AMD Blocks/Disables Overclocking the Memory for the Radeon R9 Fury X

[kipper]

Well well... Nvidia owner OP made this thread, when in reality AMD have done nothing wrong, it's a new memory subsystem, Nvidia on the other hand have zero room to talk on dated technology.

[Notional]

Again evading the question asked... I am only going to repeat once more:

 

If a matter is not in context, they why bring it up at all?

 

Until you can actually answer the above question without going off the tangent to outer space, then you are outright wasting everybody's time.

 

It is literally the first 3 words of my initial question: "out of curiosity". Because I don't understand his hatred towards Huddy. If I made similar headlines and pictures about Tom Petersen, would you not be curios as to why? I find it very odd.

 

Not understanding 3 words, is indeed wasting everyone's time

[Misanthrope]

16 Pages and we're still discussing a card that's not even out or properly reviewed to find out if the memory overclocking that's not possible would even do anything at all for performance (not looking like it)

 

Seriously sometimes I feel like just reporting all of this ridiculously stupid 10+ pages of bile on all AMD threads.

[Colonel_Gerdauf]

It is literally the first 3 words of my initial question: "out of curiosity". Because I don't understand his hatred towards Huddy. If I made similar headlines and pictures about Tom Petersen, would you not be curios as to why? I find it very odd.

 

Not understanding 3 words, is indeed wasting everyone's time

 

Interesting claim, especially when I already said this:

 

Nice evasion, but you have not answered my question: If a matter is not in context, they why bring it up at all? "I said out of curiosity" is about the weakest argument I have heard this year (and that actually says something).

 

Like I said, it is a weak argument. Do you ask "out of curiosity" about the recipe of mac and cheese in a funeral? Of course not, because it is out of context and in bad timing. Unless the question has context to the event or discussion at hand, then it is best not to ask such question.

 

Now do we have an actual answer to the question or not?

[Notional]

Interesting claim, especially when I already said this:

 

 

Like I said, it is a weak argument. Do you ask "out of curiosity" about the recipe of mac and cheese in a funeral? Of course not, because it is out of context and in bad timing. Unless the question has context to the event or discussion at hand, then it is best not to ask such question.

 

Now do we have an actual answer to the question or not?

 

That is your subjective opinion, that I am not responsible for. I answered, not my problem you want a different answer.

[EChondo]

Look, I'm fine with AMD locking this on their overdrive software in Drivers, but do we know if it's locked outside of that software?

 

I think this is a retarded move from AMD if they are blocking this all around.

[EvilWiffles]

It's new, so why risk idiots overclocking and breaking shit?

Even if it can't be overclocked, isn't the performance of the new VRAM enough?

[kacper]

AMD dying sooner rather than later is best for consumers. I've hashed that out a number of times, and most of the forum is convinced.

 

I've read some of your posts regarding that matter - it does seem logical .

Could you point me into some long and detailed read about this? 

[alphaproject]

I don't see the NEED for watercooling then....Unless the Core gets THAT hot when Overclocked...what do I know.

 

"So the "overclockers dream" is a big LIE! False advertisement. Shame on you AMD. " ROTFL

 

It sort of is unless the core goes insane high.... We won't know.

[NateGSR117]

Well that is unfortunate, but it is new memory tech and maybe AMD is already pushing the HBM chips too far so any further overclock will cause problems for the card. However, it still doesn't make it right if the card crashes on me because the overclock is to high then I should be able to deal with it, not be prevented from it.

[othertomperson]

Source: http://www.legitreviews.com/amd-fiji-arrives-radeon-r9-fury-x-details_166515#OpQZsfoAmLt9ajjt.99

 

Remember that amazing performance everybody said they were going to gain from even slightly overclocking the memory on HBM, well apparently you won't get any performance improvements from doing that because there will be no memory overclocking on the card at all. AMD Disabled the feature, therefore you will be limited to Core overclocking entirely. I know a lot of people are going to say, well HBM is new blah blah blah blah. Listen, if NVIDIA blocks overclocking on Laptops and people throw a fit, then people should throw the same fit when a company tells you cannot experiment with overclocking their new memory system because it's better for you. It's all about freedom of choice, that's the beauty of using PC. Here a company is saying you don't have that freedom to experiment yourself. There should be equal treatment for AMD doing this. Mind you this might change in the future, but for the time being when the cards release you won't be able to do any memory overclocking at all. Of course it's not like you are limited by bandwidth from HBM, but that's not the point, right?

