• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

[wccftech] rumor: amd hawaii benchmarked in 3dmark11 firestrike AMD attacks titan

Page 14 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
hawtdawg said:
Memory bandwidth is already plentiful with the current design and is as fast as the Titan and the 780. There is no reason for them to change it for a modest core update. They are also clearly gunning for the 780 at that price point. They've already taken out the Titan with the reduced price of the 7990.
Click to expand...

The 780 took out the Titan unless you're a nutter or ill informed.
 
hawtdawg said:
go ahead and write it down folks:

28nm
2560 sp's
160 TMU's
40 ROP's
1000 GPU clock
6000 MHZ RAM clock
3 Gig DDR5
384 bit memory bus
Click to expand...

Considering that rops are tied to memory bus in some way, a 40 rop config would point to a 320bit memory bus, which would mean that the 9970 will likely not have more bandwidth compared to the 7970, strange considering the greater processing power on display.
 
SiliconWars said:
This is why AMD is going with a big 28nm this time. After Hawaii launches there will be no response by Nvidia for at least 8 months.

Maxwell can launch in June 2014. June is a very long way away for Nvidia to be 2nd best for all that time, which they will be. AMD counters in October with Pirate Islands.

Now imagine this scenario - Nvidia assumed back then that AMD would stick to their "sweet spot" strategy, and decide to go with a midrange die a-la GK104 in June that can't beat the 9970 (because they didn't anticipate AMD going big on 28nm so late). Then AMD counters 3 months later with a huge Pirate Islands die for the mother and father of all smackdowns.
Click to expand...

So they go from competing with Nvidia's 3rd best (GTX 770) to everlasting AMD total world domination.
Reminds me of the chess analysis where you freeze opponent and only you are allowed to move pieces.
Very realistic
/sarcasm.


There's only one problem - while AMD has to spend tens if not hundred(s) of $M pulling magic out of the thin 28mm air,
all NV has to do is release full GK110 and move GK104 to it's natural mid-range place - for a total grand cost of approximately 0(zero) $(dollars).

How's that for mother and father of all smackdowns :twisted:

So instead of praying for total world domination, I hope for at least another Cayman. I.e. decent perf/price.
And also I pray for GPU division not to burn any more bridges with CEO Mr Rory. Because that would REALLY be unhealthy.
 
3DVagabond said:
Since when did DP matter to anyone on these forums? It's about as important as OpenCL applications where the 7970 smashes Titan for ~1/3 the price?
Click to expand...

Oh, so now you are the gude of what matters....not.

Answer my question ^^

As long as CUDA is out there...OpenCL is iirrelevant to me ^^
 
GK110 isn't a good point of comparison, cache hierarchy is a good deal more advanced on it and as such memory bandwith efficiency is higher.

I expect lots of smaller scale changes to stick more execution units in the same power envelope, resolving the ROP bottleneck, adding some ACEs and merging in improvements from the console SoCs instead of radical large scale changes. This is all just conjecture, of course.
 
Last edited:
f1sherman said:
There's only one problem - while AMD has to spend tens if not hundred(s) of $M pulling magic out of the thin 28mm air,
all NV has to do is release full GK110 and move GK104 to it's natural mid-range place - for a total grand cost of approximately 0(zero) $(dollars).

How's that for mother and father of all smackdowns :twisted:

So instead of praying for total world domination, I hope for at least another Cayman. I.e. decent perf/price.
And also I pray for GPU division not to burn any more bridges with CEO Mr Rory. Because that would REALLY be unhealthy.
Click to expand...

you want Nvidia domination and are happy with it as I see from your post. The others want the GPU crown to switch from time to time as its in the best interests of consumers that no single company dominates an industry. As soon as there was no competition for GK110 Nvidia pounced on the opportunity to push flagship single GPU prices to an all time high of USD 1000 which is frankly ridiculous.

Hawaii XT aka HD 9970 matching Titan would mean the current Titan will be replaced by a similar 2688 sp chip as a GTX 785 with 3GB VRAM and 1/24 double precision. the fully enabled GK110 aka Titan II or Titan Ultra will take the top spot.

the question is can AMD Hawaii beat Titan , because that means Nvidia will have to scramble to match AMD. is it possible ? we will have to wait and see.
 
Last edited:
Lonbjerg said:
Oh, so now you are the gude of what matters....not.

