Calling Thrax..... How long until graphics cards are forgotten tech?
Never. Literally never for users like us. Mom and dad? Your Sims-playing little sister? Not too long, now. But for as long as there are enthusiasts like us, there will never be enough room for all the transistors that make us happy.
There's also the small matter of GPU transistor advancements more or less matching that of the CPU, which means any die shrink that makes room for more GPU will be matched on the GPU side with a more powerful graphics card. In other words, the chasm between $100+ GPUs and IGP will remain, but performance will continue to improve on both sides.
All that said, I am seriously excited about Llano for my home theatre PC. Because AMD has finally migrated PCI Express lanes onto the processor, the need for a large northbridge has been obviated. In sweeps the Hudson FCH with its native SATA6/USB3/etc/etc, and the era of a two-chip AMD platform has finally begun. This will allow motherboard manufacturers to develop serious mITX solutions, whereas older designs based on the AMD 800 Series had compromises--SODIMM slots or no PCIe--because that fat northbridge was just taking up too much room.
My personal HTPC is an mITX Core i3, a Radeon HD 6670 and 8GB of RAM. Because the decode/post-processing performance of Llano's on-die GPU is so high, easily on par with anything in the 6000 Series (which continues to lead the pack in video decode quality/performance!), I can eliminate the discrete GPU in favor of a Llano to reduce system power consumption, noise and size.
I understand that thought. What I could see though, is a home enthusiast PC market that has parts that scale like a multi chip server board? I'm letting my imagination run wild, but imagine a point where you would have a motherboard with two, three or even four standard sockets all interlinked to the architecture in such a way that its doubling, tripping or quadrupling your performance. I can almost paint that vision in my head, how it would work without the need for clunky expansion cards. Maybe not in the immediate future, but I can see if becoming a reality.
Why are expansion cards and video cards a bad thing? Don't you really enjoy assembling a new computer?
On the rare occassion that I get to build a new PC either for myself or someone else, I make a big deal about putting each piece in place and tightening the bolts down. It can take me a good hour or more to put the parts in because I make a major construction project out if it. But it is fun to do.
I agree Tim, from a certain perspective, if it ever came to that, I'd miss doing the card installs, but if you could leverage the same amount of power just by dropping in another chip and slapping a heatsink on it, you would have to admidt, it would just be a more elegant solution from a certain perspective.
Just wondering if anyone has any thoughts on how the APU will affect heat management in laptops. Currently if i'm not mistaken the usual deal for a good laptop is to either make it large with a full speed chipset, or to make it smaller and thinner and clock down the chips used.
If i'm not mistaken, using an APU decreases the power draw (as compared to 2 chips), therefore for the same amount of heat/power you should be able to run more things faster.
In short I'm saying this is good for desktops, but potentially game changing for laptops.
AMD A8/A6/A4 eliminates the northbridge, and eliminates the need for discrete GPUs in every class of notebook except for dedicated "gaming" notebooks that will always have discrete GPUs. It also dramatically raises the GPU performance bar for notebooks that don't have discrete GPUs.
Intel has already eliminated the northbridge last year with Westmere, but their IGP still blows, and Llano's doesn't. So really, not only does Llano dramatically reduce platform power draw (1 less chip on the chipset, no discrete GPU circuitry), it also saves on BOM cost by reducing the size of the internals and the number of required components--that's savings for you and me.
Just wondering if anyone has any thoughts on how the APU will affect heat management in laptops. Currently if i'm not mistaken the usual deal for a good laptop is to either make it large with a full speed chipset, or to make it smaller and thinner and clock down the chips used.
If i'm not mistaken, using an APU decreases the power draw (as compared to 2 chips), therefore for the same amount of heat/power you should be able to run more things faster.
In short I'm saying this is good for desktops, but potentially game changing for laptops.
Indeed it is game changing. The GPU-side configuration is identical to the A8-3850 with the exception of speed: 444MHz vs 600MHz for the desktop. Considering that most laptops using this APU will have a max resolution of 1600x900 this shouldn't be a problem. I'd wager that power and heat won't really be an issue either. I have the A8-3850 on the review bench right now. It should be finished very soon - power and heat are the last remaining tests. Sabine (the mobile version) has a 45W max TDP, so I'd expect it to use less than half of the test CPU. Heat shouldn't be an issue either.
mertesn
I am Bobby MillerYukon, OK Icrontian
Comments
Besides, there's nothing saying you can't add a modern GPU to the system.
How long until we can take a single chip and plug it into a mobo that supports high bandwidth DDR5??
Calling Thrax..... How long until graphics cards are forgotten tech?
Never. Literally never for users like us. Mom and dad? Your Sims-playing little sister? Not too long, now. But for as long as there are enthusiasts like us, there will never be enough room for all the transistors that make us happy.
There's also the small matter of GPU transistor advancements more or less matching that of the CPU, which means any die shrink that makes room for more GPU will be matched on the GPU side with a more powerful graphics card. In other words, the chasm between $100+ GPUs and IGP will remain, but performance will continue to improve on both sides.
All that said, I am seriously excited about Llano for my home theatre PC. Because AMD has finally migrated PCI Express lanes onto the processor, the need for a large northbridge has been obviated. In sweeps the Hudson FCH with its native SATA6/USB3/etc/etc, and the era of a two-chip AMD platform has finally begun. This will allow motherboard manufacturers to develop serious mITX solutions, whereas older designs based on the AMD 800 Series had compromises--SODIMM slots or no PCIe--because that fat northbridge was just taking up too much room.
My personal HTPC is an mITX Core i3, a Radeon HD 6670 and 8GB of RAM. Because the decode/post-processing performance of Llano's on-die GPU is so high, easily on par with anything in the 6000 Series (which continues to lead the pack in video decode quality/performance!), I can eliminate the discrete GPU in favor of a Llano to reduce system power consumption, noise and size.
It's awesome.
On the rare occassion that I get to build a new PC either for myself or someone else, I make a big deal about putting each piece in place and tightening the bolts down. It can take me a good hour or more to put the parts in because I make a major construction project out if it. But it is fun to do.
To have a PC the size of a gamecube that can do the things my PC does now would be so badass
I like AMD and their APU direction
If i'm not mistaken, using an APU decreases the power draw (as compared to 2 chips), therefore for the same amount of heat/power you should be able to run more things faster.
In short I'm saying this is good for desktops, but potentially game changing for laptops.
AMD A8/A6/A4 eliminates the northbridge, and eliminates the need for discrete GPUs in every class of notebook except for dedicated "gaming" notebooks that will always have discrete GPUs. It also dramatically raises the GPU performance bar for notebooks that don't have discrete GPUs.
Intel has already eliminated the northbridge last year with Westmere, but their IGP still blows, and Llano's doesn't. So really, not only does Llano dramatically reduce platform power draw (1 less chip on the chipset, no discrete GPU circuitry), it also saves on BOM cost by reducing the size of the internals and the number of required components--that's savings for you and me.