 

It's actually worse. Nvidia blocked overclocking on laptops where, for power and thermal reasons, it's ill advised at the best of times. AMD are blocking memory overclocking on desktop hardware, and watercooled desktop hardware at that.

 

If this ends up being true then the shitstorm should be worse than Nvidia stopping overclocking on laptops. Whether it actually does happen remains to be seen

[Popingheads]

Has anyone considered that there may be an engineering or hardware reason why locking the clocks makes sense? As in, it is either not possible to change them at all or changing them at all period could make it unstable (even by a few MHz)?

 

It might not be as simple as "I should be allowed to overclock it anyway", it is a brand new memory technology and maybe overclocking it like you do traditional memory just isn't something that you should expect to do in the future.

 

But its just speculation, hell if I know.

[alphaproject]

It is a new tech....they've been working on it for so long perhaps AMD does know what they are talking about. The more I thought about it the less I think it's a big deal.

 

I'm just hoping for them the Fury is a long lasting card in case you want to pick one up used or something.... You never know I guess.

 

Maybe sometime down the road when they do release the 8 Gb version the 4 Gb version will come down a bit in price.

[Fetzie]

Has anyone considered that there may be an engineering or hardware reason why locking the clocks makes sense? As in, it is either not possible to change them at all or changing them at all period could make it unstable (even by a few MHz)?

 

It might not be as simple as "I should be allowed to overclock it anyway", it is a brand new memory technology and maybe overclocking it like you do traditional memory just isn't something that you should expect to do in the future.

 

But its just speculation, hell if I know.

 

It could well be like the base clock on Sandy Bridge and Haswell processors, where changing it by even a couple of MHz causes major instability. If you tried to overclock LGA1155 like you could with LGA1366 (when we went from base clock overclocking to multiplier overclocking), you'd either crash all the time or simply kill the chip.

[spnidel]

I currently own a sapphire tri-x r9 290 OC, and am going to get a fury X once it's out

I seriously don't see the issue of not being able to overclock the memory, especially since:

1. the bandwidth already is huge and is more than enough, coupled with the data compression technology they have for GCN 1.2, this becomes even less of an issue

2. it is new tech, there are bound to be limitations/shortcomings and in general it takes time to perfect any technology

 

with a bandwidth of 512gb/s, you'd get more performance out of your GPU by pushing the core clocks rather than the memory clocks

 

additionally - the "nvidia laptop overclock lock" argument being used as an example in this situation is stupid, GDDR5 has been around for years, HBM has just seen it's first implementation in a modern-day GPU, if they're locking it like that, there's a good reason

[patrickjp93]

I've read some of your posts regarding that matter - it does seem logical .

Could you point me into some long and detailed read about this?

The problem is it's my opinion synthesized from information from many different sources about movement in the industry, the IP everyone holds, and observing the behavior and successes/failures of our big 4 firms (still counting IBM). There's not one place I could point you. It's probably more than 100 articles and just watching the firms go for the last 3 years. I could try drawing up an in-depth report if you're interested.

[LAwLz]

It could well be like the base clock on Sandy Bridge and Haswell processors, where changing it by even a couple of MHz causes major instability. If you tried to overclock LGA1155 like you could with LGA1366 (when we went from base clock overclocking to multiplier overclocking), you'd either crash all the time or simply kill the chip.

And yet, Intel let us go ahead and increase the baseclock, and I never heard anyone complain about them letting us do it.

More power to the users, especially if it's supposed to be an enthusiast product.

additionally - the "nvidia laptop overclock lock" argument being used as an example in this situation is stupid, GDDR5 has been around for years, HBM has just seen it's first implementation in a modern-day GPU, if they're locking it like that, there's a good reason

I still don't get that argument. Would you have said the same thing if Nvidia locked down core overclocking on Pascal? It is a new architecture so by the logic of "it's new so it's okay to lock it down" that should be fine too.