Answer my question ^^

As long as CUDA is out there...OpenCL is iirrelevant to me ^^
Click to expand...

What question? The DP rate of the 780, 1/24.

BTW, the bold part of your post is completely unnecessary. It only shows your need to make things personal rather than just replying to the discussion. What's a "gude"?
 
f1sherman said:
There's only one problem - while AMD has to spend tens if not hundred(s) of $M pulling magic out of the thin 28mm air,
all NV has to do is release full GK110 and move GK104 to it's natural mid-range place - for a total grand cost of approximately 0(zero) $(dollars).
Click to expand...

Actually, I'm not too sure about that. Sure, once you mod the bios Titan can stretch its additional shader units, but with the power limits on, the difference between the Titan and Gtx 780 becomes very small. It's likely that the additional shaders from a fully unlocked GK110 are offset by the loss in frequency needed to stay within spec.

(That's ignoring the amazing logic of a 28nm refresh costing AMD tens and hundreds of millions while GK110 R&D, manufacturing and etc. was somehow free)
 
3DVagabond said:
What question? The DP rate of the 780, 1/24.

BTW, the bold part of your post is completely unnecessary. It only shows your need to make things personal rather than just replying to the discussion. What's a "gude"?
Click to expand...

So 1/3 vs 1/24...so the DP performance is~8 times better on the Titan vs the 780...is the price 8 times higher?

If not, I must submit you don't understand the Titan SKU.
 
Lonbjerg said:
So 1/3 vs 1/24...so the DP performance is~8 times better on the Titan vs the 780...is the price 8 times higher?

If not, I must submit you don't understand the Titan SKU.
Click to expand...

How many people on these boards who own Titans use it's DP capabilities?

Edit: Or even how to activate the 1/3 instead of the 1/24 it comes out of the box with?
 
3DVagabond said:
How many people on these boards who own Titans use it's DP capabilities?

Edit: Or even how to activate the 1/3 instead of the 1/24 it comes out of the box with?
Click to expand...

So now we are at fallacies...because the facts don't agree with your opinion...nice, my work is done now ^^
 
Jaydip said:
People buy Titan for DP? I hope this is a joke.
Click to expand...

Why?
So people can use CUDA and DP...it's a "joke"..we need to pay for a QUADRO SKU before it makes sense to you?

I hope you don't usse a GPU, but only use a IGP...because otherwse you are a joke for using so much more money on so Little extra performace...right?! ^^

Thanks for confirming my point 😉
 
Lonbjerg said:
Why?
So people can use CUDA and DP...it's a "joke"..we need to pay for a QUADRO SKU before it makes sense to you?

I hope you don't usse a GPU, but only use a IGP...because otherwse you are a joke for using so much more money on so Little extra performace...right?! ^^

Thanks for confirming my point 😉
Click to expand...

It seems you have no clue as usual.I am in medical imaging and use CUDA on a daily basis.People who really need DP use quadro because they need ECC too.DP without ECC is a joke.DP is not some child's toy which you will showoff at every occasion.
 
Jaydip said:
It seems you have no clue as usual.I am in medical imaging and use CUDA on a daily basis.People who really need DP use quadro because they need ECC too.DP without ECC is a joke.DP is not some child's toy which you will showoff at every occasion.
Click to expand...

Not for my use...again, I hope you are using an IGP right?
 
Lonbjerg said:
Perhaps you should actually read soeme reviews, the anwer is out there:
http://anandtech.com/show/6760/nvidias-geforce-gtx-titan-part-1/2

Even if some poster don't like the GPGPU part of the Titan...because it dosn't fit their "agenda"..and they cannot locked the Titan Down for pure gaming...no matter how much they really, really, really, really, really, really, really, really, really would like to ^^
Click to expand...

I read the reviews. Doesn't answer my question. If you use CUDA so much, why don't you have a quadro?
 
cmdrdredd said:
Something to think about...they need the drivers to make use of any card. They cannot waste 6 months fixing it. Gotta be right there on day one.
Click to expand...

I don't see why they'd suddenly regress in driver quality as long as they are keeping with the GCN architecture. The drivers themselves weren't bad for single card 7000 series launch, just there was untapped performance and room for general driver improvements at launch.
 
I like how this thread is about AMD's upcoming cards and Lonbjerg derailed it into some DP performance argument that no one cares about

What a surprise... Not
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top