Modding D-Tek Fuzion GFX for more flow/less restriction?

[Digital Nirvana]

Just wondering if it is possible, I am thinking about trying to do it myself but unfortunately I don't have the funds nor the machinery to try any modifications to this block as I am still in high school. Does anyone have an idea of what could possibly make it less restrictive while maintaining temps?

I have thought about devising a system where instead of the barbs on the front of the card, have the one barb on top and shooting straight into the block while still maintaining the location of the outlet barb. Does anyone have any ideas of what could help?

[yonton228]

Honestly cant tell you what was done or what thread I saw it in (helpful huh? LOL) However I did see something posted here roughly 4-5months ago with someone with a modified D-Tek GFX block...I will search around a bit to make myself a bit more helpful.



-yonton228/timmy

[Martinm210]

I think alot of restriction occurs with the protruding inlet nozzle. Relating it to culvert design, that would be equivalent to a protruding entrance loss which is the most significant kinds of losses.

http://www.fsl.orst.edu/geowater/FX3...oefficient.htm

K=.9 protruded CMP pipe
K=.2 rounded headwall

You could shave that nozzle down flush and radius it to give you roughly 1/4th the restriction there. It probably won't help temps any, but it may help other blocks in your loop.

Also radius the inlet/outlet around the barbs for some efficiency gains.

[Digital Nirvana]

while I thought about shaving down the nozzle, wouldn't that hurt my temps alot? When you think about it, since it is that high up it would help the water going into the nozzle and head straight down onto the die, however when you think about shaving it down, the water would have more of a tendency to just go in at an angle, missing the core completely or hitting the edge of it.

[Martinm210]

while I thought about shaving down the nozzle, wouldn't that hurt my temps alot? When you think about it, since it is that high up it would help the water going into the nozzle and head straight down onto the die, however when you think about shaving it down, the water would have more of a tendency to just go in at an angle, missing the core completely or hitting the edge of it.

Yeah that could happen, it's a risk you'd have to take.

I have suspicions though that because an 8800 GPU die is so HUGE, that the concentrated center heat removal provided by a nozzle is less important than that for a CPU.

The MCW-60 is only a couple of degrees behind the GFX in performance and it has the much larger pin style of the old apogee non gt blocks, so I think that also supports the design need around something that removes heat from a larger area.

I'm really tempted to abandon my CPU block project for a bit to focus on a GPU block, I think there is much more room for improvement here.

[Digital Nirvana]

Yeah that could happen, it's a risk you'd have to take.

I have suspicions though that because an 8800 GPU die is so HUGE, that the concentrated center heat removal provided by a nozzle is less important than that for a CPU.

The MCW-60 is only a couple of degrees behind the GFX in performance and it has the much larger pin style of the old apogee non gt blocks, so I think that also supports the design need around something that removes heat from a larger area.

I'm really tempted to abandon my CPU block project for a bit to focus on a GPU block, I think there is much more room for improvement here.

If you do, I will have your babies.

Well maybe not, but I will definitely buy a block :p

But does anyone know if the 8800s have an IHS, or is the die actually that big?

[Nate P.]

If you do, I will have your babies.

Well maybe not, but I will definitely buy a block :p

But does anyone know if the 8800s have an IHS, or is the die actually that big?

It does have an IHS. I believe the 8800's are based on the 80nm process (except the 8800GT), correct? Or is it the 90nm?

[septim]

isnt what most gpu blocks designed as 1 in 1 out style not really hitting the actual core area, like maze4 gpu block and some others...

[Digital Nirvana]

isnt what most gpu blocks designed as 1 in 1 out style not really hitting the actual core area, like maze4 gpu block and some others...

yes, except the Fuzion is designed so that it shoots straight at the core.

http://www.dtekcustoms.com/ProductIm...fx/GFX_G80.jpg
http://www.dtekcustoms.com/ProductIm.../GFX_inner.jpg

that shows how it is set up.

[jerrydog]

Take the plastic mid plate apart and drill small holes around the volcano. I call this my SALT SHAKER solution. Don't touch the volcano this causes maximum jetting effect on the hot spot of the chip.

