Temperature/Block Comparison For 8800GTX

[Snyxxx]

I have done a little testing to convince myself what the best cooling solution for this beast of a video card. Used Everest Ultimate for all temperature
mesurements and the AC is already on for the year so ambient is pretty constant 72°F (~20°C).

Multi-core Prime 95 25.2 (Small FFTs) used for heating CPU and ATI Tool used for heating GPU. I found OCCT does not produce as much heat as above. I lapped the Xeon 3070 for peace of mind that the concavity is gone. All CPU temps are at the core (max) per Everest.

Watercooling CPU Only With Stock Air Cooled 8800GTX:

http://www.xtremesystems.org/forums/...1&d=1174450293

CPU Idle: 24°C
CPU Load: 34°C

8800GTX Idle: 63°C (Diode)
8800GTX Load: 84°C (Diode)

Observations - 8800GTX Temps are a little disturbing. Happy with CPU temps.


Watercooling CPU And 8800GTX Using MCW60 With Heatsinks:

http://www.xtremesystems.org/forums/...1&d=1174450656
http://www.xtremesystems.org/forums/...1&d=1174450693

CPU Idle: 27°C
CPU Load: 39°C

8800GTX Idle: 36°C (Diode)
8800GTX Load: 44°C (Diode)

Observations - I hated dealing with all those heat sinks. The heatsink on the lone naked chip that looks like a NB or SB got so hot I could not put my finger on it more than half second. if I were to keep this option, a bigger heatsink would be required or a fan. The GPU (ambient) got to 51°C. Maybe it was from this little chip. All the ram heatsinks were cool to the touch. The VRM heatsinks were warm, but not too bad.


Watercooling CPU And 8800GTX Using EK Full Cover Block:

http://www.xtremesystems.org/forums/...1&d=1174451064

http://www.xtremesystems.org/forums/...1&d=1174451099

CPU Idle: 30°C
CPU Load: 41°C

8800GTX Idle: 39°C (Diode)
8800GTX Load: 48°C (Diode)

Observations - Results show a little more heat than the GPU only solution above. I guess some of the VRM, Memory and little chip's heat is getting into the water. 3-4°C difference is not that bad. Additional flow resistance may not help either. The GPU (ambient) temperature so hot above was only 40°C for this case. I felt all over the block and felt no hot spots. I like the tube routing with the EK block better as well.


Conclusion:

I knew I was going to use the EK full cover block in the end, but always wanted to know what the difference would be. These types of things bug me for not knowing.

I already had a MCW60 laying around and just had to buy the G80 bracket. I kept everything the same as much as possible. I believe that just about any water cooling solution used for the 8800GTX would be vastly better than the stock air cooler. Temps were almost cut in half.

By the way the 8800GTX at full load added about 7-8°C the the CPU core temperature.

Hope this helps someone.

[serialk11r]

I never thought a fullcover block would hurt flow that much...

[AuroraProject]

How much did the EK increase the cpu temps at idle?

[Omastar]

Glad I got that MCW60 now. Presents me with an upgrade path and better flow. I don't really care about having to deal with 12 heatsinks for the RAM/PWM/NVIO. And for me, the NVIO chip on my GTS never got hot with an MC14, even under ATI_Tool.

[Eddy_EK]

I am sure all would be laughing if you added a DangerDen full cover block

[Snyxxx]

From what I have read, the DD full cover block would be worse. Don't get me wrong, I love the EK block and planned on using it from the beginning. That is why I bought two. I personally did not like the MCW60 option since it only cooled the core, made my tubing runs more difficult and the other hot points got real hot.

Also, I am not sure if the temps increased from flow resistance or the summation of individual heat sources being added to the water. By using a Storm, I consider flow restriction a minor point.

To me, it does not matter. I consider the EK block a work of art and prefer using it.

[IanY]

To me, it does not matter. I consider the EK block a work of art and prefer using it.

Are you building art or are you building a watercooled computer? It seems like you are basing your decisions largely on appearances.

[lasse.j5]

Good job mate.

What does that NVIO chip actually do? Has is anything to do with Overclocking results?

[Snyxxx]

Ian Y,

I like to have both appearances and functionality.

If I can have appearances for a few degrees, so be it. Its just a hobby, but I am a little particular in the looks of things.

lasse.j5,

Thanks. I do not know what that chip does, but it is hot, hot, hot. I have not tried any overclocking yet. Trying to establish a base line currently.

[masaville]

Thanks for doing the test. Very useful info!!

[masaville]

I am sure all would be laughing if you added a DangerDen full cover block

I suspect my DD blocks are worse. My temps are slightly higher, but it's negligible for the most part. We're talking a few degrees at most. EK blocks seem to be machined better though from the pics I've seen.

[IanY]

Thanks for doing the test. I just put an order in for Swiftech's MCW60's at Petra's myself. I'd rather have the raw performance even though the EK's are much more attractive. The only thing I'll miss is the ease of setting up an SLI setup without having to bend the tubes full round.

Does anyone know if the 8900 GTX's will require a different mounting addon than he one used for the 8800's for the MCW60's? I'm planning on using the step up program and don't want these waterblocks to go to waste.

OOOooo... $10 million question!!! I'm personally waiting to see the 8900 cards before buying more video blocks hehehe

Yes.. I want 8900 cards.. My 7900s were so much better than my 7800s.. so I expect the 8900s to be cooler running and have a higher overall product quality.

[IanY]

I suspect my DD blocks are worse. My temps are slightly higher, but it's negligible for the most part. We're talking a few degrees at most. EK blocks seem to be machined better though from the pics I've seen.

I love the smack talking.

