390X coming soon few weeks

Hmm Titan X is 30% faster than a GTX 980. Wonder if the 390X will be atleast that much faster. It better hurry and be released, so tempted to sell both my 980s and grab a pair of Titan Xs.

Only thing really holding me back is paying the huge premium for that extra compute power I will never use.
 
well, with the Titan X, you are paging that huge premium WITHOUT the extra compute performance

lol
 
Hmm Titan X is 30% faster than a GTX 980. Wonder if the 390X will be atleast that much faster. It better hurry and be released, so tempted to sell both my 980s and grab a pair of Titan Xs.

Only thing really holding me back is paying the huge premium for that extra compute power I will never use.

what compute power? the Titan X doesn't have DP compute enabled. Its not a prosumer card like the original Titan was. The only compute it has are cuda cores for the few applications that have cuda acceleration, and SP compute.

in your case I would rather have a 3rd 980, over having 2 titan x's, would be cheaper and perform the same!
 
If you read AMD's Q4 earnings call transcript, you'll see their top priority for Q1 2015 is still to normalize inventory.

With them specifically talking about 1st and 2nd Gen APUs...
Don't get me wrong, they still have a decent amount of GPUs but you don't mark down a bunch of inventory unless it is essentially unmoveable.
 
Can these come out so i can buy a 290/x for cheaper? thinking about going 2560x1440 or 2560x1080... I really try to stick to reference coolers... I hope the 390 isnt a fuck up like the 290 was... The xfx 7970 isnt cutting it.

Why do you prefer reference coolers ? Unless you're watercooling the card I see zero benefit over a aftermarket cooler.
 
Well, after 8 years AMD has lost me as a future customer.
Catalyst 15.3 seems to have been delayed, after 3 months of waiting. Also, as per my post here the early notes showed FC4 and Dying Light crossfire profiles, but no Rogue profile despite the fact it would have taken AMD 30 seconds to implement it.

I can't in good conscience support them anymore.
Let's hope the new Fiji cards can push the GTX 980 down to $400 for me... I'm counting on you one more time, AMD. Just do it for me out of good faith.
 
Was wondering if you guys think the HBM would be a boon to 4K, like help maintain higher minimums? I mean we have seen the technical data as in bandwidth but little in how that will affect the outcome compared to GDDR5.
 
I remember when I had my HD 7950's that initially overclocking the core made the biggest difference. Then after a bit the core gave diminishing returns. You could go higher, but the frame rate didn't increase as much as the first couple of hundred MHz.

So then I started experimenting with overclocking the memory. I kept getting great returns all the way out to 1800MHz. It wasn't as dramatic as the core initially, but it was consistent until you hit the point where the memory starting clipping. So what I ended up doing was increase the core to say 1050 and go crazy on the memory to 1700. That gave me a ridiculous frame rate in comparison to my nephew's GTX 670.

I hope that this HBM memory tech does the same for the new AMD and future Nvidia cards. Now if I could have stopped earlier with the 1800 suicide runs on that HD 7950, I'd still be rocking it in my case. Those blocks it throws up nowadays makes me really sad. :)

R.I.P. Sapphire HD 7950 with the sick 7970 launch cooler.
 
Was wondering if you guys think the HBM would be a boon to 4K, like help maintain higher minimums? I mean we have seen the technical data as in bandwidth but little in how that will affect the outcome compared to GDDR5.

normally a user might be bandwidth starved which then wont be the case with HBM ram.
8gb cards should also make sure the textures have np either fitting it.
the power save with HBM is a big thing.
same with the size as we might see smaller cards down the line.

but MIN fps should be having a boost with it.
 
I think the HBM will be a huge boon to a much smoother frame rate, possibly increasing minimums. Nvidia is gonna have a hard time keeping up using 384 bit bus against a 4096bit, no matter how high you crank the memory.
 
I think the HBM will be a huge boon to a much smoother frame rate, possibly increasing minimums. Nvidia is gonna have a hard time keeping up using 384 bit bus against a 4096bit, no matter how high you crank the memory.
Memory bandwidth should pretty much scale with clock speed. I've still yet to see a single review that shows without a doubt that increased memory bandwidth will yield more than marginally better results.
 
