PDA

View Full Version : An end to online game lag?


fixn granys cpu
08-14-2006, 10:07 AM
http://www.killernic.com/KillerNic/

Any comments on this new technology?

fixn granys cpu
08-14-2006, 10:09 AM
http://www.killernic.com/KillerNic/
Check this out...
What do you guys think of this?
How much could it really help? If it does what it says you think it will be worth the cost?

PrntRhd
08-14-2006, 11:01 AM
merged threads to appropriate forum

saphalline
08-14-2006, 03:59 PM
Most impressive. If they can get support for this on your standard PC (ie great drivers) then this could really take off! Off-loading of the entire TCP/IP stack and UDP handling? Onto a 32-bit 400MHz CPU with 64MB of DDR RAM? All on a NIC?? Good god!

i an not a nerd
08-14-2006, 09:18 PM
Well, if they put it onto a PCIex1 or x4 slot, I will be happy, but if they stick with PCI, i'm screwed (sound card, physx card, hard drive controller taking up my pci slots)

saphalline
08-15-2006, 02:57 AM
Yeah, PCIe x16 was fast to take off, but PCIe x1 is slow. Most expansion cards would never need more than an x1 slot. I would like see mobo manufacturers go for broke and offer one PCI slot, and the rest PCIe. That would force card manufacturers to switch to the better PCIe. X-Fi and PhysX should be on PCIe, anyway. I'm hoping we'll see PCIe versions soon.

PCI slots should die! Grrr!

i an not a nerd
08-16-2006, 03:27 PM
PCI slots should die! Grrr!

Yes! and the worst part is, the PCI slots all share bandwidth!

Kraven
08-16-2006, 04:31 PM
They look cool, but the price is a little steep!

jlreich
08-16-2006, 04:49 PM
and the worst part is, the PCI slots all share bandwidth!
So does PCIe. There's just much much more available bandwidth.

saphalline
08-17-2006, 01:57 PM
PCIe doesn't share bandwidth, though. An x16 slot gets all 16 lanes to itself - it doesn't share that with any other PCIe card installed. That's why modern chipsets have 48 lanes in them! They need to support two x16 slots, as well as a couple x4 and x1 slots. So the latest mobo's don't even use all the PCIe links available to them and are in no danger of running out of PCIe bandwidth!

That's why PCI should die!

jlreich
08-17-2006, 03:33 PM
Oh OK. I was thinking somewhere along the line the bandwidth is shared? Is it shared at the chipset level? No, I would think not if each lane has it's own lead... :confused:

i an not a nerd
08-17-2006, 03:48 PM
Hm, are you thinking of the old SLI? Two vid cards with x8 lanes apiece?

jlreich
08-17-2006, 05:43 PM
No that wasn't it. It seems like during one of the big Q&A threads with Saphalline he said something about shared bandwidth on the chipset, something about the 48 leads... But I probably just misunderstood or just plain forget what it was... :p

It's a shame that my focus has been on other things lately. :( It's the hardware side of things that I really find so fascinating. You know things like transistors, the new stretchable silicon, CPU/GPU/VPU architecture, ram etc.

BTW, anyone heard about stretchable silicon? :cool:

i an not a nerd
08-17-2006, 10:21 PM
Isn't that something that AMD made to make the transistors more efficient? (I might be thinking of stressed silicone...)

jlreich
08-17-2006, 10:33 PM
Sounds like you are thinking of strained silicon.

Have a look at this. http://www.pcmag.com/article2/0,1895,1988185,00.asp

It's part of an article about the ten coolest upcoming technologies. All ten are pretty cool . Worth a read.

saphalline
08-24-2006, 07:49 PM
Interesting...

azzey
08-24-2006, 07:59 PM
Interesting indeed.

jlreich
08-24-2006, 09:46 PM
Yeah very interesting. Did you see the one in the same article about transparent transistors? Can you say "computer screens in your windshield and other glass". :cool:

And the one about chaos computing?

We live a very interesting time indeed. :cool: :cool:

mjc
08-24-2006, 11:29 PM
Not quite sure I like the transparent transistors...unless I can get some transparent aluminum to go with them.

saphalline
08-25-2006, 04:07 AM
AI still worries me. Fascinating stuff, but I'm not sure we should be messing with it just because we can make it. Configurable hardware matrices powering high-def 3D neural networks? The firmware implications are intriguing on a reactive level if one considers the behavioral randomness of genetic algorithms in any given small time frame, but how do we know what these things will do over large time frames? What will these advanced paralleled systems learn over trillions of cycles trying to solve the "highest hill" problem?

I'm still not sure about it...

jlreich
08-25-2006, 09:21 AM
It is a little scary. Some of the things conjure images of certain science fiction movies. Some good, some bad. Someone out there will find a way to exploit any new technology that comes our way.

Then there's the AI thing. Good, bad? I guess we are going to find out before to long...

I think probably the one that fascinates me the most at this particular time is chaos computing. The ability of a processor, or what ever you want to call it at that point, to quickly reorganize it self to the current needs. If you need more ram, it changes, if you are playing a game, it will reorganize to a GPU. One day we may have an entire computer on one chip. :cool:

mjc
08-25-2006, 12:34 PM
Then there's the AI thing. Good, bad? I guess we are going to find out before to long...

Reistance is futile...

saphalline
08-26-2006, 12:26 AM
Yes! That's it! We shall become the BORG! I'd like to pick my number now, please. Something like 17 of 2.18 x 10^38. :p What do you think?

mjc
08-26-2006, 01:28 AM
It's all yours......

i an not a nerd
08-26-2006, 01:27 PM
ah the borg... bring back memories of watching a whole season of Star Trek TNG in a few days...

jlreich
08-26-2006, 04:56 PM
Yes! That's it! We shall become the BORG! I'd like to pick my number now, please. Something like 17 of 2.18 x 10^38. :p What do you think?
Hey that was the number I wanted!

Well, OK you can have it. I will take #3. :p