AMD Fusion Info Thread

I'd LOVE to see Llano in an attractive Acer or ASUS frame. If there was a TimelineX with a 400SP IGP and 6770, I'd buy it in a second.

 

Call of Duty? Sure. World of Warcraft, however, is very mainstream. And the Sims for PC is one of the greatest selling titles of all time. There are a lot more gamers out there than you think. In fact, I bet most people think males game more than women... and yet studies have shown that women are the larger # of people who play games. Maybe not intensive BFBC2, but there are enough mid-ranged titles that won't do well with the current IGP systems

There are plenty of PC gamers - people are in a rage because high-end gaming has gone downhill due to consoles, but consoles have also pushed a lot more mainstream games to PC's.

You guys are seriously underselling how many people use GPUs out there, and how many *would* play on a notebook if they had the right equipment.

Seriously? Are you trying to make stuff up?

That'd be amazing

 

I think some here are underestimating the IGP in Sandy Bridge. You can play WoW, Sims and a whole bunch of other games with it. Sure you can`t play them on high details though. But it is good enough for casual gaming

 

The people who play WOW already have desktops in their mothers' basements. The glory days of the Sims were 5-10 years ago when it was a fresh idea and they sold the millions of copies, and SB IGP would have no problem playing it today. Few care about the newest iteration of Sims anymore, and it's not a demanding game to begin with.

What are these mid-ranged titles that women like? Most of them games play on facebook or flash games. The number of people who play games that require a mid-range GPU is a tiny fraction of the number of laptops that are sold every year. Serious gamers get discrete GPU's and/or consoles.

Gaming has gone mainstream with the consoles. No need to buy a PC when the console graphics are good enough for people this generation. Next generation consoles will drive high-end PC gaming to an even smaller niche. Where's L.A. Noire on the PC by the way if it's so mainstream? Consoles have taken over and they're only going to get more popular with time. That's why Microsoft is giving away a free Xbox with the purchase of a PC, they know PC gaming's a small niche now.

Once they heard the fans spin up, felt the notebook getting hot, and only their battery being drained in 30 minutes, not many, not when they have to mess with graphics settings and deal with driver updates and games crashing to the desktop.

The people who use GPU's are mainly on consoles, and the other elite pc gamer niche are on desktops. That's a minority compared to how many laptops are sold every year.

 

I don't know about you but I use my laptops (4 atm) for all purposes but especially gaming. Laptops are cheap and disposable to me at 700-1000 a pop.

I guess this stems from the $4,000 business computer that I had back in 94. In any case laptops are cheap to run portable and will do just about everything most people need. At the university where I work we all game on laptops and you don't have to spend an arm and a leg.

The upcoming llano series will give us a relatively cheap computer that you can game on as well as work. I would gladly pickup AMD's highest offering paired with a 6770m for less then a grand.

Npc5

 

One other thing, I have replaced all of our desktops at home with laptops of both AMD and intel flavors. The fact that they use far less power then the equivalent number of desktops makes them attractive.

AMD may not have the most powerful processors by far, but I play Black Ops, Dirt 2, Shogun II with no problems on my N930 paired with a 5650m. Comfort isn't a problem, I just use an external keyboard and mouse with a 23 inch monitor.

So if AMD can keep the cost down, then alot of people will buy the llano series.

Npc1978

 

Good for you, most laptop buyers don't give a crap about Black Ops, Dirt 2, or Shogun on the PC. They care about price and that's the only way AMD can compete.

 

Shogun II on a Llano? Sign me up! If they can fit it in a nice 13.3 format like Toshiba's 835 protege, even better. Almost jumped on that HP DMz1, but i figured i might as well wait for Llano and get something with more oomph. And a USB 3.0 slot.

 

I think we can all agree that Llano is going to offer much worse CPU power than SB but much better graphics and similar battery life at a lower price, right?

So what is the arguing in this topic even about?

 

There's a lot of misinformation floating around these forums as of late.

