Foxconn better than SCV GC68

[OulOat]

According to Tom's Hardware, the Foxconn is a decent cooler, and the best part is, it's only 2 bucks. The problem is, where do I find one?

 

[Nerdwannabe]

Foxconn fans are mostly sold to system builders. They do not sell to retail market.

 

[drx9175l]

Its relatively low operating noise level, a mere 48 dB(A)

and 62C with a Duron 1200???


Stick with the GC68 for $6.

 

[Mikewarrior2]

except that this THG review was done with a synthetic test platform, not some crap-ass socket-thermistor.

There was a recent generic GC68 heatsink review, done with the internal diode. The GC68(one hte same mb, comparing socket-thermistor versus internal diode readings) shows lower socket-thermistor temps by 17C. Clearly yet another clever design (ala thermoengine) that takes advantage of poor temp reading on the majority of amd motherboards. The GC68 isn't the killer cooler than socket-thermistor's "lie" it to be.


Mike

 

[Lizardman]

The other good thing about the GC68 is that it take 8cm fans. This means less noise.

 

[Jhhnn]

I don't think anybody has claimed that the gc68 is some kind of killer cooler. They are, however, cheap, effective and quietly non-annoying for XP's in the 1600-1800 range.

While Tom's article acknowledges that the on-die vs socket thermistor delta is at least 15 degrees C, for any cooler, I've never seen any independent tests showing that the gc68 has a greater delta than anything else, nor did I find it reviewed while skimming the article.

One thing I find objectionable in such reviews is the use of total noise numbers rather than a more sophisticated sound graph. Two fans of the same total noise factor can have radically different effects on the listener, higher frequencies being much more annoying.

I also object to the use of the synthetic test bench- While Tom's crew obviously has the best of intentions, and the apparatus and comparisons are accurate, it doesn't necessarily follow that their placement of the lines denoting the different processors is accurate at all. I'm more inclined to trust figures derived from actual processors running actual programs...... just old-fashioned, I guess...

 

[Mikewarrior2]

except that using existing processors with inaccurate measuring sources is far worse than a synthetic test platform.

and let's face it. Most motherboards dont' allow the end-user to see the diode temps. the A7v333 and Epox 8k3a+/8k3a are among the rare ones that can be setup to show internla diode readings to the end user. Epox has disabled this on the newer motherboards.

Irregardless, there are numerous people who have fallen in love with the GC68 as some kind of magic "near" alpha, thermalright performance cooler. Whereas the truth is this heatsink design allows airflow off of it in a way, that with another heatsink may be 15C off socket-thermistor versus die, the GC68 will be 20-25C off between socket-thermsitor and die.


Mike

 

[Jhhnn]

Not to rain on the parade, Mikewarrior2, or attempt to offend you, but those deltas are only assertions on your part, rather than the results of actual testing.

Get an A7V-333, Xp1800, gc68 and run prime95, measure it both ways. I'll believe the results, even if I suspect you harbor some bias......

 

[Mikewarrior2]

Finally, I found the review.

Not quite the GC68, but looks to be from the same OEM manufacturer (not a shock, SVC isn't big enough to manufacture their own heatsinks!).

GC68 clone heatsink test at overclockers.com

and, to give you an example with some other recent tests at overclockers.com

Example test 1

Example TEst 2

If you compare the tests, you'll note the following results:

GC68 clone = 16C delta between on-die reading versus socket-thermistor.

Example 1 = 2C delta between on-die reading versus socket-thermistor.

Example 2 = 5C delta between on-die reading versus socket-thermistor.

And lastly, hte Swiftech MCx462 had a 6C delta between the two readings.

SO if you really think its my assertion, it isn't. And if you're trying ot argue that socket-thermistor readings are accurate, well, I've argued that many times before, and on-die testing has more than proven the socket-htermistor to be a completely inaccurate test device. Hell, if you look at the above tests, the GC68 clone, if tested via socket-thermistor, performs better than the example test 1 heatsink.

Designing a heatsink to perform well with socket-thermistor testing isn't a phenomenon. Its far too common, IMO. Just like the above example....

So you've got hte gc68 clone altering socket-thermistor readings 10-12C versus the other 3 heatsinks in the list. THat gives it a 10-12C advantage in 99% of the heatsink tests out there.

Mike

P.S. It isn't personal bias. Just like back with socket-thermistors, true readings have been what i have been itnerested in. I raised hell when the thermoengine came out, because it had a similiar socket-thermistor "adjusted" reading effect. Wonder where thermoengine is now?? Their heatsinks handled the 1.2ghz tbird well, but not hte 1.4. How is that possible when it had such "great" temps? Because the temps were lies.

P.S.2 This is third party testing. You can't even claim personal bias.

 

[JokerF15]

As always mike, Good info in your threads =).


