A little bit of news on AMD Lano

Comin in 2011 to fight Intel Sandy Bridge and 22nm shrink Ivy Bridge.

• 32nm
• 4 Core cpu combined with integrated Directx 11 gpu
• 2.5-25W TDP
• Supports DDR3 Memory

Ars Technia
EETimes
Anandtech

 

Wait what? 2.5-25W TDP? 4 Cores? Am I reading right?

 

Sounds a little like Turbo Boost?:

Link

 

Very interesting.
This might bode well for AMD ... and not a moment too soon.
Still, it would have been better if it was THIS year instead of next.

Will be intriguing to see how it compares to Intel.

And even if it sounds like 'Turbo Boost' ... well, let's just hope it's efficient enough in what it does, so it can benefit the consumer.

 

Llano may not put AMD in the lead but at least they're closing the gap between them and Intel.

 

I think their HT implementation that will bring AMD back into the game.

 

that is bulldozer. i dont think llano has the dual int registers that bulldozer will have. unfortunately amd has not yet mentioned bulldozer based laptop cpu.

 

This sounds promising. I'm already starting another savings account for a new notebook in 2011.

 

I didn't say Lano's has HT.

 

IMO , its a little too late... i doubt it's gonna make AMD more attractive...

 

AMD is already attractive.
Number of people buying them has increase because
1)They are cheap with better graphics features
2)People know that they are totally compatible with x86 code
3)People don't care what brand it is as long as it can run Windows

 

I really wish that amd succesfully comes up with this. the more fierce competition there will be(vs intel), the better it will be for consumers.

 

Eh... IDK...
1) The are cheap with better graphics than intel.
2) Most shoppers don't even know what x86 code is...
3) A lot of people I know, that had bad previous experience with AMD and their heat issues in notebooks, vow only Intel now.

But I'd say more people are buying them now because they have made a big push for the lower end market segment, where they are better value than Intel's package with GMA 4500/HD, but won't out-perform them.

 

Well once you get the whole "you make bad CPUs" image, it's hard to get back from it, but unfortunately, consumers will be consumers. I mean, the soldering issues haven't stopped people from buying Nvidia nor did Windows Vista's initial bad press stop people from buying Windows 7 so in the end, AMD will live through its past "bad name" concerning heat output issues(hopefully).

 

The 2) and 3) kinda sounds contradicting, don't ya think.

 

Uh P4s were TERRIBBBBLEEE at heat management, and AMD killed them, now the tables have turned and everyone forgot about the heat that p4s put out. It's bound to happen again..

 

Like I said, it'll be hard at first, but if they manage to provide a decent product and stick with it, then everyone will eventually move on and forget.

 

They don't know what IA-32 is but they know they want Windows and AMD runs Windows very nicely.
Moreover Integrated Radeon HDs are miles ahead of Intel Graphics especially in rendering DirectX although Intel Graphics put up rather nice looking numbers in video decoding they still lack many features of 3D Rendering Gaming suitable graphics.

 

Ok, lets cut the crap already.

http://www.notebookcheck.net/Intel-G...D.23065.0.html
http://www.notebookcheck.net/NVIDIA-...G.11949.0.html

(GMA HD vs 9400M)

Sims 3 82.1 vs 77
Fear 2 45 vs 44
HL2: 47.4 vs 42
CSS: 79 vs 44

There is a CPU difference between the two, but its not that big since at that performance its 50/50 CPU/GPU bound, and the GMA HD fps is also higher to make up for having a faster CPU. So far as anybody's concerned, it's a parity.

Intel has a CPU advantage and AMD has a GPU advantage, and doesn't look like it'll change in 2011, unless Larrabee-then pulls a surprise, even then won't feature in IGPs till somewhere like 2014+ with heavily castrated variant.

 

I wouldn't take the benchmark for the the truth.
Ion includes ATOM Processors in some of the benchmarks it is the CPU limiting the performance, when you average the result it becomes crap.
You should look at this a much more accurate benchmark.

 

Actually, if you bothered to click on the link you'd notice the numbers are for the 2.1-2.5GHz Core 2 models. The difference in performance between Atom 1.6GHz and 2.xGHz Core 2 seems to be around 2x. Considering Core 2 is likely 3x faster in CPU applications, it is still 50/50 CPU/GPU bound.

