Overclock.net banner

1 - 20 of 54 Posts

·
Premium Member
Joined
·
1,601 Posts
Discussion Starter · #1 ·
Quote:
In the last five years, the classic PC and server computing market has begun to inexorably merge with mobile devices, consoles and other alternative form factors that rely on computation. This is an incredibly wide range of form factors, spanning <1W for a smart phone to 150W for high-end servers. AMD's architects realized that this range was too wide for a single processor core to address the entire market, thus spawning a strategy that relied on the Bulldozer architecture at the high-end, and the Bobcat and Jaguar cores (collectively the cat cores) for lower power devices.

AMD saw yet another axis for differentiation. Fundamentally, AMD is the only company outside of Intel that is capable of designing, validating, and shipping x86 microprocessors. Given the massive installed software base, this is tremendously valuable. Yet Intel as a semiconductor manufacturing company is largely unwilling to use foundries, which makes integrating external IP blocks challenging. AMD was already being forced to move to foundries by the spinoff of Global Foundries, but the Bobcat and Jaguar teams had aspirations on markets where customers, not manufacturers, defined the silicon. In particular, the graphics group at AMD had a strong presence in console gaming, dating back to ArtX's relationship with Nintendo and had already won the lion's share of console graphics, including the Xbox360 and Wii. To take advantage of this unique opportunity, AMD's architects designed the cat cores to be easily synthesizable and portable between foundries.
Read the Report

Like always with RWT, it is the most detailed technical review available. If you like to go deep, this is the article for you.
 

·
Been here for a while
Joined
·
866 Posts
This is cool, only means more competition in the market, and thereforce, It will obligate other companies like Intel and ARM to desing newer architectures in order to keep up the innovation
 

·
Registered
Joined
·
4,919 Posts
give credit where its due regardless of who's who
thumb.gif


you fanboys before you start - what price would you be willing to pay for your cpu/ gpu?

cos without ying you cant have yang
biggrin.gif
 

·
Registered
Joined
·
3,291 Posts
Oh man I want an AMD tablet with a GCN core in it! Or an updated version of that Acer 11 inch touchscreen with the quad core Temash in it. This time with Bobcat and a GCN core or two.
 

·
Registered
Joined
·
2,025 Posts
Good read, but I've always felt that RWT had an underdog bias. Not necessarily a bad thing, but they don't really present the facts. Way more objective than most tech journalism out there though.
 

·
Banned
Joined
·
3,485 Posts
Ok read it and say it was a touch above my level but good enough to see and understand the improvements. I like the conclusion and the changes the consoles might play into the life of the Jaguar SoCs
 

·
Politically incorrect
Joined
·
9,464 Posts
Can't wait for AMD tablets with GCN cores, hopefully matching Intel's Atoms in CPU power and surpassing in GPU. There are lots of cheap Intel powered x86 Windows tablets but I'm waiting for an AMD one simply because of their superb GPU's.

Edit: after reading the article twice I can say it was way over my head.
 

·
Iconoclast
Joined
·
31,010 Posts
Quote:
Originally Posted by Nintendo Maniac 64 View Post

I didn't realize there were that many things that were similar to K8; that may explain why IPC is so similar to K8 as well.
K8 is wider and generally has better IPC than Jaguar, at least when running applications that don't see much benefit from instructions it doesn't support.

Anyway, some similarities aren't unusual, and you can look at almost any current architecture and see how it evolved, or borrows, from much older ones.
Quote:
Originally Posted by HanSomPa View Post

Good read, but I've always felt that RWT had an underdog bias. Not necessarily a bad thing, but they don't really present the facts. Way more objective than most tech journalism out there though.
Well, this article doesn't mention performance much at all and the bulk of comparisons are between Jaguar and Bobcat; not much room for bias here.
 

·
Registered
Joined
·
2,116 Posts
Quote:
Originally Posted by Blameless View Post

K8 is wider and generally has better IPC than Jaguar, at least when running applications that don't see much benefit from instructions it doesn't support.
On the contrary, I find that my Brisbane performs slightly slower per-GHz than Jaguar in the few CPU benchmarks I've tested, in particular I distinctly remember being ever so slightly slower in Cinebench r11.5.
 

·
Iconoclast
Joined
·
31,010 Posts
Quote:
Originally Posted by Nintendo Maniac 64 View Post

On the contrary, I find that my Brisbane performs slightly slower per-GHz than Jaguar in the few CPU benchmarks I've tested (Cinebench 11.5 comes to mind).
Are you taking into account that most Jaguars have twice as many cores as Brisbane? The results I've seen do put Cinebench IPC very similarly to Jaguar, but Cinebench isn't the only test out there. A 2x2GHz Brisbane is only slightly slower than a 4x1.5GHz Jaguar in 7-zip.
 

·
Registered
Joined
·
2,116 Posts
I said IPC in my first post in this thread. IPC typically refers to performance per GHz per core.

(yes I realize it technically stands for "Instructions Per Clock" or "Instructions Per Cycle")
 

·
Iconoclast
Joined
·
31,010 Posts
Quote:
Originally Posted by Nintendo Maniac 64 View Post

I said IPC. IPC typically refers to performance per GHz per core.
Yes, typically it does. Just making sure we are on the same page.

What sort of Cinebench score are you getting on your Brisbane anyway?
 

·
Registered
Joined
·
2,116 Posts
Here's my post on the subject from a year ago:
http://www.overclock.net/t/1364086/sweclockers-amd-temash-apu-a6-1450-in-cinebench-r11-5/30#post_19371626
Quote:
Originally Posted by Nintendo Maniac 64 View Post

That Cinebench result is darn near exactly the same as what I get on my Brisbane. I think it's safe to say that Jaguar's IPC is nearly equal to K8.

64bit, 1 core(s), Brisbane @ 3.0GHz:: 0.717530
64bit, 2 core(s), Brisbane @ 3.0GHz: 1.425410

Do some math...
Bisbane multi-core scaling: 99.32755425%

Now divide the single-core result by 3 for an estimated 1GHz score:
64bit, 1 core(s), Brisbane @ 1.0GHz: 0.2391766667

Multiply by 4 to simulate 4 cores: 0.9567066667

Finally multiply by 0.9932755425 to simulate multi-core effeciency:
64bit, 4 core(s), Brisbane @ 1.0GHz: 0.9502733334

Needless to say, a 0.04 point difference between Brisbane and Jaguar is pretty spot-on. I would imagine that a K8 CPU that had the full 1024KB of L2 (such as a Toledo 4800+) would make up that point difference and match Jaguar.
 
1 - 20 of 54 Posts
Top