Swiftech's New Apogee GTZ!

[Anubis_386]

Just saw this news post up on techPowerUp (http://www.techpowerup.com/68351/Swiftech_Releases_Apogee_GTZ_Processor_Water_Block .html) about Swiftech's newest high performance CPU water block, the Apogee GTZ .. According to the product page on the Swiftech site the new block uses a 'micro pin matrix' on the plate ..

The results (well those provided by Swiftech) are pretty spectacular for 65nm Quads...


Ambients: 30°C / Flow Rate: 1.5gpm / Q6600 G0 @ 3500 Mhz - 1.38/1.50 V (Actual/Bios)

Apogee GTX: 58°C
Apogee GTZ: 50°C


An 8°C drop! The temp decrease with a Q9450 (@3500 Mhz - 1.27/1.40 V) and a E6600 (@3000 Mhz - 1.47/1.55 V) isn't as great, 3°C compared to the GTX ..

I also found this bit interesting:

Other types of sockets such as AMD® socket F, AM2, and 940 as well as upcoming sockets similar in design to that of socket 775 may also benefit from the enhanced contact area.

It would seem that this block will be compatible (when combined with a new mounting system) with Nehlem .. So, it seems Swiftech has taken back the CPU waterblock crown once again .. Pricing wise - $69.95 .. And here's a couple of pics:

http://www.swiftnets.com/assets/images/products/apogee%20GTZ/apogeeGTZ-x400.jpg

http://www.swiftnets.com/assets/images/products/apogee%20GTZ/mirror.jpg

http://www.swiftnets.com/assets/images/products/apogee%20GTZ/base-plate-pin-matrix.jpg

[Nebulous]

Hmmm, verrry nice. Not too crazy about the micro pin matrix plate tho. It looks like it was designed for low pressure pumps. I could be wrong.

Other than that, the block itself looks really nice.

[Sam__]

If those results are close to true then it looks great. Will wait for some external reviews before i make a final judgement.

Liking the look of the "micro pin matrix". have to agree with neb saying it looks like its for low pressure pumps.

[Anubis_386]

Indeed, here's the two Flow Rate and Pressure Drop graphs up on the Swiftech site ..

http://www.swiftnets.com/assets/images/products/apogee%20GTZ/apogee-GTZ-PD-vs-FR.png

http://www.swiftnets.com/assets/images/products/apogee%20GTZ/gtz-actual-FR.jpg

[muddocktor]

Well, I'll see how well this does compared to a Fuzion V1 with quad nozzle and a Storm rev 2. I just ordered it from Sidewinder Computers. They are showing it as arriving on 8/14/08, so I went ahead and preordered it now. :D

I will be able to test it in 2 systems. My main rig, which has a high pressure Iwaki MD20RTZ pump and in my old loop, which has an Eheim 1250 for the pump.

[jokers_greg]

Well, I'll see how well this does compared to a Fuzion V1 with quad nozzle and a Storm rev 2. I just ordered it from Sidewinder Computers. They are showing it as arriving on 8/14/08, so I went ahead and preordered it now. :D

I will be able to test it in 2 systems. My main rig, which has a high pressure Iwaki MD20RTZ pump and in my old loop, which has an Eheim 1250 for the pump.

yay :). I'd like to see the difference. I'm tired of my whitewater :o.

[voigts]

I'll be watching the results also whenever the testing gets done over at XS from Martin, Niksub, and mcoffey.

I respect Gabe for being actively involved over there. It shows their commitment to quality and feedback.

[maxspam]

wait a second... Kind of a water cooling noob but how does the water get to the center pins? Is there a water channel that takes one of the barbs on the side and puts it ontop of the middle pins?
also there is a boundary around the center pins, it looks as though if the water were on the outside where the straight channels are it wouldn't be able to get to the center pins.
I am sure swiftech knows what they are doing but what?

(thanks mod/admin)

[meionm]

Well, hopefully who ever gets doesn't mess up inlet with outlet. On previous apogee it didn't matter which barb was used as inlet, unlike on new one.
http://lib.store.yahoo.net/lib/sidewindercomputers/gtz12.gif

http://lib.store.yahoo.net/lib/sidewindercomputers/gtz11.jpg

[Jinu]

i have been waiting for its release for a hot min now. i was going to buy a fuzion v2, but will wait to see independent results before i buy another block.