However, the GC-68 isn't 'built' for performance. It is basically a heatsink that will perform to the specs of or better than the STOCK amd cooler. It fairly cheap at ~$6.00.

I personally think it is a good quiet replacement for any stock AMD cooler.

That's it.

After that your money can be better spent.

-JokerF15

 

[Mikewarrior2]

ehhh... its slightly better than the latest retail AMD fans.

but again, there are poeple (and many of them) that believe the GC68 is within 5C of the top dogs. Of course, that is incorrect, but nonetheless, a fairly common belief.



Mike

 

[OulOat]

Too bad Foxconn isn't availible at retail stores. Hm, a fan that performes at GC68 level at 1/3 the price, what is the world coming too...

 

[ShinSa]

Originally posted by: OulOat
Too bad Foxconn isn't availible at retail stores. Hm, a fan that performes at GC68 level at 1/3 the price, what is the world coming too...

Too bad GC68 isnt available at retail store. HSF that performs at ALPHA level at 1/5 the price.

 

[Mikewarrior2]

Shinsa,

have you not seen the links above? The GC68 is nowhere close to an Alpha in real performance, only in faked on socket-thermistor temps.

Jhhnn,

See, that's exactly what i meant why i posted the first time.



Mike

 

[zeroidea]

Funny. I just replaced an SK6 with a GC68 today, both with Panaflos. My duron's temp dropped 11C accordning to MBM5 and no longer crashes during lengthy MP3 encoding sessions..

THG's results seem very odd. I don't think I trust "die simulators".

 

[Mikewarrior2]

you trust socket-thermistor testing over internal diode/ simulated die testing (which is perfectly fine).

was your sk6 mounted with an 80MM panaflo? Criss-cross mounted 80mm fans on the sk6 do not seem to work particularly well. In addition, it is quite a common problem for the SK6 to catch part fo the plastic socket-lip. That would drastically drop its performance.


Mike

 

[deerslayer]

My sk6 is mounted with an 80mm panaflo, and I can verify that it really doesn't do that great of a job. I have the GC-68 too and it performs much better in comparison to the sk6 w/ 80mm Panaflo.

you trust socket-thermistor testing over internal diode/ simulated die testing (which is perfectly fine).

I have no clue what that means. I have the Epox 8k5a2+ motherboard and under a SETI load, I get temps of about 51C with the GC68. Whether those are socket-thermistor readings or internal diode readings I have no clue.

The processor is a 1600+ oc'ed to 1750mhz (2100+).

 

[zeroidea]

Originally posted by: Mikewarrior2
you trust socket-thermistor testing over internal diode/ simulated die testing (which is perfectly fine).

was your sk6 mounted with an 80MM panaflo? Criss-cross mounted 80mm fans on the sk6 do not seem to work particularly well. In addition, it is quite a common problem for the SK6 to catch part fo the plastic socket-lip. That would drastically drop its performance.


Mike

I bought my SK6 from a forum member and it came (at my request) with a 60mm panaflo. It did a very good job at cooling an FCPGA celeron super quietly, but when I switched to an AMD system I found it wasn't performing well at all. Another poster here on AT suggested the 80mm 'trick', so I ran it for awhile with an 80mm panaflo (both the 60mm and 80mm were the quietest models). I tried it blowinf onto the heatsink, and sucking off of. I replaced it with one of the Antec 80mm fans which came with my SX1030 case. If there was any improvement in cooling, it was negligible. Temps were always at least 54C, and when the machine was stressed for too long it would lock up.

Like I said, MBM reports an 11C decrease with the GC68 and the Crystal fan (which doesn't seem to report RPMs. I assume from the noise level it is operating somewhere near its minimum..). In addition to that, the exaust at the back of the case is much less warm than it was previously.

Granted, this isn't apples to oranges, comparing a cooler designed for 60mm fans to one designed for 80mm fans, but you still have a situation where a cooler which was, very recently, considered one of the best, and which you're lucky to find used for $25, is easily, hands down, beaten by a very inexpensive all aluminum heatsink.

Mostly, though, I think it's unlikely that there's this conspiracy of clever design based on exploiting thermistor read outs. My mind could be changed easily enough I were to see your claims validated by a significant number of worthwhile reviewers, but... It's not something I particularly feel like joining a crusade for. For me, I have a cooler which keeps my machine stable, is a very noticeable improvement over its predecessor, is not noisy, and best of all, cost a fraction of what a more high profile cooler from the likes of Thermaltake, Zalman, or Alpha would have cost.

 

[OulOat]

Originally posted by: ShinSa

Originally posted by: OulOat
Too bad Foxconn isn't availible at retail stores. Hm, a fan that performes at GC68 level at 1/3 the price, what is the world coming too...

Too bad GC68 isnt available at retail store. HSF that performs at ALPHA level at 1/5 the price.

Soo that means Foxconn performs at Alpha's level at 1/15 the price. Wow.