 

1. I think the "similar battery life" part remains to be seen. A high power CPU that's idle most of the time could indeed consume less than a lower power CPU that's more utilized more often. We'll have to just wait.
2. I believe 400SP Llano's have about 2-2.5X SB graphics performance. There are also 240SP parts, which should be closer to SB.

Everything else you said is correct. The argument is about whether mass market needs anything more than SB graphics performance, and my answer to that is a resounding no.

 

I think your mistaken. There are some people who have a need for speed half the time, and a need for power the rest of the time, like me. This is why I have an M11x. But if there is an IGP that can provide good battery life, power, without the problem of Optimus, and possibly, with a 6770M, reach the levels of a 5850 or so, then I think that will appeal to ALOT of people.

 

But then again we can argue that while people don't need powerful graphics, they don't need that powerful a CPU either. The fact that the netbook market thrives shows that even Atom-class CPUs, as poor as they are, suffice to a large portion of users.

Therefore, the only thing AMD need to secure is battery life and price. Performance wise only a fraction of users need powerful and they'll go Intel either way so AMD just needs to start competing on the other [larger] fraction of people who just want the cheapest computer possible that can still do what they want.

The problem is getting the units out there. Zacate was good compared to the Atom platform but it wasn't cheap and it wasn't as widespread as the Atom which in the end didn't help all that well.

 

Based on the battery life increase from AMD's last-gen ultrathin processors, Nile, to Brazos, though, I think we can assume that AMD has improved idle/low power consumption significantly.

They should still be slightly ahead of SB. They will also be cheaper, supposedly around $500. The cheapest SB notebooks cost around $600.

I agree but does the mass market need anything more than Llano CPU performance either, though? The answer is no.

 

Hey, I agree with you that most people don't need more than Llano CPU performance either. However, as SSD's become more common, general Windows responsiveness increases with CPU, at least that's what I noticed when I went from a C2D to SB. However, we're a long way away from SSD's becoming commonplace.

I still think that people who need a more powerful CPU outnumber the people who need a more powerful GPU, but that doesn't matter, since both pale in comparison to the number of people who just want a cheap laptop like Melody said That's where AMD is going to compete, I still think hardcore gamers will still prefer a proper GPU and a SB CPU.

 

Most people don't know what's best for them and what is enough to cover their needs. When they go an buy a system they think that faster components (CPU or and GPU) are better. With netbooks people want small and cheap.

 

Wasn't Nile was a ULV version of the pitifully inefficient K10.5? Brazos is a brand-new architecture designed for low power usage. If the gains in efficiency come from the manufacturing process, then yes, you can assume the properties carry over to Llano, but if the gains came from the architecture, then Llano won't be much better off than its predecessors.

 

Brazos/Zacate uses the same 40nm bulk silicon that GPU use not the 32nm HKMG SOI that Llano will use. Bobcat's efficiency comes from the arch Llano's will come from the process and whatever tweeks AMD made to the arch to get it onto the new silicon.

Trinity will bring some of Bobcat's efficiencies to Llano's silicon along with Bulldozer's performance increase.

 

Here is an interesting set of slides. The claim made in the slides is that they get better battery life than Sandy Bridge (by a lot, roughly 50% or so). If this is the case, then I think Llano will be a big success.

 

Just saw those as well and that is a pretty bold claim. It looks like Intel is finally going to have some serious competition this year. If AMD actual can deliver on these claims then Llano will offer midrange graphics, low-end SB processor power and ULV battery life for significantly cheaper than SB. It will be a huge success

 

If a Quad core can get 3.5 hours more battery life than a Dual core from Intel, then it is a great achievement. I am a bit suspecious against tests and claims from themselves though. I`ve seen enough manipulation and skewed reviews from Intel about various products, so I want to see it with my own eyes first.

 

It's not so much the quad-core that is puzzling as the 400 stream processors. And yes, I'll believe it when I see it... but even if they only deliver something roughly equal to Sandy Bridge in battery life, that will still be a big deal. Most people won't notice the relatively weak CPU or the relatively strong integrated GPU, but battery life is very obvious even to the most casual consumer.