[nikhsub1]

I should be getting mine anytime now... I'm sure my results will be the last published as I always take too much time LOL.

[muddocktor]

I'll probably be finished testing after you, nik. I won't be in from the rig until next Wednesday evening, so I won't even be able to start until Thursday. :(

EDIT: I figure that my testing should help show the performance differences between dual and quad core parts and also how it works with both high psi and med/low psi pumps. My Iwaki is rated at around 14 psi pressure and the Eheim is only around 2.83 psi, according to the little info I could gleam from Eheim's website.

[baditude_df]

Ooooh, crap. I just bought a V2 w quad nozzle. I think I'll just pick up one of these for the heck of it and see how it stacks up cooling my ol' q66.

[911us]

any review yet guys ?....

[voigts]

any review yet guys ?....

It takes a while to do this kind of testing, not to mention the fact that these guys aren't exactly getting paid to do so.

[xtkxhom3r]

**** i cant wait till some better reviews i was thinking of buying a v2....i have a apogee gt this will be a good upgrade right?

[ziggo0]

**** i cant wait till some better reviews i was thinking of buying a v2....i have a apogee gt this will be a good upgrade right?

thats what im wondering...i'd like to knock a couple C off

[xtkxhom3r]

hell yea man maybe even 10 :P

[Jinu]

well, if swiftech's results are anything to go off of, 8c off from the gtx is impressive.

[Anubis_386]

I just find it interesting that they say the GTZ gives an 8°C drop with a Q6600 but then only a 3°C drop from a Q9450 .. Either they designed it to better handle higher heat outputs or something is a miss ..

[911us]

I don't know how it is compared to D-tek Fuzion with quad nozzle.If it can surpass the D-tek then everything will be a different story:)..Cheer

[muddocktor]

Anubis_386, I think part of the difference is that most Q9450's are limited in their overclock by the fsb limitations that most of the 45nm quads seem to run into. A more appropriate 45nm quad to check with would be the QX9650 or the new Q9650 procs, since they both have a higher default multi and the QX is totally unlocked anyways. That way you can get a much higher overclock out of it without running into the fsb wall like most Q9450's do. With the Q6600's there generally isn't any problem with a fsb wall with their multi when going for max stable overclock. But with the Q6600 and the Q9650/QX9650 procs, it should be fairly easy to get 200+ watt loads on the waterblocks for testing purposes.

EDIT: Just did a little number crunching on the 2 procs in the oriiginal post and assuming 1.25v vcore for the Q9450 and 1.30v vcore for the Q6600 as the stock vcore, the overclocked wattages are 156 watts for the Q6600 and 128 watts for the Q9450. With the Q9450, both waterblocks are having to deal with less heat so I would expect to see less temp differential. It will be interesting to see how this block does with >200 watts though.

[Blazing fire]

I don't know how it is compared to D-tek Fuzion with quad nozzle.If it can surpass the D-tek then everything will be a different story:)..Cheer

Or an ek-supreme with LOWER RESTRICTION!

[nikhsub1]

Or an ek-supreme with LOWER RESTRICTION!
Why do you think you need less restriction?

[Mother Goose]

Man-o-man. Water cooling has gone a long way since when I was last active.

Jeez. While I don't really know if they've improved in design, you can't argue they're much cheaper and shinier these days :)

[muddocktor]

Yeah, they are prettier and they are more efficient too. My Fuzion v1 keeps my 203 watt load QX9650 at around 44 C while crunching Seti. :cool: I want to see if this new Swiftech design can beat that (and I wanted to upgrade the Storm block anyways).

[Jason Hilton]

I ordered one for my QX6850. It should be here on Monday, I'll post results as soon as I get it setup!

[Jason Hilton]

Just received my GTZ! In the pictures, the copper isn't scratched, just the plastic sticky thing.
The chrome is also very nice in person, the camera/light makes it look bad, but in person the quality is quite nice.

The back plate is a snap. It has sticky insulation, and the bolts are welded to it. That way they won't turn/come out/etc so it will be much easier to work with.

