No Thermal Paste On GPU RAM

[John699]

So I dismantled my gpu cuz the fans needed some lubrication and to my surprise the gpu ram doesn't have any thermal paste so is it normal cuz my heatsink is similar to gtx 960 with dual fans mounted and it's xfx hd6850 OC edition with 1 GB ddr5 ram.

 

[Rogue Leader]

Yes thats normal, even heat spreaders on ram are just held on by thin contact paste, they really don't need thermal grease on them.

 

[junkeymonkey]

pads are used in them applications not paste . paste should be only for the chip

 

[JackNaylorPE]

So I dismantled my gpu cuz the fans needed some lubrication and to my surprise the gpu ram doesn't have any thermal paste so is it normal cuz my heatsink is similar to gtx 960 with dual fans mounted and it's xfx hd6850 OC edition with 1 GB ddr5 ram.

Yes, unfortunately that's "normal" among some manufacturers. In Bit-tech's tear down article where they yore dorn 3 different 970s, they noted substantial differences in componentry and cooling.

Asus http://www.bit-tech.net/hardware/graphics/2014/09/19/nvidia-geforce-gtx-970-review/2

On its custom PCB, ASUS places all eight Samsung memory chips on the front side, meaning that the backplate does not directly cool any of them. In fact, the chips are left without any contact plate or heatsink touching them, relying solely on air from the fans to cool them.

We also find a 6-phase power delivery system for the GPU, a 50 percent upgrade from stock specifications. It also uses ASUS's DIGI+ VRM controller for precise, digital voltages, as well as high quality Super Alloy Power components for buzz-free choke operation, longer capacitor lifespan and MOSFETs with a 30 percent higher voltage threshold than standard. Sadly, the memory has not been granted the same treatment. It is fed by a single phase found at the other side of the PCB, and this one does not use any special components.

The MOSFETs of the DIGI+ power phases are cooled by a small heatsink, but the VRM controller and the MOSFETs for the memory power phase are left, like the memory chips, to fend for themselves without direct cooling.

EVGA http://www.bit-tech.net/hardware/graphics/2014/09/19/nvidia-geforce-gtx-970-review/3

ACX 2.0 is a new cooling solution from EVGA that comprises a triple heat pipe heatsink and a pair of 85mm fans. The copper, U-shaped heat pipes are nickel plated and are designed to make direct contact with the GPU. However, as you can see from the thermal goo outline, only two actually touch the GPU, which will significantly reduce the effectiveness of the third one.
EVGA sticks with the reference power input design, equipping the GTX 970 SC ACX2 with dual 6-pin power connectors.

EVGA sticks with the reference power input design, equipping the GTX 970 SC ACX2 with dual 6-pin power connectors. Examining the PCB reveals a 4+2 phase power design – four phases near the rear I/O for the GPU, and two in the bottom right corner for the memory. This is a slight upgrade from the 4+1 stock specification but unlike MSI and ASUS, EVGA does not use any specially crafted components.

The GPU MOSFETs are directly cooled by the main heatsink, which has a thermal strip on to draw heat up into the fin stack. On the other side of the GPU is a metal contact plate that partially cools two of the four memory chips on this side, leaving the other two exposed. It also cools the MOSFETs of the power phases serving the memory, but no thermal pads are used, so heat transfer is likely to be limited.

MSI http://www.bit-tech.net/hardware/graphics/2014/09/19/nvidia-geforce-gtx-970-review/4

The power delivery is the best of any here; it's a 6+2 design. Further, MSI uses its own improved components for both the GPU and the memory power phases. The components are referred to as Military Class 4 since they meet MIL-STD-810G regulations. Specifically, we find Hi-c CAP and Solid CAP type capacitors and Super Ferrite Chokes, which are designed to provide higher stability, lifespan and efficiency.

A miniature heatsink takes care of cooling the main GPU MOSFETs near the rear I/O, while a metal contact plate equipped with thermal pads is used to cool three of the four front PCB memory chips as well as the remaining power circuitry including the memory MOSFETs and the VR controllers.

Reading the above, it's easy to see why one of those cards consistently outperforms the other two. Some are left just to "fend for themselves" other takes slight to significant steps (pads, heatsinks, etc) to improve cooling.

