Its a toss up between 2 8800gt

[Threeflow]

this PT Boats game is one such game that can take advantage of more than 512MB of memory on the video card. Seems rather simple to me.

 

[JRMBelgium]

And again, what happens when you OC the the GT?

An OC does not increase texture usage. An OC will give you 22FPS instead of 20

1024MB requires more shaders, faster core and at least 512mbit if you want acceptable framerates.

The 9800 will be 1GB, that one will allow playable framerates on settings that are unplayable with the 8800gt, I think that we can all be sure of that...

 

[Wile E]

An OC does not increase texture usage. An OC will give you 22FPS instead of 20

1024MB requires more shaders, faster core and at least 512mbit if you want acceptable framerates.

The 9800 will be 1GB, that one will allow playable framerates on settings that are unplayable with the 8800gt, I think that we can all be sure of that...

My point is that it doesn't necessarily require more shaders if you OC. All you need is to perform more ops in the same amount of time. OCing, more shaders, a faster interface (in some cases), etc. can all lead to this. This is why I really want to see the difference when you introduce OCing. Somehow I think the gap between 600/1800/1500 and 775/2100/1902 is more than 2fps. lol.

And I'm sticking by my claim that a 512bit interface will make no difference at all.

I'm really starting to think I may have to put a better quad on the back burner, and buy a hi-res screen so I can actually test this stuff on my own.

 

[ChillyMyst]

WileE, maby you should just get a nice projector or hdtv that can do 1920x1080 and use that to test

and what WileE says is true, overclocking a 256mb 9800pro let it take advantege of the extra 128mb ram when normaly the diffrance between 128 and 256mb 9800pro cards was NILL, and i agree wileE your listed clocks will give more then 2fps boost

as to crysis, i tryed it again today, it really needs more patching to get it optimized, but really so did farcry when it came out so im not to worried about that at the moment.

If a game can take advantege of 1gb then a 1gb card will help, same was true back when doom3 came out, ultra settings only worked on 512mb cards, it loaded uncompressed textures into the ram, giving slitly better quility, 256 and 128mb cards of the day couldnt cope with those settings without using agp texturing and that kiled perf.

please Jelle Mees learn more about videocards b4 you post FUD as fact, i have been a pc gamer since b4 the voodoo1 came out, i have had cards from the s3 virge to the 8800gt, i have vast experiance with what extra ram can do for even a chip that isnt the greatist, example i had a 256mb 9600nonpro, the card couldnt in most cases use the extra 128mb ram it had, BUT in some games the extra ram put me in a better possition then a 9600xt 128mb card(i was overclocked past pro core speeds) one of the beta games i was a tester for at the time acctualy reccomended 256mb because the game was able to buffer all its textures to videoram for each level/area using the extra 128mb of ram, this really showd when u cranked quiltity up at 1280x1024(high res for those days) i was as much as 15fps faster then the 9600xt users in the same beta.

granted the extra ram isnt usefull in most tested games but new games come out all the time, and with 1gb avalable im sure some companys will start taking advantege of it..

sall good i have a 8800gt 512 tho, (720/1775/2100) just need a better then stock cooler, or to go over and kick wileE's arse and steal his 8800gt 1gb card

 

[JRMBelgium]

You can say all you want. The fact remains that the 8800GT 1024MB only gives an advantage on unplayable settings...an OC isn't gonna change that...

 

[Wile E]

You can say all you want. The fact remains that the 8800GT 1024MB only gives an advantage on unplayable settings...an OC isn't gonna change that...

That's in one, poorly optimized game. And still, have you seen an OCed test between a 512MB and 1GB GT to confirm?

EDIT: If not, you can't say for sure the 1GB is a waste of money. And to be fair, I can't say it is worth the money.

