• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

The ultimate question.

What should I do?

  • 6800GT -> Ultra Speed

    Votes: 155 70.1%

  • x800Pro -> Flash / XT Speed

    Votes: 34 15.4%

  • Wait, get x800XT later.

    Votes: 32 14.5%
  • Total voters
    221
Please I have said what I needed to say. I don't want to go on with 2 man thread.

If you want to discuss this further email me. I'm trying to grasp more and more but you just sit there tease and get your little nephew involved. What is this tag team wrestling?

Peace I'm out.

And Squirt come back here when your grown. Let the adults handle this.
 
Now that Marvelous has managed to nullifiy any credibilty he had, can we move on? Really, the last page of this thread has been a tennis match and I for one fucking hate tennis.
 
I wasn't forfeiting anything. But just tired of rambling and it's freaking weekend. I want to relax. If he's down to email me we can discuss further.
 
this has been a fun thread to read

i knewn marv didn't know what he was talking about. next marv if i where you i would do alot of research before you even try and challnge rancor. he knows alot.

all in all
the 6800 line up owns the x800 line up
hell ati didn't even come close to winning any rewards with there x800's
shit ati needs to get on the ball here and start making new cores. not some old 2 year old shit and rename it. to me ati sounds like 3dfx. only time will tell now
 
hey Marv don't worry about little squirt, he won't be posting here for a while shouldn't have posted that last comment at all, parents don't allow him to use the internet without them being in the room, bad influence, but anyways, I wish he would play with his friends more then spend time with his comp, hopefully that will change when his hormones kick in :D

He might only be 10 but he is very mature and damn good at math for his age, don't know many 30 years olds that can do a complex derivative in thier head, shit I can't really do them without writing down different stages.

Fillrates only come into play if there is no bottleneck else where, and the bottle neck for the nv line of cards is vertex shader performance, and for ATi its pixel shader performance (when compairing the cards from both companies side by side)

And about the mumbo jumbo, I don't know how to say what I said in any simplier terms sorry
 
Marvelous said:
Seems like you guys need to grow up.

I didn't goto sleep for 2 days because I got a new video card and I was busy playing Farcry which I just got.

I didn't want to sit there and argue. Being the adult that I am.

If you want to continue I will on Monday. Ejects.
Click to expand...


usual comeback for a guy who has just been defeated by a 10 year old :eek:
 
rancor said:
Thx little squirt, anyways back to the topic



http://www.hothardware.com/viewarticle.cfm?articleid=438
http://www.devhardware.com/c/a/Video Cards/FX-5900-Ultra-vs-R9800-Pro


Wrong again!

The fx line had very poor pixel shader execution which the texel fillrates couldn't over come.



Can you grasp all of the logic or only half the half you want to hear?

Poorly designed the 4x2 TMU's still provided a huge bandwidth advantage too. Bandwidth is out of the picture!

Games that utilize sm 3.0 will work just fine with this cards, but just not at the highest of settings. As the 9700's show its tech will last for a good 2 years, and this really has nothing to do with fill rates, well very little, as you can see performance on these cards are very close, and by using x number of passes will slow down these cards x number of times, even far cry goes above 100fps at 1200x1024 with no aa and af, cut that down by 3 you can do shaders up to 3 passes and they will still work fine (with branching and looping there really shouldn't be any need to go more then 2, so now you can increase geometric complexity)

http://www.planetquake.com/features/articles/editorials/pixels-1.shtml

As they say fillrates don't matter!
Click to expand...

While FX line had huge bandwidth but it was like hooking up a already maxed out 4 pipes and and can't utilize the memory bandwidth at all. How you going to hook a 4 pipe card to a 256 bit memory pathway? It had too many issues like heat because of more core mhz. It was just bad decision on Nvidia’s part.

The FX5800 was nothing more than a turble charged 4 pipe card with 128bit memory bus. LOL Do the math. More mhz does not mean more fps. More pipes with bigger memory pathways will hurt anything with 4 pipes that run at faster mhz. With todays games utilizing more bandwidth and fillrate FX cannot utilize it's full pontential because it was poorly designed. It just had too many issues to start.
Your telling me a game has nothing to do with fillrate and bandwidth? LOL Okay whatever. It has everything to do with fillrate.