I think AMD's biggest problem, even if the 390X beats the Titan X, is Nvidia Gameworks. Card beats the Titan X in FPS but the Titan X still provides excellent frame rate AND has additional game options turned on.

Also that 4096bit memory bus wont go far if indeed the 390X is limited to 4GB.

Then there's the issue of drivers. As usual for AMD - Excellent hardware, no drivers to back it up.
 
I think AMD's biggest problem, even if the 390X beats the Titan X, is Nvidia Gameworks. Card beats the Titan X in FPS but the Titan X still provides excellent frame rate AND has additional game options turned on.

Then there's the issue of drivers. As usual for AMD - Excellent hardware, no drivers to back it up.

Well AMD are suppose to be making FreeWorks to battle nvidias GameWorks.

And in my opinion, AMD have excellent pricing and good hardware. Where as Nvidia has the excellent hardware but not so good pricing.
 
Well AMD are suppose to be making FreeWorks to battle nvidias GameWorks.

And in my opinion, AMD have excellent pricing and good hardware. Where as Nvidia has the excellent hardware but not so good pricing.

Now let’s talk about Richard Huddy, AMD’s Gaming Scientist. Huddy has been crusading against Nvidia’s GameWorks and talking up the company’s technologies like Mantle and FreeSync. He’s also teased the brainstorming phases of something called “FreeWorks,” a theoretical open-source library of technologies that would counter Nvidia’s “Black Box” set of technologies. AMD still hasn’t confirmed if FreeWorks is actually in development.

http://www.forbes.com/sites/jasonev...-next-generation-lineup-of-pc-graphics-cards/

Thats from August of last year. Still nothing concrete on "Freeworks". Plus I wonder how many devs are going to use it...(TressFX anyone?)
 
Well AMD are suppose to be making FreeWorks to battle nvidias GameWorks.

And in my opinion, AMD have excellent pricing and good hardware. Where as Nvidia has the excellent hardware but not so good pricing.


Marketing. AMD's marketing absolutely sucks. Really the Never Settle Bundle was probably the best marketing ploy they've done in awhile. AMD is falling into a similar trap that they fell into with the CPU market. Selling your products for dirt cheap with cut throat margins is NOT a sustainable business model long-term.

Nvidia stopped playing the price wars even when the market share was more even because they treat their product as a luxury item and sell it as such. Whats worse is that it WORKS. Also all this, "Free-Sync" this and "Free-Works" that not only makes them look reactionary, but also look cheap. They desperately need a psychologist on their payroll.
 
Marketing. AMD's marketing absolutely sucks. Really the Never Settle Bundle was probably the best marketing ploy they've done in awhile. AMD is falling into a similar trap that they fell into with the CPU market. Selling your products for dirt cheap with cut throat margins is NOT a sustainable business model long-term.

Nvidia stopped playing the price wars even when the market share was more even because they treat their product as a luxury item and sell it as such. Whats worse is that it WORKS. Also all this, "Free-Sync" this and "Free-Works" that not only makes them look reactionary, but also look cheap. They desperately need a psychologist on their payroll.

They need for people to quit being so stupid as not to realize the benefit to them and PC gaming for features to be open and how bad it is to be proprietary. While it's in early development, sure. Too many cooks can spoil the pudding. But locking it up and refusing to allow anyone to use it unless they play the game to your advantage is bull shit, pure and simple.

Then you have the viral marketing attacking. Why is nobody concerned that DX12 is not open source and free for everyone to use and adapt for their hardware? Lots of people seemed real concerned about Mantle not being open. DX12 is offering the same improvements Mantle offered, but everyone is like DX12 is so great when they weren't impressed with Mantle. Why isn't everyone who wanted AMD to make Mantle open source pulling for OpenGLNext? Because, that's what it is. Why is 4gig on AMD cards not enough? 3.5gig+ on nVidia is supposed to be fine. Even if they lie to you about it. We don't even need a true 4gig on nVidia cards. If AMD drivers suck so bad why has Hawaii pulled past GK110? With nVidia's superior drivers the GK110 should be twice as fast as Hawaii by now. Especially considering nVidia made those magic drivers that offered op to 65% performance increase. nVidia's multi thread DX11 drivers were all we were going to ever need when Mantle was first conceived. Remember all that talk? Yet you look at relative performance between the two brands on Kepler vs. GCN and it looks like it's AMD who's making bigger improvements with their drivers. All of this while developing entire new API's even though the spin on the Web is they can't afford to keep up with nVidia.

