New Llano leaks

[Manicdan]

Sorry but he didn't hit 5.4Ghz... We have already discussed this,it's a bug.Real clock is approximately 2x lower than that.
But why you guys read about ES when we have real thing(retail) already tested here ? OCed to 3.77Ghz on CPU side and 45% on the GPU side. I summarized the performance difference Vs SB here.

wow 6k is awesome for vantage, compared to my score:
http://3dmark.com/3dmv/2099582
12k for 4850 xfire with a 3.6ghz phenom II (i think thats the speed i was using), its clear these APUs are very well rounded.

[dartaz]

If you think Llano's top desktop parts won't have turbo you are quite deluded. Also, Donanimhaber has shown about 5 different slides about Llano in the past month alone, constantly upgrading them as the launch nears.

A8-3850 is 2.9GHz and it doesn't have turbo.

Here is review for this CPU from anandtech
http://www.anandtech.com/show/4448/a...rmance-preview


Let me tell you that both CPU & GPU performance of Llano is not impressive. CPU performance is almost identical to Athlon II. While GPU performance is slightly faster than intel HD3000. We are talking about Llano with 400 shaders.

[jimbo75]

A8-3850 is 2.9GHz and it doesn't have turbo.

Here is review from anandtech
http://www.anandtech.com/show/4448/a...rmance-preview


Let me tell you that both CPU & GPU performance of Llano is not impressive. CPU performance is almost identical to Athlon II. While GPU performance is slightly faster than intel HD3000. We are talking about Llano with 400 shaders.

Did you also read the ECS thing on the front page? 33% overclock on their mobos? Looks like AMD are just going to let the OEM's decide what to do with the TDP headroom instead.

As for "slightly faster" than HD3000? What benchmarks have you been looking at lol. It's massively faster in everything except Anands pet Dragon Age Origins at laughably low settings. Oh btw, it does that while sucking half the juice.



In perf/watt terms Llano is about 3.5 to 4x that of HD3000. It's a massacre.

[Oliverda]

A8-3850 is 2.9GHz and it doesn't have turbo.

Here is review from anandtech
http://www.anandtech.com/show/4448/a...rmance-preview


Let me tell you that both CPU & GPU performance of Llano is not impressive. CPU performance is almost identical to Athlon II. While GPU performance is slightly faster than intel HD3000. We are talking about Llano with 400 shaders.

Anand for some mystical reason used DDR3-1333 modules for that sloppy test. So that review is irrelevant.

[dartaz]

Did you also read the ECS thing on the front page? 33% overclock on their mobos? Looks like AMD are just going to let the OEM's decide what to do with the TDP headroom instead.

I didn't find find the word "ECS is front page ?

As for "slightly faster" than HD3000? What benchmarks have you been looking at lol. It's massively faster in everything except Anands pet Dragon Age Origins at laughably low settings. Oh btw, it does that while sucking half the juice.

Lest than twice as fast HD3000 is not massive. People expected much more than that

Even at higher settings the difference between Llano (with 400 shader GPU) and HD3000, the difference is much less 2x in most cases
http://www.tomshardware.com/reviews/...u,2959-14.html

In perf/watt terms Llano is about 3.5 to 4x that of HD3000. It's a massacre.

3.5 to 4x in 3dmark ?

Do you think an average consumer is going to buy Llano for 3dmark ? Show us real gaming benchmarks

[jimbo75]

Updated benches with 1866 memory

http://www.anandtech.com/show/4448/a...ance-preview/3

Needless to say "slightly faster" won't be getting used again.

[jimbo75]

3.5 to 4x in 3dmark ?

Do you think an average consumer is going to buy Llano for 3dmark ? Show us real gaming benchmarks

I'm sure you can figure out that if Llano is outperforming HD 3000 by a large margin while drawing about half the power then it's a massively superior part. You want more performance? Put in some decent memory like Anand was forced to do and see what the gap is. It's about twice as fast at lowest possible (intel favouring) settings and it will still be drawing a helluva lot less power.

[dartaz]

Updated benches with 1866 memory

http://www.anandtech.com/show/4448/a...ance-preview/3

Needless to say "slightly faster" won't be getting used again.

So ?

I saw DDR3 1866 from the beginning. It was already updated when I first looked at the review

Even with DDR3 1866, it was less than 2 times the speed of HD3000 most of time. And HD3000 was using DDR3 1333. At least they should compare it with similar DDR3 speed. Not to mention that not many people use DDR3 1866 specially laptop users, because a decent DDR3 is going to cost you money $$ (I rather buy a dictated GPU than waste money on a decent DDR3)

EDIT: The word "slightly" is subjective. THe reason why I used this word is because Llano was expected to be several times faster than HD3000. I never imagined that llano is going to be less than 2 times the speed of HD3000. I was shocked.

[xVeinx]

Anand noted that they didn't see the performance speed up that AMD had promised in terms of IPC. They note the memory speed it part of it, but fail to mention that they only used 1333 till later on. I'm guessing that will make up the difference (if not drive it over what was estimated). That said, it isn't going to push it past the 2500K, but hey, not bad at any rate .