And those links mean nothing except 1 person's view. I showed you benchmarks that is clear by night and day where XT hurts a GT. You give me some rants about about some fan sites??? LOL
 
rancor said:
We weren't talking about sme 3.0 games were we, SM 3.0 will speed things up to a certian point but most of thier shaders will require 2 passes. It will be perfectly possible with nV's current technology.LOL you can't understand a single thing I said because I'm not talking about the 3d render pipeline, I'm not only talking about the 3d render pipeline, 3d render pipeline is only a small part of it. Just for your information performance has nothing to do with 3d render pipeline
Click to expand...

No we were talking about sm30 games. And farcry with the patch does this. While Radeon might need more calculations it's fill rates and bandwidth overcome this. It shows on benchmarks as well.
 
Marvelous said:
While FX line had huge bandwidth but it was like hooking up a already maxed out 4 pipes and and can't utilize the memory bandwidth at all. How you going to hook a 4 pipe card to a 256 bit memory pathway? It had too many issues like heat because of more core mhz. It was just bad decision on Nvidia’s part.

The FX5800 was nothing more than a turble charged 4 pipe card with 128bit memory bus. LOL Do the math. More mhz does not mean more fps. More pipes with bigger memory pathways will hurt anything with 4 pipes that run at faster mhz. With todays games utilizing more bandwidth and fillrate FX cannot utilize it's full pontential because it was designed. It just had too many issues to start.
Your telling me a game has nothing to do with fillrate and bandwidth? LOL Okay whatever. It has everything to do with fillrate.

And those links mean nothing except 1 person's view. I showed you benchmarks that is clear by night and day where XT hurts a GT. You give me some rants about about some fan sites??? LOL
Click to expand...


Dude keep on talking and you will oh wait already lost crediabitly its as I said before its a waste of time.....

what fansites? the links I showed you were not for benchmarks (only the digit life one was) the other stated theorotical fill rates of fx line vs the the 9800's.

The FX5800 was nothing more than a turble charged 4 pipe card with 128bit memory bus.
Click to expand...

Was it because of the 4 pipelines to a 128 bit architecture, what? What do you think the 9600 is, and the x600? hmm wait what about the gf 4's? The 9500's?, The 9550's?

Bigger memory path ways don't hurt the 4 pipelines, what the hell is the a 6600 with a 128 bit pathway doing with 8 pipelines, as is the x700's?

enough said dude, not even worth the time and effort, sure hoped everyone enjoyed the show!
 
rancor said:
Dude keep on talking and you will oh wait already lost crediabitly its as I said before its a waste of time.....

what fansites? the links I showed you were not for benchmarks (only the digit life one was) the other stated theorotical fill rates of fx line vs the the 9800's.



Was it because of the 4 pipelines to a 128 bit architecture, what? What do you think the 9600 is, and the x600? hmm wait what about the gf 4's? The 9500's?, The 9550's?

Bigger memory path ways don't hurt the 4 pipelines, what the hell is the a 6600 with a 128 bit pathway doing with 8 pipelines, as is the x700's?

enough said dude, not even worth the time and effort, sure hoped everyone enjoyed the show!
Click to expand...

Planetquake is not a fansite? LOL Nvidia=Carmack

LOL does any of those 4 pipe cards beat out a Radeon 9800pro? NOT

LOL I never said bigger memory pathways hurt peformance. Just that 4 pipe cards can't fully utilize 256bit memory pathway than 8 pipe card can.

Is that what your doing. putting on a show?
 
Marvelous said:
Planetquake is not a fansite? LOL Nvidia=Carmack

LOL does any of those 4 pipe cards beat out a Radeon 9800pro? NOT

LOL I never said bigger memory pathways hurt peformance. Just that 4 pipe cards can't fully utilize 256bit memory pathway than 8 pipe card can.

Is that what your doing. putting on a show?
Click to expand...


Peak operations it's peak fill-rate is 4x2 pixels, it gets to that when two biliear textures are require, or when trilinear filtering is used. In the fragement pipeline it is more flexible, traditionally associated with a 4 (or what every number)x2 pipeline arrangement. Looking at peak rates, what you are really looking for is aboslute or "pure" fill-rate, (the number of pixels it can write and the ROP's take care of this)

A ROP is the pixel engine, a single ROP is capable of half a pixel blend, one color write or 2 Z/Stencil writes.

there are 8 ROP's in the 5800's. And this is the same as the 9800 pro

So it comes down to this

8 stencil ops per clock
8 textures per clock
8 shader ops per clock
8 z pixels per clock, forgot to add this.