 

[ChillyMyst]

WileE give up,-removed-, just let it go, he wont ever see that just because one poorly optimized game dosnt get an advantege from the extra ram that other games do, he also thinks that poorly writen/done reviews by sites like [H] are valid........its like somebody whos used vista for 5 minutes saying it has no problems at all because they where able to check their email just fine.

some people will belive what they want to, just like a chick i know belived her BF when he said an 8600gts was better then a 3850 because radeons suck......ROFL

im just helping a buddy get his father in law setup so he can game, he has a crappy old celeron 2.4gz with a 9200.....this is gonna take some work

 

[Tatty_Two]

The reason why 1GB on the 8800Gt does not perform well is fairly simple, at a GPU speed of 600mhz there is just not enough GPU power to process the larger flow of data quick enough to bring any benefit and as the charts that have been posted show, it can have a detremental effect...... whoever says that overclocking the GPU wont help is wrong (IMO), at a GPU speed of 700Mhz (with shaders linked) the performance of the 1GB increases at 16xx x 10xx by upto 17% in crysis for example, how do I know that? well I had an Asus 512MB 8800GT before the Palit 1GB in the same system and in the Crysis Demo with all settings the same I saw an improvement on max detail with 4x AA of upto (max) 11FPS and an average of 7FPS.......now thats fairly substantial but I stress thats max everything and with max everything (in Crysis demo) the difference is almost beetween a non playable game and a playable one. I used FRAPS to measure that, crysis I beleive, even at 1280 x 1024 with high detail but NO AA uses more than 512MB of memory.

I will also go so far as to say, my Palit 1GB card performs better with high detail in Crysis than my G92 8800GTS with it's extra shaders and faster clock speeds. I appreciate that I am talking just one game here and I did not try the comparison at the time on any others so the same MAY NOT apply to other games, although, contrary to what one person said, I would find it hard to beleive that any modern 1st person game on a resolution of 16xx x 10xx and above on max detail and max AA would not use more than 512MB GDDR period.

If anyone would like me to make comparisons between the 1GB 8800GT and the 512MB 8800GTS (clocked at same speeds) in any other game (I dont still have the Asus 8800GT) I have I would be more than happy to do so, the games I have at the moment that are fairly recent, apart from Crysis are:

Supreme Commander
COD4
Stalker
Company of Heroes

It may be interesting to see if the slower 1GB card can beat the faster 512MB card in any of these games with max everything at 1600 resolutions, just shout if you want some benches

 

[JRMBelgium]

It's funny that you instult me, me, someone who uses reviews to defend his statements. Someone who has owned and/or tested practicly every card that gets released.

I did simular tests as HardOCP to test if they were honest.

But you guys, the only thing you can say is "I think that" because you haven't tested ****, youf all ignorant. You instult me while you don't have a single review that shows you might my correct.

I got reviews to backup your statements, what do you guys have? God??? Are you guys all phychic???

 

[ChillyMyst]

You are lucky. I've build a PC last week with the 4600+ and the 3870 and it had about 9400 3DMarks in 3DMark06, overclocked it had about 9960 3DMarks.

8800GT should give you about 1000-1500 more in 3DMark06.

PS: The 4600+ was 8% overclocked. ( simple bios setting on the Asus mainboard )

Is there a bottleneck? Not really. Will the card run much better on faster CPU, offcourse!

Your next upgrade should be a Phenom CPU

8% overclock......pathetic, you need to learn to overclock better.

you also need to learn that 3dmark dosnt mean shit in REAL GAME PERFORMANCE.

let me explain, you see 3dmark is a synthetic benchmark tool, it dosnt use situations that real games use, its fine for comparing tweaks and stability testing, infact thats what i use it for, i run a loop over night and if the systems still going in the morning then i know my video overclocks stable.

