beggerking
Golden Member
- Jan 15, 2006
- 1,703
- 0
- 0
Originally posted by: beggerking
Originally posted by: josh6079
You were simply trying to prove the point that the hardware does not necessarily need a driver to throw an image on your screen. That is correct..
good. End of discussion.
Originally posted by: beggerking
redbox, I'm really wondering if you are just Josh's "other" account now....
anyways, let me clarify something for you.
1. NO DRIVER PATH! it doesn't exist and is technically incorrect. Please use the correct technical term " data path ".
"In this regard the software(I include drivers and bios in the software description) reacts to the card as if it was an SLI setup which untill now has been two cards"
so, does that make 7950(hardware) not a single card solution?
Originally posted by: beggerking
2 GPU, not 2 cards.
technically, software don't see.. but thats complicated. Lets agree on this, dualcore single card.
redbox, BFG disagree with 2 GPU though. He thinks its 2 cards, why don't you debate with him now?
Originally posted by: beggerking
well, it does act as a dualcore GPU.
Originally posted by: redbox
Originally posted by: beggerking
well, it does act as a dualcore GPU.
Wrong it acts like a dual card SLI system. It would be fine by me to call it a dual gpu solution, but the fact remains that it still acts like a dual card system when regarding software. Maybe at the micro level it doesn't but as far as the user end of software it acts like two cards.
Originally posted by: beggerking
Originally posted by: redbox
Originally posted by: beggerking
well, it does act as a dualcore GPU.
Wrong it acts like a dual card SLI system. It would be fine by me to call it a dual gpu solution, but the fact remains that it still acts like a dual card system when regarding software. Maybe at the micro level it doesn't but as far as the user end of software it acts like two cards.
please explain to me the difference between dual GPU solution and a dualcore solution, in regards to software. I don't see there is a difference in these solutions when it comes to software.
Originally posted by: redbox
Originally posted by: beggerking
please explain to me the difference between dual GPU solution and a dualcore solution, in regards to software. I don't see there is a difference in these solutions when it comes to software.
Well it depends I would think that a dualcore solution would be able to address all of the information in memory instead of having it's own buffer. The 7950GX2 has 512mb of ram on for each gpu but they can't draw from the info on the other core's 512mb. Some dualcore cpu's act this way but they are moving to where there is 2mb that both cores pull from instead of just 1mb per core. I don't know wether you get any performance benifits from having a collective pool of memory, but there is a difference there. 7900gt in SLI can't write or pull any info from the other cards memory can it?
Originally posted by: beggerking
Originally posted by: redbox
Originally posted by: beggerking
please explain to me the difference between dual GPU solution and a dualcore solution, in regards to software. I don't see there is a difference in these solutions when it comes to software.
Well it depends I would think that a dualcore solution would be able to address all of the information in memory instead of having it's own buffer. The 7950GX2 has 512mb of ram on for each gpu but they can't draw from the info on the other core's 512mb. Some dualcore cpu's act this way but they are moving to where there is 2mb that both cores pull from instead of just 1mb per core. I don't know wether you get any performance benifits from having a collective pool of memory, but there is a difference there. 7900gt in SLI can't write or pull any info from the other cards memory can it?
that is hardware, not software...
anything in regards to software ?
Originally posted by: beggerking
well, it does act as a dualcore GPU.
Originally posted by: beggerking
"software doesn't need hardware to run", or "software has nothing to do with hardware."
because software can be fooled/emulated. software can be fooled to see dual cpu when there really is only 1 single CPU. (p4 ht) .
Originally posted by: spank
Originally posted by: beggerking
well, it does act as a dualcore GPU.
Would you call, for example a renderfarm as a single computer, it acts like one so therefor it must be a single one, right ?
Originally posted by: spank
Originally posted by: beggerking
"software doesn't need hardware to run", or "software has nothing to do with hardware."
because software can be fooled/emulated. software can be fooled to see dual cpu when there really is only 1 single CPU. (p4 ht) .
but nvidia does the other way around use software to emulate a single gpu
EDIT:
when sli dont work, what you got then, a half card ??![]()
Originally posted by: spank
Originally posted by: beggerking
"software doesn't need hardware to run", or "software has nothing to do with hardware."
because software can be fooled/emulated. software can be fooled to see dual cpu when there really is only 1 single CPU. (p4 ht) .
but nvidia does the other way around use software to emulate a single gpu
EDIT:
when sli dont work, what you got then, a half card ??![]()
This is nothing more than semantic nitpicking and you're simply clutching at straws because your simpleton arguments are going nowhere.well, there is no such thing as driver path, its called a DATA PATH!!!
