Zalman's fanless WC rig reviewed: Reserator1

Uberapan · Post by **Uberapan** » Sat Aug 07, 2004 1:31 am

Is the Athlon XP really that much hotter than a P4 in idle mode? Looking at the temps you reported in this article makes me wonder.

I have a reserator connected to a 3,2 ghz P4 Northwood and a Radeon 9800 Pro. When totally idle, the temps are around 28-30 degrees C. It has been on a WinXP SP2 bittorrent for 6 hours now, averaging around 20 % CPU usage, and the CPU temp is 35 C.

I will stress test the entire setup later and see what the max temps are.

charliek · Post by **charliek** » Sat Aug 07, 2004 1:48 am

Uberapan wrote:Is the Athlon XP really that much hotter than a P4 in idle mode?

The early Athlon XP 1900+ chips were very hot indeed. More recent chips have got faster, and also more heat efficient - so it doesn't necessarily follow that a newer faster chip will run hotter.

Now, I have a question: My Gainward 6800 Ultra Golden Sample has arrived, Hurrah! But the dual fans are mounted on a frame that just covers the SATA sockets on my Asus K8VSE Deluxe, bah!

Does anyone know of a 6800 waterblock that is compatible with the 6800 Ultra and the Reserator1?

jojo4u · Post by **jojo4u** » Sat Aug 07, 2004 3:27 am

Uberapan wrote:Is the Athlon XP really that much hotter than a P4 in idle mode?

Looks like the disconnect feature was not enabled for the Athlon.

http://forums.silentpcreview.com/viewto ... highlight= shows some compareable figures. A XP 2200+ with a Nvidia 5200 needs 69W AC fed by a Enermax EG465AX-VE. The efficiency if the Enermax is lower than the one of the Aurora.

Edward Ng · Post by **Edward Ng** » Sat Aug 07, 2004 6:04 am

Uberapan wrote:I have a reserator connected to a 3,2 ghz P4 Northwood and a Radeon 9800 Pro. When totally idle, the temps are around 28-30 degrees C. It has been on a WinXP SP2 bittorrent for 6 hours now, averaging around 20 % CPU usage, and the CPU temp is 35 C.

Have any of you guys done testing for diode report accuracy and properly calibrated your motherboards? Those numbers seem quite obviously missreported; what motherboards are you guys using?

I'm not blaming any of you; it's a problem with motherboards in general, and the mass majority of people just don't know that their motherboard's readings are off. It is a well known fact, for example, that ASUStek P4P800 and P4C800 mainboards report temperatures 8C short of reality (so, Uberapan, if you're using one of them, your true idle temps are around 36-38C and under your XP Bittorrent conditions and 20% load, more like 43C). The mainboard I use for my new heatsink reviews reports 7C short by default as well; all of us at SPCR have performed extremely extensive testing to precisely calibrate the readings from our diodes to ensure accurate and truthful reporting.

Not because we think we're better than anyone, but because we know, for a fact, that mainboards are almost always off. It's only a matter of how off, and being sure to properly compensate for this missreport.

-Ed

EDIT: Btw, charliek, I will be trying out a Dangerden Maze4 GPU block on my 6800GT in the coming week; keep an eye on the water cooling forum for my post; it should fit, as several people have told me that the FX and 6800 series GeForce cards use the same hole pattern, but I won't know for a fact until I actually try it.

charliek · Post by **charliek** » Sat Aug 07, 2004 9:46 am

Edward Ng wrote:EDIT: Btw, charliek, I will be trying out a Dangerden Maze4 GPU block on my 6800GT in the coming week; keep an eye on the water cooling forum for my post; it should fit, as several people have told me that the FX and 6800 series GeForce cards use the same hole pattern, but I won't know for a fact until I actually try it.

Thanks, Ed., I'll keep an eye out for it!

