used, limiting installation to either left-to-right or right-to-left. And this is where things go horribly wrong. The supplied bracket is a poor fit, with its fulcrum sitting about 3mm too low. This results in extremely high pressure needed to mount the cooler. Once I got it in on, there was more than enough tension to secure the cooler without locking the bracket into place. In fact, I needed to call a buddy to help me remove the stupid thing. Most fortunately for me, the Zalman mounting bracket is nearly a perfect fit in its place. Later, I found the clip from the stock AMD cooler will work also, but I?ll continue to use Zalman?s bolt-through bracket. Although the cooler can be installed with the motherboard in a case, the fan must be mounted to the cooler first.
The HDT-SD964 is an unorthodox heatsink, and that calls for some unorthodox methods. Usually, I apply a small blob of TIM in the middle of the chip and smoosh thoroughly. Doing this with the Xigmatek results in a thumbprint-looking smudge in the center that misses at least 1 of the 2 outer pipes. Next, I tried laying a bead on the bottom of the cooler and spreading it ultra-thin with a razor blade. This gave me a ?very heavy tint? and the gaps in the bottom were filled in perfectly. As awesome as it looked, the resulting temps were 5C too high. The best results came from laying a 1-2mm bead along each pipe followed by the traditional smooshing. This is about 8 times the amount of TIM I normally use, but the gaps seem to soak it up, while giving even coverage. Yes, the gaps are kind of a pain to clean.
?Same as the old boss?
The Zalman CNPS9500LED, and its family, is pretty well known, so we won?t spend too much time re-treading old ground. The construction is all copper, save the upper part of the base. It?s fitted with a non-replaceable (mostly) 3 pin, 90mm, blue LED fan. The CNPS stands for ?Computer Noise Prevention System.? Unfortunately, this is contaminated, expired, discount baloney. At 100%, the fan sounds like a hair dryer, with a pinch of sand in it, muffled by a pillow. At 60%, the fan still has a little growl to it.
Installation is excellent. The cooler can easily be mounted in any of 4 orientations while the motherboard is in a case, provided a backplate and bolt-through retention module are in place. Personally, I think the 9500 is an attractive piece of engineering. Let?s see if all that glitters is gold?
Testing Methodology:
Tests are conducted on a real-world computer, with a real-world installation of Windows. Tests conducted in the vertical orientation show no change in temperature. I am recording what Speedfan reports as ?K8? for CPU temps and Temp 2 for what I?m interpreting as northbridge temps. Ntune is used to log GPU temps. Prime95 (In-place large FFTs) is used alone and in conjunction with rthdribl (1280x720). LGA775 installation was not tested. Ambient temperature was measured to be 20° for all but 1 test (19°, normalized).
Relevant hardware:
AMD Opteron 165, overclocked to 2.5 GHz
Asus A8N32-SLI in stock form, BIOS 1303
Corsair 2x 1GB DDR400
XFX 7950GT (factory OC at 570 MHz, passive cooling)
Silverstone SST?ST60F power supply (Noctua 120mm fan, custom wiring harnesses)
Silverstone LC-20MB HTPC case (2 Nexus 80mm fans, brace and drive cage removed)
Tuniq TX-2
Relevant software:
Windows Vista Ultimate 32, all updates, some tweaks
Latest official Nvidia video and motherboard drivers
Prime95 v25.5
Rthdribl v1.2
Speedfan v4.33
Core Temp v0.96.1
Nvidia ntune 5.05.54.00
Call of Duty 4
Temps are taken at the end of each phase, with highest values being kept.
0 min: power on, no lid
5 min: idle 7V
10 min: idle 100%, start prime95 and rthdribl
30 min: kill rthdribl
50 min: put the lid on
60 min: restart rthdribl
80 min: loads removed
90 min: idle 100%
100 min: idle 7V
At the conclusion of the synthetic cycle, I remove the lid, play CoD4 Multiplayer for the remainder of the map in progress and the entirety of the following map, then repeat with the lid on.
It's getting hot in herre *Now with slightly better graphs for free!* *again*
West-East Orientation
East-West Orientation
South-North Orientation, Zalman only
HDT-SD964 Stock versus NMB-MAT
Well, well, well. It seems the Zalman, which isn?t a bad cooler, completely falls apart. I had to stop 2 of the 3 tests, before the fire department stopped them for me. Following the curve on Speedfan?s graph, there were still a few degrees left to go. Meanwhile, the Xigmatek is hardly breaking a sweat. Given TCaseMax of 65°, the HDT-SD964 enables me to sustain a 40% overclock straight through the dog days of August! And does it quietly. I?m guessing South to North installation, getting maximum draw from the case exhaust fans, would drop CPU temp by at least 5°C. Xigmatek?s great results are due to the combination of the heatsink and its superior fan.
The northbridge temps, the Zalman?s claim to fame, don?t really suffer with the Xigmatek. Or is that not really benefiting from the Zalman?
Notice how the GPU temps are positively affected by the Xigmatek? That?s again showing how well the fan deals with back pressure. For all its screaming, the Zalman just has no torque. This is particularly sad since its heatsink design is as wide open as it conceivably gets.
*UPDATE*
In order to compare the HDT-SD964's stock fan to something commonly available, I re-tested the cooler with a NMB-MAT FBA09A12H -1BX. Formerly Panasonic Panaflo, and formerly made in Japan, this example is known as the Chinaflo. With 2800 rpm, it's just as loud as the Zalman fan, but it has a more gritty, mechanical sound. At 60% it is still louder than, and stands out from, the power supply. Blech!
The Chinaflo was installed by tying it to the tower with 1mm round black StretchCord. Without any source for replacing the stock rubber mounts, I didn't want to risk losing any to damage. One pass of cord around the frame decoupled (padded, really) the fan from the tower. This stuff is great for improvising fan installation, and can be used to suspend hard drives. $1 to $1.50 gets you all you'll need.
The test was performed at 20C. The cooler was mounted once for the two fans, in the East-West orientation.
HDT-SD964 Stock versus NMB-MAT
I cannot explain how the Chinaflow dropped the CPU temp by 4 while raising the GPU by 6. The only variable was the fan. Either way, the noise is intolerable. Open air testing showed almost identical results for both fans.
And in the end?.
If the HDT-SD964 looks like this, then the CNPS9500 looks like this.
I hope you?ve enjoyed and benefited from this little article as much as I have.
Facebook
Twitter