[jimbo75]

So ?

I saw DDR3 1866 from the beginning. It was already updated when I first looked at the review

Even with DDR3 1866, the difference is less than 2x most of time. HD3000 was using DDR3 1333. At least they should compare it with similar DDR3 speed.

By all means petition anand to use better memory...like the HD3000 is bandwidth limited I don't think.

It's close to 2x performance at ROCK BOTTOM SETTINGS.

What happens at higher settings?



Almost 3x faster than the fastest mobile i7?



Twice as fast?



Almost 3x faster?

The higher the settings go, the more Llano pulls away. It's not even a contest. In ALL games Llano is making playable settings while the I7 is unplayable.

[Motiv]

Looks decent if it's matched with some good memory. Will be interesting to see how it performs once the drivers are a bit more stable and working (talking about the weird Xfire thingy).

I'm in need of a laptop, these may just be what I need. A laptop, good battery life and can play games (and cheap!!).

[DarthShader]

I'd love to see MLAA tests. Being compute intensive, this method is perfect for bandwidth limited situations.

Can Intel's IGP do MLAA?

How about an IQ comparision between those two?

[dartaz]

jimbo75,

You have picked the best case scenario for Llano and you did that on purpose. I'm talking about average performance.

I saw those charts, so I wonder why you trying to show them to me when I already looked at the review

You said "The higher the settings go, the more Llano pulls away". But do you realize that at high settings Llano gets about 30 fps or less ? Do you think the game is going to be playable if you play at even higher settings. You are ignoring the fact that the game won't playable anyway.

[jimbo75]

jimbo75,

You have picked the best case scenario for Llano and you did that on purpose. I'm talking about average performance.

I saw those charts, so I wonder why you trying to show them to me when I already looked at the review

You said "The higher the settings go, the more Llano pulls away". But do you realize that at high settings Llano gets about 30 fps or less ? Do you think the game is going to be playable if you play at even higher settings. You are ignoring the fact that the game won't playable anyway.

Plenty of games are very playable at 30fps. When Crysis was first released people soon started saying that "30 fps in Crysis is great". Then again nothing could play it much higher. Isn't that strange?

Yes I picked the best benchmarks on Anandtech but you based your scores on his worst case scenario for Llano. ~2x faster with ~1/2 the power draw is all that needs to be said. AMD designed the part to be like this, do you really think they couldn't just have upped the clocks by 50% and totally blown the HD 3000 away? Of course they could have, but that would have been at the expense of power draw. This is a balanced part designed for balanced loads and on that metric it totally owns SB so hard it's almost embarrassing.

Forget the comparison and look at those last 3 graphs. A 35W LAPTOP is capable of playing modern games at high settings. That's unreal.

[Oese]

its not unreal its exactly the laptop i would buy if i would not have two amd last gen and one i5 laptop here LOL

look at the pricing in germany they came up with even an dedicated gpu 1gb ON TOP for less then 600 € THATS INSANE

finally it comes out it was the right decision to postpone bulldozer since AMD is getting to roll over mobile/htpc/micro mass segment first now with that price/perf/watts^^ It all makes perfect sense to me now. i was not that excited about a new tech strategy for long compared with that move amd took now :O I think many people will think very different about amd soon as much as they are surprised and obliged to think that cant be true now

[LightSpeed]

Dartaz, bandwidth limitations mean its around 50 - 120% faster than HD3000. If thats slightly in your books, good for you since im sure it makes you happy.

Im actually impressed with the graphics performance of Llano, especially on the mobile variants. Almost matching 6630m with the mem b/w limitations is extremely impressive.

[flippin_waffles]

I'm sure you can figure out that if Llano is outperforming HD 3000 by a large margin while drawing about half the power then it's a massively superior part. You want more performance? Put in some decent memory like Anand was forced to do and see what the gap is. It's about twice as fast at lowest possible (intel favouring) settings and it will still be drawing a helluva lot less power.

The results would humiliate intel further if sites would test AMD's graphics at ultra low quality, comparable to intel's 'graphics'. Of course that wouldn't happen with intel's marketing dollars in play. Anand retests intel hardware with hand delivered intel drivers hot off the press (as insignificant as they were), meanwhile manufactures excuse after excuse for using 10.12 drivers in many of their AMD reviews. No doubt they used Vision drivers that were as outdated as possible. I see 11.6 has been released, I wonder if they plan on retesting...

[Nintendork]

Do you think an average consumer is going to buy Llano for 3dmark ? Show us real gaming benchmarks

From toms data:

Twice the battery life with over twice the gaming performance (SB + HD3000 vs A8 + HD6620G)

Is that not impressive?

[FlanK3r]

Llano is awesome as value gaming complet PC....APU in games rocks! core to core is Llano a bit better than Athlon x4 (example look at Cibennch score)