3dmark and other synthetic benches are VERY prown to inconsistancys from system to system, you can change 1 setting in bios or drivers that wont have any effect on real games and you can loose 1500 3dmarks.

example, i have 2 systems here, both are socket 754 3700+@2.65gz, x800xt pe cards, 2gb hign end ram, everything the same except the motherboard, ones a chantech vnf3-250 the other has an MSI nf3-250 in it, guess what, even with the same memory settings and other bios tweaks the chantech board system gets over 1000points on the msi system, when you look deeper into it you find that the async latancy is 1ms diffrant between the 2 boards, that setting has NO effect on games, both systems play farcry,fear,hl2,quake4,doom3,prey and a host of other games with the same avg fps, oh and yes the os's are the same infact they are cloned from drive to drive(was to lazy to do 2 installs so i did 1, same hardware other then board brands...why work harder then u need to? )

i understand you think you know alot about videocards and whats better, but from what i can see you lack real world experiance dealing with a wide spectrum of hardware, i dont, i deal with this shit every day, 8600,8800,3850,3870, blah blah blah, i have had about every card you can think of at one point or another, i had compaired 64 to 128 128 to 256 256 to 512 and so on, and you know what, at first the extra ram dosnt tend to make a huge diffrance but then games come out or are patched to take advantege of the extra ram, also companys like ati/amd and nvidia tend to take a bit of time to optimize the drivers to allow the cards to truely take advantege of the extra ram.

example, ati's x800 drivers, at first the cards perf was kickass, but then after a couple driver revisions they put out a driver that was made to optimize the use 256mb of ram the x800pro/xt cards had on them, it also boosted 9800xt 256mb perf as well tho!!!, there was up to a 90fps boost in some games(FARCRY!!!!!!) i think it was the 4.12 rivers that did it, cant remmber its been a few years, but point is that after they did that the cards perf stomped the 128 me cards of the day, it also made an already fast card even faster.........same can happen here, i dont give nvidia alot of credit in the driver update market, mainly because i have seen how they can cock that up, but in this case they could very well put out a driver that will give these 1gb cards an easy boost.

the 2900xt 1gb got a nice boost from a driver update from what one review update i read said, will look for the site later, my net is being slow as turtles stampeeding thru peanutbutter tonight.

as my german friend would say "you fail it" meaning you lack proper knowlage to continue this argument.
to use a like from tv "you are the weakist link, good buy"

ps dont take this wrong, i just get sick of people talking out their asses about stuff they dont have any personal experiance with.

 

[JRMBelgium]

The reason why 1GB on the 8800Gt does not perform well is fairly simple, at a GPU speed of 600mhz there is just not enough GPU power to process the larger flow of data quick enough to bring any benefit and as the charts that have been posted show, it can have a detremental effect...... whoever says that overclocking the GPU wont help is wrong (IMO), at a GPU speed of 700Mhz (with shaders linked) the performance of the 1GB increases at 16xx x 10xx by upto 17% in crysis for example, how do I know that? well I had an Asus 512MB 8800GT before the Palit 1GB in the same system and in the Crysis Demo with all settings the same I saw an improvement on max detail with 4x AA of upto (max) 11FPS and an average of 7FPS.......now thats fairly substantial but I stress thats max everything and with max everything (in Crysis demo) the difference is almost beetween a non playable game and a playable one. I used FRAPS to measure that, crysis I beleive, even at 1280 x 1024 with high detail but NO AA uses more than 512MB of memory.

I will also go so far as to say, my Palit 1GB card performs better with high detail in Crysis than my G92 8800GTS with it's extra shaders and faster clock speeds. I appreciate that I am talking just one game here and I did not try the comparison at the time on any others so the same MAY NOT apply to other games, although, contrary to what one person said, I would find it hard to beleive that any modern 1st person game on a resolution of 16xx x 10xx and above on max detail and max AA would not use more than 512MB GDDR period.

If anyone would like me to make comparisons between the 1GB 8800GT and the 512MB 8800GTS (clocked at same speeds) in any other game (I dont still have the Asus 8800GT) I have I would be more than happy to do so, the games I have at the moment that are fairly recent, apart from Crysis are:

Supreme Commander
COD4
Stalker
Company of Heroes

It may be interesting to see if the slower 1GB card can beat the faster 512MB card in any of these games with max everything at 1600 resolutions, just shout if you want some benches

Finnally. A decent argument.