[ranker]

OOOooo... $10 million question!!! I'm personally waiting to see the 8900 cards before buying more video blocks hehehe

Yes.. I want 8900 cards.. My 7900s were so much better than my 7800s.. so I expect the 8900s to be cooler running and have a higher overall product quality.

I agree. I only bought the 8800GTX's so that I could either RMA them for the 8900 or use the step up program. My 7900's were great at a much lesser cost than my 7800 SLI's. I'm hoping it stays the same pattern with more performance/features at a 10-20% less cost.

My only worry is that the 8900's won't be compatible with the MCW60's. I just ebayed off my EK covers and recouped 80% of my investment so I'm feeling good for now.

Let's hope the 8800 mounting kit is all we need. I'd hate to get nickled and dimed more just to run the 8900's.

[serialk11r]

Hmmmm... I like the non-fullcover look more. Anyways, there is no way that small chip you are talking about plus the ram and everything will increase water temps by that much. It has to be the restriction.

[[cTx] Nooc]

jesus I'm using the EK block on my 8800 because only about 100% of you suggested it over mcw60. now it's 50/50...oh well, I don't regret it! beautiful block, and comes with chrome barbs (deal breaker right there).

[serialk11r]

jesus I'm using the EK block on my 8800 because only about 100% of you suggested it over mcw60. now it's 50/50...oh well, I don't regret it! beautiful block, and comes with chrome barbs (deal breaker right there).

Lol chrome barbs... If Swiftech tapped all their stuff to 1/4" BSPP, that would be cool. It was the reason I got my Coolrad 22T not the swiftech thingy, because it had metal barbs Performance (heat dissipated/flow) is something like either a few percent better or worse. Of course, when you can't afford to spend an extra 20 bucks on metal barbs (i.e. me) then plastic is the way to go.

[masaville]

OOOooo... $10 million question!!! I'm personally waiting to see the 8900 cards before buying more video blocks hehehe

Yes.. I want 8900 cards.. My 7900s were so much better than my 7800s.. so I expect the 8900s to be cooler running and have a higher overall product quality.

Ian, I'm sure the 8900s will actually decrease performance. Nobody told you???? haha j/k

[masaville]

Hmmmm... I like the non-fullcover look more. Anyways, there is no way that small chip you are talking about plus the ram and everything will increase water temps by that much. It has to be the restriction.

It's the sister of the SB MCP on a 680i board. =P Super hot too!!

[serialk11r]

It's the sister of the SB MCP on a 680i board. =P Super hot too!!

Hot is one thing, heat is another. The chip gets hot because the cooling is not great. You can have a great deal of heat and cool the heat source down, and you can have something that doesn't put out much heat with bad cooling, and get it very very hot. I'm sure those mobile processors that draw very very little power would get hot with only an MC14 ramsink on it lol.

[pplapeu]

I might be the only Innovatek full card guy you hear from. Mine has been on for a week. This on a XFX-XXX GTS 580 clocked card.

OK.....before I shoot out a load number, realize i have a huge external cooler. I run a 2g res, two cores and two inline rads with a 5gpm pump and run only distilled water.

I live in a cold latitude and lately the room temp is 70F.

41C seems to be the load temp I see after an hour or so of gaming and that number is not varing much yet.....so see you in summer!

I bet you hate advice. BUT HEY! .... here you go. Get some more water. Inline a real gallon of extra juice outside the case [forget LAN parties unless you use quick connect] and inline a heater core or two. I predict the block does better.

PS.....might be getting EK if they become available here and the price stays were they are. My Innovatek cost beaucoups.

[Snyxxx]

Nooc,

Look at my picture of the MCW60. I just used DD High Flow barbs. Also, on the reservoir. The NPSM threads work on G1/4 barbs since the threaded part is very short.

[serialk11r]

Nooc,

Look at my picture of the MCW60. I just used DD High Flow barbs. Also, on the reservoir. The NPSM threads work on G1/4 barbs since the threaded part is very short.

Yea, people use the DD barbs on Swiftech stuff, but I wouldn't because they weren't meant to.

[pplapeu]

FORGOT TO SIGGIE IN MY RIG!!!!
Processor
Model : Intel(R) Pentium(R) 4 CPU 3.20GHz
Speed : 3.21GHz-------Mod # SL7AA
Performance Rating : PR4271 (estimated)
L2 On-board Cache 512kB L3 On-board Cache 2MB
Asus P4GPL-X -- Total Memory : Kingston KHX3200AK2/2G
Floppy disk drive
Hard Disk : WDC WD740GD-00FLC0 [69.2GB]
PIONEER DVD-117R
SONY CD-RW CRX175E
Microsoft Windows XP/2002 Home (Win32 x86) 5.01.2600 (Service Pack 2)
Marvell Yukon 88E8053 PCI-E Gigabit Ethernet Controller
D-Link 4100 ["GameFuel" enabled}
Monitor/Panel : Plug and Play Monitor Dell 24"
XFX-XXX PV-T80G-GHD9 Cool-Matic nVidia 8800
Innovatek Mod# 501254
Aspire ATX-AS680W-BK
Logitech G15 -- Razer Diamondback -- JBL Creature

[Fairydust]

Hmmmm... I like the non-fullcover look more. Anyways, there is no way that small chip you are talking about plus the ram and everything will increase water temps by that much. It has to be the restriction.

Without actually measuring water temp that is a bold statement. Whats the efficiency of the voltage regulation? Power consumption of ram? What fans at what voltage were used?

Pa 120.2 with Nexus at 7 V would net a 2,5 Degree water temp increase for ~35 Watts additional heat dump, 1.