I agree and I think they will release a faster card, no reason looking at specs that is shouldn't be 15% faster but we will have to wait and see.
-
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.
Possible Radeon 390X / 390 and 380X Spec / Benchmark (do not hotlink images!!!!!!)
- Thread starter Boomstick777
- Start date
- Status
Permabanned
That's GCN, not GNC - Graphics Core Next.
My bet is that it'll be surprisingly good with current API's (has to be with that core count) and absolutely scream at lower level API based stuff. At the end of the day it's stepping off the GDDR safety zone into the next standard so there are a lot of unknowns.
I hope that the 16th onwards is a good day for all gamers regardless of team, but until the reviews hit it's all a bit pointless really.
My bet is that it'll be surprisingly good with current API's (has to be with that core count) and absolutely scream at lower level API based stuff. At the end of the day it's stepping off the GDDR safety zone into the next standard so there are a lot of unknowns.
I hope that the 16th onwards is a good day for all gamers regardless of team, but until the reviews hit it's all a bit pointless really.
Soldato
Il say 100% equal to Titan x and beating TI
Associate
- Joined
- 9 Jul 2013
- Posts
- 2,186
I don't think they do need a card to beat a titan x to get there sales up or make them relevant. Titan x is a very niche market. If they can make a card around ti performance which is cooler quieter at £100 less they will sell like hot cakes.
Soldato
But if after all this time they only manage to get Ti performance the people that criticised the 900 series for the small performance leap due to efficiency improvements will likely go nuts.
I mean these cards have been in development for quite a lot longer than the 900 series and don't seem to offer the efficiency benefits. How could they possibly be happy with the same thing they complained about Nvidia doing, but without the upside Nvidia had in a smaller timescale?
I mean these cards have been in development for quite a lot longer than the 900 series and don't seem to offer the efficiency benefits. How could they possibly be happy with the same thing they complained about Nvidia doing, but without the upside Nvidia had in a smaller timescale?
Permabanned
They've taken the efficiency budget offered by HBM and spent it on the GPU, Jesus H, it's only been repeated a few dozen times.
The fury will eat about the same as a 290X for about or over the performance of the TX. On 4GB.
If i'm wrong I'll say sorry I'm wrong and you'll buy the card of your choice.
Not ******* hard is it?
The fury will eat about the same as a 290X for about or over the performance of the TX. On 4GB.
If i'm wrong I'll say sorry I'm wrong and you'll buy the card of your choice.
Not ******* hard is it?
Exactly what I've said earlier, much of the latency comes from the PCI-E bus, main memory etc, no amount of fancy GPU memory technology is going to get around that. You can be as quick as you like at emptying and filling buckets but if the guy handing the buckets to you takes an age what does it matter?
Why do people think things are moving towards unified memory architecture? PC's are a long way behind consoles at the moment in terms of the whole memory subsystem, they're still just brute forcing a decades old inefficient architecture.
Last edited:
I don't understand what you are trying to say. Why should releasing or not releasing a chip (of whatever arbitrary performance level) 6-9 months ago dictate that todays chip should be worse or better?
If they are matching performance at the same power consumption then it is offering efficiency improvements over previous gen.
In the grand scheme of time, R&D for big die Fiji for all intents is completing at the same time as big die GM200.
Last edited:
Caporegime
It will be interesting to see if AMD has had to removed much of the DP compute ability with Fiji,as that is what Nvidia has done with the GM204 and GM210,to improve gaming effiency and stuck with a mildly reworked GK110 for its higher tier markets.
If AMD has still kept considerable DP ability with Fiji,then I expect it will be less efficient in terms of power consumption than the GM210 as it is still trying to cover two different markets.
It will be very interesting to see what they decided and how they want to approach the diff markets. In some ways I want them to have kept DP as I would love them to make headway in HPC with large contracts and firepro margins in the future. It would also improve their lead even further over the reworked kepler GK210 in DP. But that may only serve to make Fiji a good showcase to attract future custom rather than necessarily making it as great a hit now.
Wasn't there are a estimation somewhere that Fiji die was around 520mm? Think it was a estimation from pictures based of the size HBM dies (god knows what the cumulative error is going to be there). If that is so it seems small so maybe reduced DP? But I really have no idea how much die area is saved moving the memory controller under the HBM stack or what sort of bus is being used to shuttle data within the gpu die and what the area implications are for that. Excited to find out.
Last edited:
Caporegime
You can calculate the area yourself by taking the known dimensions of a HBM stack (5x7mm) and going from there. By my calculations it's 500-520mm2.
As far as DP goes now would be a good time to introduce a new Firepro with record FLOPs. NV won't have anything new for a year so perhaps they can make some headway in workstation/HPC. Ol' Brer Rabbit gon' get where he need to, by and by.
As far as DP goes now would be a good time to introduce a new Firepro with record FLOPs. NV won't have anything new for a year so perhaps they can make some headway in workstation/HPC. Ol' Brer Rabbit gon' get where he need to, by and by.
Last edited:
HBM is 5.48x7.29mm.
No way that die is smaller than 560mm2. It's most likely closer to 600mm2.
Soldato
I think if Titan has proven anything its that people will buy something just to get the best performance (exception being the Titan z which was a novelty imo), Nvidia has received recognised this and capitalised but AMD have been asleep at the wheel. If ATI were still independent I doubt Nvidia would be in such a dominant position.
I think AMD made a mistake with settling as slightly slower and cheaper alternative. It worked for a while but in the long run it destroyed their image.
I really hope that changes and they always go for the kill no matter what it takes. Just like Nvidia.
The 980 was only a small performance increase over the 780ti. The 980ti is a much larger increase, though it only released last week..........
You seem to be forgetting the 980ti didn't launch the same time as the 980.
Soldato
Ah, so you're comparing the 980 to the 780Ti not the 780?
Fair enough, I'll compare the 390X to the 295x2 then if we're not doing like-for-like.
See how much of an increase it is...
- Status