So it appears the extra 256mb of RAM on the GTX does almost nothing (in current games)...

[Keysplayr]

Originally posted by: Justin343563611
Well, hasn't 512MB on a video card pretty much been the same with the 256 variant all this time, remember the x800 XL 512, yeah, nothing there, 6800 Ultra 512, yeah, nothing there, they need basically make 512megs the native memory capacity for these chips to support in my opinion before they can implement them onto a board.

What do you mean by "native memory capacity"? The GTX512 has a new PCB to utilize 16x32 chips as opposed to 8x32 chips on the GTX/GT.. According to ATI, the X1800series can handle up to 1GB of memory.

EDIT: If you meant to say that these cards: X1800xx/7800GT/GTX/512 arent powerful enough to utilize 512MB of memory, well, I'm pretty sure they are powerful enough to use it fully.

 

[dug777]

Originally posted by: BFG10K

Why does everyone else have to stop what you started?

What I started was posting facts and logical arguments and I make no apologies for that. This topic was flawed from the onset and I've already repeatedly explained why, not to mention the numerous benchmarks from multiple posters that have confirmed my comments.

Now we're reduced to the stage of certain people arguing why in their opinion they think I meant 3-4 seconds when I said "few", and also that using a dictionary is invalid. It's the most blatant clutching at straws I've seen in quite some time.

Ignoring this ridiculous seconds thing, which you started by saying AT only benched for a 'few seconds' which to most people means well less that 10 seconds, theres nothing flawed about the topic. You PROVE the 512mb GTX is smoother (with benchmarks) and then you can show me, in a civil fashion, that both mine and AT's conclusion is flawed, but going on teh AT data, as i am, there's nothing flawed about it, i merely restated the conclusion AT came to...

I suggest you look at Hardocp, since they show minimum fps that ought to prove or disprove your ideas pretty quickly

 

[Keysplayr]

I give up!!! WTF Dug? You show no interest in ending this by taking it one step further for the last word. END IT!!! IT'S a DUMB argument beyond measure.

 

[dug777]

Originally posted by: keysplayr2003
I give up!!! WTF Dug? You show no interest in ending this by taking it one step further for the last word. END IT!!! IT'S a DUMB argument beyond measure.

meh, i simply resent being told my topic was flawed, by someone who has nothing to prove otherwise. I got exactly back on topic by pointing out that Hardocp's benchmarks should show the hitching with their max/min fps graphs...

 

[kylebisme]

Originally posted by: keysplayr2003
I understand your position, but the details really don't concern me now after seeing what this "discussion" evolved into. I just want it to stop. You all constantly try to one up each other and have the "never say die" mentality to go along with it. Point is, no one ever wins, both sides lose face, and the whole discussion was really ridiculous anyway. So, I don't think it's worth it for either side to continue at this point.

You don't understand his position becuase to do so you would have t consern yoursself with the details and understand that what BFG said is simply a matter of fact. It's up to you if you want to take the time to understand these facts, and myself as probably others will be willing to help you if you are having trouble understanding it. If you don't want to understand it, if you think this is just some pissing contest and you felt like steping in to shower on a few people then you are in the wrong place and you are the one who needs to get out of this thread.

 

[kylebisme]

Originally posted by: dug777
Ignoring this ridiculous seconds thing, which you started by saying AT only benched for a 'few seconds' which to most people means well less that 10 seconds, theres nothing flawed about the topic. You PROVE the 512mb GTX is smoother (with benchmarks) and then you can show me, in a civil fashion, that both mine and AT's conclusion is flawed, but going on teh AT data, as i am, there's nothing flawed about it, i merely restated the conclusion AT came to...

I suggest you look at Hardocp, since they show minimum fps that ought to prove or disprove your ideas pretty quickly

He means a "few seconds" as in not nearly long enough to account for all the swapping of textures that can happen across the course of playing though a level of a game. Why are you being so thick about this?

 

[BFG10K]

meh, i simply resent being told my topic was flawed, by someone who has nothing to prove otherwise.

This is a joke, right? If not I see one of several possible possible scenarios here:

(1) Dug doesn't understand how hyperlinks work, therefore he's incapable of getting to the linked evidence.
(2) Dug is burying his head in the sand and refuses to accept any evidence that contradicts his flawed claims.
(3) Dug has issues with reading and/or comprehension.
(4) Dug is blind.
(5) Dug is trolling.

