Go Back   AnandTech Forums > Hardware and Technology > CPUs and Overclocking

Forums
· Hardware and Technology
· CPUs and Overclocking
· Motherboards
· Video Cards and Graphics
· Memory and Storage
· Power Supplies
· Cases & Cooling
· SFF, Notebooks, Pre-Built/Barebones PCs
· Networking
· Peripherals
· General Hardware
· Highly Technical
· Computer Help
· Home Theater PCs
· Consumer Electronics
· Digital and Video Cameras
· Mobile Devices & Gadgets
· Audio/Video & Home Theater
· Software
· Software for Windows
· All Things Apple
· *nix Software
· Operating Systems
· Programming
· PC Gaming
· Console Gaming
· Distributed Computing
· Security
· Social
· Off Topic
· Politics and News
· Discussion Club
· Love and Relationships
· The Garage
· Health and Fitness
· Merchandise and Shopping
· For Sale/Trade
· Hot Deals with Free Stuff/Contests
· Black Friday 2014
· Forum Issues
· Technical Forum Issues
· Personal Forum Issues
· Suggestion Box
· Moderator Resources
· Moderator Discussions
   

Reply
 
Thread Tools
Old 11-11-2009, 06:00 PM   #1
thilanliyan
Diamond Member
 
Join Date: Jun 2005
Posts: 8,154
Default AMD sheds light on Bulldozer, Bobcat, desktop, laptop plans

http://techreport.com/discussions.x/17948

Wasn't Thuban supposed to be AM2+/AM3 and therefore DDR2 capable as well (in the slide it only says DDR3 whereas Phenom II says DDR2/DDR3)? If not what a disappointment (for me at least)...I was going to hold off a CPU upgrade until it came out.

Laptop plans:
http://techreport.com/discussions.x/17945
__________________
Intel 4670k
Asus Z87-Expert
8GB Kingston HyperX DDR3-1600
3x7950 3gb (mining)
Corsair HX1050 Gold

Last edited by thilanliyan; 11-11-2009 at 06:03 PM.
thilanliyan is offline   Reply With Quote
Old 11-11-2009, 06:22 PM   #2
Phynaz
Diamond Member
 
Phynaz's Avatar
 
Join Date: Mar 2006
Posts: 6,650
Default

OMG, Bulldozer is horribly misbalanced.

I didn't think AMD could shock me any more, but they did.

They had better wish for two things to happen in the next two years:

1) General purpose FP loads are moved to GPU's.

2) Applications become much, much more threaded.

If that doesn't happen, then AMD will be in big, big trouble.
__________________
Quote:
Originally Posted by Abwx View Post
I saw good resolutions pics of the so called lunar module, heck, that s quite a piece of garbage with badly jointed metalic and litteraly hammered plates, seriously, you think that this piece of metalic junk actualy landed on the moon..??
Phynaz is offline   Reply With Quote
Old 11-11-2009, 06:29 PM   #3
theevilsharpie
Platinum Member
 
Join Date: Nov 2009
Location: Southern California
Posts: 2,320
Default

Quote:
Originally Posted by Phynaz View Post
If that doesn't happen, then AMD will be in big, big trouble.
ROFL

If AMD can pump out cheap low-power desktop and mobile chips in volume, they'll be just fine.
theevilsharpie is offline   Reply With Quote
Old 11-11-2009, 06:31 PM   #4
Phynaz
Diamond Member
 
Phynaz's Avatar
 
Join Date: Mar 2006
Posts: 6,650
Default

Quote:
Originally Posted by theevilsharpie View Post
ROFL

If AMD can pump out cheap low-power desktop and mobile chips in volume, they'll be just fine.
There's a saying in business - "He who lives by price dies by price".

Being able to compete only on price is death, especially in an industry that requires the kind of funding that developing CPUs requires.
__________________
Quote:
Originally Posted by Abwx View Post
I saw good resolutions pics of the so called lunar module, heck, that s quite a piece of garbage with badly jointed metalic and litteraly hammered plates, seriously, you think that this piece of metalic junk actualy landed on the moon..??
Phynaz is offline   Reply With Quote
Old 11-11-2009, 06:42 PM   #5
Falloutboy
Diamond Member
 
Falloutboy's Avatar
 
Join Date: Jan 2003
Posts: 5,915
Default

I think amd is taking a risk with this design but I think its something they have to do. they aren't like intel where they can have several parellel R&D programs. they have to pick something and hope it pans out the way they think.
__________________
"I contend we are both atheists, I just believe in one fewer god than you do. When you understand why you dismiss all the other possible gods, you will understand why I dismiss yours."
-Stephen F Roberts
Heat: FallOutBoy525
Falloutboy is offline   Reply With Quote
Old 11-11-2009, 06:53 PM   #6
nismotigerwvu
Golden Member
 
