Which manufacturer is going to be ready with 5850/5870 blocks?

[leo_bsb]

terrible VDDC temps... I don't want to mount it again but I think I'll have no other option
this is after 15 min under OCCT GPU test

[Eddy_EK]

terrible VDDC temps... I don't want to mount it again but I think I'll have no other option
this is after 15 min under OCCT GPU test

Take the block off the card and take picture of the contact.
Also tell us your watercooling system setup.
The overall temps are high which might mean your system is overloaded.

[Waterlogged]

@ Beast200 - Simple, cost. They would have to use a piece of copper that was as thick as the parts with standoffs and then machine all the copper off. It's a waste of copper, time consuming and very expensive. You have two choices, include spacers like EK or add the standoffs on afterwards like Koolance did. Both are effective at keeping the PCB from being warped.

. . .or it could be that Eddy still doesn't believe that his GPU blocks need spacers and is only throwing them in for the "idiots" that don't know how to mount a block.

[beast200]

. . .or it could be that Eddy still doesn't believe that his GPU blocks need spacers and is only throwing them in for the "idiots" that don't know how to mount a block.

Hi

i wonder the same
i have had 3 other GPU EK waterblocks without spacers
last one was the 4870 and i must admit the card could of bent badly
with out care

[Vapor]

ok it just got worse.
i uninstalled then reflashed the card with asus5870 bios and redid everything. This time however i am seeing 90A+ in VDDC current draw....
Anyone have any ideas on whats wrong? I am just going to flash back to the stock bios for now see if itll fix the problem.

EDIT: With stock bios at 900/1300 i am seeing 60-65A... You think my card is a bad one?

I think what you're seeing is 'right' and is how the card is responding to the VRMs being pushed too hard (because of FurMark) and why FurMark is not the right test for these cards.

As I posted before, the new throttling mechanism checks on the VRMs and whether or not they're being pushed too hard. If they report they're being pushed too hard, the card will internally back down PowerPlay settings (this is when you see 40A instead of higher numbers). When the VRMs report no problems is when you see the higher numbers (the correct numbers).

You're getting 60-65A at lower settings because you're at lower settings and that allows the VRMs to not be stressed (which allows FurMark to run fully unthrottled).

Do your GPU temps increase when amperage goes up? If they do, then the amperage reading is correct. I suspect while the GPU can do 1050+MHz with 1.3+V through FurMark, the VRMs cannot handle that much load (even when cooled properly) and that's why you're seeing the invisible hardware throttling.

[Vapor]

Guys, I don't think FurMark/OCCT is a good test for these cards....at low settings they'll almost definitely run unthrottled and have the really high temperatures. But at high settings they'll throttle (VRM's OCP, not GPU instability) and run at lower temps/amperage.

Maybe something like 3DMark with GPUz running in the background would be a good solid point? Or AtiTool? FurMark/OCCT are just too brutal

FWIW, the hardware OCP is a new feature implemented because of OCCT/FurMark

[Bojamijams]

Furmark and OCCT are just stupid in general. Nobody should be using it to prove ANYTHING. No game, transcoding, 3d rendering or anything people ACTUALLY use during the life of their system generates that kind of load.

[Waterlogged]

Furmark and OCCT are just stupid in general. Nobody should be using it to prove ANYTHING. No game, transcoding, 3d rendering or anything people ACTUALLY use during the life of their system generates that kind of load.

+1

[santiagodraco]

doing now - will run them at 5v instead of 12v. In fact, I'm going to set up the card EXACTLY as he is, and let the fun roll! good or bad, I will publish the result.



The reason why I'm so interested in this is that these loops are reasonably similar other than the fans; its the same heat exchanger, he is using DDC, and I am using MCP650, but I am also using quick-disconnect fittings (horribly restrictive); so if anything, my flow is much more restricted than his.

regarding the fans, I do concur; see above.



Nah.. I'm cool. No one's claiming this is a scientific test here, not me anyways. These are users first impressions. I'm seeing something of interest in what I see as a comparable environment and I'm just pointing it out; sorry if it's showing my block in a favorable light for now. Should I throw in a couple more blocks in that loop to make it look worse? J/K

No you are comparing and trying to promote your product based on the results of your tests vs. "the other guys". It has nothing to do with you being "interested" in the difference, at least other than for "wowee" factor. At least be that honest with yourself and us

Hopefully we'll see someone do a comparison test with the same conditions/equipment for each block. I'm very interested to see how such a test will go.