On the off chance that he genuinely missed the data the first two times here is the evidence for the third time.

HL2 Benchmarks
Call of Duty 2
Quake 4
BF2

And a nice quote from B3D:

With our Quake 4 test we see that the performance of the 512MB GTX is over 50% higher than the 256MB which, given that this exceeds any of the theoretical metrics, is likely due to the increase in local RAM and the quantity of data Quake 4 likes to store locally with the Ultra detail settings enabled - when the quantity of data exceeds the local framebuffer space then more will be pushed across to system RAM thus having to be addresses over the PCI Express interface, which is much lower performance then the local memory.

That's it. I can't make it any simpler than that I'm afraid.

 

[dug777]

Originally posted by: TheSnowman

Originally posted by: dug777
Ignoring this ridiculous seconds thing, which you started by saying AT only benched for a 'few seconds' which to most people means well less that 10 seconds, theres nothing flawed about the topic. You PROVE the 512mb GTX is smoother (with benchmarks) and then you can show me, in a civil fashion, that both mine and AT's conclusion is flawed, but going on teh AT data, as i am, there's nothing flawed about it, i merely restated the conclusion AT came to...

I suggest you look at Hardocp, since they show minimum fps that ought to prove or disprove your ideas pretty quickly

He means a "few seconds" as in not nearly long enough to account for all the swapping of textures that can happen across the course of playing though a level of a game. Why are you being so thick about this?

most benchmarks i'm aware of play through a significant proportion of a level, look at the farcry ones for example...

 

[jiffylube1024]

Wow this thread got nasty in a hurry. Guys chill out! Also, this is a video card forum, not English 101 . Now I'm going to go chill out for the next few seconds. See you in an hour .

 

[BFG10K]

I notice Dug is continuing to argue his flawed claims while ignoring my evidence for the third time.

Folks, take note: this is trolling 101.

 

[dug777]

Originally posted by: BFG10K

meh, i simply resent being told my topic was flawed, by someone who has nothing to prove otherwise.

This is a joke, right? If not I see one of several possible possible scenarios here:

(1) Dug doesn't understand how hyperlinks work, therefore he's incapable of getting to the linked evidence.
(2) Dug is burying his head in the sand and refuses to accept any evidence that contradicts his flawed claims.
(3) Dug has issues with reading and/or comprehension.
(4) Dug is blind.
(5) Dug is trolling.

On the off chance that he genuinely missed the data the first two times here is the evidence for the third time.

HL2 Benchmarks
Call of Duty 2
Quake 4
BF2

And a nice quote from B3D:

With our Quake 4 test we see that the performance of the 512MB GTX is over 50% higher than the 256MB which, given that this exceeds any of the theoretical metrics, is likely due to the increase in local RAM and the quantity of data Quake 4 likes to store locally with the Ultra detail settings enabled - when the quantity of data exceeds the local framebuffer space then more will be pushed across to system RAM thus having to be addresses over the PCI Express interface, which is much lower performance then the local memory.

That's it. I can't make it any simpler than that I'm afraid.

I'm aware of this for q4, didn't you yourself say one game means nothing at the beginning of this? I'm glad to see trolling is coming to teh same conclusion that AT did tho

AS i said most benchmarks will play through a reasonable amount of a level (my example being farcry ones (which if you use the preset ones last a few mins at least), but every hl2/d3 demo i've seen goes a decent way through a level, plenty enough to simulate real life gameplay)...

In an apples to apples comparison, at very high res/aa, the extra ram did almost nothing, which is all i ever said, and all the data i referred to.

take it easy on the insults as well, i haven't insulted you once.

EDIT: i'm glad to see you insulted me again while i was replying to you, since you clearly can't have an argument in a civil fashion without resorting to personal insults...

 

[kylebisme]

Originally posted by: dug777
most benchmarks i'm aware of play through a significant proportion of a level, look at the farcry ones for example...

I don't know what Far Cry benchmarks you expect me to look at, but that is completely beside the point of what BFG has been trying to get across to you. Again, why are you being so thick about this?

 

[dug777]

Originally posted by: TheSnowman

Originally posted by: dug777
most benchmarks i'm aware of play through a significant proportion of a level, look at the farcry ones for example...