"Need" is a relative word, Market Leader EQ Waterblocks, advises that you add thermal paste to each side of the thermal pads when installing their water blocks. The 9xx series cards are so power efficient that the GPU really doesn't need an "special cooling". This is why the presence of hybrid coolers on the 9xx line is a complete mystery to me since no non-reference card, (even the EVGA one above where 1/3 of its cooling system misses the GPU), ever gets anywhere near its throttling point. At say 85C throttling point, whether you see 70C or 60C on air or 50C on a hybrid has no effect on card performance. If anything, the card's OC will be limited by VRM temps ... which is why we see the ones that do make special efforts to cool VRM and memory (Gigabyte and MSI for example) dominating the performance charts. Also some hybrid options actually decrease cooling to the VRM compared to the original cooler.

Since you have it apart, and TM is cheap ... I'd make the effort to apply TIM to all thermal pads and any other **contact** points. Ofcourse of a plate or heatsink makes no contact with the chip, TIM won't help you. You can buy thermal pads (Fujipoly) at most water cooling web sites.

 

[John699]

So I dismantled my gpu cuz the fans needed some lubrication and to my surprise the gpu ram doesn't have any thermal paste so is it normal cuz my heatsink is similar to gtx 960 with dual fans mounted and it's xfx hd6850 OC edition with 1 GB ddr5 ram.

Yes, unfortunately that's "normal" among some manufacturers. In Bit-tech's tear down article where they yore dorn 3 different 970s, they noted substantial differences in componentry and cooling.

Asus http://www.bit-tech.net/hardware/graphics/2014/09/19/nvidia-geforce-gtx-970-review/2

On its custom PCB, ASUS places all eight Samsung memory chips on the front side, meaning that the backplate does not directly cool any of them. In fact, the chips are left without any contact plate or heatsink touching them, relying solely on air from the fans to cool them.

We also find a 6-phase power delivery system for the GPU, a 50 percent upgrade from stock specifications. It also uses ASUS's DIGI+ VRM controller for precise, digital voltages, as well as high quality Super Alloy Power components for buzz-free choke operation, longer capacitor lifespan and MOSFETs with a 30 percent higher voltage threshold than standard. Sadly, the memory has not been granted the same treatment. It is fed by a single phase found at the other side of the PCB, and this one does not use any special components.

The MOSFETs of the DIGI+ power phases are cooled by a small heatsink, but the VRM controller and the MOSFETs for the memory power phase are left, like the memory chips, to fend for themselves without direct cooling.

EVGA http://www.bit-tech.net/hardware/graphics/2014/09/19/nvidia-geforce-gtx-970-review/3

ACX 2.0 is a new cooling solution from EVGA that comprises a triple heat pipe heatsink and a pair of 85mm fans. The copper, U-shaped heat pipes are nickel plated and are designed to make direct contact with the GPU. However, as you can see from the thermal goo outline, only two actually touch the GPU, which will significantly reduce the effectiveness of the third one.
EVGA sticks with the reference power input design, equipping the GTX 970 SC ACX2 with dual 6-pin power connectors.

EVGA sticks with the reference power input design, equipping the GTX 970 SC ACX2 with dual 6-pin power connectors. Examining the PCB reveals a 4+2 phase power design – four phases near the rear I/O for the GPU, and two in the bottom right corner for the memory. This is a slight upgrade from the 4+1 stock specification but unlike MSI and ASUS, EVGA does not use any specially crafted components.

The GPU MOSFETs are directly cooled by the main heatsink, which has a thermal strip on to draw heat up into the fin stack. On the other side of the GPU is a metal contact plate that partially cools two of the four memory chips on this side, leaving the other two exposed. It also cools the MOSFETs of the power phases serving the memory, but no thermal pads are used, so heat transfer is likely to be limited.

MSI http://www.bit-tech.net/hardware/graphics/2014/09/19/nvidia-geforce-gtx-970-review/4

The power delivery is the best of any here; it's a 6+2 design. Further, MSI uses its own improved components for both the GPU and the memory power phases. The components are referred to as Military Class 4 since they meet MIL-STD-810G regulations. Specifically, we find Hi-c CAP and Solid CAP type capacitors and Super Ferrite Chokes, which are designed to provide higher stability, lifespan and efficiency.