Like Fetzie pointed out before, when Intel did something new and tied all clocks in the computer to a single clock they didn't lock it down to us users. They still gave us control over it despite the fact that raising it just a few MHz could fry your SATA ports among other things.

Before I get attacked by the AMD defense force, I am not saying that AMD don't have a reason for locking it down. I am just saying that the argument "it's new so it's justified to lock it down" is a very poor one I don't get. They could have reasons for locking it down but I don't think we have enough info to judge whether or not it is justified yet.

[kacper]

The problem is it's my opinion synthesized from information from many different sources about movement in the industry, the IP everyone holds, and observing the behavior and successes/failures of our big 4 firms (still counting IBM). There's not one place I could point you. It's probably more than 100 articles and just watching the firms go for the last 3 years. I could try drawing up an in-depth report if you're interested.

I'd love to read that.

[Fetzie]

And yet, Intel let us go ahead and increase the baseclock, and I never heard anyone complain about them letting us do it.

More tower to the users, especially if it's supposed to be an enthusiast product.

 

 

I still don't get that argument. Would you have said the same thing if Nvidia locked down core overclocking on Pascal? It is a new architecture so by the logic of "it's new so it's okay to lock it down" that should be fine too.

Like Fetzie pointed out before, when Intel did something new and tied all clocks in the computer to a single clock they didn't lock it down to us users. They still gave us control over it despite the fact that raising it just a few MHz could fry your SATA ports among other things.

 

Before I get attacked by the AMD defense force, I am not saying that AMD don't have a reason for locking it down. I am just saying that the argument "it's new so it's justified to lock it down" is a very poor one I don't get. They could have reasons for locking it down but I don't think we have enough info to judge whether or not it is justified yet.

 

Yeah, they could have done what MSI did with Afterburner, where you can only change certain settings like the voltage beyond a minimal threshold by typing something along the lines of "I know this can kill my gpu and I don't care that this could void my warranty" into an ini file. That would be enough to stop anybody that shouldn't overclocking their memory.

 

What I would like to know is if this is locked down generally, or only in the CCC overclocking tool (Overdrive). When nVidia pulled the plug on voltage increases of more than ~30mV that was a BIOS level change (IIRC) - is this also a BIOS level change (i.e. we could put a different BIOS on the card as it has two BIOS chips on board) or would we simply need to install Afterburner/any other overclocking suite to get around the restriction?

[spnidel]

I still don't get that argument. Would you have said the same thing if Nvidia locked down core overclocking on Pascal? It is a new architecture so by the logic of "it's new so it's okay to lock it down" that should be fine too.

Like Fetzie pointed out before, when Intel did something new and tied all clocks in the computer to a single clock they didn't lock it down to us users. They still gave us control over it despite the fact that raising it just a few MHz could fry your SATA ports among other things.

 

Before I get attacked by the AMD defense force, I am not saying that AMD don't have a reason for locking it down. I am just saying that the argument "it's new so it's justified to lock it down" is a very poor one I don't get. They could have reasons for locking it down but I don't think we have enough info to judge whether or not it is justified yet.

 

now that you put it like that, you're right

however I still am completely OK with not being able to overclock the memory, because the memory bandwidth is high enough as it is and I'd rather not risk my entire GPU over something as trivial as an increase of 10GB/s in memory bandwidth, the performance gain of which would be miniscule at the current bandwidth

[Mo5]

There could just be a 2nd BIOS for "extreme" tweaking

[Fetzie]

There could just be a 2nd BIOS for "extreme" tweaking

 

"extreme" tweaking in this case being just "tweaking"?

[TheKDub]

I agree with AMD here 100%.

You can't have your customers overclocking something if it is unstable will most likely harm the component and then have the customer tell you the card is broken and demand a refund or exchange. Same goes for the NVIDIA laptop situation.

They could be less of hole diggers and just say that they dont recommend ocing the memory, and if you break your card doing then that is your fault.

[Mr.Dingle]

 

Just out of curiosity .... does an average Joe buying a Fury X really need more than 512GB/s ??

 

 

[Fetzie]

 

 

Just out of curiosity .... does an average Joe buying a Fury X really need more than 512GB/s ??

 

 

I doubt it. Overclocking memory doesn't really do anything to your frame-rate anyway, the only real difference it makes is your graphics card runs hotter.