Fury unlocked cores/volts and LN2'd to 1450/1000 (100% HBM OC!!)

N4CR

Supreme [H]ardness
Joined
Oct 17, 2011
Messages
4,947
Finally it has happened.
Can't wait to see air/water results and if people manage to get the Fury-X scooting along to.
The HBM clocks are off the wagon..

http://forum.hwbot.org/showthread.php?t=142320

With so much negativity around these cards from a certain camp, it's good to see something positive. HBM can OC and OC extremely well by the looks of things. I guess AMD didn't lie after all?

Enjoy!
 
while i did enjoy the read, I cant really say im impressed with what they had to do to get a 30% performance bump. I would like to see how much they could manage with the included cooler
 
Finally it has happened.
Can't wait to see air/water results and if people manage to get the Fury-X scooting along to.
The HBM clocks are off the wagon..

http://forum.hwbot.org/showthread.php?t=142320

With so much negativity around these cards from a certain camp, it's good to see something positive. HBM can OC and OC extremely well by the looks of things. I guess AMD didn't lie after all?

Enjoy!

Define "lie", because LN2 and voiding the cards warranty is normal for daily operations :rolleyes: Very practical for the everyday consumer.
 
Define "lie", because LN2 and voiding the cards warranty is normal for daily operations :rolleyes: Very practical for the everyday consumer.

i think he might be a bit new to the overclocking industry and doesn't understand exactly how LN2 works.
 
I would not do that with my cards, even if I spend an arm and a leg on graphics cards every few years.
 
CPU is at stock clocks only, looks like he's about to shatter some world records. Also 1Terabyte of bandwidth, that has to be a world first for consumer GPUs (and possibly anywhere) :eek::cool:
 
Yea still waiting for some impressive gains that's more practical....The unlocking shaders is very cool but not exactly new. Something that doesn't necessarily void the warranty and reaches a 20% real world gain 24/7 stable would be good to see but I guess its to much to ask (with furys)
 
Finally it has happened.
Can't wait to see air/water results and if people manage to get the Fury-X scooting along to.
The HBM clocks are off the wagon..

http://forum.hwbot.org/showthread.php?t=142320

With so much negativity around these cards from a certain camp, it's good to see something positive. HBM can OC and OC extremely well by the looks of things. I guess AMD didn't lie after all?

Enjoy!

AMD did in fact lie since the link you posted requires hardware mods and LN2. Can you do this out of the box with a Fury/Fury X? Nope.
 
Finally it has happened.
Can't wait to see air/water results and if people manage to get the Fury-X scooting along to.
The HBM clocks are off the wagon..

http://forum.hwbot.org/showthread.php?t=142320

With so much negativity around these cards from a certain camp, it's good to see something positive. HBM can OC and OC extremely well by the looks of things. I guess AMD didn't lie after all?

Enjoy!

AMD didn't lie :eek: Because LN2 setups are very common. :rolleyes:
 
I'm curious how this compares to the air-cooled Zotac 980 Ti AMP Extreme...
 
Looks like the scores are inflated because of the ratio of the gaming test scores from tessellation being disabled.

Also the scores are about 1300 points higher on the lesser overclock amounts(1100mhz) than they should be without tessellation disabled.

Without the modification, this score probably falls down to score that could be achieved with a gtx 980 ti on water.

This is hardly an overclockers dream since disabling tessellation is hardly practical in games.

AMD has got to beat Nvidia legitimately if it wants to be taken seriously for overclocking and world records.
 
The amount of complaining in here is amazing.

This isn't "How to OC your Fury (non-X)", its going to extremes to see what the card is capable of, not what most people will get.

He took a Fury non-X, unlocked it to be closer to a full X and then went crazy with LN2 to completely max out clocks.

Will regular users go there? No not at all, but it does show you that the hardware is capable of it, and people will find a middle ground that works well for most people.

Are you going to say that the 980 can only OC using LN2 because some people have done it? No.
 
The amount of complaining in here is amazing.

This isn't "How to OC your Fury (non-X)", its going to extremes to see what the card is capable of, not what most people will get.

He took a Fury non-X, unlocked it to be closer to a full X and then went crazy with LN2 to completely max out clocks.

Will regular users go there? No not at all, but it does show you that the hardware is capable of it, and people will find a middle ground that works well for most people.

Are you going to say that the 980 can only OC using LN2 because some people have done it? No.

