Featured Articles

IHS teardown reveals Galaxy S5 BOM

IHS teardown reveals Galaxy S5 BOM

Research firm IHS got hold of Samsung’s new flagship smartphone and took it apart to the last bolt to figure out…

More...
Galaxy S5, HTC One M8 available selling well

Galaxy S5, HTC One M8 available selling well

Samsung’s Galaxy S5 has finally gone on sale and it can be yours for €699, which is quite a lot of…

More...
Intel lists Haswell refresh parts

Intel lists Haswell refresh parts

Intel has added a load of Haswell refresh parts to its official price list and there really aren’t any surprises to…

More...
Respawn confirms Titanfall DLC for May

Respawn confirms Titanfall DLC for May

During his appearance at PAX East panel and confirmed on Twitter, Titanfall developer Respawn confirmed that the first DLC pack for…

More...
KFA2 GTX 780 Ti Hall Of Fame reviewed

KFA2 GTX 780 Ti Hall Of Fame reviewed

KFA2 gained a lot of overclocking experience with the GTX 780 Hall of Fame (HOF), which we had a chance to…

More...
Frontpage Slideshow | Copyright © 2006-2010 orks, a business unit of Nuevvo Webware Ltd.
Tuesday, 23 November 2010 14:56

Bulldozer hints at 3.5GHz 8MB L3 and Turbo 2.0

Written by Fuad Abazovic
amd

213 M transistors for 2 CPUs

 

German-based Computerbase has spotted a few interesting details about AMD's upcoming Bulldozer CPU. The chaps spotted that at ISSCC 2011 conference, between 20th and 24th of February, AMD plans to talk a bit more about its Bulldozer architecture.

The interesting part is that the chip should work at 3.5GHz or a bit more and with the help of Turbo (Turbo Core 2 ) it should be able to overclock its cores for additional 500MHz making it to 4GHz. It will need between 0.8 and 1.3V which looks good and the chip as you might know by now has 8MB L3 cache.

The Bulldozer module consists of two Integer cores with total of 213 million transistors while some parts of the chip get shared. This saves you some transistors but might bring some additional problems, like more heat, we’ve learned.

Here is what AMD will be saying at ISSCC 2011:

“The Bulldozer 2-core CPU module contains 213M transistors in an 11-metal layer 32nm high-k metalgate SOI CMOS process and is designed to operate from 0.8 to 1.3V. This micro-architecture improves performance and frequency while reducing area and power over a previous AMD x86-64 CPU in the same process. The design reduces the number of gates/cycle relative to prior designs, achieving 3.5GHz+ operation in an area (including 2MB L2 cache) of 30.9mm2 …

An 8MB level 3 cache, composed of 4 independent 2MB subcaches, is built on a 32nm SOI process. It features column-select aliasing to improve area efficiency, supply gating and floating bit lines to reduce leakage power, and centralized redundant row and column blocks to improve yield and testability. The cache operates above 2.4GHz at 1.1V. ….”

Eight-core Bulldozer at 32nm aims for 852 Million transistors and probably some other parts will take a bit more. Have in mind that six core Thuban at 45nm has 904 million transistors on a 346 square millimeter die. The twelve core server version is looking at 1278 million transistors if not a tad more.

The ISSCC program is here. (PDF)

Also read

AMD „Bulldozer“: CPU-Takt 3,5+ Ghz, L3-Cache mit 2,4 GHz

Last modified on Tuesday, 23 November 2010 15:08
blog comments powered by Disqus

Comments  

 
+58 #1 Steve-O 2010-11-23 15:25
Excellent numbers! AMD doesn't seem to be shooting for the moon in terms of a gigantic die.
 
 
+43 #2 1d10t 2010-11-23 15:38
this surely will bulldoze 6 core 12 threads 980X...
 
 
+39 #3 Nerdfighter 2010-11-23 15:59
Sounds pretty awesome. If the stock Turbo is 4 gigs, it should be able to overclock beyond that. Let's just hope they get the pricing right too ;)
 
 
+44 #4 Freakazo_ 2010-11-23 16:11
Woah, if it truly will only be ~852 Million transistors this will end up being an amazing cpu.

But I think that figure is without l3 cache which would boost it's transistor count significantly. When looking at he transistor difference between propus and deneb (the main if only difference is l3 cache) we can see that 6MB L3 consists of 450 million transister, or 600 million transistors for 8MB L3 cache

giving a grand total of 1.5Billion transistors for a 8 core Bulldozer cpu with a total of 16MB cache (4* 2MB per module + 8MB l3 cache) with a die area only slightly larger than a six core i7 980X on the same process node.
 
 
+34 #5 Freakazo_ 2010-11-23 16:16
Also a something that wasn't pointed out was that the IPC is being upgraded compared to the phenoms putting it on the same level as I7's (got this from the same source as the other details)

And for the 12 core version you just edited in, it would end up being 1.9Billion transistors, 20MB cache on that one hopefully.

Also, stupid comment lenght problem >:(
 
 
+24 #6 prod 2010-11-23 16:49
If only more game designers started writing multi-threading titles, I would actually be able to make proper use of these Sweet CPUs.
 
 
+1 #7 Bl0bb3r 2010-11-23 17:33
Quoting prod:
If only more game designers started writing multi-threading titles, I would actually be able to make proper use of these Sweet CPUs.



Recent games are multithreaded AFAIK, the problem lies with Microsoft's scheduler that doesn't appoint the threads to the proper cores.


The speeds are nice... but hopefully, that 3.5+ GHz figure won't be for the single module only. Price wise, they will cost more than your average Thuban, I'm guessing... will be interesting to see if they put it in the i7 range.
 
 
+5 #8 Drecker 2010-11-23 17:49
“an 11-metal layer 32nm high-k metalgate SOI CMOS process..."

Really? The processor is made from 11 metals?

Sad that so few people know how to correctly use a hyphen. :sad:
 
 
+12 #9 Freakazo_ 2010-11-23 17:56
Quote:
Recent games are multithreaded AFAIK, the problem lies with Microsoft's scheduler that doesn't appoint the threads to the proper cores.

Uhm.... lol wut. You have it wrong man, the windows scheduler jumps threads around to all/most of the cores unless specifically specified in the program what it's affinity should be. That's why on a quad core machine with one thread in a very compute intensive loop each core will only use 25% usage (a bit of each core). two threads 50%... you get the idea. The windows scheduler is perfect(good, maybe not perfect) for multithreaded tasks.

Continued... (stupid comment length once again)
 
 
+14 #10 Freakazo_ 2010-11-23 17:58
The only cases I know of where when something is supposed to be multithreaded but doesn't use more than one core is in python due to it's global interperter lock or if the software implements a system where it switches to different 'threads of code' (not real threads) within the single thread, which is mostly done for responsiveness for the user. Game's actually do this allot, think about when it's loading a model for example, your game doesn't stop until it is finished loading the model but switches to the game engines code and the model loader code frequently so that you can still walk around in the game while it's loading the model.
Multithreaded games are multithreaded and will use multiple cores unless because of something listed above.
Continued (again,stupidco mmentlength)
 

To be able to post comments please log-in with Disqus

 

Facebook activity

Latest Commented Articles

Recent Comments