[TechReport] AMD's Carrizo will face off with Broadwell-U, won't land in desktops - Page 13 - Overclock.net

Forum Jump: 
Reply
 
Thread Tools
post #121 of 209 Old 01-13-2015, 10:57 AM
PC Gamer
 
SpeedyVT's Avatar
 
Join Date: Jan 2009
Posts: 4,885
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 140
Quote:
Originally Posted by epic1337 View Post

no not really, HSA doesn't work well with typical tasks, the CPU is still over-burdened with a lot of things, otherwise we'd be seeing AMD ditch CPU and stick in some ARM cores instead.
to point out, i5 still outperform anything AMD FM2+ on both minimum and maximum frame-rate using the same discreet GPU.

in general, i've never seen anything AMD FM2+ outperform an i5 on any tasks, so long as its not IGP vs IGP.

They are building 2016 cpus with both ARM and x86 on the same chip. CPUs are often taxed by pure math operations and a GPU can accelerate that. However nothing yet. nothing major has been compiled for HSA.

SpeedyVT is offline  
Sponsored Links
Advertisement
 
post #122 of 209 Old 01-13-2015, 11:25 AM
Not just position, SUPERposition!
 
maarten12100's Avatar
 
Join Date: Aug 2012
Location: Hillegom, Netherlands
Posts: 10,254
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 378
Quote:
Originally Posted by Pro3ootector View Post

There is so many things going on now inside this company, they are changing all their business model, executives leaving.. . Possible takeover?
Nope backed by Arabs.


I think this is Carrizo: (actually I'm sure because they say they went from SB(FCH) + APU to SoC)
EDIT:
I was wrong it is not Carrizo the G-series should be Carrizo-L or Nolan/Amur:
maarten12100 is offline  
post #123 of 209 Old 01-13-2015, 01:03 PM
Overclocker in Training
 
iSlayer's Avatar
 
Join Date: Jun 2009
Location: Chitown
Posts: 4,112
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 144
Quote:
Originally Posted by ebduncan View Post

carrizo won;t be 3.5ghz all the time, it will be only single thread probably.

My richland laptop is the same way. has a 3.5ghz turbo, but all the cores only run at 2.3ghz when they are loaded.

Anyway Carrizo mobile won't really be anything special. It will just be a general improvement from kaveri. Which was nothing to write home about.

HSA definitely shows promise if it can ever take flight.
I like the concept and I like the direction but HSA just isn't designed to take flight. Just about everything conspires against it. Even the main concept.
Quote:
Originally Posted by Nintendo Maniac 64 View Post

It'd probably be more practical once Zen comes around since HSA main benefit seems to be in productivity-based things, though games could possibly take advantage of it.

Speaking of games, any future console design wins would definitely take advantage of it, and it's even thought the PS4 and XB1 have some lesser HSA-like functionality already. I know that Nintendo in particular seems to love memory bandwidth when it comes to how they design their console hardware, so HSA would be right up their ally.
HSA doesn't exactly help with memory bandwidth though tongue.gif.
Quote:
Originally Posted by epic1337 View Post

java partly uses GPGPU, one could then imagine how vast java had already spread.
its not really a joke, nor was it meant to be offensive, it was simply a point-out at how GPGPU had already been preferred over general compute.
Which is already know but, consider this. What high performance software is done in, of all things, Java? Especially GPGPU?

GPGPU in Java is like adding a turbo charger to a golf cart.
Quote:
webpage (browser) rendering are GPGPU heavy (hw acceleration, of course you could disable some of it
Yah my HD 4000 comes to its knees rendering tabs in Chrome. And by knees I mean it wonders if that was supposed to be some kind of challenge.
Quote:
likewise video decoding and database operations had recently preferred GPGPU.
The former doesn't need more performance so much as it needs better efficiency. The latter? Databases really aren't my kinda thing, but it'll depend on the database and operations and whether they can be done in parallel.