A miniature heatsink takes care of cooling the main GPU MOSFETs near the rear I/O, while a metal contact plate equipped with thermal pads is used to cool three of the four front PCB memory chips as well as the remaining power circuitry including the memory MOSFETs and the VR controllers.

Reading the above, it's easy to see why one of those cards consistently outperforms the other two. Some are left just to "fend for themselves" other takes slight to significant steps (pads, heatsinks, etc) to improve cooling.

"Need" is a relative word, Market Leader EQ Waterblocks, advises that you add thermal paste to each side of the thermal pads when installing their water blocks. The 9xx series cards are so power efficient that the GPU really doesn't need an "special cooling". This is why the presence of hybrid coolers on the 9xx line is a complete mystery to me since no non-reference card, (even the EVGA one above where 1/3 of its cooling system misses the GPU), ever gets anywhere near its throttling point. At say 85C throttling point, whether you see 70C or 60C on air or 50C on a hybrid has no effect on card performance. If anything, the card's OC will be limited by VRM temps ... which is why we see the ones that do make special efforts to cool VRM and memory (Gigabyte and MSI for example) dominating the performance charts. Also some hybrid options actually decrease cooling to the VRM compared to the original cooler.

Since you have it apart, and TM is cheap ... I'd make the effort to apply TIM to all thermal pads and any other **contact** points. Ofcourse of a plate or heatsink makes no contact with the chip, TIM won't help you. You can buy thermal pads (Fujipoly) at most water cooling web sites.

Hey I'm planning on buying gtx 960 4gb graphics card so according to you I should rather go for Gigabyte and msi right but do these have a metal cover on their back cuz that's a deal breaker for me if no metal cover then bye I'm opting another company and hey how's Zotac?

 

[junkeymonkey]

look at reviews that give a tare down look at things then make your call ?

[for examples]

https://www.techpowerup.com/reviews/MSI/GTX_960_Gaming/images/cooler2.jpg
from
https://www.techpowerup.com/reviews/MSI/GTX_960_Gaming/5.html

https://www.techpowerup.com/reviews/EVGA/GTX_960_SSC_ACX_Cooler/images/cooler2.jpg
from
https://www.techpowerup.com/reviews/EVGA/GTX_960_SSC_ACX_Cooler/5.html

to me on a card like a 960 a backplate is more for looks and as you see evga gives a full card internal cooling plate anwway and should act as like a backplate would for sagging ??

 

[JackNaylorPE]

While corporate practice tends to "go down the line", we can't depend on that assumption as there are many exceptions. That's a 970 comparison ... for a definitive answer on the 960s. Unfortunately very few sites are willing to do complete tear down articles because a) it's a lot of work, and b) it tends to piss off their advertisers when they point out that 1 card is clearly better than the others, 1 guy is happy.. the others ain't.

I can't speak to the 960 as I have 0 personal experience with it, all of the users we've build for have chosen 970's or 980 Ti's. I do have to say that in this price category, the 380 / 380x performs better but with the usual list of smaller disadvantages (PhysX, Shadowplay, driver support, multi-card configs, heat and power). NVidia was addressing this shortfall with the 960 Ti which was supposed to droip last year and then January, but so far why it's not her remains a mystery.

Overall, after using Asus almost exclusively for 10 years.... since the 7xx series GFX cards and from Z87 thru A97, we have been using primarily MSI MoBo / GFX card products. With Z170, Gigabyte has been our first "Go To" board but we have shied away from their GFX cards because of the hi rate of unhappy card owners as shown by the number of 1 egg reviews on newegg.

So as to which to pick, until i have one in my hands that I can take apart ... I generally look at the reviews ... the ones that consistently perform the highest are the ones that generally have the best component and cooling. Of course nothing beats a good tear down article.

 

[junkeymonkey]

I look at whats out there see what I can from reviews and end user reviews like from newegg figure that's the one that looks to fit my needs the best add to cart stick it in and hope it works as expected and did not get the lemon in some way

what work good for me - him- and the guy down the road may not for you and your needs ? that's the chance we all take with any part form any brand any time

you buy you try and you hope it don't make you cry

 

[John699]

While corporate practice tends to "go down the line", we can't depend on that assumption as there are many exceptions. That's a 970 comparison ... for a definitive answer on the 960s. Unfortunately very few sites are willing to do complete tear down articles because a) it's a lot of work, and b) it tends to piss off their advertisers when they point out that 1 card is clearly better than the others, 1 guy is happy.. the others ain't.