I'd be particularly interested in knowing if the waterblock can be comfortably installed in my Reserator1 loop, and if the stock Reserator 1 is able to effectively dissipate the heat of that and the CPU under game loads.

Edward Ng · Post by **Edward Ng** » Sat Aug 07, 2004 10:31 am

charliek wrote:I'd be particularly interested in knowing if the waterblock can be comfortably installed in my Reserator1 loop, and if the stock Reserator 1 is able to effectively dissipate the heat of that and the CPU under game loads.

Well, that I'm not sure if I can answer for you definitively, because I'm using a Pro 120 heater core from D-Tek Customs as well as two CSP750 pumps, so I'd say that's just, "a bit," more powerful than a Reserator.

acaurora · Post by **acaurora** » Sat Aug 07, 2004 12:04 pm

I wouldn't try it. My reserator gets pretty warm with the CPU alone..., and I have a Barton, which supposedly runs cooler than the older ThoroughBreds (sp?).

Uberapan · Post by **Uberapan** » Sat Aug 07, 2004 12:26 pm

Edward Ng wrote: Have any of you guys done testing for diode report accuracy and properly calibrated your motherboards? Those numbers seem quite obviously missreported; what motherboards are you guys using?

I'm not blaming any of you; it's a problem with motherboards in general, and the mass majority of people just don't know that their motherboard's readings are off. It is a well known fact, for example, that ASUStek P4P800 and P4C800 mainboards report temperatures 8C short of reality (so, Uberapan, if you're using one of them, your true idle temps are around 36-38C and under your XP Bittorrent conditions and 20% load, more like 43C). The mainboard I use for my new heatsink reviews reports 7C short by default as well; all of us at SPCR have performed extremely extensive testing to precisely calibrate the readings from our diodes to ensure accurate and truthful reporting.

Not at all. I have in fact suspected that the temp readings were off for quite some time, since they are a bit unrealistic. As you suspected, I use an Asus P4C800-E Deluxe.

Now, how do you go about calibrating the temp diodes? Also, just how do you "calibrate a motherboard"? I would be most interested in getting more accurate readings.

Edward Ng · Post by **Edward Ng** » Sat Aug 07, 2004 2:05 pm

Uberapan, you've got PM!

charliek · Post by **charliek** » Sun Aug 08, 2004 2:11 am

Edward Ng wrote:Have any of you guys done testing for diode report accuracy and properly calibrated your motherboards?

I certainly haven't - I didn't even know that motherboard diodes could be calibrated.

How would one go about getting accurate readings from CPU, Motherboard and VGA thermal diodes?

And, while I'm on, how do I know what temps are acceptable?

Edward Ng · Post by **Edward Ng** » Sun Aug 08, 2004 5:51 am

charliek, you've got PM...

ekotan · Post by **ekotan** » Sun Aug 08, 2004 5:32 pm

Another real-world case study:

I had a Pentium 4 3.0E Prescott PC watercooled by the Koolance EXOS. It was stable, but the noisy fans when the Koolance went into accelerated mode (temps >= 45C) were driving me mad. The UK is having an unusual heatwave right now, so room temperature is nudging 30C even at night with the windows open (this explains why I can't sleep and instead find myself typing this message). Noise is subjective and I am more sensitive than most. My PC has been carefully crafted out of quiet components like a 500W SilentMaxx fanless PSU, Maxtor hard drives with FDB motors and undervolted Pabst case intake and exhaust fans. I then added a GeForce 6800 Ultra AGP card. A Prescott and a GF6800U, can you think of a more heat-generating combination of PC hardware? At this point, the Koolance was pretty much running in accelerated mode all the time. Also at this point, my PC started crashing randomly and it was due to overheating because leaving the case sides open always fixed the problem. So I traded my CPU in for a Pentium 4 3.0C Northwood and took this opportunity to order a Zalman Reserator 1. I used two litres of de-ionized water and a cupful of Koolance anti-corrosion/anti-algae agent to fill her up. I used the Zalman GPU block (large type) with Koolance GPU screws to cool my 6800U. The screws Zalman provides are too fat to fit through the holes on the GF card, which is why it wasn't compatible. Using the slimmer screws from the Koolance kit fixed the problem. Other than that, the Zalman GPU block is a great fit, so I don't know why Zalman doesn't add two slimmer screws into their kit to make it compatible with the 6800 cards. Unfortunately, I couldn't re-use my existing Koolance GPU waterblock (which is beefy: rated at 180W) because the pipe sizes were not identical (Exos uses 1/4 inch tubes, Zalman uses 3/8 inch tubes).

