AMD Fusion Info Thread

[abaddon4180]

So in other words a A8-3830MX could hit 2.6Ghz with all cores if using a discrete graphics card like a HD 6770M with CrossFire disabled (no IGP).

Theoretically, yes. With the IGP disabled the entire 45W TDP can be used for the CPU and the highest Danube quad had a 45W TDP and ran at 2.4GHz, the 32nm die shrink should allow for 2.6GHz on all 4 cores.

I just need a review on any of the MX CPU's with a discrete graphics card.

Link 1: http://h10010.www1.hp.com/wwpc/ca/en...2-5128782.html

Link 2: http://www.amazon.ca/HP-DV6-6140CA-1.../dp/B0053OSH8K

From Link 1 HP has the NoteBook listed as A8-3510MX CPU and 1GB DDR3 for the 6755G2 Graphics Card.

From Link 2 Amazon has the NoteBook listed as A8-3500M CPU and 1GB GDDR5 for the 6755G2 Graphics Card.

I'm willing to be the first guinea pig on a Llano laptop for the sake of the thread if the laptop above is indeed a A8-3510MX with 1GB GDDR5.

I assume it is the 6775G2 (6750M + 6620G) with GDDR5 because I am pretty sure the 6750 doesn't have a DDR3 version. I would also guess it has the A8-3510MX

Seriously, you're asking that question? We're all governed by physical constraints in the world and CPU's aren't any different.

You don't want to OC the CPU in a hot room in the first place! Intel will allow as much OC as it's safe under all conditions. Since you don't know the room temperature in AMD's case, even a room at the north pole would have to be treated as if it was too hot, limiting performance in such cases.

AMD turbo breaks the laws of physics? It can't change semiconductor process or cooling system variations from system to system. Intel makes sure that no matter what CPU and cooling system you use, you're getting the best performance out of it that's still thermally safe. Improve your cooling or be lucky to get a CPU that was more in the sweet spot of the die, and you'll get more performance. Intel Turbo Core is better under all circumstances than AMD's version unless they start measuring the actual, real time temperatures on the chip.

Again, providing that AMD tested these in the max ambient operating temperature for the APU, which I am pretty sure they would have done, I don't see how it isn't "thermally safe". As long as you aren't using the notebook outside the max operating temp for the APU, it shouldn't be a problem. AMD doesn't have to worry about the cooling systems from system to system because their turbo is not affected by temperature. Althernai already summed it up better than I could.

To some extent, yes. The benefit of AMD's approach is the results are more consistent: barring extreme temperatures which force all processors to throttle, you should get the same Turbo results regardless of the ambient temperature and specific cooling solution. The benefit of Intel's approach is that you're getting the most performance out of your environment. It's not as consistent (note that the 2630QM results in the AnandTech article vary depending on the laptop), but the CPU is always performing as well as it can.

[Nemix77]

I'm contacting HP Canada to ask about the real specs on the DV6-6140CA.

If it's indeed a A8-3510MX with HD 6750M 1GB GDDR5 then I'll buy it with a US (non-bilingual) keyboard installed direct from HP Canada.

[soguxu]

Again, providing that AMD tested these in the max ambient operating temperature for the APU, which I am pretty sure they would have done, I don't see how it isn't "thermally safe"

Then you're not turboing as much as you can when it's cooler, leaving performance on the table. For AMD, it's either burning CPU's or weak turbo, as long as they don't measure the CPU temps. We all see they went with the weak turbo route.

[Nemix77]

So apparently HP Canada has no info on the model and says it's a retail model.

Amazon Canada has no stock yet.

[abaddon4180]

Then you're not turboing as much as you can when it's cooler, leaving performance on the table. For AMD, it's either burning CPU's or weak turbo, as long as they don't measure the CPU temps. We all see they went with the weak turbo route.

They are turboing the same amount whether it is cool or hot. What is so hard to understand about that?

The only thing we all see right now is that the turbo on the test models did not work like it is supposed to. AMD has admitted that there is a problem with the BIOS.

[Phinagle]

Seriously, you're asking that question? We're all governed by physical constraints in the world and CPU's aren't any different.