 

Wait a bit here. I read the little line under the battery life graph It says:

First of all, I do think that they are lying about that AMD CPU vs the i5 2410m, because even with a notebook that is being used with IGP (surfing, movies etc) and without a discrete GPU card, you can easily get 5+ hours on it. They said they tested it on idle. Sandy Bridge is already very effective in that state. Hardly think that 6.5 hours is all you can get with a SB.
Second of all, it doesn`t matter if it had 1000 stream processors. Most new CPUs today, C2D etc, have technology that shuts off and downclock when idling. So these stream processors are not in use at all.
Third, does it really matter to anyone how much battery life you can get out of a CPU when it is not in use?

 

6.5 idle is about right for dual-core SB, maybe even a little higher than average. Most mainstream SB notebooks will get less than that when idle.

 

Depends on the battery. The Lenovo W520 reviewed here gets more than that when browsing the web with a Core i7-2920XM. It also depends on the rest of the notebook -- even a CPU that consumes no energy at all won't save you if you want to use a 17" display at maximum brightness.

Looking at the slides again, to some extent it is obvious marketing drivel aimed at people in finance who are only vaguely familiar with the technology they're dealing with. However, the claim about battery life is big enough to absorb the inevitable clash with reality. That is, even if it turns out that "idle" means "turn WiFi off and don't touch the notebook" and the real battery life (i.e. the one you get when browsing the web) is actually 7-8 hours, that will still be a lot more than most people expected from Llano.

 

From those slides mentioned above , that dual graphics option intrigues me. Could you have a dual graphics option on an intel laptop at the same price point?

 

If they're making that claim, they have indeed mastered the art of clock gating and idling transistors. Still, when at moderate to full load, K10.5 has no chance against SB in Flops/Watt because it's the same old architecture.

The claims of more battery life than SB are most likely either using different size batteries or idling the discrete GPU on the intel chip or something like that. My friend's Portege R705 with an i3 gets 10 hours at idle, so it's a good thing now they can idle all day, as you can see from my friend, most people with Intel CPU's also idle them all day

 

LOL forget about using it. Idle battle is all that matters.

 

You can't get dual graphics that use Intel's IGP...you would need a notebook that has two discrete cards from AMD or Nvidia.

 

That is also with a huge battery. Companies might stretch the truth to make their products seem better but I think AMD at least used notebooks with the same size battery for the test. With a 6-cell, 4.5h is probably about what the W520 gets.

The thing is, though, that notebooks aren't usually at moderate to full load on battery power. Most people just browse the web when on battery and the processor is almost always under low load. My last notebook, a dv6z with a Phenom II N970, almost never went past its third P-State, 1.5GHz, when browsing or watching Youtube, even 1080p. 90% of the time is was running at 800MHz.

The test was probably done using MobileMark2007, which all manufacturers use, with two notebooks of the same size with 6-cell batteries. The result would be in-line with the average SB MobileMark2007 score. Obviously there are notebooks that get more than that, but on average they don't.

As I said, everyone uses idle battery numbers when advertising. Go look at battery life estimates on product pages and go read the actual reviews if you don't believe me. The real, moderate use battery life is almost always around 3/4 to 4/5 of what the manufacturer advertises. But, like althernai said, a claim of 10.5 hours still means a real battery life of 7 hours under moderate use.

 

Those two sentences directly contradict one another. I also remember a presentation where Nvidia claimed Tegra 2 was faster than a C2D @ 2Ghz, so that's a huge assumption to say same size batteries were used.

The people who do nothing more than surfing and Youtube get netbooks. I'd like to think some people use their CPU a bit more. Go to audiusa.com in Firefox and see your CPU go to its highest state. Play around with the slider on the banner which rotates the car around in the main page, and as long as you keep scrolling from side to side, your CPU should be maxed out. I just did that side by side with CPU-Z. A couple seconds I stop scrolling, it goes back to idle 800Mhz. This would be worse on a slower processor.