You see, GPGPU is great but not everything can just encorporate GPGPU.
Quote:
i hate to quote wiki for context but...
http://en.wikipedia.org/wiki/Heterogeneous_System_Architecture
while HUMA functions as a bridge to join two different processors to function on the same ram interface, HSA functions to allow direct access without needing allocated space.
basically put, hUMA without HSA falls back to being NUMA.
http://www.pcper.com/reviews/Processors/AMD-Details-hUMA-HSA-Action


and on that site says this, in a sense they can implement HSA on discreet GPUs, just that its not practical in their views.
edit: on that note i think i get where you're getting at, i do understand HSA that this is only one part of it's function
I think you're getting closer but my main point is that GPGPU by traditional means (dGPU) far outclasses what an APU is capable of.

That GPGPU is best utilized in professional work as you are showing, where professional GPUs are faster, more accurate, more efficient and better supported.

That AMD is in NO position to dictate the market or see the spread of HSA in desktop applications due to the necessity of specialized hardware that ranks among the minority and that development requires a special compiler system.

HSA is sort of like a stepping stone in research, not important for immediate use, but as a gateway to things that are.

There doesn't seem to be any nice solution to the faults in current GPGPU using systems.
Quote:
HSA main function is to make GPGPU accessible as a general compute core like CPUs.
AMD, Nvidia, people in enterprise, consumers, etc... WISH that were the case.

HSA isn't even close to doing that. It wishes to use them as fully as GPGPU can but it doesn't make any breakthroughs there.
Quote:
Originally Posted by SpeedyVT View Post

They are building 2016 cpus with both ARM and x86 on the same chip. CPUs are often taxed by pure math operations and a GPU can accelerate that. However nothing yet. nothing major has been compiled for HSA.
You mean MANY math operations which can be run in parallel. Quick and dirty example, matrix multiplication.
iSlayer is offline  
Sponsored Links
Advertisement
 
post #124 of 209 Old 01-13-2015, 02:33 PM
Audiophile
 
Join Date: Aug 2010
Location: Northeast Ohio
Posts: 2,077
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 77
Quote:
Originally Posted by iSlayer View Post

HSA doesn't exactly help with memory bandwidth though tongue.gif.
It...doesn't? wth.gif

OK, either you're saying that it's not HSA but hUMA that does, or you're being technical and saying the bandwidth isn't improved but rather the actual memory speed and throughput is (like as in latency).

....either that or I've been a derp for the last couple years or so in my understanding of HSA/hUMA.

Overclock.net's resident straight-male kuutsundere
(because just "tsundere" makes people think of "Shana clones" *shivers*)

Nintendo Maniac 64 is offline  
post #125 of 209 Old 01-13-2015, 02:41 PM
PC Gamer
 
SpeedyVT's Avatar
 
Join Date: Jan 2009
Posts: 4,885
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 140
Quote:
Originally Posted by iSlayer View Post

HSA isn't even close to doing that. It wishes to use them as fully as GPGPU can but it doesn't make any breakthroughs there.
You mean MANY math operations which can be run in parallel. Quick and dirty example, matrix multiplication.

While I disagree with you most of the time, this is true. Yes parallel, even it'll get to the point of non-parallelism and isolate the difference in each one. Really what we need is a single core processor with many sub cores the single core doesn't act as a real processing core, but a task core. I'd love to see that. Imagine a 4 task core 4 sub core processor, that'd be gorgeous. We use a sorting method to handle the different threads but imagine a dedicate core to sorting, that'd be gorgeous.

SpeedyVT is offline  
post #126 of 209 Old 01-13-2015, 02:43 PM
Not just position, SUPERposition!
 
maarten12100's Avatar
 
Join Date: Aug 2012
Location: Hillegom, Netherlands
Posts: 10,254
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 378
Quote:
Originally Posted by Nintendo Maniac 64 View Post

It...doesn't? wth.gif

OK, either you're saying that it's not HSA but hUMA that does, or you're being technical and saying the bandwidth isn't improved but rather the actual memory speed and throughput is (like as in latency).

