PDA

View Full Version : 64 bit capable gfx cards?? release??


fantomas
01-05-04, 05:51 PM
Heres a question: with athlon64 now affordable, and duel opterons out there for the rich, and the new mobo's that nvidia has made, when are gfx cards coming out that can take advantage of the supersized bus available with these cpu's?

Does anyone have any info on what might be happening? I think this will be a determining factor when i upgrade from my outstanding geforce 3ti golden sample in June. Thanks for the help :)

reever2
01-05-04, 06:18 PM
Uhhh, the cards out now are already doing whatever you want them to be doing

|JuiceZ|
01-05-04, 06:39 PM
Originally posted by fantomas
Heres a question: with athlon64 now affordable, and duel opterons out there for the rich, and the new mobo's that nvidia has made, when are gfx cards coming out that can take advantage of the supersized bus available with these cpu's?

Does anyone have any info on what might be happening? I think this will be a determining factor when i upgrade from my outstanding geforce 3ti golden sample in June. Thanks for the help :)

I think you're a bit confused. The real question you should be asking is when will more developers begin to write more software to take advantage of 64bit hardware architecture. IMO most are probably waiting for WinXP 64bit edition to be released & M$ is probably waiting for Intel to jump onboard since they still own most of the market share :rolleyes: This could, of course, all change this year, hopefully...

fantomas
01-05-04, 07:00 PM
I think you're a bit confused.

Maybe i wasnt as clear as i should have been. Lets use my geforce3ti as an example. It has a bus speed of 266mhz, thats 2x133 mhz. Now if I want to send massive amounts of rendered textures from my gpu to my monitor the bottleneck is this bus speed. So if, for example, nvidia releases a card with 500mhz clock and 1000mhz memory, and its bus speed is 266mhz then the memory on the card is waiting to send the 1gb of textures each time the bus fills up... i could be totally wrong but i think thats how it works. So what im asking is: when are cards with bus speeds as fast as the fastest mobo's coming out? I mean if you have a pentium4 3.2ghz ee with a 800mhz bus it seems (assuming you have 800mhz ram) that the graphics bottleneck would be the gfx card due to the slow bus speed.

gstanford
01-05-04, 07:17 PM
You are confused.

The bottleneck between graphics cards and the rest of the PC architecture is the AGP port which hase a base speed of 66mhz whic then gets expanded by signal pumping technologies to 2x 4x or 8x effective.

The technology you want to look into is PCI express.

fantomas
01-05-04, 07:30 PM
The bottleneck between graphics cards and the rest of the PC architecture is the AGP port which hase a base speed of 66mhz whic then gets expanded by signal pumping technologies to 2x 4x or 8x effective.


Yeah this is what i mean... 2x66=133, 4x=266! and 8 = 512...i will do a search for pci xpress info. Im thinking about getting a dual opteron, so im trying to eliminate the bottlenecks in my specs before i order it. Many thanks for the info :)

bkswaney
01-06-04, 03:22 AM
Originally posted by fantomas
The bottleneck between graphics cards and the rest of the PC architecture is the AGP port which hase a base speed of 66mhz whic then gets expanded by signal pumping technologies to 2x 4x or 8x effective.


Yeah this is what i mean... 2x66=133, 4x=266! and 8 = 512...i will do a search for pci xpress info. Im thinking about getting a dual opteron, so im trying to eliminate the bottlenecks in my specs before i order it. Many thanks for the info :)


There is nothing even using the 8x AGP power at this point.
4x is still as good as u need right now.

-=DVS=-
01-06-04, 04:33 AM
Originally posted by fantomas
I think you're a bit confused.

