*HP dv6z AMD Llano (6XXX series) Owners Lounge*

I wasn't aware of that but there are versions of the 911 with 400-600hp, that is actually what most of them have. On average, the difference between a 911 and Focus is greater than the difference between the i5-2410m and A8-3530mx.

This analogy would be better applied to the 6620G vs. HD 3000, to be honest.

 

Lmao, I have an urge to be the car nut in the room. A base 911 has 365HP ish, base focus 120HP. The fastest focus is 300HP (Focus RS), fastest 911 is 619HP (911 GT2 RS). Haah, and, yes, its better for the 6620G and HD3000.

 

Not really, making fast CPU's is like designing formula 1 cars where every tradeoff and each transistor are crucial, but making GPU's is just a matter of building bigger and bigger trucks that can carry more.

 

Truthfully making GPUs is a matter of building smaller and smaller trucks that can carry more.

Brain explodes!

 

I like the Formula 1 analogy but, seeing as the CPU/GPU are now combined, it applies to it as a whole. Chip makers have to choose how much space to use for IGP performance and how much space to use for the CPU. AMD chose to have an IGP that takes up half of the chip while Intel used less than a quarter. Intel used half of the space on the CPU compared to less than one-third for AMD. Like I said, AMD could have used less space on the IGP and added more cache or increased clock speeds. They still wouldn't match SB clock-for-clock but they could have made the difference in processor performance way less while still maintaining a lead in graphics performance. AMD truly believes that the CPU performance of Llano is enough for their target market.

 

On the other hand...that's a waste of chip, given that no one really needs more than the HD 3000 to do anything a typical consumer would want to do.

CPU power is more applicable than GPU power to facebook and all that junk.

If you wanted to play games or something, it's been shown to be far more cost effective to get a discrete card (better performance for less $ than llano).

 

No one really needs more than a Llano CPU to do anything a typical consumer would want to do, either. But more and more people want to play newer games on their laptops while something like 3d rendering or video encoding is not growing as quickly.

The cheapest A8 looks to sell from $600-700 normally, without coupons. How many new notebooks are there that have better GPU performance than it in that price range? Not many. Sure, if you want to spend $100-150 more you can get an increase in graphics performance higher than the increase in price but many don't have that $100-150.

 

you can get a Dell XPS with a 540m and 2410m for $699

you can get a lenovo for $699 with a 555m and 2410m

you can get a vostro + acer aspire amd 6630 for even less

 

Where? The XPS 15 and Y570 both start at $799 and to customize a Vostro with the 6630m is going to cost about the same. There are a few Acer models with a 6630m/GT540m at around $700 but there are going to be A8-3500m models around $600, like the Gateway NV55S05u.

 

There are coupons + sales

e.g. $100 (maybe its $150?) coupon on the xps gets it to 699

a8-3500m without a discrete is nowhere near the power of a 540m 630m + sandy bridge

You're trying really hard to justify AMD by referring to an APU that benefits almost no one - mainstream consumers don't need the 6620g and will go for Intel for same or less $ (and be better off). budget gamers have better solutions with a discrete card. hell you could match up an older AMD processor with a discrete and be better off than with llano - this excerpt is from anandtech's desktop review, but the same idea is applicable to notebooks

The only reason that AMD is playing the APU card is because they can't keep up in x86. They were forced to integrated their only strength to stay competitive.

And it's only going to get worse, unfortunately. Can you imagine Bulldozer trying to beat a 22 nm Ivy Bridge next year? Performance per watt will be heavily in Intel's favor.

 

For one, I said without coupons. With coupons you can get the dv6z/A8-3500m down to $650 right now. Given current prices I would not be at all surprised to see A8-3500m notebooks at $600 by the end of the month. Like I said before, you could spend $100 more for a bigger increase in performance than the price increase but many don't have that $100. Even if they did why would you spend more on extra performance you don't need? By the way, the XPS 15 that is $799 base has the GT525m. The GT540m starts at $899.