....either that or I've been a derp for the last couple years or so in my understanding of HSA/hUMA.
He means that Kaveri is horribly bottlenecked by DDR3 in terms of gpu which is true. Despite the unified memory (which only means cpu and gpu can access the same memory space thus it doesn't help)
there is still a huge bandwidth limitation.
maarten12100 is offline  
post #127 of 209 Old 01-13-2015, 02:47 PM
Audiophile
 
Join Date: Aug 2010
Location: Northeast Ohio
Posts: 2,077
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 77
Quote:
Originally Posted by maarten12100 View Post

He means that Kaveri is horribly bottlenecked by DDR3 in terms of gpu which is true. Despite the unified memory (which only means cpu and gpu can access the same memory space thus it doesn't help)
there is still a huge bandwidth limitation.
That's great and all but...

1. Kaveri =/= HSA, especially when you consider that his comment was in response to my comment saying that HSA would be right up Nintendo's ally for any new game systems.

2. Wouldn't that still provide even the teeniest amount of increase, like maybe 1%?

Overclock.net's resident straight-male kuutsundere
(because just "tsundere" makes people think of "Shana clones" *shivers*)

Nintendo Maniac 64 is offline  
post #128 of 209 Old 01-13-2015, 03:45 PM
Not just position, SUPERposition!
 
maarten12100's Avatar
 
Join Date: Aug 2012
Location: Hillegom, Netherlands
Posts: 10,254
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 378
Quote:
Originally Posted by Nintendo Maniac 64 View Post

That's great and all but...

1. Kaveri =/= HSA, especially when you consider that his comment was in response to my comment saying that HSA would be right up Nintendo's ally for any new game systems.

2. Wouldn't that still provide even the teeniest amount of increase, like maybe 1%?
Lower latency would increase troughput a very minor amount. Just like with Internet the latency is the time to wait before a connection is made while your bandwidth it the amount of data/unit of time you can get in. (or send out)

data/time > data/(time+ping)
While this limit approaches data/time = data/(time+ping) for time is a very big number there is still a minor gain in throughput(bandwidth)

So yes lower latency helps bandwidth just not that much generally speaking. I have no idea what Nintendo is planning.
maarten12100 is offline  
post #129 of 209 Old 01-13-2015, 04:09 PM
PC Gamer
 
SpeedyVT's Avatar
 
Join Date: Jan 2009
Posts: 4,885
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 140
Quote:
Originally Posted by maarten12100 View Post

Lower latency would increase troughput a very minor amount. Just like with Internet the latency is the time to wait before a connection is made while your bandwidth it the amount of data/unit of time you can get in. (or send out)

data/time > data/(time+ping)
While this limit approaches data/time = data/(time+ping) for time is a very big number there is still a minor gain in throughput(bandwidth)

So yes lower latency helps bandwidth just not that much generally speaking. I have no idea what Nintendo is planning.

Also hUMA would reduce resource usage which improves bandwidth usage doesn't increase it at all but gets more out of it. HSA would allow typical x86 threads to be processed use by GCN cores as their highly programmable and can be tailored to a developers needs where Maxwell or other NVidia cores are constrained by NVidia's licensing terms. AMD is more appealing to consoles because of this.

SpeedyVT is offline  
post #130 of 209 Old 01-13-2015, 04:47 PM
New to Overclock.net
 
Join Date: Feb 2011
Posts: 6,364
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 5 Post(s)
Liked: 215
Quote:
Originally Posted by iSlayer View Post

Warning: Spoiler! (Click to show)
Quote:
Originally Posted by ebduncan View Post

carrizo won;t be 3.5ghz all the time, it will be only single thread probably.

My richland laptop is the same way. has a 3.5ghz turbo, but all the cores only run at 2.3ghz when they are loaded.

Anyway Carrizo mobile won't really be anything special. It will just be a general improvement from kaveri. Which was nothing to write home about.