Maybe i wasnt as clear as i should have been. Lets use my geforce3ti as an example. It has a bus speed of 266mhz, thats 2x133 mhz. Now if I want to send massive amounts of rendered textures from my gpu to my monitor the bottleneck is this bus speed. So if, for example, nvidia releases a card with 500mhz clock and 1000mhz memory, and its bus speed is 266mhz then the memory on the card is waiting to send the 1gb of textures each time the bus fills up... i could be totally wrong but i think thats how it works. So what im asking is: when are cards with bus speeds as fast as the fastest mobo's coming out? I mean if you have a pentium4 3.2ghz ee with a 800mhz bus it seems (assuming you have 800mhz ram) that the graphics bottleneck would be the gfx card due to the slow bus speed.


The bottleneck between graphics cards and the rest of the PC architecture is the AGP port which hase a base speed of 66mhz whic then gets expanded by signal pumping technologies to 2x 4x or 8x effective.


Yeah this is what i mean... 2x66=133, 4x=266! and 8 = 512...i will do a search for pci xpress info. Im thinking about getting a dual opteron, so im trying to eliminate the bottlenecks in my specs before i order it. Many thanks for the info



Im not expert here but i think you mixing stuff or i just read wrong becouse some of that makes no sense to me atleast. This is what i think.

1. Pentium 4 with 800 Mhz BUS sends data to PC system memory not to Video card it feeds CPU so to speak (on motherboard DDR ram example) witch must be equal speed to be optimum .
2.Now Geforce Fx with 1000 mhz memory sends data to its own GPU at that speed not CPU, what speed GPU is make it 100mhz or 500mhz in case of Geforce FX have no difference in exchange of memory data.
3. Now CPU and Video Card talk to each other over AGP bus 2x 4x 8x or PCI

Now what speed CPU BUS is ( exmp. P4 800Mhz bus ) doesn't matter becouse of AGP beeing locked to one of these 2x 4x 8x speed configurations , also todays cards hardly use FAST-WRITES . ( GPU directly writes to System Memory Bypasing CPU ) as i understand in that case haveing very fast AGP would be usefull , but no todays programs require it heck nearly all games are happy with 128 meg Video Card. If you had unlimited AGP speed Fast-Write feature would be very usefull thought.

You know old (Pentium 3 had 133mhz bus and AGP 4x) so (P4 with 800Mhz bus and AGP 4x) have no advantage here as it make as Vid card and Cpu talk to each other over AGP 4x....

I think i made to many examples but whatever , getting Dual Opterons would be cool but you still are locked to one Bottleneck AGP here is best example yet :D
| System DDR <-Motherboard CPU BUS-> CPU <--AGP--> GPU <-Video Card bus 256bit->Vid card memory |

bkswaney
01-06-04, 04:55 AM
I think he is thinking about the Athlon 64 and windows 64bit.

I think he is asking if anyone makes a videocard that uses 64bit
like the cpu and os do.

MUYA
01-06-04, 07:36 AM
Dude, do u mean 64 bit environment? Which u know needs a 64 bit OS and 64 bit driver. Then only will gfx card be able to take full advantage of the 64 bit environment. What they are....the more techincally adept can tell ya ;)

As you might know, the GPUs from NV and ATI post TNT and i dunno for ati are all 256bit. Again, experts in this forums can comment more on what that means. :D

AnteP
01-06-04, 08:32 PM
Originally posted by fantomas
Im thinking about getting a dual opteron, so im trying to eliminate the bottlenecks in my specs before i order it.

The botleneck is as mentioned the AGP Port, not neccesarily the bandwidth but the tech in general. PCI Express in general will help.
But that's still unrelated to taking advantage of 64 bit processors.

A question I have though is if 64 bit display drivers have a native benefit over 32 bit ones. Ie: does a display driver do any computing that would better off in 64 bit mode by a considerable amount?
Does anyone know? :)

SurfMonkey
01-06-04, 08:39 PM
Or do you mean when will the PCI\AGP bus move over to a 64bit datawidth from the current 32bit?

AnteP
01-06-04, 09:31 PM
Originally posted by SurfMonkey
Or do you mean when will the PCI\AGP bus move over to a 64bit datawidth from the current 32bit?

PCI-X is already 64 bit at up to 133 MHz IIRC.