We have gone back and forth on this so many times that it is getting ridiculous. The mainstream consumer won't need the 6620g, I am not denying that, but the same user doesn't need SB CPU power either. Refer to my earlier post. At $400-600 Llano is definitely competitive with Intel. The Pentiums and i3's are not going to beat the A4's and A6's by nearly the amount that the 6480/6520g are going to beat those Intel IGP's.

The same idea is not applicable to notebooks if looking at current prices.

Obviously they can't match Intel clock-for-clock but they could have improved their CPU performance by adding more cache or increasing clock speeds at the expense of using half of the die on the IGP.

If Trinity offers a 50% boost over Llano like AMD has said then the difference between Ivy Bridge, which Intel says will be 20% faster than Sandy Bridge, and Trinity will be less than the difference between Sandy Bridge and Llano. Ivy Bridge will undoubtedly be better in power consumption but under internet use/most idle conditions I doubt the difference will be much. Trinity will take an even bigger lead in IGP performance than Llano has over SB as well.

 

My argument is 1 thing. This generation, Intel beats AMD at most price points. There's only 1 place where you could justify llano IMO, and that's if you have a strict inflexible budget and you are a gamer. Even there I argue that you can get a discrete card combo with comparable price. Plus, almost no one who requires a card for gaming will have a budget so inflexible that he is stuck in the 100 dollar range between 550-650 where llano is the best option. Outside of that miniscule niche it makes almost no sense to go AMD.

What kind of grudge you got against Intel? You said "I had bad experiences with 4 laptops" but then neglected to mention how any of them were related to Intel. Sounds like you are rather unlucky as Intel is very reliable for the past 5+ years.

Most people who keep up to the date on this news agree that the 50% is coming from the GPU which is Cayman (6900 desktop series) because bulldozer allows for a smaller CPU part of the die and bigger GPU part of the die. Which is great but doesn't do much for a consumer who is likely better off with an Ivy Bridge + discrete 28 nm mobile part even at budget pricing options

AMD is clearly the #2 in x86 and they need to prove themselves if they want to regain market share. Does AMD pay you to defend them in this thread? Cause you're really grasping for straws

 

Even if I didn't personally prefer AMD I would disagree with that. In the $400-600 price range it basically comes down to the notebook. Battery life is the same, CPU difference is not going to be noticeable for the mainstream consumer and GPU difference is not going to be noticeable for the mainstream consumer. It comes down to individual preferences on looks, brand, size, etc. But, if you add in the fact that there are more mainstream users who want to play current games than there are that encode videos, Llano is a great choice.

I don't have a grudge against them, I just prefer AMD because most Intel notebooks I have owned had cooling problems. I already explained it a few pages ago. Maybe I was just unlucky with previous Intel notebooks and maybe it is irrational but, at the same price and performance as far as what I do is concerned, I prefer AMD.

Until we see something official I think that we shouldn't assume anything. We have to at least wait for Bulldozer benchmarks. Right now arguing Ivy Bridge vs. Trinity is just stupid, to be honest.

AMD is #2, that is not up for discussion. I am not grasping at straws at all, I am pointing out facts. You seem to be repeating the same argument over and over, and that is that Llano is significantly weaker than Sandy Bridge. No one is arguing that it isn't. The whole point is that it doesn't matter to the mainstream consumer that Llano is directed at.

 

It would have improved, but I am not sure by how much. If you look at the voltages they are forced to push Llano to in order to get even moderate clock speeds (i.e. the max Turbo ones), it doesn't seem very plausible that dedicating more die size or thermal budget to the CPU would do much good. At best, you'd get the usual 10% from L3 cache and a little less than the maximum Turbo speeds as default. There's only so much you can do with the K10.5 architecture -- it's a small modification of something positively ancient and it was never very good to begin with.

