Should GTX Titan's have come with a back plate from the factory?

[lavaheadache]

SO...

I started a petition over on the EVGA Titan forums asking that Titan owners be provided with the functional and protective heat sink/back plate that is being produced. I feel that the cards were incomplete as shipped and owners should be provided or reimbursed for the purchase of the heat sink/back plate

I ask that any of you fellow owners and enthusiasts alike that agree please sign the petition.


http://www.evga.com/forums/tm.aspx?m=1891078

 

[3DVagabond]

I don't own one, but for a grand I'd hope to get one installed. It's one of those things you like to get when you are buying a luxury item.

 

[lavaheadache]

I don't own one, but for a grand I'd hope to get one installed. It's one of those things you like to get when you are buying a luxury item.

you can join to sign my petition =)

 

[VulgarDisplay]

Don't take this the wrong way, but I would never have been silly enough to buy a $1000 GPU that's 25% faster than a $380 one even if it had a gold backplate on it.

 

[Grooveriding]

I have EVGA models and will post over there. This is something that should of been done by nvidia as a reference design addition though. A backplate is probably $2 to produce ffs.

The EVGA backplates are pretty snazzy. My Titans have EK backplates on them, $30 a piece...

 

[tviceman]

I think back plates are entirely overvalued. I couldn't care less if the card I buy has one or not.

 

[lavaheadache]

I think back plates are entirely overvalued. I couldn't care less if the card I buy has one or not.

On the Titan, like the GTX 280 it provides cooling to memory chips on the back of the card. Would you want the memory on the front of your card to be left bare?

 

[Grooveriding]

Backplates are pretty helpful on this card. There is 3GB of memory on the back of the card that it can help to protect and cool. It also helps prevent PCB flex and on a $1000 'boutique' card....

 

[lavaheadache]

Don't take this the wrong way, but I would never have been silly enough to buy a $1000 GPU that's 25% faster than a $380 one even if it had a gold backplate on it.

What if money were no object to you? Would you "settle" for that $380 card or would you want the best that was out?

I'm not saying that applies to me. I'm an average Joe with a wife, kid, mortgage and a couple of car payments. We do well and don't live paycheck to paycheck but that is about it. My computer hobby pays for it's self most of the time(at least that's what I tell my wife) and I like to have the best of the best.

 

[ViRGE]

No. Backplates look nice, but they're wholly unnecessary and actually cause problems. You cannot put cards in SLI/CF directly next to each other if they have backplates; the extra space occupied by the backplate makes all the difference in whether the blocked card has enough room to pull in fresh air or not.

AMD learned that one the hard way with the 6900 series, which is why the 7900 series doesn't have a backplate.

If Titan had a backplate, you'd never be able to install it triple-SLI in a standard ATX case, as it would effectively become a card that requires 3 slots instead of 2.

A backplate wouldn't provide much of a benefit anyhow. It has already been proven in a practical manner that backplates are not necessary for cooling GDDR5, so all they really do is provide shielding against damage for the back side of the card. That's hardly a good enough reason to give up tri-SLI.

 

[lavaheadache]

I guess I stand corrected by the voice of the people at EVGA forums and here.... Who'd have thunk?

 

[Eureka]

Personally, I think it's nice to have one included, whether or not it's functional. The user can always take it off, but you can't put one on without having one included in the box.

 

[lavaheadache]

No. Backplates look nice, but they're wholly unnecessary and actually cause problems. You cannot put cards in SLI/CF directly next to each other if they have backplates; the extra space occupied by the backplate makes all the difference in whether the blocked card has enough room to pull in fresh air or not.

AMD learned that one the hard way with the 6900 series, which is why the 7900 series doesn't have a backplate.

If Titan had a backplate, you'd never be able to install it triple-SLI in a standard ATX case, as it would effectively become a card that requires 3 slots instead of 2.

A backplate wouldn't provide much of a benefit anyhow. It has already been proven in a practical manner that backplates are not necessary for cooling GDDR5, so all they really do is provide shielding against damage for the back side of the card. That's hardly a good enough reason to give up tri-SLI.

I respect your stance but using the 6900 series as an example of the back plate causing issue is actually incorrect in my eyes. The problem was not the back plate but the fan shroud instead. They were foolish to shape the card like a brick. Airflow would not have been an issue if the front of the card was contoured to allow for air to get between them. Nvidia has been boing that for years. My example of the GTX 280 is a testament to this. The fan shroud was designed to create a gap between the cards

 

[ViRGE]

Both NVIDIA and AMD have stopped using backplates though. In AMD's case they specifically referenced the backplate being an issue and is why they removed it from the 7900 series. Though I completely agree making both sides of the card as flat as a pancake didn't help. NVIDIA said something similar with either the 480 or the 580; but I can't immediately find the article referencing it.

I suspect they've found that the necessary "wedge" design to do back-to-back video cards with a backplate was less effective than removing the backplate entirely.

 

[lavaheadache]

The 480 and 580 didnt have memory chips on the back....

 

[tviceman]

On the Titan, like the GTX 280 it provides cooling to memory chips on the back of the card. Would you want the memory on the front of your card to be left bare?

Is there empirical data that shows a back plate provides cooling?

 

[lavaheadache]

How can a metal plate that rests on top of the memory chip with TIM between the 2 not provide cooling? If I rest a random chunk of metal on a hot surface it will provide cooling

 

[ViRGE]

Is there empirical data that shows a back plate provides cooling?

