• 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.

AMD Working On An Entire Range of HBM GPUs To Follow Fiji And Fury Lineup – Has Priority To HBM2 Cap

Soldato
Joined
2 Jan 2012
Posts
12,408
Location
UK.
We’ve received information that AMD engineers are hard at work on a range of GPUs featuring stacked high bandwidth memory to follow Fiji. While Fiji is AMD’s first GPU to feature this new high bandwidth graphics memory standard, it certainly won’t be the last

Read more: http://wccftech.com/amd-working-entire-range-hbm-gpus-follow-fiji-fury-lineup/#ixzz3fnObQqiM

The range will span top to bottom solutions featuring SK Hynix’s stacked high bandwidth memory, HBM for short. This new lineup will also serve as the vehicle by which AMD will introduce this new memory technology to the mobile segment with the company’s first ever set of HBM GPUs designed for notebooks.
 
This might cause some trouble for nvidia, considering their Pascal launch.

But tbh, I like they are pushing more products with HBM tech.
 
HBM is waste of time for low end cards which are likely to be used @1080p where the new type of memory does not do well.

Better to stick with old fashioned GDDR5 which is probably cheaper too.
 
This might cause some trouble for nvidia, considering their Pascal launch.

Er why, NVidia will most likely also have a complete top to bottom HBM2 line up.



Edit: ok after finishing reading the article, the AMD HBM2 priority might indeed cause some issues.

We will have to wait and see.
 
Last edited:
HBM is waste of time for low end cards which are likely to be used @1080p where the new type of memory does not do well.

Better to stick with old fashioned GDDR5 which is probably cheaper too.

Low end amd can offer tiny solutions (Great for notebooks), mid end in notebooks it allows going beyond 256 bit which mobile mxm modules have been limited to.
 
Yes indeed. Can't just look at raw performance in terms of fps here. Powersaving is huge for tablets/laptops.
 
HBM is waste of time for low end cards which are likely to be used @1080p where the new type of memory does not do well.

Better to stick with old fashioned GDDR5 which is probably cheaper too.

AMD"s plan may simply to buy up all the stock to limit Nvidia. If they just purchased stock and left it in a ware-out they might leave themselves open for anti-competition litigation, if they buy up all the stock and shove them in cards that don't really need it then they are relatively safe.

I doubt that is the case because HBM is very expensive o makes no sense for a cheap card. However, at some point AMD have to decide if they want a shared memory controller over the entire range or the lower range has to be yet another architecture with GDDR5.



Anyway, if HBM is supposedly going to be in short supply next year that would explain why FuryX is vaporware and paints a very bad picture for GPUs in 2016.
 
The Fury's HBM not being good for 1080p or lower at the moment, is that possibly down to how AMD are tuning the memory to resolve the 4GB limitation, or do you think it is down to the actual HBM itself.
I seem to remember a discussion a while back about if the bandwidth is there it will work for all uses or something.
 
bru;28307346[B said:
]The Fury's HBM not being good for 1080p or lower at the moment, is that possibly down to how AMD are tuning the memory to resolve the 4GB limitation, or do you think it is down to the actual HBM itself.[/B]
I seem to remember a discussion a while back about if the bandwidth is there it will work for all uses or something.

Its nothing to do with the memory and all too with the driver overhead.
 
HBM is waste of time for low end cards which are likely to be used @1080p where the new type of memory does not do well.

Better to stick with old fashioned GDDR5 which is probably cheaper too.

Nothing to do with the memory. More to do with the driver being unable to feed the card.
 
The Fury's HBM not being good for 1080p or lower at the moment, is that possibly down to how AMD are tuning the memory to resolve the 4GB limitation, or do you think it is down to the actual HBM itself.
I seem to remember a discussion a while back about if the bandwidth is there it will work for all uses or something.

I don;t think there is any reason for HBM to perform worse at 1080p.

People look at the FuryX 1080 P performance and make unfounded allegations including:
* DX11 APi issues, DX12 will cure cancer.
* HBM is slow at 1080p
* Future drivers will bring big benefits.



They always fail to consider the willing options:
* Architectural/design issues limit performance at 1080p, at higher resolution the extra pixel shaders come into their own but at lower resolutions other issues bottle beck the card. E.g., geometry, tessellation, the same 4 shader engines.
* It is not that performance at 1080P is sub-par, but at 4K where bandwidth is frequently a bottleneck, HBM really helps out. This is somewhat like the above point




I can't think of any reason why HBM should perform worse at 1080p than 4K, it makes absolutely no sense.
 
Its nothing to do with the memory and all too with the driver overhead.

I suspect driver overheads in anything but Windows 10 is severely hindering FuryX (and will do the same with any future HBM based cards), just like it does for any games that use a lot of cpu power in combination with current AMD graphics cards.
 
Anyway, if HBM is supposedly going to be in short supply next year that would explain why FuryX is vaporware and paints a very bad picture for GPUs in 2016.
I'm not so sure, I can't see Nvidia releasing a high priced & desirable card & not having any to sell.

They like money too much. :D
 
APU's is the best place for HBM memory, AMD really need to bring out a line of motherboards with a more powerful selection of APU's and unified HBM memory... 4GB isn't going to cut it though.
 
The latencies are still higher in HBM, it's close to the core but much wider and slower. A hybrid setup with a couple of gigs for the IGP and a 128bit DDR4 interface would be interesting.
 
Back
Top Bottom