This is possible, but it's very dangerous to trust the projections of the manufacturers. All this tells us is that AMD is more aggressive than Intel about its marketing which is actually something we already knew from our discussion of Turbo. Given what I've seen of leaked Bulldozer benchmarks and the absence thereof in the retail channel, it's not obvious to me that AMD's new architecture is performing up to expectations. Also, what they're trying to do is take a server architecture and stuff it into a laptop. This did not work very well for Intel (see Clarksfield and to a lesser extent Arrandale) and I'm not sure whether it will work for AMD. Likewise, we know very little about either the Ivy Bridge battery life or its GPU.

 

All right then we'll just agree to disagree.

IMO AMD should just stick with what they are good at (GPUs) seeing as they are probably going to obliterate nvidia next year.

 

A 10% increase and a clock speed of say 2.3GHz for all 4 cores the A8-3530mx, which I think would have been plausible while still having similar IGP performance to Intel, would have put it in the range of the i5-2520m in multithreaded performance. With working turbo on it, even to say 2.6GHz with one core, would put it about 30-40% behind SB i5's, right in the range of SB i3.

I haven't seen any leaked benchmarks of Bulldozer, though I admit I haven't been looking much. Just out of curiosity do you have a link?

AMD also isn't just putting Bulldozer in the laptops. Trinity is going to be using "Enhanced Bulldozer" cores, whatever that means. I find it hard to believe that they would just stuff Bulldozer into laptops after working so hard to shake off the idea that AMD notebooks run hot and get poor battery life.

I don't exactly trust AMD's 50% performance increase estimate, based on Turbo Core like you said, but even half of that would put Trinity close to Sandy Bridge in performance. Intel has said that they expect a 25% increase over SB in both graphics and CPU. So, Ivy Bridge would still have a 25% lead over Trinity in CPU performance but Trinity will have a much bigger lead in IGP performance.

 

Right. So you'd have a 4-core CPU that can barely (if at all) match a dual-core CPU. With similar graphics, this would not have made any sense at all -- they'd have to sell them even cheaper than they currently do.

Here's one. That said, this is not really worth much because AMD is still fine-tuning the silicon.

No, of course not -- they will trim it down to something that fits in a laptop. What I meant was, it is very hard to do this and get the same level of performance as if you were designing for laptops in the first place. Clarksfield was not Nehalem; Intel reduced the clock speeds and made significant architectural changes. However, it still wasn't very good: the battery life was awful and, except for Turbo Boost, the performance was no better than the Core2Quads. To be fair, Core 2 was a great architecture whereas K10.5 is mediocre or worse so AMD should not have much trouble improving upon it, but 50% is a bit of a stretch.

Really? Do you have a link? Last I heard, they were adding 4 EUs to the 12 which are already there so naively, I would expect a 33% boost just from that. In addition, they're revising the architecture (at the very least to add support for OpenCL and DX11) so it'll probably be a little more than that. I'd be pretty surprised if they did all that and the net gain was only 25%.

 

Intel said they were targeting a 20% CPU boost and a 30% GPU boost.

Trinity is targeting a 50% improvement most of which is predicted to come from the GPU.

No matter what, I just don't see the OEM blueprints changing. Mainstream consumers will probably go with a low end Intel processor. Gamers will be best served by getting a discrete GPU (28 nm process means HUGE performance gains in next generation (Q1 2012) GPUs - can't wait!) with a powerful Intel processor which I suspect may be something like i7 - 3630qm - quad core @ 2.5 ghz.

There's a couple ways AMD can win this battle. The first is to out engineer Intel, and that ain't gonna happen. AMD's current processors perform like a lower end Core 2 Quad. They have a long ways to go. The second is to make their solutions more cost effective. Right now llano costs nearly the same as Intel's sandy bridge, but if they sell at a lower margin they could compete. At best AMD could control the entire budget market including the mainstream consumer market.

So TLDR: the way for AMD to win this battle is to cut prices.

 

Point. Maybe something with a 2.1GHz base, turbo to 2.4 and 1.5-2x the graphics performance would have been a better example. I don't think that would have been impossible for AMD to do.

