Slashdot Mirror
Unique Dell XPS M1710 Review
Searching4Sasquatch writes "Hot Hardware has just posted a unique review of Dell's flagship XPS M1710 notebook. They stumbled across some very interesting information within the BIOS which seems to indicate Dell is working on a docking station with its own discrete graphics. 'The user is given the option of using either the integrated GeForce Go 7900 GTX GPU found within the system or the extremely interesting option of using the graphics card found within a docking station. Could Dell be planning on releasing an enthusiast dock that features a high-end GPU that could not otherwise be crammed into the confinements of the notebook chassis? Perhaps an upgrade to allow for standard or even Quad-SLI would be possible with such a dock.'"
Mobile Gaming is becoming somewhat of a buzzword, but I think this idea has some potential to it in a slightly different way. For those of us who don't necessarily enjoy lugging around a 12 lb notebook just for the occasional gaming opportunity, Dell or whoever could create a very portable notebook that docked into a more sophisticated machine, thereby allowing for a small family to go back to one PC. Of course, it's probably not in the best interest of the manufacturer, since they want us all to have as many PCs as we can stand, but it makes sense for the consumer. There are docks out there with built-in hard-drives, why not built-in video cards and extra RAM, and even a bigger monitor? Having everything on one PC would be beneficial to a lot of people.
Too bad what we want and what manufacturers want us to want are often two different things.
Yes, you can dance to Radiohead.
On my Dell laptop. The dock has a dual card with two monitors.
I use one for the program, one for the debugger, and the laptop screen for email.
I've ONLY had this setup for 4 years!
Never trust a man wearing a coat and tie!
I never quite understood why someone would buy these really pricey gaming laptops. For my boys I built microATX cubes that have every bit as much performance as these high end laptops, for about 1/4th the price -- and they are easily and cheaply upgradeable down the road. When they go to a friend's house for a LANparty they just grab the cube by its handle and throw their keyboard/mouse into a bag. Monitors are not a problem -- most people have monitors leftover in their basement/attic from when they upgraded to LCD, so they just connect to the surplus monitor, plug into their network and off they go. Seriously, you can build a nice cube gaming box for about $550 (DVD writer, Athlon 64 3500+, 1GB DDR400, 300GB SATA HD, Windows XP license, box w/420W supply, motherboard) plus whatever graphics floats your boat (I find the $99 NVidia 6 series PCIe boards are more than adequate, though I have also found that many games are actually quite playable using just the embedded graphics like the NVidia 6150). Sure, you may be 10 or 20 fps slower than your buddies, with a little less detail in the shadows, but who cares (especially when most LCD monitors top out at 60Hz refresh rate anyhow ;-).
The more you regulate a company, the worse its products become.
Why carry something that can drive a display much heavier-duty than the onboard LCD?
Why not? It adds nothing to the cost, in comparitive terms.
If you've got dual monitors, you're using a dual monitor card, which isn't necessary for the much more common single monitor, and therefore much more expensive in small production quantities
Except that dual head cards are now entirely standard (virtually all cards over $30 are dual head capable) and so we're not talking small production quantities we're talking huge quantities. The cost is utterly negligible.
In other words, distributed GPU provides the same economics and efficiencies as distributed networks of host computers
No, it really doesn't. Distributed computers work well (for certain problems) because of huge scale - hundreds or thousands of nodes. This lets you scale easily by adding more nodes, and replace dead nodes without adversly affecting performance. Clustering can also be used to make use of cheaper components at the expense of some performance - 16 single CPU nodes will not perform as well as one 16-cpu beast.
On the other hand, with graphics we're looking at a typical maximum of what, 8 screens? With the average being between 1 and 2? That's just not comparable. It's not a "distributed" GPU, it's a "remote" GPU. The average user with one screen will still have one GPU, it'll just be on the end of a wire. What does that buy him? Nothing. In the case of a laptop user who occasionally uses a monitor and projector - he now needs 3 GPUs. Again, why? One works just fine for him today. Where you get into people with multiple screens in use at once, like myself, again - what does it buy me? I currently have one GPU - why have two? Each would need the same amount of texture ram as my current single card, so there's double the cost right away. They'll also need some way to communicate with the host and each other which just adds to the bandwidth requirements and complexity of your communications solution. All this to replace a simple uni-directional cable which works perfectly well. I simply don't get it.
And as a final point, I still think you underestimate the bandwidth requirements. Look at this page. We can see that PCIe x16 is 40GBit/s. That's huge - it's 8 times faster than the fastest SCSI standard and 30 times faster than fiber channel - in fact it's faster than most ram interfaces. Now I'm no electronics expert, so I'm not saying it's impossible, but getting that kind of speed in a 2 or 3 meter cable which is (a) cheap and (b) reliable seems non-trivial. The only cabled standard I can find with comparable speeds is OC768 which is far from cheap or commonplace.
---- Den ene knappen er powerknapp, den andre er Bender voice knapp "Bite My Shiny Metal Ass"