Corsair QL120 and QL140 RGB fan review
Promo: Windows 10 Pro for $13 With Office 2016 For $33
Corsair Void RGB Elite Wireless Headset review
Team Group PD400 Portable SSD review
AMD Athlon 3000G review
Team Group T-Force Delta Max 1 TB SSD review
Guru3D Rig of the Month - November 2019
ASUS ROG Rampage VI Extreme Encore review
Toshiba RC500 500GB NVMe M.2 SSD review
Promo: URcdkey Black Friday Windows 10 Pro for $12
Rumor: GeForce GTX 1080 8GB gddr5x based on Pascal GPU in May
Well, at least I have a nice juicy rumor just in time for the weekend. Asian based Benchlife claims that Nvidia will release a Pascal architecture based GeForce GTX 1080 with 8GB gddr5x in May. It would make sense as HBM2 will be available in volume only later in the year.
There's a precise date being mentioned as well, May 27th, four days prior to the Computex exhibition. Obviously Nvidia is not going to comment on this rumor.
The GP104 based GTX 1080 would get a 256-bit memory controller tied to 8GB of gddr5x memory. The usage of gddr5x over regular gddr5 is rather speculative as it isn't even in volume production that we know of? Micron (the only one that will fab these gddr5x ICs) announced in a recent blog post that they intend to commence high-volume production of GDDR5X chips in mid-2016, or sometime in the summer. But who knows, Nvidia have gotten their hands on an early batch for a couple of thousand cards right ? GDDR5X is your standard GDDR5 memory however, opposed to delivering 32 byte/access to the memory cells, this is doubled up towards 64 byte/access. And that in theory could double up graphics card memory bandwidth.
Early indications according to a presentation show numbers with the memory capable of doing up-to 10 to 12 Gbps, and in the future 16 Gbps. So your high-end graphics cards these days hover at say 400 GB/s. With GDDR5X that could increase to 800~1000 GB/sec and thus these are very significant improvements, actually they are competitive enough with HBM.
Pascal GPUs are fabbed in the Taiwan Semiconductor Manufacturing Company (TSMC) (recently hit by an earthquake) based on the new 16nm FinFET process. 16nm should result in significant power savings. The card would only need one 8-pin connector for power needs. The card would get a DVI, HDMI and two display ports. From the looks of things, and as expected .. this is not 'Big Pascal' GP100. That GPU would get 17 billion transistors made on the 16 nm FinFET+ node at TSMC and likely the HBM2 memory.
So to recap, in the latest prognosis Nvidia are rumored to launch the first high-end GP104 GPUs with models GTX 1080 and GTX 1070 (should Nvidia use this naming scheme). These chips would come in a 37.5×37.5mm BGA package, and get 2152 pins.
Lowice
Member
Posts: 45
Joined: 2014-01-24
Member
Posts: 45
Joined: 2014-01-24
#5243512 Posted on: 03/11/2016 12:00 PM
But how would it work in DX12 compaired to AMD
But how would it work in DX12 compaired to AMD
xIcarus
Senior Member
Posts: 941
Joined: 2010-08-24
Senior Member
Posts: 941
Joined: 2010-08-24
#5243514 Posted on: 03/11/2016 12:02 PM
Hah first?
Pretty much what I expected. Except the 256bit bus... not sure this is a good idea, although I don't mind GDDR5X for the cards at all, 8GB should be more than sufficient.
At least it meets the time frame I expected for first launches, let's see how things turn out.
Let's just hope this time there won't be any 224-bit bus
.
But seriously though, a 256-bit bus could be enough if GDDR5X is aimed towards speed. I mean a 15GHz GDDR5 memory suddenly doubles what we currently have.
And it's not that hard to achieve even if it sounds like a lot, remember - GDDR5 reaches around 7GHz _effective_ data rate since it's a QDR-type memory. If they somehow managed to make it octa-pumped instead of quad-pumped a 5GHz memory today becomes a 10GHz memory tomorrow.