Not really a good test of processor performance anyway.

Like I said, I don't expect a 50% increase.

Ivy Bridge to have 20 percent performance advantage over Sandy Bridge by VR-Zone.com

Says they are targeting 30% for graphics and 20% for the CPU. Don't know if that is true.

 

The only future for Llano is if they make a 20W version with 2 cores and 160 SP's and sell it for significantly less than Intel's overpriced ULV Sandy Bridge chips. $500 "gaming capable" mainstream laptops are not that impressive when there are $200 consoles, but $500 thin and light and cheap laptops with decent performance and not pathetic Atom/E-350 level performance have a lot of appeal.

 

How thin-and-light are you talking? I wouldn't be surprised at all to see A4 APU's (240 SP's) in 12-14'' notebooks less than an inch thin and under 4 pounds, maybe slightly over, for that price. The E2-3000M (160 SP's) will probably be in similar notebooks for cheaper.

 

Help, applications don't show up in the "recent applications" in the switchable graphics tab in CCC. I just played Team Fortress 2 and nothing, even the option is checked for "show popup warning for unassigned application"

Nothing ever popups up.

 

Let me just add one thing to the discussion:

You may let a video being encoded while you are doing something else e.g sleeping.
Can you sleep while you're waiting your game's frames to render?

 

Open team fortress 2.

Then ALT+TAB or windows key or whatever to get you to your desktop. Then open the switchable graphics configuration. You should see it now.

 

They have the drivers up now for the dv6z-6100CTO on HP's website. The graphic cards supported list is kind of worrisome.

AMD M880G with ATI Mobility Radeon HD 4250
AMD Radeon HD 6470M
AMD Radeon HD 6650M

One is a 512mb and the other is 1gb. I really hope these aren't the 2 cards they are using in them.

 

I wouldn't worry about that. There is no way it is right, seeing as the 4250 is the IGP on the Danube CPU motherboards and the 6470m/6650m aren't available with GDDR5. Not to mention that Notebookcheck has one in their hands and it has the 6750m

 

ULV SandyBridge (17W) level thin and light, such as the macbook air. Should be achievable with a 160 SP + dual core Llano, still going to be much slower than SB, especially without any HT, but if the price is right it will sell.

 

No way there is going to be something as thin and light as the MBA for $500. Like I said, I could see something like about an inch thin and around 4 pounds, similar to the Asus UL-series or Acer TimelineX, for around that price.

 

@ Abbadon: how did you manage to lower down the prices for your current specs.

I ordered mine thru. HP Academy and paid about $769.96

.......................
Another thing, anybody notice why the ship dates of this notebook suddenly pushed back to august 5 [3 weeks instead of the usual 2]

I was about to order yesterday and the ETA was Jul 22. a day has passed when im actually just about to order and now it says ETA is August 5.

Oh yeah here's the specs:
A6-3410mx
6g DDR3
1gb Radeon DDR5
Blu-ray
640gb 5400 rpm
....

 

Most likely there is a shortage of Llano motherboards but I still would assume it ships before the 5th. That is a worst-case estimate so HP doesn't get itself into trouble.

And I got mine with a 30% off coupon

Edit: PCMag has a review of the dv6-6135dx, in case anyone hasn't seen it.

http://www.pcmag.com/article2/0,2817,2388287,00.asp

 

Probably shortages of chips or motherboards. I ordered mine on June 29th, and it is still in production. It's "estimated" to shipping on July 12th, but I'm not holding my breath on that....

 

That review makes me happy, because everything I'm likely to do looks good (browsing, watching dvds and blue rays, and gaming, and probably a good amount of those will be done on the battery). I'm probably the target audience for the near maxed out 1080p dv6zqe version, especially with the 30% coupon....

 

Thanks for both the link and the answer. And yeah it was kinda late for me to realize the coupons doesn't stack with the academic discount.