So now, current PC has:

- P4C 3.0 GHz
- Intel D875PBZ motherboard
- inno3D GeForce 6800 Ultra
- 5 hard drives (4 in RAID-5 config on an Adaptec 2410SA controller)
- Reserator 1 cooling CPU and GPU using stock Zalman Reserator 1 waterblocks

Idle temps (after 10 hours of idling at the WinXP desktop):

CPU: 38
GPU: 52
Case ambient: 38
HDD: 42

Temps under load (after 12 hours of looping two copies of Prime 95 and 3DMark03):

CPU: 47
GPU: 68
Case ambient: 44
HDD: 46

I used the integrated temp sensors in my CPU, GPU, motherboard (for ambient) and HDD to read these temps. I understand they may not be entirely accurate, but I don't care about exact temperature values to the nearest degree. All I care about is to have a fast, 100% stable and quiet computer which is being adequately cooled.

Note that room temp was sometimes exceeding 30C during these tests, which is very unusual for England!!! That's as hot as it gets in England, so I expect the Reserator to perform even better when the English climate returns back to its normal, colder temperatures.

Overall, I could not be more impressed and pleased with the Zalman product. It's absolutely quiet and works startlingly well. I can now run my PC with the sides closed and there are no crashes. 100% stable. Best of all, the only way I can tell my PC is on is by looking at the screen to see if it's showing anything. That's the way it should be. I love quiet computers, which is one of the reasons why I love Macs. My Powerbook Al, on which I'm typing this, is absolutely silent. No fan noise whatsoever. Bliss...

If anyone would like to buy a second-hand EXOS Al in the UK, feel free to email me or watch eBay.co.uk.

E@

charliek · Post by **charliek** » Mon Aug 09, 2004 1:40 am

ekotan wrote:Another real-world case study:
(6800U) Other than that, the Zalman GPU block is a great fit, so I don't know why Zalman doesn't add two slimmer screws into their kit to make it compatible with the 6800 cards. Unfortunately, I couldn't re-use my existing Koolance GPU waterblock (which is beefy: rated at 180W) because the pipe sizes were not identical (Exos uses 1/4 inch tubes, Zalman uses 3/8 inch tubes).

Thanks for that, ekotan. It sounds like you are doing what I had in mind, but I was put off by the fact that Zalman hadn't officially produced a 6800 compatible vga block, and I was anxious that the amount of heat put into the loop by that and my A64 3400+ would be more than the Reserator could effectively dissipate. (See my thoughts on the subject in the watercooling thread). How are you measuring your temps, by the way?

I live on the French/Spanish border, so your unusual ambient temps are more common here. I am sure I could rig up a beefier pump and radiator somewhere in the reserator's loop if necessary.

Another question: the Danger Den 6800 block looks very expensive, and cools the RAM as well. Given that the 6800's GDDR3 is supposed to be more efficient than the previous generation's GDDR2 (which could be passively cooled) - surely I only need a waterblock for the GPU itself, and can passively cool the RAM? (My Gainward 6800U Golden Sample has some sizeable RAM heatsinks on it already)

Anyway, cheers ekotan, perhaps you could give us an update after a while of running with your current rig?