Here are some (8) unboxing pics:

http://img329.imageshack.us/img329/3654/96666879vm7.jpg
http://img296.imageshack.us/img296/5985/34032522rr3.jpg
http://img503.imageshack.us/img503/5428/90770913ws2.jpg
http://img411.imageshack.us/img411/6231/39529078br2.jpg
http://img530.imageshack.us/img530/7718/31953570ih1.jpg
http://img524.imageshack.us/img524/9443/64892851pc1.jpg
http://img294.imageshack.us/img294/8938/57576041dq8.jpg
http://img210.imageshack.us/img210/9017/closeupnq2.jpg

[Jason Hilton]

Almost forgot the shots with the DX48BT2!

http://img526.imageshack.us/img526/1787/mobo1lp8.jpg
http://img213.imageshack.us/img213/1959/mobo2cx1.jpg
http://img201.imageshack.us/img201/7035/mobo3pm4.jpg

[Jinu]

results, we want results. lol.
that looks hella sexy.

[z0n3]

Anxiously awaiting the results :D

[Jinu]

not sure if anyone has seen these yet, but here we go.
http://lh3.ggpht.com/mcoffeyXS/SKY--omafbI/AAAAAAAABRM/E6X1LbmIjX8/s800/Presentation1.jpg
http://lh6.ggpht.com/mcoffeyXS/SKY--3tAvtI/AAAAAAAABRU/jLrzPqiIAC4/s800/Presentation2.jpg

more info over at http://www.xtremesystems.org/forums/showthread.php?t=198681 very good write up and results from these guys.
i just want to see how it does on a q6600 though. Swiftech claims the 8c drop in a 65nm core and the test were done on a 45nm. hopefully another review comes out soon.

[hata28]

Look like I sticking with my D-tek FuZion V2:beer:

[muddocktor]

I just had 3 packages delivered to the house yesterday, according to my wife when I talked with her last night. So I imagine that one of them is my GTZ. I'll be leaving the rig in a few hours and going home, so hopefully I can start doing some comparison testing pretty soon. One of the other packages is a Q6600 and P5Q Pro that I got on sale at Newegg last week too, so I don't even know if I will test on the old E6300 at all in my second watercooled rig. I might just drop it in that rig and test the Storm against it on that rig with a quad instead of the E6300. But I do plan to test my Fuzion on that system too, since it has a lower pressure pump than my main to see if differences in the pump will make a performance difference with the 3 blocks.

[LordBaal]

Has anyone seen anything about how it performs on an AMD processor? From what I read from Swiftech, their new block was specifically designed for and tested on Intel chips and probably will not see the same performance on an AMD.

[Dj Spiel]

mmmm copper

[baditude_df]

I'm gonna add one of these to my next Petra's order just for the hell of it. Gotta tell ya though, that FusionV2 with the quad nozzle has really been doing a good job on my Q66.

More results people. Gitterdun!

[Kuroimaho]

The results on the XS aren't so good to justify an upgrade unless you use low pressure pump, for a new build looks like the best choice.
At least a step forward but I hoped for a bit more.

[Evilsizer]

Has anyone seen anything about how it performs on an AMD processor? From what I read from Swiftech, their new block was specifically designed for and tested on Intel chips and probably will not see the same performance on an AMD.

i would suspect better numbers on AMD's chips at 65nm. namely the X3/X4's since they are native tri/quad core. i wouldnt really say the WB is made for intel chips only. just from seeing the design looks like it would lend it self well to a native quad core. maybe they did this for future neha chips, who knows...

[muddocktor]

Yeah, I'm thinking that the design will work well with all processors that have large cores, whether it's 1 big die or 2 smaller ones under the IHS.

[Blazing fire]

Why do you think you need less restriction?

Why not? You can add more stuff in your loop that way.

[Jim Morbid]

Sexy...

[muddocktor]

I went and transplanted my new Q6600 in my second watercooled system last night and I'm presently working up a half decent overclock on it to test with. I will start testing with that system since it has a much weaker pump than my main rig (Eheim 1250 compared to my Iwaki MD-20RTZ) and I think I will see a greater cooling difference with a lesser pump. I presently have my Storm Rev 2 on that system as a baseline block and I will gather data from it first, then transplant the GTZ on that rig. As you know, the Storm has a fairly high resistance to flow and I think that the GTZ will allow much better flowrate at the pressures an Eheim 1250 can put out.