I don't know what Far Cry benchmarks you expect me to look at, but that is completely beside the point of what BFG has been trying to get across to you. Again, why are you being so thick about this?

most benchmarks play through a significant proportion of a level, therefore any stuttering/hitching as a result of having only 256mb RAM should show up in the avg fps at the end, or at least in the min fps (which is why i mentioned the hardocp bencjmarks).

 

[kylebisme]

Originally posted by: dug777
I'm aware of this for q4, didn't you yourself say one game means nothing at the beginning of this?

Holy crap man, you are trying to rail him for bring up one game when he linked to benchmarks of 4? The reason you are getting personaly insulted is because you are being a moron, did you eat a lot of paint chips as a kid or something?

 

[kylebisme]

Originally posted by: dug777

Originally posted by: TheSnowman

Originally posted by: dug777
most benchmarks i'm aware of play through a significant proportion of a level, look at the farcry ones for example...

I don't know what Far Cry benchmarks you expect me to look at, but that is completely beside the point of what BFG has been trying to get across to you. Again, why are you being so thick about this?

most benchmarks play through a significant proportion of a level, therefore any stuttering/hitching as a result of having only 256mb RAM should show up in the avg fps at the end, or at least in the min fps (which is why i mentioned the hardocp bencjmarks).

No it won't show up much as it only happens for a second or so while the textures swap. The longer the benchmark runs without any swaping on either card the more the average framerate will look the same. Do you need me to draw you a damn picture or something?

 

[jiffylube1024]

Originally posted by: BFG10K
HL2 Benchmarks
Call of Duty 2
Quake 4
BF2

And a nice quote from B3D:

With our Quake 4 test we see that the performance of the 512MB GTX is over 50% higher than the 256MB which, given that this exceeds any of the theoretical metrics, is likely due to the increase in local RAM and the quantity of data Quake 4 likes to store locally with the Ultra detail settings enabled - when the quantity of data exceeds the local framebuffer space then more will be pushed across to system RAM thus having to be addresses over the PCI Express interface, which is much lower performance then the local memory.

That's it. I can't make it any simpler than that I'm afraid.

It is probable that some of the increase is due to the extra RAM, but it's not for certain. It could just as easily be memory bandwidth limited on the 256MB card, which is killing the frames. Since they're comparing the 512MB GTX at its much higher clocks vs. the 256MB GTX at its stock clocks, B3D is merely speculating. Chances are the extra RAM is indeed helping some, but there is absolutely no proof that it's doing the lion's share of improvement; more like the 'icing on the cake' .

Furthermore, Anandtech's benchmarks show that there's very little increase due to the memory size - nothing until 20xx by 15xx with 4X AA. Anand has tested numbers to back up their claims, while B3D is speculating.

Yes, B3D is full of quite a few wise individuals, but AT is on the more solid ground in this case. Let's see B3D clock down that 512MB GTX and do some testing and then we will see if their speculation is actually true or not.

 

[BFG10K]

take it easy on the insults as well,

What insults are those? Saying you're trolling is not an insult.

most benchmarks i'm aware of play through a significant proportion of a level,

What does a level have to do with this? Most benchmarks are less than a minute and even the longest ones seldom exceed a few minutes. OTOH the length of a typical FPS is 8 hours or more.

look at the farcry ones for example...

Not only is this beside the point, I don't see any Far Cry benchmarks in your original Anandtech benchmarks. Why even bring up Far Cry?

I'm aware of this for q4, didn't you yourself say one game means nothing at the beginning of this? <snip>

I didn't understand a word of your response mainly because it seems to be a random collection of meaningless comments. What does any of it have to do with your original flawed claim that 512 MB does nothing and that Anand's benchmarks are all you need to prove that claim?

 

[dug777]

Originally posted by: TheSnowman

Originally posted by: dug777
I'm aware of this for q4, didn't you yourself say one game means nothing at the beginning of this?

Holy crap man, you are trying to rail him for bring up one game when he linked to benchmarks of 4? The reason you are getting personaly insulted is because you are being a moron, did you eat a lot of paint chips as a kid or something?

easy mate...the first link shows exactly what i said in my OP, but says nothing about smoothness and lack of hitching, the second link i linked to myself in the OP, and read my op aagain for my comments on that, that was mainly why i posted the thread. You didn;t even bother to read my OP before jumping in to flame did you? The third is the one i mentioned, and the fourth is a valid point, but is barely significant in terms of the differnce in fps at the range it't. If you bothred to look we've already discussed this one as well.