That's a possibility.
Hah first?
Pretty much what I expected. Except the 256bit bus... not sure this is a good idea, although I don't mind GDDR5X for the cards at all, 8GB should be more than sufficient.
At least it meets the time frame I expected for first launches, let's see how things turn out.
Let's just hope this time there won't be any 224-bit bus
.
But seriously though, a 256-bit bus could be enough if GDDR5X is aimed towards speed. I mean a 15GHz GDDR5 memory suddenly doubles what we currently have.
And it's not that hard to achieve even if it sounds like a lot, remember - GDDR5 reaches around 7GHz _effective_ data rate since it's a QDR-type memory. If they somehow managed to make it octa-pumped instead of quad-pumped a 5GHz memory today becomes a 10GHz memory tomorrow.
That's a possibility.
DribbelDog
Junior Member
Posts: 15
Joined: 2012-12-31
Junior Member
Posts: 15
Joined: 2012-12-31
#5243516 Posted on: 03/11/2016 12:05 PM
I really hope this is true.
Especially since NVIDIA seems to have started nerfing the 7xx series with (lack of) driver support. In some benchmarks the 780ti seems to perform worse than the 970, which worries me... I thought I'd be fine skipping the 9xx series, but I'm slightly disappointed by how my 780ti's are holding up. Too late to jump on the 9xx bandwagon now, so can't wait for something new to come out.
I really hope this is true.
Especially since NVIDIA seems to have started nerfing the 7xx series with (lack of) driver support. In some benchmarks the 780ti seems to perform worse than the 970, which worries me... I thought I'd be fine skipping the 9xx series, but I'm slightly disappointed by how my 780ti's are holding up. Too late to jump on the 9xx bandwagon now, so can't wait for something new to come out.
Ziggymac
Senior Member
Posts: 109
Joined: 2015-01-22
Senior Member
Posts: 109
Joined: 2015-01-22
#5243522 Posted on: 03/11/2016 12:12 PM
Ebay is about to get flooded with used 980's and 970's...
Ebay is about to get flooded with used 980's and 970's...
Lane
Senior Member
Posts: 6361
Joined: 2005-02-25
Senior Member
Posts: 6361
Joined: 2005-02-25
#5243523 Posted on: 03/11/2016 12:13 PM
Hah first?
Pretty much what I expected. Except the 256bit bus... not sure this is a good idea, although I don't mind GDDR5X for the cards at all, 8GB should be more than sufficient.
At least it meets the time frame I expected for first launches, let's see how things turn out.
Im really not sure about this story of GDDR5x who run since some times.....
Mass production of HBM2 ( 16-32GB ) should happend in summer, but tthis dont mean theres not allready HBM2 4-8GB in production.
HBM require ( as seen on Fury ) a complete different setup on the silicon side: Interposer, Different memory controller, different layout on the cache side.
Why make 2 completely different chips layout, when with HBM2 you have the possibility to got 4-8-12-16-32GB ? Specially as x104 cost not the same price as high end 4 years ago. ( 500-600 $ ) ..
Ofc the lineup will split at a moment or another, HBM are too costly for low end gpu's.. But so GP104 and GP100 will be quite different. Even on Power efficiency... I can be wrong but i imagine this split only at the GP106 level. ( for GP106 and under ).... Not between GP104-GP100. Witthout saying that GDDr5x is surely not ready yet for mass production.
Will be funny to see Nvidia reserve HBM2 for the 1000$ " Titan " 16GB GPU`s ( followed then by a 1080Ti 12GB HBM2 ).
Funny but look exactly how Nividia is selling lineup anyway
Tesla GP100: 32GB HBM2
Titan GP100: 16GB HBM2
Geforce 1080TI: 12GB HBM2
Geforce 1080 : 8GB ( GDDR5x or HBM2 )
Geforce 1050TI : 4GB GDDR5
Hah first?