Join Date: May 2004
Location: Kent, Ohio
Posts: 1,416
Default

This seems like a high risk/high reward type move. I'm actually glad they acted instead of just steering the course for once. While nV has been trying to drum up GPGPU, having both an open standard (openCL) and a mainstream worth caring about (direct compute) I can see this sector really take off. Also, getting bobcat into netbooks and perhaps high end smartphones (depending on where that wattage range hits) could be a huuuuuuuuuge source of income.
nismotigerwvu is offline   Reply With Quote
Old 11-11-2009, 07:03 PM   #7
Idontcare
Administrator
Elite Member
 
Idontcare's Avatar
 
Join Date: Oct 1999
Location: 台北市
Posts: 20,472
Default

All that talk about cores and modules and two cores, etc, just confused the hell out of a me.

So when AMD says a Zambezi CPU will have 4/8 Bulldozer Cores does that mean it will only have 2/4 of these "tightly link two-core modules"?



Idontcare is offline   Reply With Quote
Old 11-11-2009, 07:11 PM   #8
IntelUser2000
Elite Member
 
IntelUser2000's Avatar
 
Join Date: Oct 2003
Posts: 3,510
Default

It's confusing to me too, but I think this might be a hint:

"Two tightly linked cores..."

8 clusters might be too large even at 32nm.
__________________
Core i7 2600K + Turbo Boost | Intel DH67BL/GMA HD 3000 IGP | Corsair XMS3 2x2GB DDR3-1600 @ 1333 9-9-9-24 |
Intel X25-M G1 80GB + Seagate 160GB 7200RPM | OCZ Modstream 450W | Samsung Syncmaster 931c | Windows 7 Home Premium 64-bit | Microsoft Sidewinder Mouse | Viliv S5-Atom Z520 WinXP UMPC
IntelUser2000 is offline   Reply With Quote
Old 11-11-2009, 07:15 PM   #9
GaiaHunter
Diamond Member
 
GaiaHunter's Avatar
 
Join Date: Jul 2008
Posts: 3,421
Default

Well, I guess the smallest functional unit is 1 "bulldozer", which
is composed by 2 smaller cores and seems to be in reality, 1.5 cores...

Yeah it is confusing.

I guess they want to say "Hey we have HT too" but then their HT is more hardware intensive and sounds a lot better to say you have 2 linked cores than 1.5 cores or a core that can do 2 threads.

From 4/8 I read (4cores/8 threads)/(8 cores/16 threads).

On the other hand 4/8 is different from "up to 6". Fucking slides.

Last edited by GaiaHunter; 11-11-2009 at 09:51 PM.
GaiaHunter is online now   Reply With Quote
Old 11-11-2009, 07:35 PM   #10
GaiaHunter
Diamond Member
 
GaiaHunter's Avatar
 
Join Date: Jul 2008
Posts: 3,421
Default

Quote:
Originally Posted by Phynaz View Post
OMG, Bulldozer is horribly misbalanced.

I didn't think AMD could shock me any more, but they did.

They had better wish for two things to happen in the next two years:

1) General purpose FP loads are moved to GPU's.

2) Applications become much, much more threaded.

If that doesn't happen, then AMD will be in big, big trouble.
Well,



and



Shouldn't this take care of number 1 concern? (non-ironic remark)
GaiaHunter is online now   Reply With Quote
Old 11-11-2009, 07:44 PM   #11
piesquared
Golden Member
 
Join Date: Oct 2006
Posts: 1,075
Default

Quote:
Originally Posted by GaiaHunter View Post
Well, I guess the smallest functional unit is 1 "buldozer", which
is composed by 2 smaller cores and seems to be in reality, 1.5 cores...

Yeah it is confusing.

I guess they want to say "Hey we have HT too" but then their HT is more hardware intensive and sounds a lot better to say you have 2 linked cores than 1.5 cores or a core that can do 2 threads.

From 4/8 I read (4cores/8 threads)/(8 cores/16 threads).

On the other hand 4/8 is different from "up to 6". Fucking slides.
Naw, I don't think they have any plans to ever say hey we have hyperthreading too. In fact, they have spoken out on the record saying they don't plan on having anything similar to hyperthreading, as it shows performance losses in some cases. Not really worth the real estate.
piesquared is online now   Reply With Quote
Old 11-11-2009, 07:45 PM   #12
IntelUser2000
Elite Member
 
IntelUser2000's Avatar
 
Join Date: Oct 2003
Posts: 3,510
Default

Quote:
Originally Posted by Phynaz View Post
OMG, Bulldozer is horribly misbalanced.

