Bolt-on RAM/VRM sinks for 4850/4870?

[iandh]

any test using watercooling solution?, excelent job btw, love the color especialy

It will be a bit unless someone posts results... I have my loop torn down right now as I am working on a new build.

Out of interest was that using 8.8's, if so did you rename the exe? 8.8's have nerfed furmark, half the FPS, lower VRM temps and less current draw!

I actually used the 8.8's on purpose, because now it is a more representative test of what I would consider a "real world" load.

I repeatedly recommended not using Furmark before the fix because it stressed the card beyond what I would call reasonable for a stress test.

For example, On 8800GT/9800GTX's, Furmark would run about 5-10C hotter than during intense gaming use... that's about the increase I would consider normal for a gpu stress test.

On Furmark before the new driver though, it was more like a 20C+ increase between Furmark and gaming. It was to the point of ridiculous, and nowhere near anything you'd see in normal use, not even 3dmark or Crysis cranked to the max. Now it is back in to that 5-10C increase range I consider it to be a "proper" stress test for this card, and I now use it.


I'm going to be doing some gaming/3d06 benches with a day or two. Honestly though, I did not design these sinks for beyond normal gaming or bench/stress testing use. IMO ATI didn't design the card for anything beyond that either, hence why they put the lid on this abnormally abusive (for this card) stress test.

[Creig]

When will you know if you intend to make them for the 4850? You previously had stated that the 4850 sinks would be released first, with the 4870 coming about a week afterwards. I would think the 4850 being a single-slot cooler would be a more logical candidate for sinks than the 4870 with its dual slot cooler. Sort of like the 8800GT needing cooling more than the 8800GTS. Although I can understand the need for additional 4870 cooling as it runs hot, too. Just not as hot as a 4850.

[Pansuu]

I think the results pretty much speak for themselves.

Sure

Why you dont use Everest and show what readings are? So everybody else can compare...

[Gir92]

When will you know if you intend to make them for the 4850? You previously had stated that the 4850 sinks would be released first, with the 4870 coming about a week afterwards. I would think the 4850 being a single-slot cooler would be a more logical candidate for sinks than the 4870 with its dual slot cooler. Sort of like the 8800GT needing cooling more than the 8800GTS. Although I can understand the need for additional 4870 cooling as it runs hot, too. Just not as hot as a 4850.

I believe Ian stated that there was a lot less demand for 4850 unisinks than 4870 ones.

[Sparky]

Sure

Why you dont use Everest and show what readings are? So everybody else can compare...

Everest only shows GPU VRM temps apparently. I just tried it on my stock card.

[Pansuu]

Everest only shows GPU VRM temps apparently. I just tried it on my stock card.

Yes? Im only interest what is VRM temps, im not interest what is HS temp.

[Sparky]

Yes? Im only interest what is VRM temps, im not interest what is HS temp.

Oh OK, wasn't sure if there was something else too. However, I don't think it should read much different than the IR temp sensor, providing the software is reading it right.

IR temp sensor, DMM, all that is better than software for temp and voltage measurements.

[iandh]

When will you know if you intend to make them for the 4850? You previously had stated that the 4850 sinks would be released first, with the 4870 coming about a week afterwards. I would think the 4850 being a single-slot cooler would be a more logical candidate for sinks than the 4870 with its dual slot cooler. Sort of like the 8800GT needing cooling more than the 8800GTS. Although I can understand the need for additional 4870 cooling as it runs hot, too. Just not as hot as a 4850.

I had originally planned to make the 4850 sinks first, but the problem is that they are a more complicated design, and therefore more expensive to make than the 4870. That wouldn't be too much of a problem except that the 4850 is a only a $150 card, so it is pretty hard to sell $40 sinks. Once we cut the price down into the $30 range the profits for all involved begin to be hard to detect.

Once my production shop got jammed up, I told him just to concentrate on making the sure seller, the 4870, and we'd worry about the 4850 later. The main problem with the 4850 sinks now is that I can't get the shops to commit to purchasing enough to make it safe and worth our while, because they haven't seen the same demand as they've seen for the 4870 sinks.