Assumptions, assumptions. They can easily say "our notebooks can have bigger battery due to space and cost saved by not having a discrete GPU and use a bigger battery". Every company stretches the truth when advertising.

 

How do those statements contradict each other? AMD could have stretched the truth by using numbers from an idling AMD machine and an active Intel one, or by activating discrete graphics on the Intel notebook, or by setting the brightness lower on the AMD notebook,etc. I seriously doubt they used different sized batteries, though.

I have an E-350 now and it kicks up to 1.6GHz, around 70% utilization, sometimes but it quickly goes back down to its lowest P-state. It doesn't take a whole lot to browse even heavy flash pages.

It may be an assumption, but it is a valid one to make. It is better to assume the same size battery was used than what you just said. I agree that every company stretches the truth, I said it earlier, but even if 10.5 hours is the MobileMark2007 figure, real battery life should be around 7 hours

 

Wow. This anandtech thread could use the services of an Intel fanboy or two.

 

Keep moving that slider and it'll remain there. For the record my CPU usage is 22% when doing the same, while running a VM and have outlook and Access open. I have the headroom to do that, but I admit the average customer wouldn't.

Why is it better? I have zero faith in K10 architecture's ability to be competitive in performance per watt. Of course, what really hurts AMD is that they're not as fast to ramp up and get there chips out there as Intel. For every 10 Intel makes, they only make 1 or so, so they're not very common to begin with.

 

That is a bit of a skewed view. The reason why AMD got so many hours in -idle- (if it is actually true), is because they are using power gating, which shuts down almost everything on the CPU since it is not needed. Intel also use this to a certain degree. The new Sandy Bridge CPUs have very low power consumption in idle state compared to 1st gen i7s. Now there is a very big difference between idle with almost everything shut off, and actually using the CPU with the IGP. Having a good IGP pays it`s price. With 2820QM, it will draw around 15 watt more if you watch a movie. Power gating is nice, but I imagine that benefit goes away when you start using the CPU.

 

Llano's IGP will have it's own power management feature, likely similar to what's found on either Cayman desktop GPU or Bobcat's IGP, so that power draw will scale based on the workload placed on the IGP. i.e. You won't be drawing the full power of 400SPU@444Mhz to watch a movie.

Furthermore, even at full load the power profile on those 400SPU should be below 15W given that the discrete counterpart to Llano's IGP is rated at 15-19W on 40nm bulk and these APU are 32nm HKMG SOI.

In the end though, what needs be recognized, is that someone who finds themselves stressing Llano's IGP to draw max power is obviously one of those people that needs more GPU power than Sandy Bridge's IGP can offer and are therefore benefiting from Llano's more powerful IGP.



One final side note....moving forward AMD plans to introduce full power gating to all of their GPU with the goal that entire blocks of stream processors can be shut down and draw almost no power at idle just like CPU cores.

 

Yes, I find this potentially more exciting than Llano itself. If this is the type of idle power draw they are claiming with a 32nm half-new half-old architecture, imagine what they can do with the 28nm Southern Islands GPUs later this year. I was going to buy a new laptop this summer, but assuming my dying 8600M GT holds on, I think I will just wait a little longer.

 

AMD's mobile 28nm won't be launched until CES 2012 and probably won't be for sale until March'ish 2012.

http://forum.notebookreview.com/gam...28nm-mobilty-hd7000-series-coming-2012-a.html

 

Yes, I know. That's also roughly the timescale of Ivy Bridge and affordable 250GB+ SSDs. This is why it is so tempting to wait. Maybe I'll just buy something cheap like the dv6t now and only keep it for a year and a half rather than 3 years (by then AMD should finally ditch K10.5 and may have a competitive mobile CPU offering).

 

Okay, so it remains at 1.6GHz as long I am moving the slider. The E-350 is nowhere near the power of even the slowest Llano. My old N970 would have no problem with heavy flash pages even with a bunch of things happening in the background.