I didn't think AMD could shock me any more, but they did.

They had better wish for two things to happen in the next two years:

1) General purpose FP loads are moved to GPU's.

2) Applications become much, much more threaded.

If that doesn't happen, then AMD will be in big, big trouble.
It SEEMS as though FP on CPU is being deprioritized with the advent of on-die GPUs and whatnot. Ironically, its really Intel doing Fusion-idea first. The push for CPU-accelerated vertex processing on their IGPs is a case in point. It's just that no one cares currently. Only matter of time CPUs start accessing GPU resources.

Problem with thread is more complex, but Intel is currently pushing it with Hyperthreading, which by Bulldozer timeframe might be ready with more threaded apps.
__________________
Core i7 2600K + Turbo Boost | Intel DH67BL/GMA HD 3000 IGP | Corsair XMS3 2x2GB DDR3-1600 @ 1333 9-9-9-24 |
Intel X25-M G1 80GB + Seagate 160GB 7200RPM | OCZ Modstream 450W | Samsung Syncmaster 931c | Windows 7 Home Premium 64-bit | Microsoft Sidewinder Mouse | Viliv S5-Atom Z520 WinXP UMPC
IntelUser2000 is offline   Reply With Quote
Old 11-11-2009, 07:49 PM   #13
GaiaHunter
Diamond Member
 
GaiaHunter's Avatar
 
Join Date: Jul 2008
Posts: 3,421
Default

Quote:
Originally Posted by piesquared View Post
Naw, I don't think they have any plans to ever say hey we have hyperthreading too. In fact, they have spoken out on the record saying they don't plan on having anything similar to hyperthreading, as it shows performance losses in some cases. Not really worth the real estate.
Well they can call it whatever and their approach is different, that is for sure, but
from http://www.anandtech.com/cpuchipsets...oc.aspx?i=3674 ,

Quote:
A single Bulldozer core will appear to the OS as two cores, just like a Hyper Threaded Core i7. The difference is that AMD is duplicating more hardware in enabling per-core multithreading. The integer resources are all doubled, including the schedulers and d-caches. Itís only the FP resources that are shared between the threads. The benefit is you get much better multithreaded integer performance, the downside is a larger core.
when I read the above, even if they don't want to say HT, they want to say "look you have loads of CPU thingies on task manager, isn't that pretty?".
GaiaHunter is online now   Reply With Quote
Old 11-11-2009, 07:50 PM   #14
Phynaz
Diamond Member
 
Phynaz's Avatar
 
Join Date: Mar 2006
Posts: 6,650
Default

Quote:
Shouldn't this take care of number 1 concern?
No, because current GPU's are only fast at extremely parallel applications that use a rather limited instruction set. That's why I mentioned general purpose floating point.

Even if they manage to pull off a miracle and somehow create the next evolution in computing, notice on the slide it says "Next-Generation Software Ecosystem". A new generation of development tools and end user applications aren't going to happen in two years. Maybe, just maybe, you could pull it off in five if you were Intel and owned the compiler.
__________________
Quote:
Originally Posted by Abwx View Post
I saw good resolutions pics of the so called lunar module, heck, that s quite a piece of garbage with badly jointed metalic and litteraly hammered plates, seriously, you think that this piece of metalic junk actualy landed on the moon..??

Last edited by Phynaz; 11-11-2009 at 07:52 PM.
Phynaz is offline   Reply With Quote
Old 11-11-2009, 07:59 PM   #15
IntelUser2000
Elite Member
 
IntelUser2000's Avatar
 
Join Date: Oct 2003
Posts: 3,510
Default

Meh, its not like the FPU is disappearing. 2x 128-bit FMAC equals to 2x 256-bit FPU that can do ADD and MUL each. In single thread, its effectively having a 256-bit FPU, while in multi-thread its probably limited by resource contention and bandwidth to take advantage of full 256-bit so 128-bit might be enough.