Nvidia got all of us a little jumpy with the whole releasing cards every four days thing this year. The 9800GTX's sinks weren't what I would call a disaster, but it was such a short lived card that it wasn't quite the business prospect that the 8800GT's were.

That said, it isn't totally cancelled, just postponed until we make a decision.

Sure

Why you dont use Everest and show what readings are? So everybody else can compare...

Oh OK, wasn't sure if there was something else too. However, I don't think it should read much different than the IR temp sensor, providing the software is reading it right.

IR temp sensor, DMM, all that is better than software for temp and voltage measurements.

I actually didn't use everest on purpose, so that we all can compare. Diodes read a little different on every card, but a calibrated infrared thermometer and a thermocouple gauge read the same for everyone.


I may post some screenshots though just for fun.

[Pansuu]

Yes, im sure my diodes is broken or something... Becouse i have 5time bigger HS and your little tiny red HS is so more powerfull, must be that red colour

Btw. if contact is bad and HS is measured 15c, is HS then superpowerfull? (maybe you get this joke, maybe not)
It is good to have some software readings too, or those readings are unuseless too.

I hope someone else puplish his own readings too. I maybe dont belive everything what a seller or reseller say

[cegras]

Just curious; but what's the acceptable temperature for memory IC's? Let's say both vram and normal ram.

Also, pansuu just got served.

[Gir92]

Yes, im sure my diodes is broken or something... Becouse i have 5time bigger HS and your little tiny red HS is so more powerfull, must be that red colour

Btw. if contact is bad and HS is measured 15c, is HS then superpowerfull? (maybe you get this joke, maybe not)
It is good to have some software readings too, or those readings are unuseless too.

I hope someone else puplish his own readings too. I maybe dont belive everything what a seller or reseller say

Troll somewhere else please.

Ian has absolutely no reason to lie. He has stated many times that he barely breaks even with these, so it's not like there's a ton of profit that he's making.

And he has also offered that he would replace for FREE the VRM heatsinks if everyone has issues with it not being large enough. That's not something a self centered guy who's just out for money would say.

[iandh]

Yes, im sure my diodes is broken or something... Becouse i have 5time bigger HS and your little tiny red HS is so more powerfull, must be that red colour

Yeah, and nobody believed that my 8800GT sinks would work either until a customer posted benches showing that they cooled better than enzotech forged copper sinks.

I make a good design.... you make guesses about my design.

Btw. if contact is bad and HS is measured 15c, is HS then superpowerfull? (maybe you get this joke, maybe not)
It is good to have some software readings too, or those readings are unuseless too.

So you are saying that improperly installed a cooling product that I designed myself?!?!

Contact IS NOT bad. Why do you think I took measurements on the back of the board as well? What purpose could they possible serve other than this?

I hope someone else puplish his own readings too. I maybe dont belive everything what a seller or reseller say

DO NOT imply that I am a liar, or that I publish false results. My word is trusted on this forum for a reason.

I stated all test conditions... ambient temps, equipment used, driver used, benchmark used, I don't see how I could have done any better.

Troll somewhere else please.

Ian has absolutely no reason to lie. He has stated many times that he barely breaks even with these, so it's not like there's a ton of profit that he's making.

And he has also offered that he would replace for FREE the VRM heatsinks if everyone has issues with it not being large enough. That's not something a self centered guy who's just out for money would say.

Seriously... I even told Petra's that I would reimburse them the shipping cost out of my own pocket to ship the replacements to customers that already have the sinks. I completely fail to see how I have done anything wrong here.

Not only did I provide a product that looks good and works well, I put what is essentially an unconditional guarantee on its performance.

[cegras]

Yes, im sure my diodes is broken or something... Becouse i have 5time bigger HS and your little tiny red HS is so more powerfull, must be that red colour

Btw. if contact is bad and HS is measured 15c, is HS then superpowerfull? (maybe you get this joke, maybe not)
It is good to have some software readings too, or those readings are unuseless too.