You don't want to OC the CPU in a hot room in the first place! Intel will allow as much OC as it's safe under all conditions. Since you don't know the room temperature in AMD's case, even a room at the north pole would have to be treated as if it was too hot, limiting performance in such cases.


AMD turbo breaks the laws of physics? It can't change semiconductor process or cooling system variations from system to system. Intel makes sure that no matter what CPU and cooling system you use, you're getting the best performance out of it that's still thermally safe. Improve your cooling or be lucky to get a CPU that was more in the sweet spot of the die, and you'll get more performance. Intel Turbo Core is better under all circumstances than AMD's version unless they start measuring the actual, real time temperatures on the chip.

Intel's approach is more conservative but that doesn't necessarily mean it's better.

As has been said, a notebook's cooling system is already designed to monitor heat and take action to prevent a meltdown regardless of whether or not there's a Turbo feature involved or not. Intel's Turbo Boost check on CPU temp is essentially redundant, and while a redundancy is safer, Intel's approach doesn't take into account what temps a specific system's cooling can handle. If Intel sets it's Turbo cut off at say 90 degrees, and ambient temps are hot enough to have the CPU currently running at 90 degrees then Turbo isn't going to kick on...it wouldn't matter if the notebook's fans were only running at half speed and set to ramp up at 91 degrees.

AMD's approach doesn't use that redundancy, but Llano's temps at Turbo Core will be a known factor for a given ambient temp, and notebook manufacturers should design their cooling to handle those temps. There's nothing different there from how notebooks have always been designed.


Where AMD's approach could have a real advantage is user overclocking. Since Turbo temp controllers are not in the way OCing should be more like older non-turbo CPU, and depending on how the conditions for Turbo Core are set AMD could introduce a feature like Cayman's PowerTune to the AMD Overdrive utility.

[abaddon4180]

So apparently HP Canada has no info on the model and says it's a retail model.

Amazon Canada has no stock yet.

According the most places, they aren't supposed to be available until July. I have seen sources that say Q2, though, which would mean June.

[Nemix77]

Wait, AMD admitted there's a problem with the BIOS on the Compal laptops with A8-3500M sent to reviewers that affected Turbo Core?

[Althernai]

Like I said earlier, I think we have to assume right now that turbo on the samples is broken.

I don't know. When a reviewer (particularly from a major site) gets something wrong, they typically get contacted by the manufacturer and issue an update saying "Company X has written to us and it turns out that result Y was due to problem Z." Llano has been reviewed by pretty much every major site and they all get consistent CPU performance. Somebody at AMD took the trouble to send them these reference laptops, surely somebody there read at least one of these reviews and would have written to the reviewers by now had something been out of the ordinary.

If Turbo is really broken, then there is a serious failure at AMD's marketing department. The findings of the reviewers are propagated through forums like this one and word of mouth so the bad performance is essentially a seed that spreads with time. The longer they wait, the worse it becomes. By the time laptops are in the hands of users, most will have been convinced that the CPU performance is bad. Even if the real laptops perform a lot better, instead of unambiguous statements, discussion of them will turn into debates -- people will link the reviews and there will be doubt of who is right.

I'm kind of puzzled by AMD's marketing strategy in general. When Intel made a reference Sandy Bridge laptop, they packed it with their second most potent quad-core and a fairly large battery. It was designed to send a very definite message and all of the reviewers got it and repeated it. When AMD made a reference Llano laptop... well, we still have no idea what is going on with Turbo Core and hybrid CrossFire as currently implemented is obviously broken (which the reviewers had to find out the hard way; AMD didn't tell them). They got the battery and iGPU part right, but why send a machine with buggy firmware? And for that matter, why not send an MX processor with a 45W TDP? It can almost certainly Turbo higher.

[soguxu]

Wait, AMD admitted there's a problem with the BIOS on the Compal laptops with A8-3500M sent to reviewers that affected Turbo Core?

They admitted there's a problem with the bios due to hybrid crossfire not working. They did not say Turbo Core is effected, that's Anand's speculation.