Larrabee based on-die GPU might be able to do it.
__________________
Core i7 2600K + Turbo Boost | Intel DH67BL/GMA HD 3000 IGP | Corsair XMS3 2x2GB DDR3-1600 @ 1333 9-9-9-24 |
Intel X25-M G1 80GB + Seagate 160GB 7200RPM | OCZ Modstream 450W | Samsung Syncmaster 931c | Windows 7 Home Premium 64-bit | Microsoft Sidewinder Mouse | Viliv S5-Atom Z520 WinXP UMPC
IntelUser2000 is offline   Reply With Quote
Old 11-11-2009, 08:01 PM   #16
GaiaHunter
Diamond Member
 
GaiaHunter's Avatar
 
Join Date: Jul 2008
Posts: 3,421
Default

Quote:
Originally Posted by Phynaz View Post
No, because current GPU's are only fast at extremely parallel applications that use a rather limited instruction set.
Shouldn't the success of the future architecture than be dependent on future GPUs?

Rumors have the next ATI generation end of 2010, more than in time to pair with bulldozer. Additionally, we still have to see what Fermi will do (although I don't think AMD is counting on what NVIDIA has to offer or not).

Last edited by GaiaHunter; 11-11-2009 at 09:52 PM.
GaiaHunter is online now   Reply With Quote
Old 11-11-2009, 08:07 PM   #17
IntelUser2000
Elite Member
 
IntelUser2000's Avatar
 
Join Date: Oct 2003
Posts: 3,510
Default

There's no way they'll be able to pair a end-2010 GPU with Llano, unless Llano is releasing at sometime like say, September. From what I heard the GPU performance will be at Radeon 4700-ish levels and will feature a 5x00 derivative core.
__________________
Core i7 2600K + Turbo Boost | Intel DH67BL/GMA HD 3000 IGP | Corsair XMS3 2x2GB DDR3-1600 @ 1333 9-9-9-24 |
Intel X25-M G1 80GB + Seagate 160GB 7200RPM | OCZ Modstream 450W | Samsung Syncmaster 931c | Windows 7 Home Premium 64-bit | Microsoft Sidewinder Mouse | Viliv S5-Atom Z520 WinXP UMPC
IntelUser2000 is offline   Reply With Quote
Old 11-11-2009, 08:13 PM   #18
Phynaz
Diamond Member
 
Phynaz's Avatar
 
Join Date: Mar 2006
Posts: 6,650
Default

Quote:
Originally Posted by IntelUser2000 View Post
Meh, its not like the FPU is disappearing. 2x 128-bit FMAC equals to 2x 256-bit FPU that can do ADD and MUL each. In single thread, its effectively having a 256-bit FPU, while in multi-thread its probably limited by resource contention and bandwidth to take advantage of full 256-bit so 128-bit might be enough.
Yes, the FMAC is nice but the CPU still seems out of whack between INT and FP units to me.
__________________
Quote:
Originally Posted by Abwx View Post
I saw good resolutions pics of the so called lunar module, heck, that s quite a piece of garbage with badly jointed metalic and litteraly hammered plates, seriously, you think that this piece of metalic junk actualy landed on the moon..??

Last edited by Phynaz; 11-11-2009 at 09:15 PM.
Phynaz is offline   Reply With Quote
Old 11-11-2009, 08:15 PM   #19
GaiaHunter
Diamond Member
 
GaiaHunter's Avatar
 
Join Date: Jul 2008
Posts: 3,421
Default

Quote:
Originally Posted by IntelUser2000 View Post
There's no way they'll be able to pair a end-2010 GPU with Llano, unless Llano is releasing at sometime like say, September. From what I heard the GPU performance will be at Radeon 4700-ish levels and will feature a 5x00 derivative core.
Sure, but Llano isn't a bulldozer core, right? I was under the impression it is phenom II cores.

My point was that, the perceived or real imbalance of the bulldozer core, might not be one if the 6xxx architecture solves the problems highlighted by Phynaz.

Last edited by GaiaHunter; 11-11-2009 at 09:53 PM.
GaiaHunter is online now   Reply With Quote
Old 11-11-2009, 08:26 PM   #20
Idontcare
Administrator
Elite Member
 
Idontcare's Avatar
 
Join Date: Oct 1999
Location: 台北市
Posts: 20,472
Default

Wait...so now we are finding out that AMD's "the future is fusion" marketing strategy is really more like "back to the future" with CPU's being Integer processors and the FPU being shoveled into math-coprocessors?



What's next, move the CPU caches back onto the mobo? (j/k)
Idontcare is offline   Reply With Quote
Old 11-11-2009, 08:32 PM   #21
Phynaz
Diamond Member
 
Phynaz's Avatar
 
Join Date: Mar 2006
Posts: 6,650
Default

Lol, that's funny IDC.
__________________
Quote:
Originally Posted by Abwx View Post
I saw good resolutions pics of the so called lunar module, heck, that s quite a piece of garbage with badly jointed metalic and litteraly hammered plates, seriously, you think that this piece of metalic junk actualy landed on the moon..??
Phynaz is offline   Reply With Quote
Old 11-11-2009, 08:34 PM   #22
GaiaHunter
Diamond Member
 
GaiaHunter's Avatar
 
Join Date: Jul 2008
Posts: 3,421
Default

Quote:
Originally Posted by Idontcare View Post
Wait...so now we are finding out that AMD's "the future is fusion" marketing strategy is really more like "back to the future" with CPU's being Integer processors and the FPU being shoveled into math-coprocessors?

