Sign in with
Sign up | Sign in
Your question
Closed

AMD Unveils its Heterogeneous Uniform Memory Access (hUMA) Technology

Last response: in News comments
Share
a b U Graphics card
a b à CPUs
April 30, 2013 10:08:15 AM

"Lover development costs as the more efficient architecture enables less people to do the same work" I think we should lower the development costs instead of trying to love it :p 
a b U Graphics card
a b à CPUs
April 30, 2013 10:09:34 AM

The question is: when we will see this in application?
a b U Graphics card
April 30, 2013 10:12:06 AM

"the Radeon 9700 Pro could provide a performance of 31.2 GFLOPS of performance, 5 years later the Radeon HD 2900 XT offered 473.6 GFLOPS and by 2012, the Radeon HD 7970 GHz Edition was capable of computing 4301 GFLOPS - an increase of 13,700% when compared to the HD 2900 XT."
I think you meant compared to the 9700 Pro... The difference between the 7970 and 2900XT is 908%, or 9-times the GFLOP performance.
April 30, 2013 10:17:53 AM

AMD still has some aces in its sleeve. Go AMD, make this a reality, not just a press release.
a b U Graphics card
a b à CPUs
April 30, 2013 10:29:56 AM

"Lover development costs as the more efficient architecture enables less people to do the same work" I think we should lower the development costs instead of trying to love it :p 
a c 1324 U Graphics card
a c 807 à CPUs
April 30, 2013 10:40:19 AM

Certainly tech worth following.
April 30, 2013 10:44:10 AM

As usual, AMD is ahead of the curve in terms of ideas. This is trully a good leap of progress and will set the tone for HPC computing for the next decade. However, they are usually behind the curve in terms of implementing their awesome ideas. AMD get this stuff to market as soon as possible!!!
a b U Graphics card
a b à CPUs
April 30, 2013 10:51:10 AM

So which is it? Heterogeneous or uniform? Non-uniform uniform memory access! Great job naming this!
April 30, 2013 11:00:17 AM

I figured this would happen sometime. It'll be nice when the ususal APIs (OpenGL/Direct3D) support this directly but until then drivers should still be able to optimize for older applications. This tight integration has been needed for a while and should open up alot of new techniques.
April 30, 2013 11:00:52 AM

AMD Unveils its Heterogeneous Uniform Memory Access (hUMA) Technology? Of course... it will be in PS4... the 8 core + GPU + 8Gb of GDDR5...
a c 91 U Graphics card
a b à CPUs
April 30, 2013 11:10:27 AM

long story short, AMD plans to further utilize their superiority in the GPU department over Intel to boost their systems' overall performance, hoping to catch up and surpass intel in the coming years. hmm, lots of hurdles to jump through but sounds promising. nice work and good luck to you AMD!
April 30, 2013 11:15:10 AM

NUMA NUMA....
April 30, 2013 11:24:49 AM

Kaveri (which is based on Steamroller) will basically roll out with this feature.
On top of that... AMD stated that they were focusing on fixing single-threaded performance on Steamroller cores and that with all the modifications, 30% to 40% performance gains in CPU performance should be seen.
So effectively speaking, the 1 module that contains 2 cores won't behave like a single core Intel CPU, but instead will behave like 2 cores i series CPU.
Right now (at least in the mobile area), the A10 4600m is comparable to an entry level i5 Ivy Bridge in terms of overall CPU performance.
If Kaveri CPU performance reflects AMD statements, then quad core Kaveri APU's will basically be comparable to Ivy Bridge/Haswell in CPU power alone... and probably surpass Haswell in IGP performance.
Adding to all of this, seeing how consoles like PS4 and Xbox have adopted AMD x86 hardware, it hUMA will probably find itself in widespread adoption... not to mention utilization in multicore cpu's, and adoption of x86 instruction set (it will be far easier to port games from consoles to PC's... they should be far more optimized compared to what's done today).
And of course, graphics will likely experience a very high jump.
All in all, I'm interested to see what Kaveri brings.
I just hope that OEM's decide to build proper mobile platforms for those APU's instead of putting an A10 4600m into a 17" monster.
Current AMD APU's are ideal for 15" and lower form factors... but their adoption is extremely low to the point of non-existance.
Heck, even AMD's mid-range mobile 7xxx series GPU's cannot be found virtually anywhere.
April 30, 2013 11:26:59 AM

Hmmm. well it is time to AMD really use that GPU advantage it has so far compared to Intel, because in reality it is the only advantage at this moment. Just hope that programmers start allso utilising these. or we really have only one real choise... Intel... We need to have more than that just to keep the competition going. The articel says that CPU developmen has been slow... yep... we should have at least 12 core CPU prosessors at this moment. I mean 12 core Hasvell would be nice to have. In that case the ingrease in compute power would be more in scale with GPU compute power. The GPU type prosessor is better for compute, but because AMD have had production problems with CPU's compared to Intel has lead that CPU have not been used to full extend... We should have had 8 core during the time of phenemo I and II and first intel core CPU and after that 12 and 16 core by now in consumer level... But no... The CPU have become cheaper and cheaper to produce, but they stay the same with 5-10% increase with each production node... *sigh*
If AMD can put some pressure on Intel with heavy use of unified CPU GPU computing we can see somewhat faster advancement allso in CPU department...
April 30, 2013 11:27:38 AM

Respectfully, can you please have a conversation the editor who cleared this article and the person(s) who is (are) responsible for your comments section. Both are not doing their jobs properly.