But the 980 clocks to 1500-1550 pretty easily on air. The 980ti gets to about the same on air and 2100 on LN2. Given ref boost clocks are about 1150 on Maxwell, you still get about the same % OC with Maxwell on air vs Fury at -190C.

It's not complaining, it's just to post these results as a positive is almost satire.

The HBM OCs are nice to know for the future but 500Mhz was already overkill for 4GB.
 
But the 980 clocks to 1500-1550 pretty easily on air. The 980ti gets to about the same on air and 2100 on LN2. Given ref boost clocks are about 1150 on Maxwell, you still get about the same % OC with Maxwell on air vs Fury at -190C.

It's not complaining, it's just to post these results as a positive is almost satire.

The HBM OCs are nice to know for the future but 500Mhz was already overkill for 4GB.

The 980 was a newly engineered architecture, the Fury/Fury X is an old architecture with incremental improvements and a new memory interface making it stretched very thin as it is. So comparing the two with regards to overclocking limits is a bit unfair, honestly. Different approaches, different outcomes.
 
Am I the only one who sees the pre-voltmod/LN2 numbers?

Firestrike Extreme
6237 -> 8121

Thats +30%, on air.
 
That's the Fury X (8121) with tesselation off. For reasons I do not understand.
 
The amount of complaining in here is amazing.

This isn't "How to OC your Fury (non-X)", its going to extremes to see what the card is capable of, not what most people will get.

He took a Fury non-X, unlocked it to be closer to a full X and then went crazy with LN2 to completely max out clocks.

Will regular users go there? No not at all, but it does show you that the hardware is capable of it, and people will find a middle ground that works well for most people.

Are you going to say that the 980 can only OC using LN2 because some people have done it? No.

People are just responding the OP's statement saying that this somehow showed AMD didn't lie when they hyped up Fury as an overclocker's GPU, when the fact is that LN2 overclocking are not considered to be viable.

How overclockable a GPU is has to be judge on how well you can do it within a practical setting, which means no hardmodding or LN2 cooling. Fury didn't even have it's voltage unlocked, a very basic requirement in overclocking. So it's hard to see how the chip was meant to be overclockable when it's voltage is locked down.
 
People are just responding the OP's statement saying that this somehow showed AMD didn't lie when they hyped up Fury as an overclocker's GPU, when the fact is that LN2 overclocking are not considered to be viable.

How overclockable a GPU is has to be judge on how well you can do it within a practical setting, which means no hardmodding or LN2 cooling. Fury didn't even have it's voltage unlocked, a very basic requirement in overclocking. So it's hard to see how the chip was meant to be overclockable when it's voltage is locked down.

One guy made the comment "It's an overclocker's dream". Even if you want to make him the spokesperson for the entire company, he never qualified it as meaning "out-of-box on stock cooling".

Look, whatever the overclocking is, it is what it is and no amount of he said she said is going to effect it one way or the other. That's not what this thread, or the OP's source are attempting to do. Some guy unlocked and Vmodded his Fury and then O/C'd it with extreme cooling. What is interesting, being brand new tech with no prior figures to examine, is the 100% O/C of the HBM. Not what some guy who works for AMD said in passing.
 
One guy made the comment "It's an overclocker's dream". Even if you want to make him the spokesperson for the entire company, he never qualified it as meaning "out-of-box on stock cooling".

Look, whatever the overclocking is, it is what it is and no amount of he said she said is going to effect it one way or the other. That's not what this thread, or the OP's source are attempting to do. Some guy unlocked and Vmodded his Fury and then O/C'd it with extreme cooling. What is interesting, being brand new tech with no prior figures to examine, is the 100% O/C of the HBM. Not what some guy who works for AMD said in passing.

Except when that "one guy" is the CTO of AMD he IS speaking for his ENTIRE company. Let's not do AMD's bullshit PR spin for them. When he stated it he knew exactly what he meant and exactly how people would take it. As of right now it's a lie, plain and simple. AMD needs to called out on their bullshit just like any other company.
 
Except when that "one guy" is the CTO of AMD he IS speaking for his ENTIRE company. Let's not do AMD's bullshit PR spin for them. When he stated it he knew exactly what he meant and exactly how people would take it. As of right now it's a lie, plain and simple. AMD needs to called out on their bullshit just like any other company.

absolutly he was off base big time and should kept his mouth shut.
 
Except when that "one guy" is the CTO of AMD he IS speaking for his ENTIRE company.