[QuietIce]

Hey, muddocktor - whatcha' gonna' do with that Storm when you're done with it ...? ;)

[muddocktor]

I don't know for now. I might try watercooling a third machine in the future though, most probably one of my DC Intels. Basically all I will need to set up another system will be a pump and some tubing. I also have a DD RBX block too, but it's set up for socket 478 and I don't know if I want to invest $15-20 more for a 775 top for such an old design.

[baditude_df]

How would an Eheim 1250 stack up against a D5 mud? In terms of pressure I mean. They look roughly the same if you look at the specs. I ask because I have a D5.
http://www.marinedepot.com/ps_ViewItem.aspx?idproduct=EH1250
http://www.acousticpc.com/laing_d5_water_pump.html

Also, I have a DD-MC-TDX on the classies for cheap! Cheap cheap! hehe.

[muddocktor]

The D5 has a bit better head pressure than the Eheim 1250. Max head pressure on the 1250 is 6' 7" water and the max head pressure on the D5 is 10 feet with water. That works out to the following:

Eheim 1250 - 2.841 psi
D5 - 4.316 psi

So, the D5 would be much better for a restrictive block.

[Jason Hilton]

I haven't had a chance to test it on my QX6850 yet, but I did test it on a E4700 2.6 @ 3.33ghz

Idle: 35*C
Load:44*C

[Jinu]

I haven't had a chance to test it on my QX6850 yet, but I did test it on a E4700 2.6 @ 3.33ghz

Idle: 35*C
Load:44*C

what were the temps before? what block did you have before?

[Jason Hilton]

Don't know, never water cooled the E4700 until now. I just had one sitting around.

[Jason Hilton]

Ok... the results you have been waiting for! Temperatures were averaged across all 4 cores.

Intel Extreme QX6850 @ 3ghz (stock)

Idle: 44*C
Load: 51 to 52*C

Here are some results at 3.66 ghz: (averaged across all 4 cores)

Idle: 46*C
Load: 55*C (averaged)
Peaked at 61*C


Not too bad!

[Albyno]

considering my TRUE holds my QX6700 @3ghz at 54c idle thats good enough for me to get one. now to decide on the rest of the components

[Blazing fire]

That's insane temperatures. What's the load like? 90?

[Albyno]

not loading it till i get better cooling. dun wanna kill the poor thing

[Jason Hilton]

I did some MAJOR stress testing @ 3.33ghz. to get the CPU at 90-100% load, I had to run Super PI, Prime 95, and BurnIn Test at the SAME TIME.

Full Load @ 3.33ghz: 62*C (averaged across all 4 cores)
With a Peak of 67*C

[muddocktor]

I just installed mine on the Q6600 system in place of the Storm Rev 2 and you guys are going to laugh at my expense. I went ahead and installed it and mounted it up using the existing mount I was using for the Storm. Basically, it's a modified U-120 backplate retapped to 6-32 thread with 4 studs mounted in it and I am using the old DD mounting springs and knurled nuts and tighten it down to a slight coil bind on the springs. This has worked well for me in the past and I didn't feel like removing the mobo to use the new Swiftech backplate. So I boot up and start up Seti to warm up the loop and instantly my temps are 60-65 C, which is at least 10 C higher than my Storm was doing.:eek: So I shut BOINC down and start up Prime95 to see what a real load would do and temps jump to the low 70's.:drool: So after a few minutes of that I shut down P95 and just let it sit at idle and it's idling in the high 30's. All these temps are way high for this setup compared to the Storm, BTW. So I start thinking what could be wrong as the mount seemed to go well. Then I had a thought; this block was brand new and it had plastic on the base. Did I pull the plastic off?:bang head:bang head So I pulled the block off and sure enough, I left the protective plastic on the block.:screwy: It surprised me that it worked as well as it did with that plastic still covering the base.

Anyways, now that I have removed the plastic and remounted the block, I'm presently warming up the loop to prepare for recording some temps to compare to the Storm while using the Eheim 1250, which isn't a high pressure pump. BTW, I tried blowing through the Storm and the GTZ and the Storm has a noticeable amount more restriction than the GTZ.

[Evilsizer]

HAHAHAHAH
new code for next month, "mud needs to learn to remove plastic before using the WB"

now what do the temps look like mud?

[Albyno]

lol thanks for wacking my up Mud. soon as i read "this block was Bra" i knew where it was going hehe.

how does it look now that plastic is gawn!?