And again, it depends a lot on the game, but the highest end Clarkdale is on par with 790GX, which is also the highest end AMD: http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=3704&p=4
http://www.xbitlabs.com/articles/video/display/intel-hd-graphics_10.html

HD4200 500MHz=733MHz Ironlake
HD4200 700MHz=900MHz Ironlake

 

Real world computing isn't like that.
There is no 50/50 thing.
How fast something perform depends on the weakest link.

 

Not really My mother knows the difference in Intel and AMD brands, Intel being the one she now likes instead of AMD. If I were to say "While I was getting your prius serviced today, they put some x86 in it and it runs much better now..." she'd say.. "ok dear.." lol. Same for the rest of my family. We are an Intel family by and large, but none of them are really too technical...lol

 

There you go. GMA HD is faster then.

Except there is such thing as 50/50. Just because its a "bottleneck" doesn't mean you'll get 0% on CPU until you get all bottlenecks elsewhere out and vice versa. Core 2 2.1GHz is more than 2x faster than Atom 1.6GHz, yet you can see there(I bet you still haven't) that with the 9400M the difference is only 2x, therefore, the faster CPU still benefits it.

Back to the topic. As you get to things like fusion, you can not only share GPU resources for CPU work, but vice versa. If you can write a truly dynamic driver for games using CPU resources for aiding the GPU like an advanced form of hybrid sli/xfire, CPU differences will make up for some of the GPU differences.

 

The problem with your argument is you are not technical enough, CPU and GPU works differently there is no replacement of one over the other in architecture.
I don't know how you get the idea of 50/50 processing because in gaming CPU and GPU takes different calculation, GPU only return result requested by CPU there is no 50/50 workload division, it doesn't process IA-32 code, your CPU needs to have a certain level of power in order not to let the GPU have "rest time" in between.
If you still want to insist you are right then i rest my case...

 

Ok, sorry, I'll clarify about the "50/50".

If you run CPU-intensive apps like encoding/compiling or even benchmarks that simulate usage patterns like Mobilemark or Sysmark, Core 2 at 2.1GHz would be 3-3.5x faster than Atom at 1.6GHz.

With the same 9400M iGPU, the Atom and Core 2 systems get:
Anno 1404: 19.1 vs 42 fps
Sims 3: 42 vs 77 fps
Fear 2: 20.7 vs 44 fps

So for approximately 3.5x better CPU you are getting 1.5-2x better frames which means 50% "scaling".

For discussion purposes its being inaccurate but I think the point stands. Even if you use a low-end GPU, if you increase CPU performance vastly, like going from Atom to Core 2, you'll increase performance. If bottlenecks worked like black-and-white, you'd gain nothing from going between the two CPUs until at some point you get a faster GPU and voila, you get massive increase.

They'll never replace each other in the performance segment, but processors like Llano never meant to be. It's to provide balanced performance between the two.

And just like you might be able to use GPU's floating point resources for general purpose work that CPUs do, you can also use CPU's faster access to caches and optimization of serial code for graphics as well. You'll never be able to fully replace each other, but I never said that.

GPUs have branch prediction as well. Why not throw away the shoddy branch prediction unit on the GPU and use the one on the CPU instead? Or texture caches for example? Since you'll be integrating the two, it'll eventually make sense to share it.

Of course, that won't happen in Llano generation but we will, soon.

 

Bulldozer... ahahaha, must have been named that to give it an aggressive sound, and who knows maybe it means they are radically changing architecture, thus "bulldozing" over the old and building something completely fresh.

 

mmm..I dont know.

I always assumed it meant that they were going to "bulldoze" the competition (Intel...) hahaha

 

A little bit of news on Sandy Bridge. CES 2011. Then after that, bring on 22nm! I'm tired of 32nm.

 

I have a hard time believing USB 3.0 won't be integrated. If that's true it's pathetic, but I won't believe it until we see OEMs needing to put discrete USB controllers into Sandy Bridge systems.

 

Intel is trying to stall USB 3.0 in favor of their Light Peak tech.

 

Most likely. It's supposedly due out near the end of this year. However, it wouldn't surprise me if AMD's update in May (pre-Llano) does support USB 3.0, which would put some pressure on Intel to have better USB support.

The dynamic core boosting is good news - IMO that's a major advantage for Intel right now. I'd heard that the hexacore desktop AMD Thuban processors would support it, but not yet that mobile chips would. Too bad the release of all this is still so far off, though - if it were released today, AMD would be fairly competitive.