Edward Ng · Post by **Edward Ng** » Mon Aug 09, 2004 5:13 am

charliek, I was not aware that Danger Den makes a full-card block until now; I have placed my order for one.

I will do a comparison of the standard Maze 4 block versus the NV 68 block, once I get under way with the water cooling tests. I'm still awaiting a few parts.

-Ed

ekotan · Post by **ekotan** » Mon Aug 09, 2004 8:45 am

No problem charliek, glad to be of service. I'm using the temperature probes integrated into the CPU (P4 diode), the GPU (All 6800 family of GPUs have this, AFAIK) and my hard disks (Maxtor PATA and SATA drives with SMART). Case ambient temps are getting reported via my Intel D875PBZ motherboard's on-board circuits. I'm using MBM5 to monitor all of these reading in real time. Its Hi-Lo panel comes in especially handy to see temperature variances at a glance. Like I say, the values I see may not be ultra-accurate, but I'm more interested in stability and silence, not actual temp values as long as all's vaguely within spec.

I've been playing Need For Speed Underground today for a solid good two hours. Apart from being an entertaining game with cool graphics, it is notoriously heavy on the GPU as well as the CPU with lots of heavy shader effects. I experienced no slowdowns or lockups. The system can handle maximum detail at my LCD screen's maximum resolution of 1280x1024 at fluid frame rates. My old Radeon 9700 Pro used to choke unless I switched off light trails and motion blur. This 6800U is quite a beast.

It rained last night and it's thankfully slightly cooler outside today than it was yesterday, so my ambient temps had fallen as low as 35C this morning before I started gaming. The Zalman is still impressing me. I also don't have any variance in noise levels which other users have reported, it's always quiet. I would have noticed if the pump was getting louder at nights, since it's sitting in my bedroom. My unit came with a flow indicator which had metal connectors already, not plastic. I assume this is the revised version of the Reserator hardware, since Zalman has a recall notice for the first revision which had plastic connectors on the flow indicator. I run it 24x7, the way I run my PC.

Note that nothing is overclocked in my system, I never ever overclock anything for fear of introducing small, random errors into the calculations I do for my work (as you can see from the specs in my sig, this is my primary workstation which is only used for gaming occasionally). This is why I didn't bother with the expensive Danger Den full-card 6800U waterblock which also cools the VRAM. I just used passive aluminium heatsinks on the GDDR3 chips which came with the Zalman GPU waterblock (the type that you stick on with thermal tape). With default clocks on my inno3D 6800U (which seems to be 400 MHz core in 2D mode and 425 MHz in 3D mode by default, with VRAM always at 1.1 GHz), the Zalman GPU cooler seems to be able to cope admirably. It's actually a hefty heatsink with a lot of room inside for water to flow through, so I think the card retention bracket mechanism was the source of incompatibility. Besides, the 6800U GPU is designed to tolerate temps well over 110C, and I'm getting nowhere near that even after looping 3DMark03 for hours and hours. I don't even hit the 70C mark, so I think the Zalman is doing a good enough job here, at least for my purposes. Obviously, if you want to overclock, then the Danger Den would be a better bet. But it is expensive, so don't bother unless you'd like to go for extreme overclocking. It is entirely possible to passively cool the DDR3 modules. Just don't use the one big heatpipe+heatsink arrangement on the VRAM chips that the NVIDIA reference cooler uses, that only works with active air-cooling (i.e. when there's airflow). Rip the whole thing off, use a waterblock on the GPU and put some individual aluminium heatsinks onto the DDR3 chips directly. That would be my advice based on my experience.

As for your CPU, I think your AMD64 chip ought to be cooler when idling than my P4C anyway if you're using Cool and Quiet thanks to clock throttling. Even under load, I think my P4C will be dissipating slightly more heat, though please do check the specs to make sure. But I think you'd be perfectly fine with the Reserator.