Pretty much what I expected. Except the 256bit bus... not sure this is a good idea, although I don't mind GDDR5X for the cards at all, 8GB should be more than sufficient.
At least it meets the time frame I expected for first launches, let's see how things turn out.
Im really not sure about this story of GDDR5x who run since some times.....
Mass production of HBM2 ( 16-32GB ) should happend in summer, but tthis dont mean theres not allready HBM2 4-8GB in production.
HBM require ( as seen on Fury ) a complete different setup on the silicon side: Interposer, Different memory controller, different layout on the cache side.
Why make 2 completely different chips layout, when with HBM2 you have the possibility to got 4-8-12-16-32GB ? Specially as x104 cost not the same price as high end 4 years ago. ( 500-600 $ ) ..
Ofc the lineup will split at a moment or another, HBM are too costly for low end gpu's.. But so GP104 and GP100 will be quite different. Even on Power efficiency... I can be wrong but i imagine this split only at the GP106 level. ( for GP106 and under ).... Not between GP104-GP100. Witthout saying that GDDr5x is surely not ready yet for mass production.
Will be funny to see Nvidia reserve HBM2 for the 1000$ " Titan " 16GB GPU`s ( followed then by a 1080Ti 12GB HBM2 ).
Funny but look exactly how Nividia is selling lineup anyway
Tesla GP100: 32GB HBM2
Titan GP100: 16GB HBM2
Geforce 1080TI: 12GB HBM2
Geforce 1080 : 8GB ( GDDR5x or HBM2 )
Geforce 1050TI : 4GB GDDR5
Kaotik
Senior Member
Posts: 140
Joined: 2004-12-24
Senior Member
Posts: 140
Joined: 2004-12-24
#5243527 Posted on: 03/11/2016 12:24 PM
Micron, only manufacturer of GDDR5X, isn't starting mass production of GDDR5X until summer (August mentioned earlier). Thus, card with GDDR5X in May is simply impossible.
Asian based Benchlife claims that Nvidia will release a Pascal architecture basedÂ*GeForce GTX 1080 with 8GB gddr5x in May.
Micron, only manufacturer of GDDR5X, isn't starting mass production of GDDR5X until summer (August mentioned earlier). Thus, card with GDDR5X in May is simply impossible.
Kaotik
Senior Member
Posts: 140
Joined: 2004-12-24
Senior Member
Posts: 140
Joined: 2004-12-24
#5243528 Posted on: 03/11/2016 12:26 PM
Mass production of HBM2 ( 16-32GB ) should happend in summer, but tthis dont mean theres not allready HBM2 4-8GB in production.
HBM2 4GB stacks (used for 16GB / 4096bit) are in mass production already, 8GB (used for 32GB / 4096bit) won't be until late this year, probably end of Q3/early Q4.
1GB HBM2 stacks aren't even planned, 2GB HBM2 stacks haven't been announced to be in production yet.
Why make 2 completely different chips layout, when with HBM2 you have the possibility to got 4-8-12-16-32GB ? Specially as x104 cost not the same price as high end 4 years ago. ( 500-600 $ ) ..
Different chips, AMD has already confirmed they can accommodate both GDDR5 and HBM markets with Polaris as needed
Mass production of HBM2 ( 16-32GB ) should happend in summer, but tthis dont mean theres not allready HBM2 4-8GB in production.
HBM2 4GB stacks (used for 16GB / 4096bit) are in mass production already, 8GB (used for 32GB / 4096bit) won't be until late this year, probably end of Q3/early Q4.
1GB HBM2 stacks aren't even planned, 2GB HBM2 stacks haven't been announced to be in production yet.
Why make 2 completely different chips layout, when with HBM2 you have the possibility to got 4-8-12-16-32GB ? Specially as x104 cost not the same price as high end 4 years ago. ( 500-600 $ ) ..
