AMD Fusion Info Thread

[Phinagle]

The specs for the Tosh notebooks with Llano that I've seen were kind of fuzzy on whether or not they're using a discrete GPU.

[RWUK]

Where is AMD sitting with DRM relative to Intel? Sandy Bridge was the first full hardware implementation of DRM, yes? I've read some articles about AMD deciding on this in both CPUs and ATI cards back from about 2007/8 but I don't know what happened from there. Where did this head and will Llano have this feature? AMD offers little to no information about their CPUs on their website.


EDIT: also wanna add, here are some Llano benchmarks on Ubuntu 11.04
http://www.phoronix.com/scan.php?pag...no_linux&num=1

[KaranX]

AMD's TDP is a little bit strange. Here's the top-of-the-line 45W part:

A8-3530MX
6620G
2.6GHz/1.9GHz
4 cores
4MB cache
400 stream processors
444 MHz GPU clock
DDR3-1600/DDR3L-1333 RAM

And here is another 45W part:

A4-3310MX
6480G
2.5GHz/2.1GHz
2 cores
2MB cache
240 stream processors
444 MHz GPU clock
DDR3-1333/DDR3L-1333 RAM

It has half the number of CPU cores at very nearly the same clocks, half the cache, 60% of the stream processors clocked at the same frequency... and the same TDP. You would think a chip that is essentially half of another chip operating at more or less the same frequencies would have the TDP likewise cut in half, but for the A4-3310MX, it remains the same. It's not clear to me how this is possible -- the quality of the silicon would have to be very different. I wonder if the A4s are low binned A8's with sections disabled to keep the TDP down.

I read somewhere, maybe I'll get the link later, that the A4's had a section of their processor fused/disabled/locked, so maybe this explains the 45W TDP.
Maybe it is the same chip.

[Althernai]

I read somewhere, maybe I'll get the link later, that the A4's had a section of their processor fused/disabled/locked, so maybe this explains the 45W TDP.
Maybe it is the same chip.

That would not explain the TPD. When a part of the chip is disabled, it draws no power (no sense wasting energy on something that won't be used). Thus, the TDP should be substantially lower. The fact that this is not the case is indicative of variation in the quality of the silicon.

[Cloudfire]

That would not explain the TPD. When a part of the chip is disabled, it draws no power (no sense wasting energy on something that won't be used). Thus, the TDP should be substantially lower. The fact that this is not the case is indicative of variation in the quality of the silicon.

Isn`t it the maximum TDP which the system is required to dissipate?

[Karamazovmm]

Isn`t it the maximum TDP which the system is required to dissipate?

somewhat that is the amount of heat that needs to be dissipated under the load conditions of that component.

People always forget that a chip acts as a resistor

[Nemix77]

For everyone waiting for Llano, I spoke directly with AMD just the other day.

All MX CPU's with 45W TDP have a Turbo Core that works like "Bulldozer" on the desktop without the L3. Meaning it can do half the max Turbo with all cores active and maximum Turbo with half the cores active.

Example:

A4-3310MX (Dual-Core): 2.1Ghz - 2.5Ghz, 2.3Ghz with 2 Cores active and 2.5Ghz with 1 core active
A6-3410MX (Quad-Core, DDR3-1600) : 1.6Ghz - 2.3Ghz, 1.9Ghz with 4 cores active and 2.3Ghz with 2 cores active
A8-3510MX (Quad-Core, DDR3-1600): 1.8Ghz - 2.5Ghz, 2.1Ghz with 4 cores active and 2.5Ghz with 2 cores active
A8-3530MX (Quad-Core, DDR3-1600) 1.9Ghz - 2.6Ghz, 2.2Ghz with 4 cores active and 2.6Ghz with 2 cores active

Please do not quote me, all I know is Turbo Core will be able hit the maximum with the 45W TDP CPU's.

[Althernai]

That's still pretty pitiful, but it's better than what they showed in the reviewers' benchmarks. I wonder why they didn't ship an A8-3530MX with the review laptop. It would at least have been able to beat the Arrandale dual-core in multi-threaded performance (Sandy Bridge dual-core still beats it though).

[Nemix77]

I asked him about it but he didn't say.

It's obvious AMD want to to be inline with SB in battery life tests and still have a better better IGP/APU, quad-core CPU while producing less heat/noise (not in reviews).

Even with the crappy Turbo Core on the A8-3500M it's still comparable to a i3 in multi threaded applications, the HD 6620G APU clearly beat the SB HD 3000 and the battery life is as good as SB. All this in a 35W TDP CPU that has cripple Turbo Core but who's looks at the single threaded CPU application benchmarks anyway, everyone only cares about the APU performance, battery life and quad-core.

It's very smart mass marketing move by AMD, to erase the image that AMD laptops have low battery life, run hot and the on-board graphics is not as good as discrete. Enthusiasts who care about Turbo Core, DDR3-1600 and Hybrid CorssFire wouldn't mind a bit less battery life, more heat and will most likely get a discrete Radeon to go with their Llano laptop anyway.

[Althernai]

Would the MX series really have worse battery life though? I would think that when down-clocked, it wouldn't matter (obviously, when gaming it would, but nobody really cares about that).