[muddocktor]

Looks real shiny now.:D

@ Evilsizer, I will do a logged temp run tomorrow morning, just like I did with the Storm, but it's looking so far like it's running 3-5 C cooler, which is pretty good IMO, since the Storm isn't just an old hack waterblock.

[muddocktor]

Well, I have some data to post and this GTZ really shines in more than the physical sense. I haven't done like nik and did 5 mounts of each as I just don't have the time, especially with a hurricane potentially coming my way sometime next week. But I do believe I got quality mounts on both waterblocks and I don't think that doing more mounts would change results substantially.

The system specs:
Gigabyte P35 DS3R
Q6600 G0, vid 1.325
Swiftech MCW30 nb block
Thermochill PA160 radiator with Panaflo M1BX fan in push configuration
7/16 ID Tygon R3400 tubing
Home made reservoir plumbed with 1/2" brass ell and pipe to divert return 1/2" brass barb to downward position, intake 1/2" brass barb comes straight out of res with 4-5" 1/2" to 3/4" DD adapter to pump intake.
Eheim 1250 pump, outlet side plumbed with 1/2" brass ells, pipe and 1/2" brass barb to divert pressure side downwards towards my waterblocks.
Home made waterblock mount for the cpu blocks, made out of a TR backplate rethreaded to 6-32 with 4 studs sticking through the mobo holes. The blocks are held in place with a DD hold down kit something like this one (http://www.dangerden.com/store/product.php?productid=108&cat=40&page=1). I've found this hold down works very well and with a backplate it doesn't stress the mobo.

I used some Thermalright silicone based TIM for both tests that I got with a hsf because it requires no break in time and is half decent for silicone based TIM. When mounting the blocks, I run the knurled nuts down on the studs until I get a slight coil bind on the springs.

The processor is running at 3300 MHz with loaded vcore of 1.36 volts. BOINC running Seti was used to generate the load on the processor. It doesn't get everything quite as hot as Prime95 but I can also get some valid crunching done and the results are still comparable since they place the same load on the core for both waterblock tests. Core Temp was used to log the temps for at least 1-1 1/2 hours for each waterblock. I didn't check idle temps because they don't mean squat to me and my machines run at 100% load anyways crunching DC projects.

Now for the results.

Swiftech Storm Rev 2 results:
Core 0 AVG- 51.02524834
Core 1 AVG- 50.68605132
Core 2 AVG- 49.74254967
Core 3 AVG- 51.86548013

AVG Core temp- 50.82983237

Core 0 high- 54
Core 1 high- 54
Core 2 high- 54
Core 3 high- 55

Core 0 low- 46
Core 1 low- 47
Core 2 low- 47
Core 3 low- 47

Swiftech Apogee GTZ results:
Core 0 AVG- 44.18242188
Core 1 AVG- 45.35820313
Core 2 AVG- 42.55494792
Core 3 AVG- 42.09166667

AVG Core temp- 43.5468099

Core 0 high- 47
Core 1 high- 49
Core 2 high- 46
Core 3 high- 46

Core 0 low- 40
Core 1 low- 42
Core 2 low- 39
Core 3 low- 38

So it looks like I am seeing an average difference of 7.28 degrees C between the 2 blocks with nothing but a waterblock change.:attn:

I've zipped both excel files of the runs and attached it to this post if you want to look at the raw data that I got from the temp logs.

[Evilsizer]

O thats nice mud, seems that pump was having to much trouble with the storm. i might get one of those GTZ's when/if i get back into water...

[nikhsub1]

Yeah the storm is no good on IHS cpu's, your results are typical... and Mudd, please use the provided mounting hardware for the GTZ, Swiftech says it DOES make a difference for the better...

[muddocktor]

Oh, I do plan to try it. But things are happening too quickly down here right now. I just spent 1/12 hours trying to get 2 propane bottles filled. Gasoline is out at a bunch of stations except for premium and they have jumped prices 40-60 cents per gallon the last 2 days, gouging everyone that's prepping for this possible hurricane.:rolleyes: There are supposed to be laws here against price gouging during emergency times such as this but I guess our local and state officials are on the take so they don't enforce the law.:bang head:bang head

[Evilsizer]

dude that blows mud, hope you and family make it thru!