Have to agree there, as CTO it would be appreciated he clarify what he meant. In this case it probably won't cost AMD much money since Fury X cards are still out of stock and almost impossible to find. And technically it does overclock, so this probably wouldn't factor into SEC filings.
 
absolutly he was off base big time and should kept his mouth shut.

Agreed. This seems to be a common problem for AMD.

I still don't see how it's applicable to this thread, though. It's fine not to be impressed with the LN2 clocks. Obviously 1450MHz isn't anything special. The memory O/C is very interesting though. HBM shows some real potential.
 
Have to agree there, as CTO it would be appreciated he clarify what he meant. In this case it probably won't cost AMD much money since Fury X cards are still out of stock and almost impossible to find. And technically it does overclock, so this probably wouldn't factor into SEC filings.

Though I'm too lazy to look up the source, when mr. CTO mentioned the fury series as an OC'ers dream, he specifically mentioned the ease at which it will OC on stock cooling.
 
i think he might be a bit new to the overclocking industry and doesn't understand exactly how LN2 works.

Just because I have a young hardocp account, doesn't mean I have not been overclocking since the days of pencil or pin mods.

AMD did in fact lie since the link you posted requires hardware mods and LN2. Can you do this out of the box with a Fury/Fury X? Nope.

I understand where you are coming from. But this 'lie' is with CEO/marketing speak (and a big hangover, please excuse any mistakes..), with the results at hand, AMD didn't completely lie - it's a usual marketing/ceo speak white lie. This 'lie' is similar to marketing statements from other companies in this industry.
For me, all this shows, is that it is possible to OC the fury, particularly HBM (and what the future could bring with HBM silicon revisions..). Sure, LN2 doesn't count, but most records are set this way.
What I'm taking from this is that with good air/water, some OCing with unlocked volts/clocks could result in a nice bump for owners of these cards.

Either way, there does appear to be some free lunch on the table. After a month or so of heavy use, I'd be happy that my card(s) is stable/out of the failure bell curve, allowing similar modifications if necessary. Sure I'd much rather use software voltage control.. less hassle.

But hey, modding gpus, this is [H], I'm sure people here will be doing simiilar!

And with AMD cards, it's not uncommon to see voltage control a few months later, I'm looking forward to seeing this and the air/water results that come with.
.
 
Though I'm too lazy to look up the source, when mr. CTO mentioned the fury series as an OC'ers dream, he specifically mentioned the ease at which it will OC on stock cooling.

It might? We haven't seen the O/C'ing software in consumer's hands yet. The stock cooling should still make the card cool and quiet where reference blowers (either company) when cranked up to keep an over volted card stable is loud and the card runs at it's thermal limits.
 
Can't wait to see what HBM2 has in store.
That is what I'm waiting to upgrade to.
 
Just because I have a young hardocp account, doesn't mean I have not been overclocking since the days of pencil or pin mods.



I understand where you are coming from. But this 'lie' is with CEO/marketing speak (and a big hangover, please excuse any mistakes..), with the results at hand, AMD didn't completely lie - it's a usual marketing/ceo speak white lie. This 'lie' is similar to marketing statements from other companies in this industry.
For me, all this shows, is that it is possible to OC the fury, particularly HBM (and what the future could bring with HBM silicon revisions..). Sure, LN2 doesn't count, but most records are set this way.
What I'm taking from this is that with good air/water, some OCing with unlocked volts/clocks could result in a nice bump for owners of these cards.

Either way, there does appear to be some free lunch on the table. After a month or so of heavy use, I'd be happy that my card(s) is stable/out of the failure bell curve, allowing similar modifications if necessary. Sure I'd much rather use software voltage control.. less hassle.

But hey, modding gpus, this is [H], I'm sure people here will be doing simiilar!

And with AMD cards, it's not uncommon to see voltage control a few months later, I'm looking forward to seeing this and the air/water results that come with.
.

This is bad overclocking results with ln2 even.

On 28nm, basically everything has overclocked 60+%. Hawaii overclocked to over 60 percent, the gtx 680 - 980 all overclocked past 2ghz under ln2.

All fiji is overclocking is 38% which is terrible for ln2. This is less than half the overclock in percentage terms, than an ln2 maxwell overclock. In terms of percent,this is similar to a decently clocked air sample gtx 980 ti.

The only reason the score is decent for the clocked is the disabled tesselation. All things considered, the fury X overclocking is brutal all around. It litterally has the poorest overclocking potential of any GPU on 28nm. HBM is showing some overclocking potential but this is separate from the potential of the fiji chip.
 
Back
Top