I can't speak to the 960 as I have 0 personal experience with it, all of the users we've build for have chosen 970's or 980 Ti's. I do have to say that in this price category, the 380 / 380x performs better but with the usual list of smaller disadvantages (PhysX, Shadowplay, driver support, multi-card configs, heat and power). NVidia was addressing this shortfall with the 960 Ti which was supposed to droip last year and then January, but so far why it's not her remains a mystery.

Overall, after using Asus almost exclusively for 10 years.... since the 7xx series GFX cards and from Z87 thru A97, we have been using primarily MSI MoBo / GFX card products. With Z170, Gigabyte has been our first "Go To" board but we have shied away from their GFX cards because of the hi rate of unhappy card owners as shown by the number of 1 egg reviews on newegg.

So as to which to pick, until i have one in my hands that I can take apart ... I generally look at the reviews ... the ones that consistently perform the highest are the ones that generally have the best component and cooling. Of course nothing beats a good tear down article.

Sorry mate I'm having trouble understanding you here cuz now you seem quite bipolar.....the only reason I'm not opting gtx 970 is cuz nVidia played with us all as it's a 3.5gb card while the 500 mb runs slower then the rest so it's pretty much useless....I have seen at Linux gtx 970 is limited to 3.5gb....so listen suggest any card with 4gb and shouldn't be costly as I can afford gtx 970 range cards but plz don't suggest radeon cards I have this hd6850 and it's pretty hot and also I'm done with amd.

 

[junkeymonkey]

well cant argue with somr of that ''but plz don't suggest radeon cards I have this hd6850 and it's pretty hot and also I'm done with amd''

but this 7850 I replaced my 6850 with was night and day I'd say for what it is this 7850 has been one of my better cards I ever used never has got over 65c looping like heaven where that 6850 wanted to bang 75-80c in the same rig on the same day

cant say anything about the newer refresh of the refresh amd cards - cant see just replacing this 7850 until it just dies

then xfx is not known for there great coolers / fans anyway

one more thing don't them xfx 6850 cards have a life time warrantee [transferable too boot ] ???

you should of got with the xfx rep here at toms or just xfx support on sending it in that now may just of sent you a new 380 card in replacement today ??

then you could passed better judgment on things

 

[JackNaylorPE]

Sorry mate I'm having trouble understanding you here cuz now you seem quite bipolar.....the only reason I'm not opting gtx 970 is cuz nVidia played with us all as it's a 3.5gb card while the 500 mb runs slower then the rest so it's pretty much useless....I have seen at Linux gtx 970 is limited to 3.5gb....so listen suggest any card with 4gb and shouldn't be costly as I can afford gtx 970 range cards but plz don't suggest radeon cards I have this hd6850 and it's pretty hot and also I'm done with amd.

Do a bit more reading and things will be clearer.

Dozens of web sites have soundly debunked the youtubers 3.5 GB claim.... if ya gonna fall for that, despite all the published date to the contrary, I really can't help you. The following are facts:

1. The 380/380x outperform the 960. I wouldn't buy either. Two 960s can't match the performance of the 970 and, like you, the heat, power issues are more than i want to have to deal with as well as the features wise AMD comes up short (Shadowplay, G-Sync, PhysX, multi card & driver support). You are stating you want a 4 GB 960 which scores no better than a 2 GB 960 (65%) instead of a 970 which scored 105% ... 62 % faster.

http://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_960_SLI/23.html

2. You could wait for the 960 Ti
http://videocardz.com/57886/nvidia-preparing-geforce-gtx-960-ti

3. The 970 (3.5 GB as you call it) kicks the 960 4GB's tail.... If you want to believe the 4GB 960 is getting you something over the 2 GB, then don't look at any of these actual test results which show no gain to be had ... To be fair, I have seen results where it has made a difference but only in an extremely poor console port AC: Unity

http://www.guru3d.com/articles_pages/gigabyte_geforce_gtx_960_g1_gaming_4gb_review,12.html
https://www.pugetsystems.com/labs/articles/Video-Card-Performance-2GB-vs-4GB-Memory-154/
http://www.legitreviews.com/gigabyte-geforce-gtx-760-4gb-video-card-review-2gb-4gb_129062/4
https://www.youtube.com/watch?v=o_fBCvFXi0g
http://www.ign.com/boards/threads/gtx-960-4gb-performance-review.454431019/