Different chips, AMD has already confirmed they can accommodate both GDDR5 and HBM markets with Polaris as needed
TheDeeGee
Senior Member
Posts: 6235
Joined: 2010-08-28
Senior Member
Posts: 6235
Joined: 2010-08-28
#5243530 Posted on: 03/11/2016 12:32 PM
Hmmz, a bit later than i hoped (Overwatch/Legion).
Apart from that i don't mind it having GDDR5(X), this might be in favour of my favorite GPU Coolers from Arctic, as it's probably compatible out of the box.
Hmmz, a bit later than i hoped (Overwatch/Legion).
Apart from that i don't mind it having GDDR5(X), this might be in favour of my favorite GPU Coolers from Arctic, as it's probably compatible out of the box.
Mtom
Member
Posts: 26
Joined: 2015-02-09
Member
Posts: 26
Joined: 2015-02-09
#5243531 Posted on: 03/11/2016 12:33 PM
Yeah GDD5X based Pascal in May...when GDDR5X mass production starts at summer...seems legit.
Yeah GDD5X based Pascal in May...when GDDR5X mass production starts at summer...seems legit.
cowie
Senior Member
Posts: 13197
Joined: 2005-03-30
Senior Member
Posts: 13197
Joined: 2005-03-30
#5243535 Posted on: 03/11/2016 12:39 PM
stacks (used for 16GB / 4096bit) are in mass production already, 8GB (used for 32GB / 4096bit) won't be until late this year, probably end of Q3/early Q4.
1GB HBM2 stacks aren't even planned, 2GB HBM2 stacks haven't been announced to be in production yet.
Different chips, AMD has already confirmed they can accommodate both GDDR5 and HBM markets with Polaris as needed
SK Hynix hits mass production of 4GB HBM2 in Q3, 8GB HBM2 in Q4 2016
when they are ready and usable is another story.
I hope they out bench titan x and heavy clocked 980ti's I don't care too much about dx12 yet
from micron
Micron is currently ramping GDDR5X to mass production, and will be announcing sample dates later this spring. We plan to be in full volume production this summer.
you know just thinking it would be really nice if this was a
"between" card like 980-980ti and cheaper like 400-450 or so and not a high priced pos ref only card.
stacks (used for 16GB / 4096bit) are in mass production already, 8GB (used for 32GB / 4096bit) won't be until late this year, probably end of Q3/early Q4.
1GB HBM2 stacks aren't even planned, 2GB HBM2 stacks haven't been announced to be in production yet.
Different chips, AMD has already confirmed they can accommodate both GDDR5 and HBM markets with Polaris as needed
SK Hynix hits mass production of 4GB HBM2 in Q3, 8GB HBM2 in Q4 2016
when they are ready and usable is another story.
I hope they out bench titan x and heavy clocked 980ti's I don't care too much about dx12 yet
from micron
Micron is currently ramping GDDR5X to mass production, and will be announcing sample dates later this spring. We plan to be in full volume production this summer.
you know just thinking it would be really nice if this was a
"between" card like 980-980ti and cheaper like 400-450 or so and not a high priced pos ref only card.
RavenMaster
Senior Member
Posts: 1076
Joined: 2009-08-19
Senior Member
Posts: 1076
Joined: 2009-08-19
#5243540 Posted on: 03/11/2016 12:56 PM
Ha ha so true! I just sold both my Titan X cards before they depreciated in value. I'm using a EVGA 970 as a tide-me-over card until i can buy 2x 1080's.
Ebay is about to get flooded with used 980's and 970's...
Ha ha so true! I just sold both my Titan X cards before they depreciated in value. I'm using a EVGA 970 as a tide-me-over card until i can buy 2x 1080's.
Darren Hodgson
Senior Member
Posts: 15830
Joined: 2004-08-18
Senior Member
Posts: 15830
Joined: 2004-08-18
#5243552 Posted on: 03/11/2016 01:41 PM
Definitely want one though I definitely don't need one either!