Regards,

E@

msmrodan · Post by **msmrodan** » Mon Aug 09, 2004 10:12 am

just to add my 2 cents, I recieved my Reserator about a month and a half ago and since then it's been performing admirably on my athlon 64 3200+

I can concurr that Zalman have yet another winner under their hands (this is my third Zalman product

, and not just because I like their ugly logo)

35C idle and 45C load is wonderful for a machine with just one case fan and one PSU fan, both at 5V...under the desk it's inaudible in my studio.
I thought about watercooling the PSU/HD but why bother?

Just make sure you mount the waterblock properly - as Mike noted it does slide on its own axis but when mounted correctly does a great job.

Another good thing is that if you move to another machine you can take the Reserator easily with you...no radiators bolted to side of case like my PIII watercool project

DeadBySundown · Post by **DeadBySundown** » Tue Aug 10, 2004 11:08 am

I purchased this unit based upon the review; I'm a big fan of Zalman products, but I've learned to seek the wisdom of SPCR on these matters.

Couldn't be happier. Found it on sale with free shipping at $200US, which is about the level that I can slide stuff by the spousal unit without getting the hairy eyeball come credit card bill time.

Quality was what I've come to expect from Zalman, and installation wasn't bad at all. I did end up drilling the case for the hose connections, just didn't like the look of them in a slot, and it's not all that convenient to have them there. Fortunately I've got a big Antec 1080 case, so there were plenty of good spots to chooose from.

System is a 2.66/533 P4 with a Radeon 9700 Pro, and my previous cooling setup was a Zalman 7000Cu and ZM-80A, respectively. This was a good setup with respect to noise, but the 7000 required frequent cleaning, as we have 3 cats, and the ZM-80 (original model, no fan) was marginal on the 9700; I knew this when I purchased it, but the whiny stock fan just had to go, and it did perform perfectly on all but the hottest days. Replaced both the the respective Reservator water blocks.

Temps with the old setup were typically 40C idle, 65C under brutal load. Temps now are 38C idle, 41C load, not bad at all.

The unit is indeed completely silent; I'm now back here researching power supply options to replace my old Antec 430, as I can hear it now. The quest continues.

Thanks to SPCR for another great review.

Edward Ng · Post by **Edward Ng** » Tue Aug 10, 2004 12:09 pm

Okay, bad news about Maze4 and 6800...

It's no good. The holes do not line up between the two.

I guess I now have a Maze4 GPU block for sale!

The good news is, and I don't know if this is luck, fate or both, but over the weekend, I ordered a new Danger Den NV-68 block from them; it shipped out yesterday by 2-day, so it should arrive tomorrow. I'll be including numbers from it when I do the write-up comparing different water cooling circuit models.

-Ed

charliek · Post by **charliek** » Tue Aug 10, 2004 1:37 pm

Edward Ng wrote:Okay, bad news about Maze4 and 6800...

It's no good. The holes do not line up between the two.

That's too bad, but it'll be interesting to hear how the big boy you've got on order performs!

Is there no way of modding the Maze4, so it'll fit?

Edward Ng · Post by **Edward Ng** » Tue Aug 10, 2004 1:45 pm

Not without drilling, and not just drilling, but threading the holes to match the mounting components...

Definitely not a simple mod; I wouldn't want to risk cutting into the flow chamber of the block, either. The thing is expensive...

...but not as expensive if you buy mine (see Deals forum for more details)...

...sorry, but I just had to.

tomcat · Post by **tomcat** » Thu Aug 12, 2004 8:09 am

>Of course, it would be really simple to compare the innovatek passive >solution to the Zalman...

Actually, i just got one of those innovatek passive radiators, called KOM (Konvekt-o-matic). It's the KOM-Maxi and it cools two 2.4Ghz XEON's to about 45C (28C ambient). Pretty cool and very silent. Is someone interested in details or a comparison?

