Monday, May 16th 2011

AMD's Fastest Mobile Fusion Processor Detailed

One of AMD's design goals for its Fusion architecture must have been its application in mobile processors, where AMD isn't the brightest in terms of energy efficiency. The company's fastest product for the notebook platform is based on the Llano silicon, will release in 2011, and is a quad-core APU. Enter the AMD A8-3530MX.

Built on the 32 nm process with FS1 package, the A8-3530MX is an accelerated processing unit (APU), it combines a quad-core processor with a powerful GPU and northbridge component. With a default clock speed of 1.90 GHz and TurboCore speed of 2.60 GHz, the A8-3530MX packs 4 MB of L2 cache, and a dual-channel DDR3-1600 MHz memory controller, that also supports DDR3L-1333 MHz.

The GPU component is the DirectX 11 compliant AMD Radeon HD 6620G, with 400 stream processors, and engine clock speed of 444 MHz. To drive it, the northbridge component is entirely integrated into the APU, which packs a PCI-Express 2.0 hub to support discrete graphics. Despite so much machinery, the top-of-the-line chip maintains a TDP of 45W (common for notebook quad-core chips).

The AMD A8-3530MX will be backed by a single-chip chipset, which isn't much more than a southbridge (just like PCH on the Intel platform). The platform will launch with AMD A60M chipset, and later the company will introduce A70M, which packs native USB 3.0 support. That's not to say that A60M-based notebooks won't feature USB 3.0, they'll just use third-party controllers like Intel-platform notebooks do.
Source: DonanimHaber
Add your own comment

32 Comments on AMD's Fastest Mobile Fusion Processor Detailed

#26
mastrdrver
ThefumigatorTo me quad cores in laptops are special. I wasn't making my point on how long they've been available... but only taking the fact there's a quad core in them. I mean, I've never had a quad core laptop in my hand, nor even know a friend who owns one, also don't even serviced one.

Moreover, look at newegg:
quad core laptops available: 86 (core i7) + 11 (Phenom II X4) total of 97 laptops

dual core laptops available: 134 (Core i3), 233 (Core i5), 76 (Core 2 Duo) and most of the rest AMD like 50 to 100 more. Total of 400~500 laptops.

Of course Llano just makes the whole thing even more special to me.
With most laptop use just being an extension of the mass desktop use is the overly abundant dual cores really a surprise? Most people want longer battery life too and that is another spot that a quad will take a slight hit (in laptops only since they usually are a higher watt chip then a dual core version). You can get a similar watt quad as a dual but your clock speeds will take a hit and so will the performance. Just look at the quad vs dual of both AMD and Intel laptops. Quads (outside of turbo speeds) are done at ~2Ghz. Their similar family dual core brothers are clocked 2.5-3Ghz. While the watts between the two are the same the clock speeds are vastly different which will be noticed is just the normal "mass" laptop use.
Posted on Reply
#27
swaaye
All CPU manufacturers will have refined Turbo like tech soon. It is how multi core CPUs will match dual cores on low thread apps. For Intel it is already the case with the mobile quads having very aggressive Turbo.

Llano is interesting but if the cores are indeed based mostly on Phenom, Intel will destroy them on CPU performance. Even old Nehalem based Core chips were doing that let alone say Ivy Bridge. And the Llano IGP is going to be very bandwidth limited and slow compared to all but the worst discrete GPUs. To me it's a new generation of budget IGP platform but I'm not really excited by it because frankly it will be a barely adequate gaming platform and current IGPs do video aspects fine already.
Posted on Reply
#28
Thatguy
futurechipsSorry for being blunt but how is risc vs cisc even relevant to this argument? ARM is not RISC by any means and thats what Windows is being ported to.
---
futurechips.org
umm arm is a risc type architecture, it has everything to do with decoding and register capability.

every processor that approach the mips power of x86, invariably ends up with the same power draw. Its just a fact of the design.

each has its place, and the windows 8 port to arm will be a slow boat to china vrs x86 clock for clock.

Arm is only popular becuase its cheap.
Posted on Reply
#29
Thatguy
TheLaughingManThey can never have Intel's Hyper Threading because it is copyrighted and Intel don't like to share. That is called good business. And you really should say Intel's HT because AMD has a tech shorted to HT called Hyper Transport. It can be confusing when you say HT only.
hypertransport is nothing like hyper threading. hyperthreading is a secondary pipeline scheudler that attempts to cram the core pipeline stages full of instructions at all time. In apps with piss poor non optimized coding, it really helps speed them up. If you had a app with a big deep pipeline need and lots of heavy lifting, HT in a intel chip can actually degrade performance if it starts prempting the pipeline and flushing it.

Hypertransport is a cpu interconnet and other high speed interconnect for high speed data buss, similar to DMa in a few conceptual ways.
Posted on Reply
#30
Imsochobo
swaayeAll CPU manufacturers will have refined Turbo like tech soon. It is how multi core CPUs will match dual cores on low thread apps. For Intel it is already the case with the mobile quads having very aggressive Turbo.

Llano is interesting but if the cores are indeed based mostly on Phenom, Intel will destroy them on CPU performance. Even old Nehalem based Core chips were doing that let alone say Ivy Bridge. And the Llano IGP is going to be very bandwidth limited and slow compared to all but the worst discrete GPUs. To me it's a new generation of budget IGP platform but I'm not really excited by it because frankly it will be a barely adequate gaming platform and current IGPs do video aspects fine already.
GT220 from nvidia sucked c... (and it's not even a IGP)
cpu power is abit overrated, phenom II's are fine, 1.9ghz is abit low, if they managed 2.1ghz it'd be very good! (turbo 2.7ghz)
amd is on the way to more balanced computing, and I think Intel will have to aswell, when they get better graphics.
Kirsna will have 30gb/sec bandwidth that will be quite good, that is fusion A 2012 series.
if you have sideport with 70gb/sec 128bit GDDR5 with shared 28gb/sec on krisna it will be quite good for 2012 apu performance.
it all depends on sideport or not. either way it'll bash intels and nvidia's low end solutions.
Posted on Reply
#31
TheLaughingMan
Thatguyhypertransport is nothing like hyper threading. hyperthreading is a secondary pipeline scheudler that attempts to cram the core pipeline stages full of instructions at all time. In apps with piss poor non optimized coding, it really helps speed them up. If you had a app with a big deep pipeline need and lots of heavy lifting, HT in a intel chip can actually degrade performance if it starts prempting the pipeline and flushing it.

Hypertransport is a cpu interconnet and other high speed interconnect for high speed data buss, similar to DMa in a few conceptual ways.
Ya, I know what they are. I was simply stating you can't just use a phrase like, "But what about HT?" because no will know what you are talking about.
Posted on Reply
#32
devguy
TheLaughingManYa, I know what they are. I was simply stating you can't just use a phrase like, "But what about HT?" because no will know what you are talking about.
HTT = "Hyper Threading Technology"
HT = "Hyper Transport"

Solved. Reminds me of the little punks who mistakenly spread annoying rumors back in 2005; "Xb0x 36ty has no HD!!!111!1". They meant "HDD", but their stupid error resulted in a lot of misinformed people.
Posted on Reply
Add your own comment
Nov 28th, 2024 21:45 EST change timezone

New Forum Posts

Popular Reviews

Controversial News Posts