Memory chips don't have accessible diodes, so not really. But if you look at modern video card designs, it's common for GDDR5 chips (even 6GHz chips) to be free and naked. The only time you don't see this is when the heatsink for the GPU is so large that it would completely cover the memory chips and cut off airflow, at which point it becomes best to attach them to the heatsink.

 

[3DVagabond]

Memory chips don't have accessible diodes, so not really. But if you look at modern video card designs, it's common for GDDR5 chips (even 6GHz chips) to be free and naked. The only time you don't see this is when the heatsink for the GPU is so large that it would completely cover the memory chips and cut off airflow, at which point it becomes best to attach them to the heatsink.

I think you're looking at that backwards. The reason the base plate on the heatsink is large enough to cover the ram chips is to provide cooling for them. It's not that the plate was too big so they had to use it to cool the ram chips.

You are correct that some cards have bare ram chips (and vrm too). It's not a desirable thing though, it's just cheaper than putting heat sinks on them.

 

[Cookie Monster]

Agreed with Virge on this one. Also you have the problem of the backplates sometimes shorting to the PCB. I guess the only cooling it may provide is by dispersing the heat from the backside of where the GPU is located. But even then it won't do much since the method of heat transfer is very very inefficient e.g thick thermal pad.

However.. backplates do look mint and thats why most people want it.

GDDR5 chips, actually GDDR3 even dont really require passive cooling. Most ICs don't require any sort of cooling because its cooled via the PCB/copper plane unless its a power component..

Just looking at Elpdia's 2gbit GDDR5 datasheet states an operating temperature range of 0C to +95C (case temperature so the actual IC would be even moreso hotter).

 

[ViRGE]

I think you're looking at that backwards. The reason the base plate on the heatsink is large enough to cover the ram chips is to provide cooling for them. It's not that the plate was too big so they had to use it to cool the ram chips.

You are correct that some cards have bare ram chips (and vrm too). It's not a desirable thing though, it's just cheaper than putting heat sinks on them.

I can see why you'd think that's backwards, but I'll tell you why I see it the other way around.

For open-air coolers it's common to see one large aluminum heatsink running the length of the card (or more). Something like this Gigabyte 670: http://www.techpowerup.com/reviews/Gigabyte/GeForce_GTX_670_Windforce/4.html

Now with the heatsink sitting where it is, it's between the various components and the fans. The GPU makes contact through heatpipes running into the heatsink, the rest of the components aren't so lucky. The heatsink would be deflecting most of the air the fans move, so the various components would be left with little room to breathe: they aren't getting any kind of real airflow, and they wouldn't have much space to radiate heat (especially opposite that heatsink).

So what does Gigabyte do? They add a metal plate and some thick thermal pads to make contact with the RAM.

The RAM itself doesn't need cooling to operate. It sits as naked as a jaybird on the back of cards like the GTX 660 Ti; and this is the same 6GHz stuff that would be on the Gigabyte card. And if you look at a card with a smaller cooler like a GTX 650 Ti, even the RAM on the front is naked.

Ergo the only reason Gigabyte had to actually heatsink the RAM on their card was because of the large heatsink for the GPU that would otherwise be constricting the natural convection cooling of the RAM. If it was a smaller heatsink, they could have run it bare.

 

[3DVagabond]

I can see why you'd think that's backwards, but I'll tell you why I see it the other way around.

For open-air coolers it's common to see one large aluminum heatsink running the length of the card (or more). Something like this Gigabyte 670: http://www.techpowerup.com/reviews/Gigabyte/GeForce_GTX_670_Windforce/4.html

Now with the heatsink sitting where it is, it's between the various components and the fans. The GPU makes contact through heatpipes running into the heatsink, the rest of the components aren't so lucky. The heatsink would be deflecting most of the air the fans move, so the various components would be left with little room to breathe: they aren't getting any kind of real airflow, and they wouldn't have much space to radiate heat (especially opposite that heatsink).

So what does Gigabyte do? They add a metal plate and some thick thermal pads to make contact with the RAM.

The RAM itself doesn't need cooling to operate. It sits as naked as a jaybird on the back of cards like the GTX 660 Ti; and this is the same 6GHz stuff that would be on the Gigabyte card. And if you look at a card with a smaller cooler like a GTX 650 Ti, even the RAM on the front is naked.

Ergo the only reason Gigabyte had to actually heatsink the RAM on their card was because of the large heatsink for the GPU that would otherwise be constricting the natural convection cooling of the RAM. If it was a smaller heatsink, they could have run it bare.

I agree that there are cards that the ram is left naked. Do you not also agree though that there are cards that the ram is purposely cooled? I'm just saying that the latter is generally preferred by most enthusiasts. As I said, there are also VRM that is left naked. Again, I would prefer VRM with a heatsink. I only mention VRM because they are a more obvious hotter running component, but ensuring better cooling for any solid state component would have to be seen as a positive.

 

[boxleitnerb]

The memory on the backside of the PCB doesn't get very warm. Around 60°C if my IR-thermometer is correct. Hardly critical or in need for a backplate.

 

[TheUnk]

If Titan had a backplate, you'd never be able to install it triple-SLI in a standard ATX case, as it would effectively become a card that requires 3 slots instead of 2.

A backplate wouldn't provide much of a benefit anyhow. It has already been proven in a practical manner that backplates are not necessary for cooling GDDR5, so all they really do is provide shielding against damage for the back side of the card. That's hardly a good enough reason to give up tri-SLI.

Not sure how accurate that is. I had tri-sli 670s with backplates without a problem, seem to be as thick as the Titan just not as long.

 

[bleucharm28]

I expect to have backplate by default, for performance or asthetics. I expect manufacture to provide back plate for any high end like GTX 680's and above.