So stop trolling, throwing personal insults (which i haven't at any stage), and check your facts and the OP before you flame people.

 

[dug777]

Originally posted by: jiffylube1024

Originally posted by: BFG10K
HL2 Benchmarks
Call of Duty 2
Quake 4
BF2

And a nice quote from B3D:

With our Quake 4 test we see that the performance of the 512MB GTX is over 50% higher than the 256MB which, given that this exceeds any of the theoretical metrics, is likely due to the increase in local RAM and the quantity of data Quake 4 likes to store locally with the Ultra detail settings enabled - when the quantity of data exceeds the local framebuffer space then more will be pushed across to system RAM thus having to be addresses over the PCI Express interface, which is much lower performance then the local memory.

That's it. I can't make it any simpler than that I'm afraid.

It is possible, maybe even likely that some of the increase is due to the extra RAM, but it's not for certain. It could just as easily be memory bandwidth limited on the 256MB card, which is killing the frames. Since they're comparing the 512MB GTX at its much higher clocks vs. the 256MB GTX at its stock clocks, B3D is merely speculating.

Furthermore, ATI's benchmarks show that there's very little increase due to the memory size - nothing until 20xx by 15xx with 4X AA. Anand has tested numbers to back up their claims, while B3D is speculating.

Yes, B3D is full of quite a few wise individuals, but AT is on the more solid ground in this case. Let's see B3D clock down that 512MB GTX and do some testing and then we will see if their speculation is actually true or not.

exactly what i've said all along, the AT test is the only apples to apples one. The whole aspect of smoothness is pure conjecture on anyone's part without actually playing the game yourself, i never said it wasn't smoother, but that nothing concrete we have goes to prove or even suggest this...

 

[jiffylube1024]

Originally posted by: BFG10K

take it easy on the insults as well,

What insults are those? Saying you're trolling is not an insult.

Pretty much all of the stuff you were saying to dug and a couple others about not comprehending the basics of the English language, insinuating they lack Grade 6 English skills, etc. IMO that's unnecessary. Dug has 12K+ posts, and all of them have been in English, to the best of my recollection .


I think the main failure to communicate is, once again, arguments of normative usage of words.

When people are saying that the extra 256MB does nothing for frames, they mean a few things:

1. At the resolutions and IQ settings they play at, it does indeed do nothing (eg: 1280, 4X AA, 16X AF, 1600 4X AA, 16X AF)

2. At VERY high AA/AF settings, the game runs at unplayable frames per second, meaning the results are academic, but have no real value to a gamer who wants to actually play the game.

3. (and this is the key one): by "Nothing" they don't indeed mean nothing (zero, nil), simply the conversational 'nothing' : very little, eg. an improvement of under 10% at playable settings. So for someone who realizes that less than 10% of an improvement is indeed more than nothing, they are wrong. But in conversational english, they just mean a very small amount, an intangible amount, an amount which is not relevant to them.


Once again I agree with this - there's no point in whipping out the old Oxford English Dictionary! 512MB does very little for the new GTX; the bulk of the improvements is from the clockspeed.

 

[BFG10K]

...the first link shows exactly what i said in my OP,

No it doesn't and in fact all four links blatantly disprove this farce of a thread.

but says nothing about smoothness and lack of hitching

Uh uh, and what did your original AT benchmarks show with respect smoothness and hitching, Doug?

I love it, first your evidence was AT's benchmarks showing no performance difference.

Then when other benchmarks were posted to disprove your claims you then turned around and told us they don't count because you need to see hitching/smoothness, something your original claim and orignal AT benchmarks didn't even show.

This sort of childish logic is exactly why these threads eventually end up getting locked. This thread is nothing more than blatant trolling, no two ways about it.

 

[dug777]

Originally posted by: jiffylube1024

Originally posted by: BFG10K

take it easy on the insults as well,

What insults are those? Saying you're trolling is not an insult.

Pretty much all of the stuff you were saying to dug and a couple others about not comprehending the basics of the English language, insinuating they lack Grade 6 English skills, etc.


I think the main failure to communicate is, once again, normative arguing.

When people are saying that the extra 256MB does nothing for frames, they mean a few things:

1. At the resolutions and IQ settings they play at, it does indeed do nothing (eg: 1280, 4X AA, 16X AF, 1600 4X AA, 16X AF)

2. At VERY high AA/AF settings, the game runs at unplayable frames per second, meaning the results are academic, but have no real value to a gamer who wants to actually play the game.