 

Looks like a few samples of A8-3500M apus are listed on the passmark rankings

PassMark CPU Lookup


are those benchmarks legit? It scores above the sandy bridge i5 there, and it was lower in the anandtech benchmarks. I like the cpubenchmark site because it seems the most comprehensive listing of benchmarks (and best arrayed too btw, giant score cascade of EVERY cpu under the sun), but is it an accurate measure?


If there is any validity to it, it seems the new apus are just fine in terms of cpu performance if your programs can use all the cores.

Be nice to see the difference in the higher clocked chip, has anyone received theirs yet? And if so, where are the scores?!?

start posting guys, its not hard to get.

 

Passmark is pretty useless, go with Cinebench numbers for single and multithreaded benchmarks.

 

The reason it's like that is because Passmark is heavily multithreaded and as such a 4 core CPU will tend to beat a 2 core CPU.

But this is unrealistic as most tasks are optimized for single threads or maybe dual core, and do not scale as well to 4 cores as Passmark does. Look at the #1 CPU - its a server CPU thats 6 cores. But it would be absolutely and completely moronic to use this CPU to game. Some people might say that in the future that more cores will become useful, that's true and not true at the same time because the way games inherently work, you need a main render thread and there will be an inequitable reliance on a single thread to do the majority of the tasks. You can only multithread to a certain extent.

You could look at it another way - it takes 4 AMD cores to beat 2 Intel cores.

 

There is no way that the Passmark numbers for the A8-3500m and A8-3850 are correct. Even if all 4 cores were running at 2.4GHz on the A8-3500m, which hasn't happened in any other benchmarks we have seen, it would only score around 3200 tops. The A8-3850 score makes even less sense. At 2.9GHz it should be scoring no higher than 3700. The A4-3400m score could be correct, based on Danube and Deneb scores.

Passmark isn't a useless benchmark, by the way, at least once more scores get added for each individual CPU.

 

I looked at the chart the whole thing looks screwed up. A 2920XM at 2.5 ghz mobile processor beats a i5-2500k @ 3.3 ghz desktop processor. No way jose.

I will take a guess that it is because of the hyperthreading that is on the 2920xm but not on the 2500k. And if that is the case, that means that passmark is a bullsh*t benchmark because it is nowhere close to real computer usage.

 

Maxon Cinebench 11.5 Score Results


cinebench needs more scores, either that or more people posting real world results with different chips. Hopefully that will come.

 

We had this discussion earlier in the thread. As I said, Passmark is a random number generator. In your link above, forget Sandy Bridge and compare the 3500M to the processor immediately above it (the Phenom II X4 970). The difference in scores is 4452 for the Phenom to 4437 for the Llano. The architectures are similar, but the 970 is clocked at 3.5GHz and has 6MB of L3 cache whereas the 3500M is clocked at 1.5-2.4GHz and lacks L3 cache, but has more L2. The Phenom should still be superior clock for clock. Even if you assume that the 3500M is running at 2.4GHz the entire time (a big "if" given what he have seen in reviews and in this thread), it has still somehow managed to make up roughly 50% performance.

To be honest, I am kind of curious how this was done. Either AMD put some dedicated hardware in Llano that does whatever it is that Passmark is looking for (extremely unlikely and not consistent with earlier scores) or somebody has found a way to either overclock Llano or trick Passmark into thinking that they're running Llano whereas they're actually using another CPU. Has anyone tried overclocking the dv6z?

 

Except that the i7-2920xm can run all 4 cores at 3.2GHz, has more L3 cache than the i5-2500k and has hyperthreading. The i7-2920xm performing on par with, or even slightly beating, the i5-2500k is not a surprise. I don't see how hyperthreading means it us a bad benchmark or that it is nowhere near real computer usage. Cinebench takes advantage of hyperthreading as well and the i7-2920xm beats the i5-2500k in 11.5, 5.7 to 5.4. Althernai and I already had this exact discussion a few days ago.