I hope someone else puplish his own readings too. I maybe dont belive everything what a seller or reseller say

"Just enough" is ~52?

You realize those things are spec'd to run into the 100 degree area.

[iandh]

I'll see what I can do for some software readings tonight, just to put this whole thing to bed. Either way I still don't trust any on-board thermal diode as much as I trust my personal test equipment.

[Pen3]

I say just forget about it. If he doesn't want to buy it then its his loss.

[iandh]

As I said in the beginning, I've never seen temperatures reported past 90C. My ambient is even slightly higher today, and they still did well. These temps are somewhat tame compared to what I've seen some people report with stock cooling. I may remount the stock cooler and do a run to see what turns up. As I said so far in my experience these sinks are sufficient to cool the card and even leave a little room for OCing. If you want to be doing massive OC's, it is your responsibility to make sure you supply adequate fresh airflow for all of the components on the card. It's a hot card, there is no way of getting around that besides giant sinks or airflow, and my design aesthetic has always drifted towards proper ventilation rather than brute force.

[Pansuu]

"Just enough" is ~52?

You realize those things are spec'd to run into the 100 degree area.

Yes, i can see you are from USA.. That shot is idle, and idle hardly counts nothing, if you understand. It was only a example in soft, Everest.


Still waiting Everest readings.

Why make small sinks? Why not bigger, then it will perform also many users needs. Why make two times if you can make good and proper in one time, that is my point.

But ok, that nice looking tiny sink is just enought in Arctic cooling+120mm fan. But i dont belive it is enough for example Zalman 1000 coolers and watercoolers i dont even speak. And put two cards in crossfire, your readings go higher, or overclock.

[Sparky]

Pansuu, your comment "I can see you are from the USA" is that supposed to be an insult or something?

LISTEN:
Making things BIGGER costs MORE. Why cost more when you don't need to?

I shall be posting my results when I get mine next week, and my 4870 shall be under water. But you are starting to tick me off with your attitude. If you don't like the sinks, even though you have NO firsthand experience with them, then fine. But don't just go throwing crap around please.

[MrMojoZ]

But ok, that nice looking tiny sink is just enought in Arctic cooling+120mm fan. But i dont belive it is enough for example Zalman 1000 coolers and watercoolers i dont even speak. And put two cards in crossfire, your readings go higher, or overclock.

Prove it.

[iandh]

Yes, i can see you are from USA.. That shot is idle, and idle hardly counts nothing, if you understand. It was only a example in soft, Everest.


Still waiting Everest readings.

Why make small sinks? Why not bigger, then it will perform also many users needs. Why make two times if you can make good and proper in one time, that is my point.

But ok, that nice looking tiny sink is just enought in Arctic cooling+120mm fan. But i dont belive it is enough for example Zalman 1000 coolers and watercoolers i dont even speak. And put two cards in crossfire, your readings go higher, or overclock.

I think you are missing something here... we don't just chop off pieces of extrusion here, we are machining these sinks out of solid bar stock. Each fin is cut individually by a tiny cutter. They are flatter, more stable, and higher in quality than any extrusion based sinks. They are two totally different types of aluminum.

If they are twice as large,

they use twice as much machine time,

which means they cost twice as much to make,

and they cost almost twice as much to sell. Very simple.

Trying to provide quality, looks, performance, and price simultaneously is a careful balancing act.

Materials cost is insignifigant to machine time. Machined extrusions are the smartest way to go in a business sense, but that isn't why I make these.

I posted GPU-Z screenshots that clearly show the benchmark and the all GPU temperatures stabilized on the graphs, that is enough IMO.

Under water cooling, you still need air blowing over the card, such as front intake, or a side panel fan. There is no way around this, not even large sinks. It still just plain needs airflow.

My sinks from the beginning have never been designed for agressive OC'ing like v-mods, etc... even the 8800GT and 9800GTX sinks weren't either. They are designed for mild to moderate OC'ing along with providing sleek looks, and ease of use. I don't like large sinks mounted on anything, in fact the Acellero S1 is only temporary until I have the components for the watercooling loop fully prepared.


lunchtime edit:

I edited all of the argumentative crap out of my post.

