Acer: AMD's XGP External Technology No Good
Ouch. Acer didn't speak too fondly of AMD's XGP technology.
TechRadar reports that Acer isn't too keen on AMD's XGP external graphics technology. The feelings were conveyed to the website during a query about the release date of Acer's external ATI Radeon graphics card, slated to ship this year. According to the report, it appears that the card has now been placed on the back-burner. In TechRadar's actual words, Acer responded to the query with a "less than flattering description of XGP."
While TechRadar didn't offer a full disclosure of the transaction, Acer apparently still shows some interest in the external GPU technology. "Acer is interested in this solution, but at the moment it doesn't offer the best user experience," said a spokesman for the company. AMD was a little more positive, saying that it was working closely with Acer to bring the XGP product to the market.
"The Acer Ferrari One [netbook], now available for purchase, does include the XGP port," AMD told the site. "AMD demonstrated a prototype solution with this notebook at International CES 2010 in Las Vegas at the beginning of this year."
For laptops without graphical horsepower, AMD's XGP technology (short for eXternal Graphics Platform) could be a huge boost to consumers wanting to play better games or survive the flood of embedded Flash objects littering the Internet. Unfortunately, the XGP market is scarce, virtually non-existent at this present time.
That's just it though they're designing the platform and interface from scratch theres no reason they shouldn't be able to get it up to atleast the speed of a PCIEx8... as for the complaint from Acer it's new tech, show me a piece of new tech that worked right out of beta without flaws.
The on-board would still get better bandwidth, and that counts for a lot. XGP is a neat idea, but not very practical for exactly that reason.
When I first looked at this I thought that AMD had finally created external graphics cards, after reading this up at Wikipedia I'm still confused at what this is for'
there's always the ATI website.... granted it's probably got they're marketing spin on it but should give you a feel for the tech
interesting, could be eyefinity for laptops that just dotn have the space for 6 ports or even better still..... hybrid crossfire with compatible ATI IGP, mostly though i just like the idea of not having to rely on Sony/Dell/vendorX releasing half a$$ed drivers thats are not supported for the latest OS cause they want me to upgrade my laptop so i can use the TV out which worked perfectly fine on the old OS..... dont i just love Intel GMA chips........
wcooper007: PCI-E 2.0 supplies 500MB/s per direction per lane. That is 4gbps. If you are talking about a 16x 2.0 bus running in full duplex mode then you get 128gbps total bandwidth. Once PCI-E 3.0 comes into popularity that bandwidth will double. In terms of internal connections (which is what you would want for a GPU) light peak is quite slow. If they want to make a really good product they could do what another poster recommended and have an actual PCI-E slot and power connector plus a case to put your dedicated video card in.
Using a high speed interface for a video card is a low-end solution. It means you didn't put enough memory on the card. Using system memory slows things down, either way. If you use a slow interface, it slows it down even more.
GPUs get commands to do work, which are tiny. Provided they get them quickly (i.e. low latency), and work with their own memory, interface bandwidth is essentially irrelevant, for a single card. The downside is applications that use the GPU for things residing in system memory, to assist the CPU.
So, they'd have to sell these things with lots of memory if they don't have great bandwidth, and they'd probably be a bit more expensive, but, then, laptops always tend to cost more, and have more expensive parts. They'd probably also want to avoid apps that work with system memory, when possible.
GPUs get commands to do work, which are tiny. Provided they get them quickly (i.e. low latency), and work with their own memory, interface bandwidth is essentially irrelevant, for a single card. The downside is applications that use the GPU for things residing in system memory, to assist the CPU..
You forgot about texture loading. which is not tiny.. That needs huge bandwidth. Even since Hardware TNL was introduced since Geforce 1 days, Hardware shaders etc etc. Textures have to be in Graphics memory for lighting and shading processing, where they originate from! your CPU and System memory, and it is constantly moved too and removed from the GPU ram in large amounts.
Where you get your Info from? Interface Bandwidth is extremely relavant.
or how is it going to get plugged in.
probebly work good on the new fiber usb thing.