On the slide it says to see the footnotes for configuration details and, while companies try and stretch the truth, they would not put the configuration details if it had such obvious bias. The journalists this presentation was given may not be tech-savvy, or maybe they are, but they are likely not complete idiots. The smartest thing would bee to assume the systems both had the same screen size, the same battery, no discrete graphics, the same hard drive, the same amount of RAM, the same OS and were both idling.

We aren't even talking about performance per watt, we are just talking about power consumption when idle which has nothing to do with performance.

I said that they stretched the battery life numbers by using an idle test and I said that all manufacturers do the same. That is the definition of not being skewed. You seem to assume that AMD's power gating technology only works when the APU is completely not in use but why? It likely works by shutting off certain parts when not in use and if that is the case then the battery life estimate of 7 hours seems realistic under moderate use given their 10.5 resting life. Turbo Core also will help because, like the slide says, it allows for tasks to be done quickly so the processor can rest.

We can argue back and forth all day, though. Right now we just don't know and we will have to wait and see.

 

Anybody have any ideas on when Toshiba and HP are going to start pushing out the Llano laptops? Do they they usually give people a heads up a bit before they put these things out or do they just sort of dump them out there with no warning?

 

If they could shut down the stream processors, then it would be a great feat. But Llano will work the same way as IGP from SB: downclock the processors and cores (EUs) when there is little load. HD 3000 speed range from 650 MHz to 1300 MHz, and the power draw varies with it. Llano have the upper hand here because they run at lower frequency. But they are all drawing power when you are using the CPU, right? And there are much more cores than Intel`s solution which either idle or is in use.

About that 15W power draw: I found a more reliable review from Anandtech that found some different result. With a laptop with 2820QM and without discrete GPU, it drew 9.04W on DC mode (on battery) when idling. When you watched a movie H.264, the total power consumption of the system was 16.38W in DC mode. That is 7.34 W. That involves the screen, the HDD, the CPU, the IGP, everything. I still think it is very difficult to beat that by much.
http://www.anandtech.com/show/4084/intels-sandy-bridge-upheaval-in-the-mobile-landscape/10

So if AMD found a way to increase battery life in idle, great. But I don`t think we are there yet where you get much better results when you are using the laptop other than what Intel have come up with yet.

 

Notebooks with Llano will just start showing up in the next couple of weeks. Computex would be the most likley place to end any NDA and notebooks are likely all ready to start shipping.

 

Except that they'd drain the battery very quickly. Llano is pretty much the same as your old Athlon II processor, except it's married to a GPU and has clock gating at idle, so it can idle all day long. Once you start doing real work on it, battery life will go down quickly, much more than a Sandy Bridge.

They could easily compare their 20W LV part to a 2720QM and say it gets better battery life

 

Not really. Even when watching HD Youtube, my N970 would get 2.5-3 hours of battery life as opposed to 3.5 under normal internet browsing and 4.5-5 idling, all on 60% brightness with wifi on. Once again, you are assuming that the AMD's only power gating occurs when completely idle.

It says they used the A8-3510MX, 45W, versus the i5-2410M, 35W.

 

Except for the idling figure, those aren't really impressive. Youtube HD is GPU accelerated so that's really not a valid comparison. I'm also sure you're quite exaggerating all of this, since you're very biased towards AMD, because face it, no one cared about the mobile Athlons, and bought them, but it seems you did, and now you're defending them. I on the other hand, use a laptop with both Intel and AMD chips in it, as well as having AMD processors in my desktops before I got rid of them for notebooks.

Yet no explanation on the rest of the system. Maybe the Intel system had a hot power hungry NVidia GPU idling or had its brightness all the way up.

 

Looks like AMD crafted a winner with its Zacate Fusion processors: AMD ships five million Fusion chips, says it's sold out -- Engadget

 

I mean yeah, I'd take a Zacate core over any Atom, it's what happens when you make a processor for 25 bucks and make it perform worse than Pentium-M.

 

that's great news
now i just need a tablet with a fusion processor and i'll do my effort to suport AMD

 

Lenovo X120 tablet with the E450 would do perfectly. I'd get that in a heart beat.

How do the current Atoms that just came out compare to the Pentium M and Fusion?