• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Are new full fat Hawaii XT cards on the way ?

Status
Not open for further replies.
Of course Nvidia have an answer, ridiculous to think otherwise tbh :p. They have been one step ahead each time AMD drop a GPU..

AMD dropped 7970, Nvidia answered with 680 plus
'Nvidia Titan'
then followed with 780 (Took the lead)
AMD answered with 290X
Nvidia followed with 780Ti / Titan Black. (Took the lead)
If AMD drop full fat Hawaii, Nvidia will answer for sure..



Agreed. GTX 790 (Maxwell) 28nm. Plus price drops on GTX 780 / 780 Ti.

AMD dropped 7970, Nvidia answered with 680 plus
AMD sorted the drivers and sold shedloads of HD 7970s and outperformed the GTX 680.

AMD have also sold shedloads of 290Ps

The only thing NVidia could try is a GTX 790 but we already know where that is going as the Titan Z performance is very poor. The thing NVidia could do with a GTX 790 is to watercool it which would make the card err more expensive.:p:cool:

If NVidia really want to compete they need to sell their cards at realistic price points.
 
Heck even the 384bit bus on the Titan Z is showing the strain @4K with the performance dropping right off compared to Hawaii cards.

This would be more down to the lack of ROPS rather than the memory bandwidth. In the 295X2 vs Titan Z comparison the latter has more memory bandwidth (336GB/s vs 320GB/s).

The ROP throughput/Pixel fillrate on the titan Z is theoretically much lower than the 295X2.

~876Mhz x 48 = 42GB/s (x2)
~1000Mhz x 64 = 64GB/s (x2)

Back on topic the release of the "Full fat" 290XTX doesn't really change much at the top end. The 780TI will still end up (marginally) quicker in most review benchmarks, whilst user reviews / more in depth benchmarks will show rough parity across the top end cards (clocked up of course).

Overall, meh.
 
We may be in for a bit of a disappointment in 20mn, just because its a smaller node does not automatically mean its this and that.

the fact that its a smaller node could bring its own problems, it may not clock as well as 28nm, it maybe be more difficult to cool because of its more compact nature....
 
We may be in for a bit of a disappointment in 20mn, just because its a smaller node does not automatically mean its this and that.

the fact that its a smaller node could bring its own problems, it may not clock as well as 28nm, it maybe be more difficult to cool because of its more compact nature....

Can't be much worse than cards running at 95c in any case :p

Pity none of the interviews lately with Richard Huddy asked him about thermals and what amd was going to do to do a better job with regards to heatsink designs and fan noise.
 
This would be more down to the lack of ROPS rather than the memory bandwidth. In the 295X2 vs Titan Z comparison the latter has more memory bandwidth (336GB/s vs 320GB/s).

The ROP throughput/Pixel fillrate on the titan Z is theoretically much lower than the 295X2.

~876Mhz x 48 = 42GB/s (x2)
~1000Mhz x 64 = 64GB/s (x2)

Back on topic the release of the "Full fat" 290XTX doesn't really change much at the top end. The 780TI will still end up (marginally) quicker in most review benchmarks, whilst user reviews / more in depth benchmarks will show rough parity across the top end cards (clocked up of course).

Overall, meh.

You are making my argument for me

How many ROPs is the new NVidia card said to have.

3,200 CUDA cores, TMU's: 200, ROPS: 32, Memory Bus: 256, Architecture: Maxwell.

This card is starting to look like an epic failure at high resolutions and oh that 256bit bus.
 
The trouble with this post, is that if he had a dig at AMD you wouldn't mind at all because I have never seen you leap to the defence of thread when guys dig at AMD. And there is a few guys who constantly do.

That's not the case at all but believe what you want.

LambChop does it often and I pick him up on it.
 
Last edited:
How many ROPs is the new NVidia card said to have.

3,200 CUDA cores, TMU's: 200, ROPS: 32, Memory Bus: 256, Architecture: Maxwell.

This card is starting to look like an epic failure at high resolutions and oh that 256bit bus.
It's pretty much slower than GTX780Ti at above 1920x1080 res guaranteed looking at those specs...
 
They probably were not able to cool it adequately (or were not confident they could do so). This would be much like the GTX 480 which wasn't a full chip either.

My guess is that the success of the AIO cooler on the 295x2 card (when compared to the Titan Z cooler) means that AMD are confident they could cool the full chip with such a cooler.
 
You are making my argument for me

How many ROPs is the new NVidia card said to have.

3,200 CUDA cores, TMU's: 200, ROPS: 32, Memory Bus: 256, Architecture: Maxwell.

This card is starting to look like an epic failure at high resolutions and oh that 256bit bus.

Quite (I wasn't disagreeing with your underlying point BTW).

It makes me a sad panda to see a rumoured 256bit / 32 ROP config on the Maxwell cards. I want true next gen performance, not neutered slightly quicker than current gen in 80% use case.
 
Until we see retail cards, this is all speculation on rumor. Nvidia certainly aren't stupid, whatever AMD release you can be sure Nvidia will respond and most likely with more performance albeit at a higher price.

Whatever comes next, 28nm really is getting long in the tooth...

Want full fat new architecture on 20nm node :D
 
Until we see retail cards, this is all speculation on rumor. Nvidia certainly aren't stupid, whatever AMD release you can be sure Nvidia will respond and most likely with more performance albeit at a higher price.

Whatever comes next, 28nm really is getting long in the tooth...

Want full fat new architecture on 20nm node :D

*Cough* FX 5800 Ultra */Cough*

:p
 
Status
Not open for further replies.
Back
Top Bottom