[santiagodraco]

Guys, I don't think FurMark/OCCT is a good test for these cards....at low settings they'll almost definitely run unthrottled and have the really high temperatures. But at high settings they'll throttle (VRM's OCP, not GPU instability) and run at lower temps/amperage.

Maybe something like 3DMark with GPUz running in the background would be a good solid point? Or AtiTool? FurMark/OCCT are just too brutal

FWIW, the hardware OCP is a new feature implemented because of OCCT/FurMark

I'm seeing strange behavior like this with Furmark, really extreme temps compared to say running Crysis at 1080i with 85% GPU utilization where temps are really low. So there must be something that is triggering this effect. I don't have the knowledge to understand what it is but I'd sure like to know if it's the test software (ie Furmark), or if it's mounting, or block or something else.

[Utnorris]

It has already been stated in various threads about the 5800 series that Furmark is not the best test for them, haven't heard about OCCT being bad though. Personally, if it passes the benchmarks without artifacting and I can play games for 3-4 hours then I consider that stable. OCCT, like Prime95, is good to test your CPU for stability, but stability is in the eye of the beholder so to speak. If you are trying to test the temps, run multi passes of 3DMark06 or Vantage of the GPU only tests and that should give you an idea of the temps you will see.

[Vapor]

OCCT GPU test is even more brutal than FurMark, IMO.

[shazza]

No you are comparing and trying to promote your product based on the results of your tests vs. "the other guys". It has nothing to do with you being "interested" in the difference, at least other than for "wowee" factor. At least be that honest with yourself and us

...

Gabe has stated his interest in checking out the card/cooling and the differences other are seeing. Let's leave it at that, and not get into your interpretation of motives, please.

[Utnorris]

. . .or it could be that Eddy still doesn't believe that his GPU blocks need spacers and is only throwing them in for the "idiots" that don't know how to mount a block.

Pretty sure it's the cost issue:


http://www.xtremesystems.org/forums/...3&postcount=46

By the way Eddy, what ever happened to this promise?

"I promise you we will start seriously to find solution to this, to provide you the standoffs for et least next coming water blocks if not for the present ones.
Options are:
- plastic standoffs that are additionally glued to the copper
- metal standoffs that are additionally glued to the copper
- full copper one-piece standoffs like XSPC and some others have it"

That was made last year today.

[Waterlogged]

Pretty sure it's the cost issue:


http://www.xtremesystems.org/forums/...3&postcount=46

By the way Eddy, what ever happened to this promise?

"I promise you we will start seriously to find solution to this, to provide you the standoffs for et least next coming water blocks if not for the present ones.
Options are:
- plastic standoffs that are additionally glued to the copper
- metal standoffs that are additionally glued to the copper
- full copper one-piece standoffs like XSPC and some others have it"

That was made last year today.

Eddy has shown a repeated hidden disdain for standoff's and those requesting them, as if they are some how incapable of properly mounting a GPU block. Some of the ppl that requested standoff's from Eddy are far from "inexperienced" at installing GPU blocks.

From your link

Hi there,

I'll be honest with you. There are few reasons why we don't use standoffs.
1. I have mounted many blocks on cards and found no need for this as I can find optimal pressure for the block quickly.
2. for standoffs we need at lease 1,2mm more copper material (which is about 15% more material and this is $$$)
3. more machining which costs from 25€-35€ per hour.

. . .snip

Eddy

and from a couple other posts,

Hi,
There are standoffs.
I tested the fit yesterday and the standoffs work perfectly
For expert users standoffs are not necessary.
First blocks will be available in 7 days.

I believe some had problems, because of tightening the screws too much, until the bent PCB touched the block.
I believe we have to take this into our account so even the most unexperienced users can mount it easily.
So EK has started implementing standoffs into the products. Starting with 4870 GS and continuing with Acetal standoffs at 295/285/260(55nm) that can be used or not, depending to the user wishes. But I can say the experienced user can mount the block without standoff with better result.