Both cards are capable of doing the same, but due to shader performance one of them got the short end of the stick.
 
Karma said:
The X800XT (non-PE) is the same price as the X800 Pro VIVO (~450), BTW. You could think of them as Pre-modded VIVO's that you don't have to mess with.
Click to expand...

Pro VIVOs are down to the $420s.
 
rancor said:
Peak operations it's peak fill-rate is 4x2 pixels, it gets to that when two biliear textures are require, or when trilinear filtering is used. In the fragement pipeline it is more flexible, traditionally associated with a 4 (or what every number)x2 pipeline arrangement. Looking at peak rates, what you are really looking for is aboslute or "pure" fill-rate, (the number of pixels it can write and the ROP's take care of this)

A ROP is the pixel engine, a single ROP is capable of half a pixel blend, one color write or 2 Z/Stencil writes.

there are 8 ROP's in the 5800's. And this is the same as the 9800 pro

So it comes down to this

8 stencil ops per clock
8 textures per clock
8 shader ops per clock
8 z pixels per clock, forgot to add this.

Both cards are capable of doing the same, but due to shader performance one of them got the short end of the stick.
Click to expand...

Your comparing apples to oranges. 8 pipes at slower clock is faster than 4 pipes at faster clock speeds.

Radeon 9600XT has better instructions per set then say a Radeon 9500 pro. They are very close in peak fill rate performance correct? But in certain fill rate deprived games or newer games the Radeon 9500pro walks away with the crown.

http://www.techreport.com/etc/2004q3/doom3-midlow/index.x?pg=3

You see the Radeon 9500pro edging out the XT in higher resolutions even though the XT has more memory bandwidth. This proves that 8pixel pipes are more efficient than 4 pipes.

Nvidia just realized this now after test and trial. That's why they stuck out the 6600GT and deprived it with 128bit memory.
 
Marvelous said:
Your comparing apples to oranges. 8 pipes at slower clock is faster than 4 pipes at faster clock speeds.

Radeon 9600XT has better instructions per set then say a Radeon 9500 pro. They are very close in peak fill rate performance correct? But in certain fill rate deprived games or newer games the Radeon 9500pro walks away with the crown.

http://www.techreport.com/etc/2004q3/doom3-midlow/index.x?pg=3

You see the Radeon 9500pro edging out the XT in higher resolutions even though the XT has more memory bandwidth. This proves that 8pixel pipes are more efficient than 4 pipes.

Nvidia just realized this now after test and trial. That's why they stuck out the 6600GT and deprived it with 128bit memory.
Click to expand...

Doesn't the 9500 have a 256 memory bandwidth? Oh yes it does. Thats why it wins at higher settings ;)

Why are you saying that, its because the 9500 can do 8 textures per clock, and the 9600 can do only 4 textures per clock, hmm and the 9500 can do 4 color pixels per clock, and the 9600 can do only 2 color pixels per clock, wait lets see, the 4x2 arrangement does 8 textures per clock and 4 color pixels per clock, oh my god its the same as the 8 pipeline arrangement, just as I explained above.

The link I showed you before theoreical fillrates aren't even being fully used!

But you have to understand games won't utilize the bandwidth efficiently if the card isn't stressed. Plus the fact, Doom 3 uses a max of 4 textures channels at a time anyways..

I never said there is no advantage of 4 and 8 pipelines, I said there really was no advantage over 4x2 and 8 when performance is concerned. So you are talking about something else, now there is an advantage of chips being less complex with 8 pipelines over the 4x2 structure, and this wsa one of the reasons the fx chips were so expensive to make. Nvidia had no choice but to use 4x2 otherwise they would have had to redesign the entire chip, which would have taken longer then 6 months and probably would have ended up costing way too much in developement.
 
Hahaha, not my fault you guys are all crazies, I still have yet to see a real comparison between x800 Pro's flashed and OC'd to GT's OC'd, so there :p
 
joecuddles said:
Hahaha, not my fault you guys are all crazies, I still have yet to see a real comparison between x800 Pro's flashed and OC'd to GT's OC'd, so there :p
Click to expand...


the over clock gt, just look at ultra benchmarks performs exactly like an ultra, the modded pro look at xt benchmarks, performs exactly like an xt and if your lucky an xt pe.
 