Continuance of this debate is pretty much pointless now IMO. I've requested as many numbers as everyone would like to bring to the thread, we will discuss those numbers thoroughly as a group. If at that point we decide that improvement needs to be made just for general use, I will take care of it. If we decide that these are only unsuitable for higher overclocks and v-mod use, and are perfectly fine for general use, then perhaps I will make some behemoth sink that can be purchased as an option for $10 extra a set or something.

As a note, although the VRM sink is slimmer, its fins are also 40% higher than any of the other sinks I've made, so it is a bit deceiving in photos. These things by no means have massive cooling capacity, but that really isn't what they were meant for.

[evilrad]

Got my set in the mail today (a day earlier then expected)! The sinks look really nice (good job iandh)! Getting the ram sinks to be perfectly flat on the ram was a little tricky and took some time (screwing in one screw more, while unscrewing another). For anyone saying that the VRM sink is too small: it is actually much larger than what the picture would suggest. I'm using these sinks along with a D-Tek Fuzion GFX 2. Hopefully I'll be able to post some results later tonight after I finish leak testing and installing Windows on my new build. If not tonight, I'll definitely have some data by tomorrow.

[evilrad]

Well, here are my results on my Asus brand. Screen shot was taken after about 20mins of running ATI Tool. Not bad at all. I have seen people with the VRMs getting 120+ with stock cooling, so I'm happy. When I first started testing, two of the sensors were getting up to 112 while the other one was at 88. I knew something wasn't right. You have to be careful as to not tighten down the sinks too much. Too tight, and you actually bend the sinks off the VRMs in the middle. After a little tweaking of the screws, I was able to get the temps close together. So good work iandh I'll work on overclocking after I overclock my CPU (which good take a few days). I'll post my results when I finally get around to overclocking the card.

[iandh]

Well, here are my results on my Asus brand. Screen shot was taken after about 20mins of running ATI Tool. Not bad at all. I have seen people with the VRMs getting 120+ with stock cooling, so I'm happy. When I first started testing, two of the sensors were getting up to 112 while the other one was at 88. I knew something wasn't right. You have to be careful as to not tighten down the sinks too much. Too tight, and you actually bend the sinks off the VRMs in the middle. After a little tweaking of the screws, I was able to get the temps close together. So good work iandh I'll work on overclocking after I overclock my CPU (which good take a few days). I'll post my results when I finally get around to overclocking the card.

Thanks for posting those results. I'm hoping that we can get some more and get an average for what these run.

Just FYI ceramique is my TIM of choice for mounting. It isn't too thick so you can get good contact with little pressure.

You have to be careful when mounting these, there really isn't too much of a way around it unless I supplied a backplate... which would probably need to be laser cut with a ventilated hole pattern or risk hurting more than helping with temps.

What do you have as far as airflow over the card goes?

[Eternalightwith]

So I couldn't use say IC diamond on the video card

[evilrad]

Thanks for posting those results. I'm hoping that we can get some more and get an average for what these run.

Just FYI ceramique is my TIM of choice for mounting. It isn't too thick so you can get good contact with little pressure.

You have to be careful when mounting these, there really isn't too much of a way around it unless I supplied a backplate... which would probably need to be laser cut with a ventilated hole pattern or risk hurting more than helping with temps.

What do you have as far as airflow over the card goes?

Yeah, I'm using ceramique as well. As for air flow, I'm using a MM U2-UFO case with a Thermochill PA120.3 with 3 120mm Yate Loon D12SM-12 pushing air through the rad & 2 pulling air on the middle and bottom of the rad. The fan in the middle of the rad allows air to flow over the top of the card, while the bottom fans allows air to flow over the bottom (where the sinks are).

BTW, is there a program to monitor the temp of the memory on the card? I want to make sure I have good contact all around the ram. When I was putting the sink on the ram, I had a good look to see if light was getting through between the sinks and ram. So I believe have a made good contact, but I just want to make sure.