4. Dozens of websites have refuted any claim that the split 3.5 Gb / 0.5 Gb thing has had any impact on card performance unless you work hard to create a problem by doing some "really freaky stuff" and no probems at 1080 / 1440p.

http://www.guru3d.com/news-story/middle-earth-shadow-of-mordor-geforce-gtx-970-vram-stress-test.html

After some internal testing here over the weekend we could quite honestly not really reproduce stutters or weird issues other than the normal stuff once you run out of graphics memory.,,,

Thing is, the quantifying fact is that nobody really has massive issues, dozens and dozens of media have tested the card with in-depth reviews like the ones here on my site. Replicating the stutters and stuff you see in some of the video's, well to date I have not been able to reproduce them unless you do crazy stuff, and I've been on this all weekend.

At 2560x1440 I tried filling that graphics memory, but most games simply do not use more than 1.5 to 3 GB at that resolution combined with the very best image quality settings. This includes MSAA levels of up-to 8x. At the best settings and WHQD we tried, Alien Isolation, Alan Wake, BioShock Infinite, Hitman, Absolution, Metro Last Light, Thief, Tomb Raider, Asassin’s Creed Black Flag.

Let me clearly state this, the GTX 970 is not an Ultra HD card, it has never been marketed as such and we never recommended even a GTX 980 for Ultra HD gaming either. So if you start looking at that resolution and zoom in, then of course you are bound to run into performance issues, but so does the 4GB GTX 980. These cards are still too weak for such a resolution combined with proper image quality settings. Remember, Ultra HD = 4x 1080P. Let me quote myself from my GTX 970 conclusions “it is a little beast for Full HD and WHQD gaming combined with the best image quality settings”, and within that context I really think it is valid to stick to a maximum of 2560x1440 as 1080P and 1440P are is the real domain for these cards. Face it, if you planned to game at Ultra HD, you would not buy a GeForce GTX 970.

So the two titles that do pass (without any tricks) 3.5 GB are Call of Duty Advanced Warfare and of course that has been most reported to stutter is Middle Earth: Shadow of Mordor. We measured, played and fragged with COD, and there is just NOTHING to detect with the graphics memory fully loaded and in use. But we know that COD simply likes to cache a lot of stuff in VMEM, opposed to using it for rendering. So our focus for this quick test will remain Middle Earth: Shadow of Mordor.

We tested specifically with this title as **reportedly** it runs into problems (stutters) at the point where it can't use more VRAM, with our card that is 3.6 GB. Read that carefully, we can not load up the card with more VRAM even with heavier DSR - and that is the precisely point where people have been complaining about stuttering and other issues.

Overall you will have a hard time pushing any card over 3.5 GB of graphics memory usage with any game unless you do some freaky stuff. The ones that do pass 3.5 GB mostly are poor console ports or situations where you game in Ultra HD [4k] or DSR Ultra HD rendering. In that situation I cannot guarantee that your overall experience will be trouble free, however we have a hard time detecting and replicating the stuttering issues some people have mentioned.

Utilizing graphics memory after 3.5 GB can result into performance issues as the card needs to manage some really weird stuff in memory, it's nearly load-balancing. But fact remains it seems to be handling that well, it’s hard to detect and replicate oddities.

In short... to create the problem you need to use high settings at 4k resolution and do some really 'freaky stuff" to create the problem. As the author said, if you chose a 970 or 980 to play at 4k, you did not make a wise decision, the cards just aren't meant to "go there". And there are no problems at 1080p or 1440p, so again, if you want to go at 65% of the speed of a 970 with your 4Gb 960, by all means get it ... but if you buy a sns one and SLI it, it still won't catch the 970.

5. And yes, when posting for help and asking for information, always be sure to insult the people helping you and providing information ... it's sure to increase the response rate