@Althernai - Clearly someone found a way to upload a fake score. Either that or someone has overclocked the A8-3500m. I think the former is more likely.

 

Hypethreading can be useful but most of the time isn't because it turns 4 cores into 8 cores and even the most multithreaded programs are designed to take advantage of 4 cores only.

While HT should not make a lesser processor better than another one with higher clocks + turbo, L3 cache certainly could. Wasn't aware that the 2920xm had more...Still, in a real-world benchmark I think the 2500k would outperform the 2920xm. By real-world I mean typical consumer stuff, like running a game, graphical tools, media editing, etc. etc. not so much the pure computational power.

 

Those passmark scores are useless not only because of the fake uploading, but because of core utilization. Looking at Cinebench single and multi thread scores at reviews makes a lot more sense. Single threaded tasks are still very important in determining a regular user's overall usage experience with a CPU. That's why I chose a dual core at 2.6-3.3 Ghz over a quad at 2.0-2.7 Ghz, because I know it'll be faster for 99% of things I'll do with it. Of course cost and TDP also had something to do with it.

 

Most things that are CPU-intensive nowadays can take advantage of more than 4 cores to at least some extent. 3d rendering and video encoding, for example. Which is why the i7-2920xm beats the i5-2500k in Cinebench.

It depends on what you are doing but the i7-2920xm is still going to win in a few things and it won't be far behind in anything. I doubt anyone would notice a difference between the two.

Single threaded performance is arguably the most important factor in determining a regular user's overall usage experience. The regular user is more than fine even with the poor single threaded performance of Llano, though.

For the record, the reason I am so interested in the Passmark scores is because I would like to know if there is something that can make Llano turbo properly. If Turbo Core is working in Passmark, which would seem to be the case for the A6-3400m at least, I am curious as to why it is there but not in other benchmarks.

 

Of those who ordered, who's actually gotten their dv6zqe's from HP online (or at least a tracking number)?

I'm curious as to whether I can expect mine anytime soon, especially since I've heard all kinds of rumors as to parts availability (some people saying that the cpus/mobos are scarce, some saying that the 1080p screens are scarce)....

 

I ordered mine on 6/26, the specs are in my sig. It was supposed to ship on 7/7 but I got a delay email on 7/8. My new rescheduled ship date is 7/22, Hopefully it ships out sometime before then. At this point I have no idea what is going on, but I hope they clear it up soon.

 

My estimated ship date is Thursday the 14th, ordered on the 1st. I am not getting my hopes up, though. I am fairly certain that the delay is not with the 1080p screens anymore, as people who ordered the dv6/7t with them are getting their laptops, but with Llano motherboards or the actual chips.

 

The regular user is fine with a Bobcat if you're going that route. There is a reason why there are faster CPU's because people don't like waiting.

It's far more likely that the passmark scores are bogus than anything else. All the reviews put Llano through a battery of tests and TC didn't really work as you expected reading from the marketing materials.

 

Agreed. Still, I don't want to have this argument again but the average user isn't going to notice a speed difference between Llano and SB.

The scores for the the A8-3850 and A8-3500m are undoubtedly fake but the interesting thing is that the A6-3400m scores about what would expected if all 4 cores were running at 2.3GHz. I also don't understand why you seem to insinuate that it was me who expected TC to work differently than it does, as if no one else did. Everyone who actually read what AMD had said about TC expected it to work differently.

 

I disagree with this, but we probably have different definitions for "average user" so it doesn't make much sense to argue that. AJAX is a cruel thing when you have multiple tabs open on Firefox and I do get 100% cpu usage for 10 continuous seconds at times. Do you never experience windows "graying out" the currently active window and tell you that a program is not responding, but if you wait a bit it comes back? I do get that sometimes with firefox and I'm glad I have a fast CPU so that I am not forced to kill tasks and start them over again.

I didn't mean that of course, it came out wrong. I meant exactly what you've said here and I still think it's pretty much 100% bogus with those passmark scores.