AMD has been poorly managed. That, and that alone, has been their problem. They ever straighten that out and they can succeed.
 
They need for people to quit being so stupid as not to realize the benefit to them and PC gaming for features to be open and how bad it is to be proprietary. While it's in early development, sure. Too many cooks can spoil the pudding. But locking it up and refusing to allow anyone to use it unless they play the game to your advantage is bull shit, pure and simple.

Then you have the viral marketing attacking. Why is nobody concerned that DX12 is not open source and free for everyone to use and adapt for their hardware? Lots of people seemed real concerned about Mantle not being open. DX12 is offering the same improvements Mantle offered, but everyone is like DX12 is so great when they weren't impressed with Mantle. Why isn't everyone who wanted AMD to make Mantle open source pulling for OpenGLNext? Because, that's what it is. Why is 4gig on AMD cards not enough? 3.5gig+ on nVidia is supposed to be fine. Even if they lie to you about it. We don't even need a true 4gig on nVidia cards. If AMD drivers suck so bad why has Hawaii pulled past GK110? With nVidia's superior drivers the GK110 should be twice as fast as Hawaii by now. Especially considering nVidia made those magic drivers that offered op to 65% performance increase. nVidia's multi thread DX11 drivers were all we were going to ever need when Mantle was first conceived. Remember all that talk? Yet you look at relative performance between the two brands on Kepler vs. GCN and it looks like it's AMD who's making bigger improvements with their drivers. All of this while developing entire new API's even though the spin on the Web is they can't afford to keep up with nVidia.

AMD has been poorly managed. That, and that alone, has been their problem. They ever straighten that out and they can succeed.

well said

Value is what people percive it to be.
either it has or not or works or not.
I see a company that lies and decives their own customer base constantly and people accept it due to them think it was a mistake, yes once not the third or such time.
you can then stand on stage and show a card that does not exist and tell the public and shareholders here it is and people accept that?

King of the hill the replacement to the 290 series are coming.
 
AMD has been poorly managed. That, and that alone, has been their problem. They ever straighten that out and they can succeed.

Isn't AMD a bit too far down the gutter as far as their CPU division is concerned?

Don't get me wrong I would so love for AMD to return to its Athlon64 heydays, but somehow I just can't see that happening anytime soon.
 
If AMD doesn't have the money like NVidia, how will they compete in GPU department? I have started worrying, just like the way, they completely lost CPU to Intel, I hope they don't start loosing GPU market to NVidia.
 
They need for people to quit being so stupid as not to realize the benefit to them and PC gaming for features to be open and how bad it is to be proprietary. While it's in early development, sure. Too many cooks can spoil the pudding. But locking it up and refusing to allow anyone to use it unless they play the game to your advantage is bull shit, pure and simple.

Then you have the viral marketing attacking. Why is nobody concerned that DX12 is not open source and free for everyone to use and adapt for their hardware? Lots of people seemed real concerned about Mantle not being open. DX12 is offering the same improvements Mantle offered, but everyone is like DX12 is so great when they weren't impressed with Mantle. Why isn't everyone who wanted AMD to make Mantle open source pulling for OpenGLNext? Because, that's what it is. Why is 4gig on AMD cards not enough? 3.5gig+ on nVidia is supposed to be fine. Even if they lie to you about it. We don't even need a true 4gig on nVidia cards. If AMD drivers suck so bad why has Hawaii pulled past GK110? With nVidia's superior drivers the GK110 should be twice as fast as Hawaii by now. Especially considering nVidia made those magic drivers that offered op to 65% performance increase. nVidia's multi thread DX11 drivers were all we were going to ever need when Mantle was first conceived. Remember all that talk? Yet you look at relative performance between the two brands on Kepler vs. GCN and it looks like it's AMD who's making bigger improvements with their drivers. All of this while developing entire new API's even though the spin on the Web is they can't afford to keep up with nVidia.

