Llano - when will we start seeing reviews?

[AtenRa]

Looks like someone broke NDA.

http://www.legitreviews.com/news/10565/


Article incorrectly reports the i7-2600 as having HD 3000 when it actually uses the weaker HD 2000. So parts using HD 3000 (the desktop K series processors and all mobile Sandy Bridge) should perform better relative to the A8-3510MX, although will probably still be quite a ways behind it. If this is legit looks like 45W mobile Sandy Bridge parts might have roughly 1/3 to 1/2 the GPU performance of the A8-3510MX. Of course on the CPU side it will be no contest for Intel I'm sure. So no clear winner, as with most other things it will just depend on what you want to do with the system.

Definitely can't wait to see full reviews from AnandTech and other sites.

edit: And looks like some of the videos were posted on page 3. Unfortunately they're gone now, though, and this article summarizes some of the results.

Exept the Super Pi the second program running was FluidCS11.exe from Direct3D11 SDK. Since Intel's Core i7 2600 HD2000 cannot run D3D11, the program was running in software emulation in the CPU and not in GPU (That's why 100% CPU Utilization).

 

[dma0991]

Interesting to know that AMD pulled the video down due to NDA. I kinda expected that to happen so I... :sneaky:

 

[mosox]

With wikileaks and all I think AMD would make more money if in charge with the US Govt secrets. Damn, they're good at it.

 

[jimbo75]

Exept the Super Pi the second program running was FluidCS11.exe from Direct3D11 SDK. Since Intel's Core i7 2600 HD2000 cannot run D3D11, the program was running in software emulation in the CPU and not in GPU (That's why 100% CPU Utilization).

So? What do you think intel would do if AMD couldn't run mainstream software properly?

 

[exar333]

So? What do you think intel would do if AMD couldn't run mainstream software properly?

Great test. Let's run crappy IGPs with DX11 that essentially no one would use the product for anyway. :\

I really like Llano but this test proves nothing and is just marketing FUD.

 

[jimbo75]

Great test. Let's run crappy IGPs with DX11 that essentially no one would use the product for anyway. :\

I really like Llano but this test proves nothing and is just marketing FUD.

That's not the point. We're not discounting legitimate benchmarks just because intels graphics are so poor they can't do dx11.

You should read http://blogs.amd.com/nigel-dessau/2010/04/19/have-you-paid-the-“sysmark-tax”/ and see what benchmarketing FUD really is.

 

[cotak13]

Great test. Let's run crappy IGPs with DX11 that essentially no one would use the product for anyway. :\

I really like Llano but this test proves nothing and is just marketing FUD.

Wow completely missing the point!

read: http://msdn.microsoft.com/en-us/library/ff729716(v=vs.85).aspx

FluidCS11 is a simple fluid dynamic simulation for animation but it's a cousin of full computational fluid dynamics like Navier-Stokes.

What you see is the weakness of intel's CPU without heterogeneous compute features. If you want to run any OpenCL app you aren't reaching for a Intel GPU a this point.

The idea of APU isn't just about lowering costs or sticking a GPU onto a CPU. No that wouldn't be any improvement over what we have today. No. APUs are about heterogeneous computing bringing better performance for certain parallelism heavy applications.

 

[AtenRa]

So? What do you think intel would do if AMD couldn't run mainstream software properly?

Im was only talking about the technical aspects of the benchmark, one system was running in D3D11 in GPU when the other was running in software emulation with the CPU,

im not taking any sides

 

[Phynaz]

That's not the point. We're not discounting legitimate benchmarks just because intels graphics are so poor they can't do dx11.

You should read http://blogs.amd.com/nigel-dessau/2010/04/19/have-you-paid-the-“sysmark-tax”/ and see what benchmarketing FUD really is.

Sysmark is FUD? Lol, okay then!

Way to miss the point.

 

[jimbo75]

Sysmark is FUD? Lol, okay then!

Way to miss the point.

Yeah what do you think it is? Did you even read the article? 7% performance difference equating to 40% difference in score is FUD if ever there was any.