You continue to ignore the hardware side of things and that's why your understanding is so limited.So you would consider a single P4 CPU with Hyperthreading as 2 CPUs since software sees it exactly the same way it sees a dualcore CPU?
It's not a single card because:so, does that make 7950(hardware) not a single card solution?
Then demonstrate the following:Right, thats what I meant by "software has nothing to do with hardware". what software sees don't reflect what is truely presented in hardware.. because software is capable of emulations.
a mistake is a mistake, please stop trying to justify it.Originally posted by: BFG10K
This is nothing more than semantic nitpicking and you're simply clutching at straws because your simpleton arguments are going nowhere.well, there is no such thing as driver path, its called a DATA PATH!!!
And I guess you think John Carmack (et al) are all incorrect when they refer to their games as having render paths?
no. hyperthread is the ability to set aside current process( that might be waiting for IO) and start running the next process.Hyper-Threading doesn't add another physical CPU, it simply allows the utilization of unused pipeline locations from the existing CPU without duplicating an entire CPU core.
It's not a single card because:
[*]It requires the SLI driver path to function properly.
[*]It is literally two separate GPUs stuck together on two PCBs with all of the associated hardware duplication.
[*]It can't run dual-display like single cards can.
what does that have to do with anything?[*]How to get SLI AA operating on a single card.
[*]How to get dual-display operating on SLI.
Your entire posting history in this thread is a giant mistake and the sad thing is that you're the only one that thinks otherwise.a mistake is a mistake, please stop trying to justify it.
So are you disagreeing with what I said or are you simply posting useless rhetoric? Do you read what you respond to? And why are your replies at a child-like level when you think you're so smart?no. hyperthread is the ability to set aside current process( that might be waiting for IO) and start running the next process.
If it doesn't depend on hardware then explain how to operate SLI on a single card. Also explain why nVidia calls two 7950s Quad SLI and also explain how to get the "scheduler" to get SLI AA working on a single card.SLI is a software data scheduler. It is not hardware and does not depend on hardware and does not determine whether its a single card or not..
The "software has nothing to do with hardware" steaming pile you keep producing. If that's the case then demonstrate how to get SLI AA working on a single card.what does that have to do with anything?
But I thought the "hardware has nothing to do with software"? So why then do we need to change the software to make it behave like a single card - something that is not needed on an actual single card - despite your claim the 7950 is already a single card?disable SLI so the same data is distributed to both GPUs.
Originally posted by: BFG10K
So are you disagreeing with what I said or are you simply posting useless rhetoric? Do you read what you respond to? And why are your replies at a child-like level when you think you're so smart?no. hyperthread is the ability to set aside current process( that might be waiting for IO) and start running the next process.
1. explain to me then, what is your explaination of SLI software? how does it work that would makes it require hardware?If it doesn't depend on hardware then explain how to operate SLI on a single card. Also explain why nVidia calls two 7950s Quad SLI and also explain how to get the "scheduler" to get SLI AA working on a single card.SLI is a software data scheduler. It is not hardware and does not depend on hardware and does not determine whether its a single card or not..
The "software has nothing to do with hardware" steaming pile you keep producing. If that's the case then demonstrate how to get SLI AA working on a single card.what does that have to do with anything?
little kid, please, please do some research before you reply.But I thought the "hardware has nothing to do with software"? So why then do we need to change the software to make it behave like a single card - something that is not needed on an actual single card - despite your claim the 7950 is already a single card?disable SLI so the same data is distributed to both GPUs.
Originally posted by: beggerking
please, please read do some research before you reply.
hardware works as long as it runs as it is designed for. 7950 is designed to take 2 input data and process them on 2 GPUs, so as long as there are 2 input data coming into the card, the hardware is working and is working as it was designed for.
the problem is, we don't want 2 GPUs to work on same set of data. We want them to each work on a different set of data. that is why we have SLI/CPU scheduler.