AMD has been poorly managed. That, and that alone, has been their problem. They ever straighten that out and they can succeed.

Really awesome post!

It's all marketing.
 
Mantle was developed and under full control of one of gpu makers - that made it impossible for any competitor to use it or they would be at the mercy of AMD.

DirectX is under control of Microsoft and they don't have any interest in playing favorites with hardware makers - plus they have heritage of saving game development from API wars of 3dfx era.

OpenGLNext is open source standard done the correct way - but previous versions were mess so I'd expect a lot of programmers to stick to more familiar DX at least initially.
 
Gibbo over at OCUK has sorta kinda confirmed thats what they were doing and that its been ready for a while, and they are holding off to clear inventory.

He couched it as a personal opinion, but since he runs one of Britains largest enthuiest etailers, im inclined to take his "opinion" as closer to the truth.

There was also some funky Italian site that said he had inside sources at some of the AIBs that implied the leaked performance numbers for the 390x WCE are actually low.

Now THAT would be interesting.

50% over 290x is LOW?

Jebus...

edit: source - https://translate.google.com/transl...-aspetta-che-i-driver-siano-maturi&edit-text=

Right now I have my Titan X performing 38% higher than my OC'd 980. That's 71% over a OC'd 290x according to the [H] 980 OC vs 290x OC review. I didn't even get my waterblocks yet. :cool:

If AMD wants the 390x to match the Titan X, especially OC vs OC which is all I care about, they have quite the hill to climb.

I personally like the AIO by default, or better yet I wish you could buy cards without a cooler, but this Titan X cooler struggles after you get around +20% on the OC.
 
The Chiphell benchmarks actually show 37% over the 980 @ 4K stock, which would be the leaks the EU sites are referring to. Not such a big hill to climb.
The latest 390 non-X leak puts it about 10% slower than the Titan X stock, using 3dmark11. http://www.chiphell.com/forum.php?mod=viewthread&tid=1256211

But we're optimistic in this thread. The amount of people on AMD's side who are willing to spend $1,000 on a GPU is precisely zero.
Unless Nvidia cuts the price of the Titan X or launches a 980 Ti, whatever the OC potential of the 390X ends up being, the Titan X is not an option.
 
Last edited:
Right now I have my Titan X performing 38% higher than my OC'd 980. That's 71% over a OC'd 290x according to the [H] 980 OC vs 290x OC review. I didn't even get my waterblocks yet. :cool:

If AMD wants the 390x to match the Titan X, especially OC vs OC which is all I care about, they have quite the hill to climb.

I personally like the AIO by default, or better yet I wish you could buy cards without a cooler, but this Titan X cooler struggles after you get around +20% on the OC.

what are these numbers?

With my overclocked 290 (1275/1600 under water) in 3dmark firestrike I get this.
Graphics Score 14197

So to bust your bubble I highly doubt your Titan X is 71% faster than my 290. Probably along the lines of 35-45% faster

I'm almost certain the 390x will be faster than the Titan X, if not extremely close. Overclock to Overclock the 390x should stomp it. Hardocp hasn't had much luck with overclocking the R9-290/X's Their results kinda suck. Slap a water block on them and they are entirely different beasts. Almost all water cooled 290/X's manage 1200 core, with the average being closer to 1300. Yes they suck down the power and are horribly inefficient compared to the maxwell counter parts but they are fast. (Maxwells do suck the power to when you start to get into bios mods and such as well, but not nearly as bad as the hawai's)

oh well I'm not going to play the speculation game, we will see when it gets released how fast it is and how well it overclocks. I know one thing though the water cooled edition should overclock very well, compared to the Titan X's stock cooler/tdp limits.
 
Back
Top