You want to talk about unrealistic workloads then lets talk about the 35000 row spreadsheet. When was the last time you saw one of those?

The point is quite simple actually. AMD isn't going to play it by intels rules, and the minute intel put that POS graphics core onto their cpu, they became fair game for all of its flaws. You're only as good as your weakest part, and contrary to what you might think, the vast majority of SB buyers will not be getting a discrete gpu to go along with it.

 

[Phynaz]

Yeah what do you think it is? Did you even read the article? 7% performance difference equating to 40% difference in score is FUD if ever there was any

Another butt hurt fanboy doesn't like a benchmark and starts shouting "Intel FUD".

You want to talk about unrealistic workloads then lets talk about the 35000 row spreadsheet. When was the last time you saw one of those?

Today, as a matter of fact.

the vast majority of SB buyers will not be getting a discrete gpu to go along with it.

Really? Wow, thanks for pointing that out! I'm sure nobody here thought of that.

 

[Accord99]

That's not the point. We're not discounting legitimate benchmarks just because intels graphics are so poor they can't do dx11.

Basically, this LLano video is the equivalent of comparing Linpack between a Sandy Bridge and a Deneb. We use AVX (DirectX 11) on Sandy Bridge but because Deneb doesn't support AVX, we go all the way back in time and use an x87 kernel (CPU emulation) rather than say a SSSE3 (DirectX 10.1) kernel.

And to add to it, we have Deneb run another CPU intensive, multi-threaded application at the same time (the i7 system was running 8 threads of HyperPi versus 4 threads on LLano).

 

[jimbo75]

Another butt hurt fanboy doesn't like a benchmark and starts shouting "Intel FUD".

I think you missed the fact that it's intel users crying foul over AMD's current benchmarking tactics?

Today, as a matter of fact.

Checking out sysmark hm?

Really? Wow, thanks for pointing that out! I'm sure nobody here thought of that.

Therefore it's a completely legitimate benchmark. For the past two months all I've read about is "butt hurt" intel fanboys claiming the benchmarks are invalid because...SB can't run them properly? AMD is completely within their right, in fact I'd go as far as saying it is their duty to point out SB's horrible graphic flaws because it's sold as a complete cpu+gpu and it clearly isn't on a par with Llano.

 

[Mopetar]

Not to get in the way of a good Intel/AMD mudslinging competition, but seriously, every time someone misuses FUD I die a little inside.

 

[Idontcare]

You want to talk about unrealistic workloads then lets talk about the 35000 row spreadsheet. When was the last time you saw one of those?

Today. And yesterday, and the day before that.

I use Excel to import csv files for rank-sorting purposes on a near-daily basis. The files have anywhere from 20k to 60k rows of data.

That said, the impractical part is that today's processors manipulate the data in such a spreadsheet with so much speed that it hardly makes a material difference to me whether my current processor could be replaced by one that is 20% faster...everything I do in Excel with my 30k+ rows of data happen in less than a second anyways.

Once I upgraded to SSD I gave up even bothering to benchmark Photoshop anymore. Who gives two craps if it now only takes 22.345 ms to fully rotate an 18MB pic versus the 36.738 ms it used to take?

Apps that fully load my cpu for sustained periods of time (>1 min) in order to complete a task that I need completed before I can proceed to do something else with the output are my main concern.

Video transcodes, financial modeling/backtesting, computational chemistry.

This office-productivity stuff is mostly bogus in my experience. You want to improve office productivity? Block facebook and myspace at work, office productivity will magically improve as the time-wasters disappear. No CPU upgrade required.

 

[Phynaz]

This office-productivity stuff is mostly bogus in my experience. You want to improve office productivity? Block facebook and myspace at work, office productivity will magically improve as the time-wasters disappear. No CPU upgrade required.

You forgot Anandtech forums!

 

[SickBeast]

A little bird told me that a new CPU review is imminent. I think it has to be Llano. I don't know what else it could possibly be.

 

[podspi]

A little bird told me that a new CPU review is imminent. I think it has to be Llano. I don't know what else it could possibly be.