[Utnorris]

I don't doubt that Eddy has a disdain for standoffs, as I already pointed out, he promised at the very least plastic ones glued on, which still has not happened. The question originally asked was why EK did not fittied/machine the standoffs into the block, which would be cost. That was my first point, my second was the promise made by Eddy one year ago today about having the standoffs either machined or glued onto the block. Quite honestly, it's a pain and messy to use TIM to hold plastic spacers in place while you line up the block. My first FC blocks were EK's for my HD3870's and I had no issue mounting them without the spacers/standoffs and therefore am capable of doing so, but I still would rather have them on the block anyway.



Edit: By the way, I am pretty sure we agree that the blocks should use standoffs and that even experienced folks have had issues with blocks not having standoffs.

[Waterlogged]

I don't doubt that Eddy has a disdain for standoffs, as I already pointed out, he promised at the very least plastic ones glued on, which still has not happened. The question originally asked was why EK did not fittied/machine the standoffs into the block, which would be cost. That was my first point, my second was the promise made by Eddy one year ago today about having the standoffs either machined or glued onto the block. Quite honestly, it's a pain and messy to use TIM to hold plastic spacers in place while you line up the block. My first FC blocks were EK's for my HD3870's and I had no issue mounting them without the spacers/standoffs and therefore am capable of doing so, but I still would rather have them on the block anyway.



Edit: By the way, I am pretty sure we agree that the blocks should use standoffs and that even experienced folks have had issues with blocks not having standoffs.

I've asked this before but don't recall ever getting an answer but, is screwing the spacers in place and then using just a couple dabs of super glue on the outside of the spacer to hold it in place feasible?

[Pen3]

I tried playing Aion in windowed mode at max everything till i was getting under 45fps and gpu load to 95%. My vddc current is normal again at ~40-45A at 1060/1300 1.31v and i had the game running 6hrs+ no problem.

[Utnorris]

I've asked this before but don't recall ever getting an answer but, is screwing the spacers in place and then using just a couple dabs of super glue on the outside of the spacer to hold it in place feasible?

Funny, I actually didn't think of doing that. I will try that tonight after work since I have to install mine. Do you think that the super glue would hurt the nickel plating? I doubt it and wouldn't really see it, but stranger things have happened before.

[Waterlogged]

Funny, I actually didn't think of doing that. I will try that tonight after work since I have to install mine. Do you think that the super glue would hurt the nickel plating? I doubt it and wouldn't really see it, but stranger things have happened before.

Haven't a clue but super glue seems to be fairly benign with metal IIRC.

[gabe]

I think what you're seeing is 'right' and is how the card is responding to the VRMs being pushed too hard (because of FurMark) and why FurMark is not the right test for these cards.

As I posted before, the new throttling mechanism checks on the VRMs and whether or not they're being pushed too hard. If they report they're being pushed too hard, the card will internally back down PowerPlay settings (this is when you see 40A instead of higher numbers). When the VRMs report no problems is when you see the higher numbers (the correct numbers).

You're getting 60-65A at lower settings because you're at lower settings and that allows the VRMs to not be stressed (which allows FurMark to run fully unthrottled).

Do your GPU temps increase when amperage goes up? If they do, then the amperage reading is correct. I suspect while the GPU can do 1050+MHz with 1.3+V through FurMark, the VRMs cannot handle that much load (even when cooled properly) and that's why you're seeing the invisible hardware throttling.

Eric, go back to my screenshots, and take a look at my VDDC temps at 1.35v.. I'm averaging 59C under Furmark.

[Utnorris]

Haven't a clue but super glue seems to be fairly benign with metal IIRC.

I'll let you know then.

[Vapor]

Eric, go back to my screenshots, and take a look at my VDDC temps at 1.35v.. I'm averaging 59C under Furmark.

Yeah, your results are throwing a wrench in my throttling theory

Pen3 might just have borked current sensors causing the odd 70+A readings

[gabe]

Yeah, your results are throwing a wrench in my throttling theory

Pen3 might just have borked current sensors causing the odd 70+A readings

right so, can we not reasonably assume that these high VDDC temps are the result of poor mounts? But then the pertinent question would be to ask: why poor mounts? user error, or something else? scrupulous inspection is needed.

[Vapor]

If it were a poor mount, I think that would show in games though?

(though pen3 didn't say what the temps were in Aion, just the amperage)

Not enough data to draw conclusions yet