Markus

HammerSandwich · Post by **HammerSandwich** » Thu Aug 12, 2004 3:15 pm

tomcat wrote:It's the KOM-Maxi and it cools two 2.4Ghz XEON's to about 45C (28C ambient). Pretty cool and very silent. Is someone interested in details or a comparison?

Please!

charliek · Post by **charliek** » Fri Aug 13, 2004 1:09 am

For those who are interested, I've had a reply from Zalman to my request about compatibility of their waterblock kit with the nVidia GeForce 6800:

Thank you for your interest in Zalmanâ€™s products.

Weâ€™re happy that youâ€™re happy with it, too.

But unfortunately we have yet to develop the proper mounting kits for Geforce 6800 series.

Itâ€™s underway and the moment itâ€™s completed, weâ€™ll have it known on our website.

Thank you for your patience in advance and please keep an eye on our new release section on the website.

To me, that would imply that the only issue is with mounting. As ekotan has mentioned earlier, the block will fit with narrower screws, and does an admirable job of cooling the system, with a stock Reserator, P4 and 6800. Hurrah!

In comparison to other GeForce 6800 blocks on the market, this unit would have the advantage of being much cheaper, and galvanically compatible with the rest of the system (the block is anodized aluminium).

I suspect it may be pushing things to overclock, without beefing up the reserator, though.

burcakb · Post by **burcakb** » Sat Aug 14, 2004 12:32 am

I've noticed that benchmark programs don't really stress the CPU or the GPU enough. They stress a "component part" of the processor one at a time so the processor both draws less power and has time to cool another part of itself. That also holds true for Prime95 (remember you can tune it to use more memory, etc. Makes a difference on temps)

I read somewhere that the game Far Cry running at full everything was very good at stressing both the CPU and the GPU to the max. Could we try to get results running Far Cry?

Uberapan · Post by **Uberapan** » Sun Aug 15, 2004 10:26 am

I just noticed that the Reserator has a tiny hole at the top. Does anyone know why this is preferable to a totally closed loop?

zds · Post by **zds** » Sun Aug 15, 2004 10:42 am

I just noticed that the Reserator has a tiny hole at the top. Does anyone know why this is preferable to a totally closed loop?

I assume it's there to balance pressure. I further assume that closed loop works better, if it has same pressure than atmosphere outside.

acaurora · Post by **acaurora** » Sun Aug 15, 2004 11:29 am

It also could help with the heat - the minute amount of air inside the Reserator needs to cool off somehow, right?

Tibors · Post by **Tibors** » Sun Aug 15, 2004 3:48 pm

If you heat water it expands.

Closed loop watercooling systems have no reservoir or just a little reservoir. Thus there is a relative small volume of water. So the increase in volume of the water when it heats is small. This is probably "absorbed" by the system by a little stretching of the tubes.

The Reserator contains a large volume of water. No amount of stretching of the tubes can "absorb" the extra volume if all this water is heated. So this pressure is released by the air escaping through that little hole.

Tsorovan · Post by **Tsorovan** » Sun Aug 15, 2004 5:42 pm

Uberapan wrote:Is the Athlon XP really that much hotter than a P4 in idle mode? Looking at the temps you reported in this article makes me wonder.

I have a reserator connected to a 3,2 ghz P4 Northwood and a Radeon 9800 Pro. When totally idle, the temps are around 28-30 degrees C. It has been on a WinXP SP2 bittorrent for 6 hours now, averaging around 20 % CPU usage, and the CPU temp is 35 C.

I will stress test the entire setup later and see what the max temps are.

Downloading a torrent... 20% CPU usage... funny post.

acaurora · Post by **acaurora** » Sun Aug 15, 2004 5:46 pm

Yes, interesting at the CPU usage. I download BT all the time with next to 0% usage. I use Azureus.

I have an Athlon XP 3200, and at idle, with room temp being around 25-30 C, mine idles at around 40-45C, and I'm on watercooling with a Reserator, which provides slightly lower than air-cooled temperatures.