 

[Mikewarrior2]

In addition to that, the exaust at the back of the case is much less warm than it was previously.

Granted, this isn't apples to oranges, comparing a cooler designed for 60mm fans to one designed for 80mm fans, but you still have a situation where a cooler which was, very recently, considered one of the best, and which you're lucky to find used for $25, is easily, hands down, beaten by a very inexpensive all aluminum heatsink.

Mostly, though, I think it's unlikely that there's this conspiracy of clever design based on exploiting thermistor read outs. My mind could be changed easily enough I were to see your claims validated by a significant number of worthwhile reviewers, but... It's not something I particularly feel like joining a crusade for. For me, I have a cooler which keeps my machine stable, is a very noticeable improvement over its predecessor, is not noisy, and best of all, cost a fraction of what a more high profile cooler from the likes of Thermaltake, Zalman, or Alpha would have cost.

Look at your first quoted statment. the air in your case is cooler, while hte only change is the heatsink.

YOUR cpu did not miraculously become cooler. Rather, it is retaining more heat than before. The SK6 was transferring more of the heat from the CPU to the heatsink than the GC68 is. (evidenced by your exhaust air being cooler iwth the GC68). If hte GC68 were really more efficient of a design, your case airflow would be warmer, as more heat would be transferred between the GC68 and the CPU.

I wish there were more websites that did true CPU temp testing. Ufortunately, the vast majority still use socket-thermistor testing.

But your own observation very much proves that the gc68 is doing something weird to your temp "readings". If it were cooler, then your case exhaust should be warmer. Not cooler. Warmer exhaust air means the heatsink is doing the job. Unless you changed your CPu when changing heatsinks, this is physically true. Let's take an analogy of that. Let's take your stove, set to the same temp. With Skillet A, the skillet gets incredibly hot, and puts out fairly warm air to the room. With skillet B, the skillet stays fairly cool, and the "heat" coming off of it is quite a bit lower. The same is happening with your SK6 versus GC68. HEat retained in the CPU isn't miraculously going to be purged via heat expulsion off the CPU pcb.


Mike

 

[Lucky]

Its relatively low operating noise level, a mere 48 dB(A)

that's a joke, right?

 

[nemo160]

the sk6 heatsinks are still top performers, they require a fairly high flow fan to be effective though. My sk-6 with delta 60mm 38cfm keeps my tbird 1.4 at around 5-6 c over ambient idle, no more than 10c over ambient full load(thermistor temps, but its the best i can do with a tbird) , even adding a factor of 10-15 c to the 38 i'm getting under full load right now that's still beating the foxconn in toms review by 27c, the new sk6+ is supposed to be even a bit better still, not to mention the slk-800

 

[OulOat]

Originally posted by: Mikewarrior2

YOUR cpu did not miraculously become cooler. Rather, it is retaining more heat than before. The SK6 was transferring more of the heat from the CPU to the heatsink than the GC68 is. (evidenced by your exhaust air being cooler iwth the GC68). If hte GC68 were really more efficient of a design, your case airflow would be warmer, as more heat would be transferred between the GC68 and the CPU.

He's right, ya know.

Edit ---
I think a better test of how effective your cooling system is to test both the ambient air temp and CPU diode temp. Low ambient and diode temp = awesome case and CPU cooling. High ambient and low diode temp (CPU temp > air temp), good cpu cooling but lacking in the case department. Etc, etc...

 

[Jhhnn]

Some of what's been said about case temp vs cpu temp bothers me, as I think it's incorrect, or misleading at best. Heat output from the cpu is continuous, virtually constant when running any distributed client, regardless of what hsf is employed. From a cold start, the case exhaust temp will rise faster using a more effective hsf, but the continuous full load case exhaust temp will be the same as a less effective hsf unit once equilibrium is achieved. Pretty basic- the heat output doesn't change, the case airflow doesn't change, therefore the temp must remain the same according to the conservation of energy principle...... With a less effective hsf, the cpu temp is higher in order to overcome the greater thermal resistance thru the mechanism of a higher Delta T, still dissipating the same amount of heat.

 

[Mikewarrior2]

not true.

A less efficient heatsink will reach its "saturation point" at a lower temp than a more efficient heatsink. In addition, a more efficient heatsink is more effective at transferring both heat from CPU to itself, but from itself to the Air.

Conservation of energy remains true since the CPU is running hotter. The system itself still has the same overall energy, it just isn't being trasnferred as efficiently to the case exhaust air. Hence why he "feels" that the case exhaust is cooler with the gc68. If the CPU is running hotter, Case airflow is running cooler, Conservation of energy is the same as before, where the CPU was running cooler, and case airflow was slightly warmer.


Mike

P.S. you set up the scenario yourself when you said that CPU temp was warmer. In order for conservation fo energy to remain equal, the other variable (case exhaust temp) must be lower.