Rancor what is your view on 9800se type2 at stock compared to 9600xt, gimme some technical details plz. *flame baits* :D Let's see those naysayers for the se's come out when Rancor let's hem down :D. No but really I'm curious as to what your views on it are.

~Adam
 
CleanSlate said:
Rancor what is your view on 9800se type2 at stock compared to 9600xt, gimme some technical details plz. *flame baits* :D Let's see those naysayers for the se's come out when Rancor let's hem down :D. No but really I'm curious as to what your views on it are.

~Adam
Click to expand...


hmm which one was type 2 4x1 or 8x1?
 
By 8x1 you mean 8 pipes and 1 shader ..? I'm not too technical on gpu cores as you can tell but I do know that it has 8 pipes with 4 disabled and 256 bit ram interface.

~Adam
 
well the whole point of it being an se is that it had 4 pipes disabled, from my knowledge anyhow, mine was a 4piper with 4 disabled. I was under the impression if it was an 8 piper enabled it'd just be a 9800np

~Adam
 
rancor said:
Doesn't the 9500 have a 256 memory bandwidth? Oh yes it does. Thats why it wins at higher settings ;)

Why are you saying that, its because the 9500 can do 8 textures per clock, and the 9600 can do only 4 textures per clock, hmm and the 9500 can do 4 color pixels per clock, and the 9600 can do only 2 color pixels per clock, wait lets see, the 4x2 arrangement does 8 textures per clock and 4 color pixels per clock, oh my god its the same as the 8 pipeline arrangement, just as I explained above.

The link I showed you before theoreical fillrates aren't even being fully used!

But you have to understand games won't utilize the bandwidth efficiently if the card isn't stressed. Plus the fact, Doom 3 uses a max of 4 textures channels at a time anyways..

I never said there is no advantage of 4 and 8 pipelines, I said there really was no advantage over 4x2 and 8 when performance is concerned. So you are talking about something else, now there is an advantage of chips being less complex with 8 pipelines over the 4x2 structure, and this wsa one of the reasons the fx chips were so expensive to make. Nvidia had no choice but to use 4x2 otherwise they would have had to redesign the entire chip, which would have taken longer then 6 months and probably would have ended up costing way too much in developement.
Click to expand...

Shows how much you know. Radeon 9500pro is 8pipe card with 128bit memory bandwidth.
 
Marvelous said:
Shows how much you know. Radeon 9500pro is 8pipe card with 128bit memory bandwidth.
Click to expand...


LOL 1 mistake with a 2 year old card, which I could careless because we weren't talking about 4 pipeline vs. 8 pipeline.

4x2 vs 8 please.
 
CleanSlate said:
well the whole point of it being an se is that it had 4 pipes disabled, from my knowledge anyhow, mine was a 4piper with 4 disabled. I was under the impression if it was an 8 piper enabled it'd just be a 9800np

~Adam
Click to expand...


true the 4 pipeline version is very close to the 9600xt, don't know why they did that wierd.
 
LOL
Marv has just lost...give it up dude. i have looked at all of rancor's stuff and he is right. and you are wrong....
 
[RIP]Zeus said:
LOL
Marv has just lost...give it up dude. i have looked at all of rancor's stuff and he is right. and you are wrong....
Click to expand...


Marv knows no bounds of defeat :D , its not a defeat if he learned something, but looking how the thread progressed so far I don't see that ;)
 
rancor said:
true the 4 pipeline version is very close to the 9600xt, don't know why they did that wierd.
Click to expand...

It kinda makes economic sense, if you have a big batch of cards with the same fabrication flaw on 1 certain pipe and you can disable pipes you could cut the pipe (and in doing so cutting the flaw) and making the product still worth some money on the market as a less performing but still good card that people would buy.

~Adam
 
CleanSlate said:
It kinda makes economic sense, if you have a big batch of cards with the same fabrication flaw on 1 certain pipe and you can disable pipes you could cut the pipe (and in doing so cutting the flaw) and making the product still worth some money on the market as a less performing but still good card that people would buy.

~Adam
Click to expand...


Very true
 
You must log in or register to reply here.
Back
Top