[Digital Nirvana]

do you have any pictures or performance differences? I would love to see it if you do

[jerrydog]

The thermal performance was not effected but the flow was as good or better than the Swiftech GPU block...i did this a while ago and foget the numbers, but I will check my notebook when I get into my office tomorrow

I will post picture tomorrow

[MorGo7h]

The thermal performance was not effected but the flow was as good or better than the Swiftech GPU block...i did this a while ago and foget the numbers, but I will check my notebook when I get into my office tomorrow

I will post picture tomorrow

I might buy FuZion gfx too, since it's hard to MCW 60's here
So if you can fix that pressure drop it would be great

[Philly_Boy]

Yeah that could happen, it's a risk you'd have to take.

I have suspicions though that because an 8800 GPU die is so HUGE, that the concentrated center heat removal provided by a nozzle is less important than that for a CPU.

The MCW-60 is only a couple of degrees behind the GFX in performance and it has the much larger pin style of the old apogee non gt blocks, so I think that also supports the design need around something that removes heat from a larger area.

I'm really tempted to abandon my CPU block project for a bit to focus on a GPU block, I think there is much more room for improvement here.

Back at the end of last winter I had the chance to try out the MCW-60 (which uses the regular Apogee base plate) with the Apogee GT base plate (much smaller pins). My temps with the "regular" MCW-60 were identical (35c idle and 38c loaded) to the MCW-60GT. I think if there were big differences, Gabe would've released a MCW-60GT version already.

[jerrydog]

Here is pictures of my modification. This increased the flow 5X over stock performance. NOW it flows better than MCW60!

[Digital Nirvana]

thank you sir

[Digital Nirvana]

so since it has an IHS, would a bow in the block help any? although how would you put a bow in it?

[Martinm210]

so since it has an IHS, would a bow in the block help any? although how would you put a bow in it?

Yeah they do have an IHS, but both the MCW-60 and the GFX come with a step, so bowing is not necessary. I have seen a few nekkid 8800s though which is supposedly worth some improvement.

The GFX has something like a 30mm square step, and the MCW-60 a 24mm square deep step. They really did this to avoid conflicts on other blocks, but its also good to avoid bridging on the IHS perimeter.

[Polizei]

Five times the flow? Hard to believe, but Im sure it did increase it quite a bit. Got any numbers?

I like the looks more of the Fuzion GPU than the MCW60, but the pressure drop is a killer.

[Digital Nirvana]

how exactly did you measure your pressure drop? as long as it wasn't with a meat thermometer....

[carholmes]

Jerrydog, did you get any improvements on temps?

Measuring chipset temps with a meat thermometer. LOL

[Digital Nirvana]

so I went and took apart my block today, much to my disdain (ugh, those O-rings are so small, and the ridges are tiny) and I found out that instead of the block being 1/2" all the way though, turns out the jet in the middle is a 3/8 hole instead, and that is probably why it is so restrictive. so I can see why drilling holes would make it flow better

[Martinm210]

so I went and took apart my block today, much to my disdain (ugh, those O-rings are so small, and the ridges are tiny) and I found out that instead of the block being 1/2" all the way though, turns out the jet in the middle is a 3/8 hole instead, and that is probably why it is so restrictive. so I can see why drilling holes would make it flow better

The CPU block is also 3/8" and flows just fine, I think most of it is the sharp extended entrance. The other thing I thought about was making some sort of doughnut shaped ring on the inlet side just to give the water a more radiused inlet. That's something you could try that would be completely reversible.

[Digital Nirvana]

The CPU block is also 3/8" and flows just fine, I think most of it is the sharp extended entrance. The other thing I thought about was making some sort of doughnut shaped ring on the inlet side just to give the water a more radiused inlet. That's something you could try that would be completely reversible.

while the Fuzion CPU might be 3/8, it is also straight down into the 3/8 hole, whereas the GFX is not. I figure I will try and hook a hose straight into the fitting since I am running 3/8 and 1/2 split system (haven't made any barbs yet)