Now, can you repost some framerates? Before/after OC and stuff like that?

 

[ChillyMyst]

The reason why 1GB on the 8800Gt does not perform well is fairly simple, at a GPU speed of 600mhz there is just not enough GPU power to process the larger flow of data quick enough to bring any benefit and as the charts that have been posted show, it can have a detremental effect...... whoever says that overclocking the GPU wont help is wrong (IMO), at a GPU speed of 700Mhz (with shaders linked) the performance of the 1GB increases at 16xx x 10xx by upto 17% in crysis for example, how do I know that? well I had an Asus 512MB 8800GT before the Palit 1GB in the same system and in the Crysis Demo with all settings the same I saw an improvement on max detail with 4x AA of upto (max) 11FPS and an average of 7FPS.......now thats fairly substantial but I stress thats max everything and with max everything (in Crysis demo) the difference is almost beetween a non playable game and a playable one. I used FRAPS to measure that, crysis I beleive, even at 1280 x 1024 with high detail but NO AA uses more than 512MB of memory.

I will also go so far as to say, my Palit 1GB card performs better with high detail in Crysis than my G92 8800GTS with it's extra shaders and faster clock speeds. I appreciate that I am talking just one game here and I did not try the comparison at the time on any others so the same MAY NOT apply to other games, although, contrary to what one person said, I would find it hard to beleive that any modern 1st person game on a resolution of 16xx x 10xx and above on max detail and max AA would not use more than 512MB GDDR period.

If anyone would like me to make comparisons between the 1GB 8800GT and the 512MB 8800GTS (clocked at same speeds) in any other game (I dont still have the Asus 8800GT) I have I would be more than happy to do so, the games I have at the moment that are fairly recent, apart from Crysis are:

Supreme Commander
COD4
Stalker
Company of Heroes

It may be interesting to see if the slower 1GB card can beat the faster 512MB card in any of these games with max everything at 1600 resolutions, just shout if you want some benches

should add the ut3 demo in there, that would be a good little addon. and yeah, if you have the cards to compair i would lke to see some benching, specly if u can give the cards a little overclocking, my stock 8800gt is at 720/1775/2100 and quite happy

 

[JRMBelgium]

8% overclock......pathetic, you need to learn to overclock better.

you also need to learn that 3dmark dosnt mean shit in REAL GAME PERFORMANCE.

let me explain, you see 3dmark is a synthetic benchmark tool, it dosnt use situations that real games use, its fine for comparing tweaks and stability testing, infact thats what i use it for, i run a loop over night and if the systems still going in the morning then i know my video overclocks stable.

3dmark and other synthetic benches are VERY prown to inconsistancys from system to system, you can change 1 setting in bios or drivers that wont have any effect on real games and you can loose 1500 3dmarks.

example, i have 2 systems here, both are socket 754 3700+@2.65gz, x800xt pe cards, 2gb hign end ram, everything the same except the motherboard, ones a chantech vnf3-250 the other has an MSI nf3-250 in it, guess what, even with the same memory settings and other bios tweaks the chantech board system gets over 1000points on the msi system, when you look deeper into it you find that the async latancy is 1ms diffrant between the 2 boards, that setting has NO effect on games, both systems play farcry,fear,hl2,quake4,doom3,prey and a host of other games with the same avg fps, oh and yes the os's are the same infact they are cloned from drive to drive(was to lazy to do 2 installs so i did 1, same hardware other then board brands...why work harder then u need to? )

i understand you think you know alot about videocards and whats better, but from what i can see you lack real world experiance dealing with a wide spectrum of hardware, i dont, i deal with this shit every day, 8600,8800,3850,3870, blah blah blah, i have had about every card you can think of at one point or another, i had compaired 64 to 128 128 to 256 256 to 512 and so on, and you know what, at first the extra ram dosnt tend to make a huge diffrance but then games come out or are patched to take advantege of the extra ram, also companys like ati/amd and nvidia tend to take a bit of time to optimize the drivers to allow the cards to truely take advantege of the extra ram.