Will this card support DX12's asynchronous compute?
The GTX 980 Ti is still a superb card even with the upgrade in resolution from 1920x1200 to 2560x1440. I am only interested in single GPU configurations (SLI and CFX are too annoying to deal with and sparsely supported by developers to even interest me these days) so I prefer to opt for a top of the range card that doesn't have Titan in its name (since those are poor value in my humble view). Fortunately, multi-monitor setups and VR do not interest me, which keeps my budget down to a reasonable level. Also 60 fps is all I need so I do not need the additional horsepower for higher framerates either. Maxed out settings at 2560x1440 with decent anti-aliasing and ideally a locked 60 fps are what I want and hopefully these new Pascal cards will deliver.
Whether I buy one at launch is debatable though. I bought the GTX 980 at launch and that was followed by the far superior GTX 980 Ti so common sense is telling me that I would be better waiting for the next cards after these.
Definitely want one though I definitely don't need one either!
Will this card support DX12's asynchronous compute?
The GTX 980 Ti is still a superb card even with the upgrade in resolution from 1920x1200 to 2560x1440. I am only interested in single GPU configurations (SLI and CFX are too annoying to deal with and sparsely supported by developers to even interest me these days) so I prefer to opt for a top of the range card that doesn't have Titan in its name (since those are poor value in my humble view). Fortunately, multi-monitor setups and VR do not interest me, which keeps my budget down to a reasonable level. Also 60 fps is all I need so I do not need the additional horsepower for higher framerates either. Maxed out settings at 2560x1440 with decent anti-aliasing and ideally a locked 60 fps are what I want and hopefully these new Pascal cards will deliver.
Whether I buy one at launch is debatable though. I bought the GTX 980 at launch and that was followed by the far superior GTX 980 Ti so common sense is telling me that I would be better waiting for the next cards after these.
BangTail
Senior Member
Posts: 3590
Joined: 2006-10-15
Senior Member
Posts: 3590
Joined: 2006-10-15
#5243555 Posted on: 03/11/2016 02:00 PM
Probably best to wait until Nvidia puts all its cards on the table (of course that can change very quickly on launch day
)
I too bought 980s and then 980Tis but I prefer to get one set of cards per generation so I might hold out for the HBM SKUs as these Ti cards are still making short work of everything.
Probably best to wait until Nvidia puts all its cards on the table (of course that can change very quickly on launch day
)
I too bought 980s and then 980Tis but I prefer to get one set of cards per generation so I might hold out for the HBM SKUs as these Ti cards are still making short work of everything.
-Tj-
Senior Member
Posts: 16437
Joined: 2012-05-18
Senior Member
Posts: 16437
Joined: 2012-05-18
#5243556 Posted on: 03/11/2016 02:01 PM
@Darren
That would be a stupid move, full gp100 might be better with 10-12tflops (980ti 1400mhz has 8.2 -8.4tflops). This gp104 wont be fast enough, even 8gb is not worth mentioning.
And async will be the same as now, in cuda. Im pretty sure about that.
If gp100 is only 30-40% faster from oc'ed gm200 according to tflop difference then i'll wait for Volta, screw pascal.
@Darren
That would be a stupid move, full gp100 might be better with 10-12tflops (980ti 1400mhz has 8.2 -8.4tflops). This gp104 wont be fast enough, even 8gb is not worth mentioning.
And async will be the same as now, in cuda. Im pretty sure about that.
If gp100 is only 30-40% faster from oc'ed gm200 according to tflop difference then i'll wait for Volta, screw pascal.
Click here to post a comment for this news story on the message forum.
Senior Member
Posts: 11021
Joined: 2014-07-21
Hah first?
Pretty much what I expected. Except the 256bit bus... not sure this is a good idea, although I don't mind GDDR5X for the cards at all, 8GB should be more than sufficient.
At least it meets the time frame I expected for first launches, let's see how things turn out.