Please see below:
"In addition to the "Top 10 Reasons" noted in the above slide, AMD cites an additional six benefits that its hUMA technology brings to both developers and consumers:" -> Horrible sentence structure -> Additionally, AMD cites six benefits to its hUMA technology for both developers and consumers.

"Lover development costs as the more efficient architecture enables less people to do the same work" -> Lower development costs come from improved efficiency and enable smaller resources for the same job.

Better experiences through "radically different user experiences" -> What does this mean? Sounds like marketing gibberish.

"Further information on this technology" -> Further information about this technology.

Respectfully, please correct the above.
April 30, 2013 11:30:49 AM

Are the AMD folks Dragonlance fans? Will they name the first hUMA based platform "Dragonbane"?
April 30, 2013 11:31:15 AM

Time will tell whether it's a success or flop, but based off all that hard work AMD put into Fusion it isn't yet paying off. Hopefully this or something can turn AMD around again.
April 30, 2013 11:31:31 AM

Are the AMD folks Dragonlance fans? Will they name the first hUMA based platform "Dragonbane"?
April 30, 2013 11:41:14 AM

Sounds promising.
April 30, 2013 12:26:40 PM

So is this groundbreaking stuff? Like a game changer on how future PCs will be built? Seems that way.
April 30, 2013 12:30:19 PM

I'm glad I didn't buy into the APU architecture already. The idea of integrated graphics on a desktop didn't really appeal to me at first, but with advancements like this, and the fact that you can add discrete cards in addition to the IGP, makes it all look pretty promising.
a b U Graphics card
a b à CPUs
April 30, 2013 1:02:08 PM

They couldn't add one more word, starting with "N"? Maybe "Node?"
Then the acronym would be...
April 30, 2013 1:04:00 PM

WOW, just like what Amiga did years ago when the memory cold be directly accessed from any device.
a b à CPUs
April 30, 2013 1:48:42 PM

Putting all your eggs in one basket? We will see in the next 2 years how it turns out. I seems good, but you never know.
April 30, 2013 1:53:31 PM

slomo4sho said:
So which is it? Heterogeneous or uniform?

Both. It's uniform memory access for heterogeneous architectures. The two aspects are complementary, not mutually exclusive.

a b U Graphics card
a b à CPUs
April 30, 2013 3:32:17 PM

agnickolov said:
Both. It's uniform memory access for heterogeneous architectures. The two aspects are complementary, not mutually exclusive.


I am aware of what it does. I was just commenting on the oxymoronic naming of this technology. They could have done a better job naming it.
April 30, 2013 5:18:27 PM


How ironic, this is in some ways similar to what SGI did with its O2 and VW320/540
workstations, ie. a single pool of memory for all subsystems, though back then they
pushed GE/lighting calculations onto the CPU. Still, for its time, the results were
impressive, with a maxed-out VW320 able to match a GF3, while O2 held the Shake
& Chalice records for several years. Here's an old O2 diagram:

http://sgidepot.co.uk/o2-block-diag-2.gif

and a more detailed version:

http://sgidepot.co.uk/ip32.gif


I hope AMD can pull this off, it should finally give them something to shout
about if the end result shows useful advantages & performance gains just
as SGI did back in the 1990s.

Ian.

PS. For those who post comments and find their text mangled eg. (line
breaks removed), select Tracked Threads from your profile drop-down
menu, click on the thread to which you've just posted, you'll then be on
the .co.uk forums site and can edit your post to reinsert the line breaks,
etc. Just a temporary solution until the site wombles fix all these posting
& forum bugs. :) 

a b U Graphics card
a c 102 à CPUs
April 30, 2013 10:08:04 PM

Sounds a lot like a variant or extension of DirectX's texture streaming where the IGP/GPU can process textures directly from system RAM instead of making a local copy first.
May 1, 2013 2:39:17 AM

If you add an 'N' in the end.. it becomes 'hUMAN'.
May 1, 2013 8:02:53 AM

Does this mean we can see some truly stellar AMD based tablets soon?
August 17, 2013 10:10:55 AM

this sounds good and all, but CPU performance on AMD CPU's need match Intel, as of moment even if HSA+hUMA , it still no chance with Intel SB/IV/HS + discrete top gaming GPU, the brute force of what intel and AMD GPU and NV GPU have out weight what this has to offer, unless AMD CPU/APU can catch intel, and from what i can see use Discrete GPU with HSA+hUMA + the APU, if that is case then this would be future, but it has work with discrete cards as well, they new overhual how Motherboard work and how PCI lanes works as well
!