Zambezi?

Either way I'm excited. Especially after the disappointing "leak" last week...

 

[nonameo]

Zambezi?

Either way I'm excited. Especially after the disappointing "leak" last week...

eh? refresh me again on last week's leak? I don't remember anything.

 

[podspi]

eh? refresh me again on last week's leak? I don't remember anything.

I think it has mostly been cleaned up, but someone from Chiphell posted benchmarks from an ES sample. Like the earlier leak it was slow, slow, slow. Like, easily beaten by Thuban slow.


From what I've heard, the ES was operated at < 1ghz. Or it was made up. Who knows...

 

[nonameo]

I think it has mostly been cleaned up, but someone from Chiphell posted benchmarks from an ES sample. Like the earlier leak it was slow, slow, slow. Like, easily beaten by Thuban slow.


From what I've heard, the ES was operated at < 1ghz. Or it was made up. Who knows...

Ah, okay... yeah, it's the same architecture so it doesn't make sense for it to be slower clock for clock. I expect it to be a tad faster than an athlon II clock for clock but that's about it. I'm more interested the way AMD is going to price it, and how it'll be price in laptops.

 

[wahdangun]

its turn out I'm right after all Llano will be using the sideport memory !!! i can't wait to grab the GDDR5 sideport version

AMD's upcoming Llano line of accelerated processing units (APUs), to compete with Intel's Sandy Bridge LGA1155 processors, is said to embed what AMD will refer to as the Radeon HD 6550 graphics core. Unlike Sandy Bridge where a processor die is simply fused with a integrated graphics northbridge onto a single die, Llano will see the GPU part of the silicon integrated with the rest of the APU in many other levels, including assisting the x86 cores with serial processing loads.

Llano's embedded GPU carries the AMD Radeon SKU of HD 6550. It will feature on AMD's Fusion A8-3550 and A8-3550P APUs, is DirectX 11 compliant, has 400 stream processors, and a core clock speed of 594 MHz. It uses memory shared from the main memory, but in all likelihood, AMD might work on SidePort-based memory support. Further, the Radeon HD 6550 can work in tandem with discrete AMD Radeon HD 6570 and HD 6670 "Turks" based graphics cards in the same way as AMD's IGPs have been known to work with entry-level Radeon GPUs using Hybrid CrossFireX. When the HD 6550 iGPU is working in tandem with HD 6670 or HD 6570, the graphics hardware will be recognized as "Radeon HD 6690".

EDIT: ups sorry here are the source linky

 

[Idontcare]

its turn out I'm right after all Llano will be using the sideport memory !!! i can't wait to grab the GDDR5 sideport version




EDIT: ups sorry here are the source linky

You read a story where someone quipped "might be using sideport" and you take that as confirmation that you are right and Llano "will be" using side-port?

Mind you the actual "story" here is from Turkish site donanimhaber and nowhere in the original article was sideport ever mentioned.

http://www.donanimhaber.com/islemci...izli-entegre-grafik-birimi-Radeon-HD-6550.htm

 

[podspi]

You read a story where someone quipped "might be using sideport" and you take that as confirmation that you are right and Llano "will be" using side-port?

Mind you the actual "story" here is from Turkish site donanimhaber and nowhere in the original article was sideport ever mentioned.

http://www.donanimhaber.com/islemci...izli-entegre-grafik-birimi-Radeon-HD-6550.htm

I hope they have the option of adding sideport memory, either in the mono or on die, but I don't think we'll see it with Llano. Trinity, perhaps. I expect (read hope) Trinity will have superior graphics and graphics compute than any ivy bridge competitors, so this may be necessary.

 

[Tuna-Fish]

I hope they have the option of adding sideport memory, either in the mono or on die, but I don't think we'll see it with Llano. Trinity, perhaps. I expect (read hope) Trinity will have superior graphics and graphics compute than any ivy bridge competitors, so this may be necessary.

It's extremely unlikely that there will be any sideport RAM on the MOBO, because that would need a socket upgrade. You might be able to fit one or two GDDR5 dies on the package.