example, ati's x800 drivers, at first the cards perf was kickass, but then after a couple driver revisions they put out a driver that was made to optimize the use 256mb of ram the x800pro/xt cards had on them, it also boosted 9800xt 256mb perf as well tho!!!, there was up to a 90fps boost in some games(FARCRY!!!!!!) i think it was the 4.12 rivers that did it, cant remmber its been a few years, but point is that after they did that the cards perf stomped the 128 me cards of the day, it also made an already fast card even faster.........same can happen here, i dont give nvidia alot of credit in the driver update market, mainly because i have seen how they can cock that up, but in this case they could very well put out a driver that will give these 1gb cards an easy boost.

the 2900xt 1gb got a nice boost from a driver update from what one review update i read said, will look for the site later, my net is being slow as turtles stampeeding thru peanutbutter tonight.

as my german friend would say "you fail it" meaning you lack proper knowlage to continue this argument.
to use a like from tv "you are the weakist link, good buy"

ps dont take this wrong, i just get sick of people talking out their asses about stuff they dont have any personal experiance with.

Oh my God. Where did you see me saying that 3DMark reflects gaming performance???

8% overclock? Yeah, so what. It was just a simple bios setting, I didn't take time to overclock the CPU! Did you see mee saying that 8% is the maximum??

You read my post and jumped to conclusions that don't make sence. You can google my as much as you want, you will always see me saying "3dmark doesn't reflect game performance!!!"

You lack the ability to read posts correctly. NEVER will you see me saying that 3DMark scores say anything about hardware!!!

 

[ChillyMyst]

It's funny that you instult me, me, someone who uses reviews to defend his statements. Someone who has owned and/or tested practicly every card that gets released.

I did simular tests as HardOCP to test if they were honest.

But you guys, the only thing you can say is "I think that" because you haven't tested ****, youf all ignorant. You instult me while you don't have a single review that shows you might my correct.

I got reviews to backup your statements, what do you guys have? God??? Are you guys all phychic???

if everybody is out to get you maby you should stop what your doing, as to reviews, there arent alot of 8800gt 1gb reviews yet that i can find, and the few idid are like the hardocp one, they are invlaid because they dont tell you what settings are use or because they use weird mixed settings that differ from test to test, thats no way to bench a card.......

and your calling us ugnorant........wileE has the card in question, i have an 8800gt 512 and have tested it side by side with a 3850 512mb and a 3870, in real games you cant tell the diffrance till ur past 1600x1200, crysis is the ONLY place you gotta go lower, and that games quite poorly optimized, it really needs some patches b4 its a good bench to use.

wheres your proof that a 9800 1gb will beable to play crysis at high settings properly? show us some benchmarks that proove it???common now where are they??? oh you cant proove it....to bad........

im sorry yall but this is getting to me, this guy insults and talks down to people dispite not having the cards in question or the background to know what hes talking about........

WileE i will leave this to you, but i would just add this guy to my iggy list, hes clearly a KIA(know it all) and isnt gonna have his mind changed even if you had video recordings of proof that 1gb card can out perform a 512mb card.

 

[ChillyMyst]

Oh my God. Where did you see me saying that 3DMark reflects gaming performance???

8% overclock? Yeah, so what. It was just a simple bios setting, I didn't take time to overclock the CPU! Did you see mee saying that 8% is the maximum??

You read my post and jumped to conclusions that don't make sence. You can google my as much as you want, you will always see me saying "3dmark doesn't reflect game performance!!!"