What's next, move the CPU caches back onto the mobo? (j/k)
Well, being in a single die has to count for innovation, right?
GaiaHunter is online now   Reply With Quote
Old 11-11-2009, 08:43 PM   #23
IntelUser2000
Elite Member
 
IntelUser2000's Avatar
 
Join Date: Oct 2003
Posts: 3,510
Default

Quote:
Originally Posted by GaiaHunter View Post
Sure, but Llano isn't a buldozer core, right? I was under the impression it is phenom II cores.

My point was that, the perceived or real imbalance of the buldozer core, might not be one if the 6xxx architecture solves the problems highlighted by Phynaz.
Yea, you are right. But the "heterogeneous computing" slide is explicitely about Fusion, which made me thought you were naturally referring to Llano.
__________________
Core i7 2600K + Turbo Boost | Intel DH67BL/GMA HD 3000 IGP | Corsair XMS3 2x2GB DDR3-1600 @ 1333 9-9-9-24 |
Intel X25-M G1 80GB + Seagate 160GB 7200RPM | OCZ Modstream 450W | Samsung Syncmaster 931c | Windows 7 Home Premium 64-bit | Microsoft Sidewinder Mouse | Viliv S5-Atom Z520 WinXP UMPC
IntelUser2000 is offline   Reply With Quote
Old 11-11-2009, 09:06 PM   #24
GaiaHunter
Diamond Member
 
GaiaHunter's Avatar
 
Join Date: Jul 2008
Posts: 3,421
Default

Quote:
Originally Posted by IntelUser2000 View Post
Yea, you are right. But the "heterogeneous computing" slide is explicitely about Fusion, which made me thought you were naturally referring to Llano.
Might be misinterpreting your post, and if so I apologize, but won't Bulldozer CPUs also have GPU on the same die, using Fusion as the controller?

High-end being APU (Buldozer CPU+GPU on same die) + Discrete GPU that will also be accessed as GPGPU if needs be, while low-end can just be the APU with no discrete GPU and/or a less powerful individual GPU.

I've the impression Llano is basically a test bed for integration of CPU+GPU on same die (and probably being a cheaper overall platform that can actually play current games), that will then be replaced by smaller Bulldozers a la i3/i5.

It will also be interesting to see where NVIDIA GPUs will fit on this - if the AMD APU will access them and/or if the GPU on the APU will be able to enhance NVIDIA GPU.

Then we have the Intel side - it is a safe bet to say Intel CPU will be faster than AMD CPU, but will the APU (what is the name for the Intel version? Sandy Bridge?any?) be overall a better buy? What will be the value of Larrabee and how will NVIDIA GPUs work with Intel APU?

2011 seems as it can be an interest year. 2010, though, seems one-sided - again...

Last edited by GaiaHunter; 11-11-2009 at 09:54 PM.
GaiaHunter is online now   Reply With Quote
Old 11-11-2009, 09:11 PM   #25
Viditor
Diamond Member
 
Viditor's Avatar
 
Join Date: Oct 1999
Posts: 3,290
Default

Quote:
So when AMD says a Zambezi CPU will have 4/8 Bulldozer Cores does that mean it will only have 2/4 of these "tightly link two-core modules"?
I believe that it means that every 2 cores will share one cache, and that the shared cache pairs will be using direct connect architecture with each other. So an 8 core will be 4 dual cores directly connected to each other on the DCA Bus...
This is similar to Intel's shared cache on the Core Duo, except that Intel had to connect the cache through an off-die FSB rather than directly connect to another cache on the die (much higher latency).

Quote:
A new generation of development tools and end user applications aren't going to happen in two years
They're talking about OpenCL...so I would disagree with your statement as they are almost ready now and developers already have initial SDK kits...
__________________
"Time flies like an arrow, Fruit Flies like a banana"

G Marx
Viditor is offline   Reply With Quote
Reply

Thread Tools

Posting Rules
You may not post new threads
You may not 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

Forum Jump


All times are GMT -5. The time now is 03:39 PM.


Powered by vBulletin® Version 3.8.7
Copyright ©2000 - 2014, vBulletin Solutions, Inc.