HSA definitely shows promise if it can ever take flight.
I like the concept and I like the direction but HSA just isn't designed to take flight. Just about everything conspires against it. Even the main concept.
Quote:
Originally Posted by Nintendo Maniac 64 View Post

It'd probably be more practical once Zen comes around since HSA main benefit seems to be in productivity-based things, though games could possibly take advantage of it.

Speaking of games, any future console design wins would definitely take advantage of it, and it's even thought the PS4 and XB1 have some lesser HSA-like functionality already. I know that Nintendo in particular seems to love memory bandwidth when it comes to how they design their console hardware, so HSA would be right up their ally.
HSA doesn't exactly help with memory bandwidth though tongue.gif.
Quote:
Originally Posted by epic1337 View Post

java partly uses GPGPU, one could then imagine how vast java had already spread.
its not really a joke, nor was it meant to be offensive, it was simply a point-out at how GPGPU had already been preferred over general compute.
Which is already know but, consider this. What high performance software is done in, of all things, Java? Especially GPGPU?

GPGPU in Java is like adding a turbo charger to a golf cart.
Quote:
webpage (browser) rendering are GPGPU heavy (hw acceleration, of course you could disable some of it
Yah my HD 4000 comes to its knees rendering tabs in Chrome. And by knees I mean it wonders if that was supposed to be some kind of challenge.
Quote:
likewise video decoding and database operations had recently preferred GPGPU.
The former doesn't need more performance so much as it needs better efficiency. The latter? Databases really aren't my kinda thing, but it'll depend on the database and operations and whether they can be done in parallel.

You see, GPGPU is great but not everything can just encorporate GPGPU.
Quote:
i hate to quote wiki for context but...
http://en.wikipedia.org/wiki/Heterogeneous_System_Architecture
while HUMA functions as a bridge to join two different processors to function on the same ram interface, HSA functions to allow direct access without needing allocated space.
basically put, hUMA without HSA falls back to being NUMA.
http://www.pcper.com/reviews/Processors/AMD-Details-hUMA-HSA-Action


and on that site says this, in a sense they can implement HSA on discreet GPUs, just that its not practical in their views.
edit: on that note i think i get where you're getting at, i do understand HSA that this is only one part of it's function
I think you're getting closer but my main point is that GPGPU by traditional means (dGPU) far outclasses what an APU is capable of.

That GPGPU is best utilized in professional work as you are showing, where professional GPUs are faster, more accurate, more efficient and better supported.

That AMD is in NO position to dictate the market or see the spread of HSA in desktop applications due to the necessity of specialized hardware that ranks among the minority and that development requires a special compiler system.

HSA is sort of like a stepping stone in research, not important for immediate use, but as a gateway to things that are.

There doesn't seem to be any nice solution to the faults in current GPGPU using systems.
Quote:
HSA main function is to make GPGPU accessible as a general compute core like CPUs.
AMD, Nvidia, people in enterprise, consumers, etc... WISH that were the case.

HSA isn't even close to doing that. It wishes to use them as fully as GPGPU can but it doesn't make any breakthroughs there.
Quote:
Originally Posted by SpeedyVT View Post

They are building 2016 cpus with both ARM and x86 on the same chip. CPUs are often taxed by pure math operations and a GPU can accelerate that. However nothing yet. nothing major has been compiled for HSA.
You mean MANY math operations which can be run in parallel. Quick and dirty example, matrix multiplication.
i think you missed the point, GPGPU = better GFLOPs per watt, AMD = mobile and consoles.
even if GPGPU is too fast (or tasks are too light) to be irrelevant, using GPGPU will still increase perf/watt over pure CPU.
consoles could use more efficiency since heat and all is an issue, while mobile is all about power consumption... and heat.

and the reason why we can't just port everything to run in GPGPU is because a lot of them are pipelined, e.g. can't be done without previous operation results.
in a sense pipelined tasks are much better off in a single fast processor core, and since GPGPU is a clustered slow processor core... well you get the idea.

trolling an adult is very dangerous, don't try it at home nor at work. you don't want to play tag with a rabid man.
epic1337 is online now  
Reply

Quick Reply

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off