You lack the ability to read posts correctly. NEVER will you see me saying that 3DMark scores say anything about hardware!!!

you are the one who started in with 3dmark scores, they are in your chart and you mantioned them saying that a 8800gt would get 1000-1500 higher then the 3870.....


Oh and back on topic, if u got a crossfire board i would consider 2 3850 512mb powercolor cards, they come stock at 720 core(higher then any other stock 3850, pretty close to the clock on a 3870!!!) also a kickass lifetime warr that they acctualy honour!!!

 

[JRMBelgium]

I am a know it all because I read reviews before I make statements.

Again, not a single test so far has prooven what you all are saying.

 

[JRMBelgium]

you are the one who started in with 3dmark scores, they are in your chart and you mantioned them saying that a 8800gt would get 1000-1500 higher then the 3870.....

He wanted to know how the 8800GT would perform on his 4600+. I just sad "you will have about 10000-11000 points in 3DMark. That's all I sad. I didn't mention in-game performance or anything. Just wanted to help out the guy with his question.

Again, I tested it myself. Again I post something based on facts...

 

[JRMBelgium]

Another 2900XT 1024MB review:
http://www.trustedreviews.com/graphics/review/2007/08/14/ATI-HD-2900-XT-1024MB-DDR4/p1

A 3850 1024MB review:
http://www.hartware.de/review_784.html

That's 4 reviews so far that all show the same thing. 1024MB only provides an advantage on unplayable settings.

Seriously, how many reviews do you need to get convinced?

Why would 4 different reviews websites fake results???

 

[Mussels]

Why would 4 different reviews websites fake results???

To get more free review samples from the other company? heh, many review sites are biased. I stopped doing reviews because every last company (except thermalright and OCZ) nags you behind the scenes to make their products look better - and if you dont, you stop getting samples.



Thats highly unlikely for this situation, but i make a request (to EVERYONE): stop posting opinions and fanboy bullshit. Either link to a review, or post about personal experience with the hardware. by posting something incomplete and missing minor details, you're only encouraging the other fanboys to pick on the one tiny flaw in your wording, and it ends up a flame war. we dont want that.

 

[Xaser04]

Another 2900XT 1024MB review:
http://www.trustedreviews.com/graphics/review/2007/08/14/ATI-HD-2900-XT-1024MB-DDR4/p1

A 3850 1024MB review:
http://www.hartware.de/review_784.html

That's 4 reviews so far that all show the same thing. 1024MB only provides an advantage on unplayable settings.

Seriously, how many reviews do you need to get convinced?

Why would 4 different reviews websites fake results???

Here are some results from firingsquads review of the Palit 8800GT 1gb at 2560x1600

http://www.firingsquad.com/hardware/palit_geforce_8800_gt_super+_1gb_review/page13.asp

Its interesting in the sense that whilst the 1gb card does indeed become quicker than its 512mb counterpart in crysis and oblivion, neither game is what I would call playable at these settings.

I think at certain settings in certain games the 1gb card would offer a slight performance advantage over the 512mb model however I wouldn't personally say that this performance difference is worth the extra outlay (you may as well stump up a little extra and get a GTS which is quicker again)

 

[Wile E]

Another 2900XT 1024MB review:
http://www.trustedreviews.com/graphics/review/2007/08/14/ATI-HD-2900-XT-1024MB-DDR4/p1

A 3850 1024MB review:
http://www.hartware.de/review_784.html

That's 4 reviews so far that all show the same thing. 1024MB only provides an advantage on unplayable settings.

Seriously, how many reviews do you need to get convinced?

Why would 4 different reviews websites fake results???

Now I agree that a 3850 1GB might be stretching it, but that 2900 review is from last August. Things have gotten even better for it.

But anyway, what in God's name does that have to do with the 8800GT, which, I might add, is a more powerful gpu to begin with? Just because there's no gian on a 3850, doesn't mean there's no gain on an 8800GT. You're comparing Apples to Oranges here.