3. (and this is the key one): by "Nothing" they don't indeed mean nothing, simply very little, eg an improvement of under 10% at playable settings. So for someone who realizes that less than 10% of an improvement is indeed more than nothing, they are wrong. But in conversational english, they just mean a very small amount, an intangible amount, an amount which is not relevant to them.


Once again I agree with this - there's no point in whipping out the old Oxford English Dictionary! 512MB does very little for the new GTX; the bulk of the improvements is from the clockspeed.

:beer: excellently put. (i'll note i said 'almost nothing' in my title as well).

 

[kylebisme]

Originally posted by: dug777

Originally posted by: TheSnowman

Originally posted by: dug777
I'm aware of this for q4, didn't you yourself say one game means nothing at the beginning of this?

Holy crap man, you are trying to rail him for bring up one game when he linked to benchmarks of 4? The reason you are getting personaly insulted is because you are being a moron, did you eat a lot of paint chips as a kid or something?

easy mate...the first link shows exactly what i said in my OP, but says nothing about smoothness and lack of hitching, the second link i linked to myself in the OP, and read my op aagain for my comments on that, that was mainly why i posted the thread. You didn;t even bother to read my OP before jumping in to flame did you? The third is the one i mentioned, and the fourth is a valid point, but is barely significant in terms of the differnce in fps at the range it't. If you bothred to look we've already discussed this one as well.

So stop trolling, throwing personal insults (which i haven't at any stage), and check your facts and the OP before you flame people.

None that changes the fact that you tried to call him out for focusing on one game when he quite obviously brought up 4.

And seriously, dig back in the arhives and you will find BFG and myself explaining this same fact to people aguing about the point of 256mb cards, and 128mb cards before that, and 64mb cards even further back. Hardly ever is the difference very apparent in average framerate benchmarks, but all the same ther will be hitching from textures swapping in and out of video ram whenever you don't have enough ram on the card to fit everything.

 

[jiffylube1024]

On hitching - doesn't hitching/stuttering usually affect timedemo scores? Since the framerate gets killed for a second or a split second, an unnecessarily low framerate occurs for a second, does it not?

If this is the case, then there should be several symptoms:

-Extremely low minimum FPS on the 256MB vs the 512MB card, probably in the single digits. If the 256MB card is hitching, it should be posting drastically lower minimums than the 512MB card, should it not?

-Slightly diminished average FPS, proportional to the length of the benchmark. The shorter the benchmark timedemo, the more a hitch (or several hitches) would drag down average fps. Of course in a long demo the hitches would be 'averaged out' and the difference would fall within the margin of error.


I think a good place to start to back up this '256MB hitches' argument is to find some conspicuously low minimum framerates on the 256MB card. An even better proof would be finding minimum framerates on a downclocked 512MB 7800GTX versus a 256MB 7800GTX @ stock.

One thing to remember is that some hitching is caused by not enough system RAM, so it's best to make sure the tester in question is using 2GB of RAM, since it is known that some games (plus a minimum of Windows background tasks) do indeed exceed 1GB of RAM usage.

 

[dug777]

Originally posted by: BFG10K

...the first link shows exactly what i said in my OP,

No it doesn't and in fact all four links blatantly disprove this farce of a thread.

but says nothing about smoothness and lack of hitching

Uh uh, and what did your original AT benchmarks show with respect smoothness and hitching, Doug?

I love it, first your evidence was AT's benchmarks showing no performance difference.

Then when other benchmarks were posted to disprove your claims you then turned around and told us they don't count because you need to see hitching/smoothness, something your original claim and orignal AT benchmarks didn't even show.

This sort of childish logic is exactly why these threads eventually end up getting locked. This thread is nothing more than blatant trolling, no two ways about it.

did you even read my OP and my questions regarding why we saw a difference for the x800xl? The AT article is the only equivalent to that, being the only apples to apples comparison out of the collection. You've twisted everything i said so much i really don;t know what you are saying anymore, and i'm sick of your insulting nonsense. Jiffylube has summed it up nicely, argue with him if you must.

You and Snowman brought up the smoothness/hitching aspect if you recall, and i merely answered you by saying at this stage that's pure conjecture. I haven't thrown one personal insult at you, and stayed civil throughout. People like you who can't have a civil argument ARE the reason perfectly valid threads get locked.