And we won't believe your reviews because we have FIRST HAND experience. Reviewers don't always have the proper time, optimal setups, or any other number of reasons, to make 100% accurate reviews. What drivers were the 8800GT 1GB reviews using? What kind of increases are to be had with newer drivers, or even game patches? The list goes on and on. At any rate, the original 8800 1GB review you posted was all over the charts. That right there denotes something is amiss. There's one point where the 1GB wins at a less intensive setting, then loses at a more intensive setting in the same game.

One test proves beyond a shadow of a doubt, that 1GB cards have their benefit, yet you toss it aside as if it doesn't exist.

Also, you said Tatty_One provided a reasonable argument, yet I already said that OCing makes a substantial difference, and you blew it off.

The bottom line is, if you 100% trust review sites using release or older drivers, not listing their settings, or in any way skewing (if even inadvertently) results, then you have a lot to learn.

 

[Wile E]

Here are some results from firingsquads review of the Palit 8800GT 1gb at 2560x1600

http://www.firingsquad.com/hardware/palit_geforce_8800_gt_super+_1gb_review/page13.asp

Its interesting in the sense that whilst the 1gb card does indeed become quicker than its 512mb counterpart in crysis and oblivion, neither game is what I would call playable at these settings.

I think at certain settings in certain games the 1gb card would offer a slight performance advantage over the 512mb model however I wouldn't personally say that this performance difference is worth the extra outlay (you may as well stump up a little extra and get a GTS which is quicker again)

But that also doesn't have OCed testing. I'm willing to bet the difference would be more substantial at 700MHz core+. Something Tatty already confirmed.

 

[Xaser04]

I don't think I have read one review that compares overclocking so I can't give an onion either way there.

I would say though even with a hefty overclock the two games where the 1gb actually shows a performance advantage in the firingsquad review would still not be playable (oblivion might be if you don't mind more choppy framerates).

I was trying to point out though for the price of the 1gb 8800GT the 512mb GTS is maybe a more viable option as it does show a nice performance advantage over the GT when compared the performance difference in reviews comparing the 1gb GT with the 512mb GT.

 

[Wile E]

I don't think I have read one review that compares overclocking so I can't give an onion either way there.

I would say though even with a hefty overclock the two games where the 1gb actually shows a performance advantage in the firingsquad review would still not be playable (oblivion might be if you don't mind more choppy framerates).

I was trying to point out though for the price of the 1gb 8800GT the 512mb GTS is maybe a more viable option as it does show a nice performance advantage over the GT when compared the performance difference in reviews comparing the 1gb GT with the 512mb GT.

Yeah, the GTS may be the better value than 1GB GT, unless you find a good deal. But my only point is that the 1GB card has it's benefits, and that some games do use more than 512MB of the buffer.

tbh, I don't even remember how this all started. lol.

 

[Mussels]

I don't think I have read one review that compares overclocking so I can't give an onion either way there.

I would say though even with a hefty overclock the two games where the 1gb actually shows a performance advantage in the firingsquad review would still not be playable (oblivion might be if you don't mind more choppy framerates).

I was trying to point out though for the price of the 1gb 8800GT the 512mb GTS is maybe a more viable option as it does show a nice performance advantage over the GT when compared the performance difference in reviews comparing the 1gb GT with the 512mb GT.

i'm pretty sure a card going from 600 to 740 core like my GT has (2nd system, before you comment) would be able to power the ram a lot better. Different archcitectures have different balances - some have more GPU power and not enough ram, others have more ram power nad not enough GPU.

Assuming the 8800GT has enough GPU power for 512MB but not 1GB, the increases would make sense - on a lower end card like the 3850, its probably already starving its ram due to the 'slow' (relatively) GPU.

I'm an ex-reviewer - hows that for first hand experience? reviews you only have one system setup, a short amount of time, and often pre-release drivers - and people will still quote your article as truth and valid even if its a year old. A lot changes in a year - CPU speeds go up, systems get more ram, drivers get tuned in.