ATI xf 4830 slaps NV 280 around

[nitromullet]

Originally posted by: SSChevy2001
4830CF should be compared to 9800GT SLI not the GTX280

http://www.driverheaven.net/re...?reviewid=654&pageid=5

Wow, SLI and CF scaling in Far Cry 2 looks pretty lousy.

 

[SSChevy2001]

Originally posted by: nitromullet

Originally posted by: SSChevy2001
4830CF should be compared to 9800GT SLI not the GTX280

http://www.driverheaven.net/re...?reviewid=654&pageid=5

Wow, SLI and CF scaling in Far Cry 2 looks pretty lousy.

512MB framebuffer just isn't enough for Far Cry 2.

 

[nitromullet]

Originally posted by: SSChevy2001

Originally posted by: nitromullet

Originally posted by: SSChevy2001
4830CF should be compared to 9800GT SLI not the GTX280

http://www.driverheaven.net/re...?reviewid=654&pageid=5

Wow, SLI and CF scaling in Far Cry 2 looks pretty lousy.

512MB framebuffer just isn't enough for Far Cry 2.

Yeah, that is probably it.

 

[MarcVenice]

Originally posted by: keysplayr2003

Originally posted by: soccerballtux

Originally posted by: keysplayr2003

Originally posted by: CP5670
More information is always better, but I would much rather know the minimum framerate alone than the average framerate alone. It gives a better indication of what is happening during the big fights, when you need the performance the most.

Agree.

More info is always better. In this particular case, minimum framerate alone would tell the tale, as you say. Actually, minimum framerate, and a report on how the game "feels". As in smooth, choppy, hitching, etc.... I do appreciate those kind of comments when reading a review article. Much like N7 did with his 4870X2 GTX280 commentary/comparison.

I prefer to stay away from the "feel" comments, they hold no weight. Completely at the mercy of the reviewer's bias.

If they're biased, what business do they have benching and reviewing for web sites in the first place? How do you know they don't falsify numbers? By your logic, there should exist zero benchmarks because of undetectability of bias.

There is something called science. They have established secure ways of emperically determining something, and through the use of statistics the are able to say that they are sure about something for 90/95% of the time, depending on the alfa they use.

Feel can not be measured, and is thus always inaccurate, non reproducable and thus not objective. AVG, MIN or MAX FPS can be measured, and remeasured, it can be averaged, the worst and the best run can be substracted. Hence they are meaningfull. Feel should have no place in a review, save for an editorial comment, that everyone should look at with a lot of scepcis.

Btw, I do agree about the MIN FPS, they should have been included. On of the downfalls of many scientists is their obscure use of statistics, you can make anything look interesting, but if you dig a little deeper you often notice there's more going on then meets the eye.

 

[Keysplayr]

I just asked for a reviewer to describe the "feel" of the game when using one card or another. If there is no difference in feel, then there should be no comment. If the feel of the gameplay is obviously different, a comment should be made. This is within a human beings capability, to detect something like this. We all can when there is notable difference. Your position is that the feel of a game has no place in a review. My position is the exact opposite. Being a reviewer yourself, you might take into consideration what people want and not dismiss things so easily. If a game is "clunky", shouldn't you comment on it? If it's smooth and works without bringing your attention to anything, then I would expect no comment because the game works like it was expected to, which means normally.

 

[MarcVenice]

The problem is keys, that some people notice microstuttering, and some don't, some people notice a crappy FOV, and some don't. I think the benchmarks should speak for themselfs, and unless there really is something noticable going on, then the reviewer could make a fraps movie, and show us. Or he could make a editorial comment and say that even though said graphics card shows great performance, it just doesn't feel right, but that we should see for ourselfs. Thats what I meant with a lot of scepsis.

 

[BenSkywalker]

Feel can not be measured, and is thus always inaccurate, non reproducable and thus not objective.

And I think this points to a much larger difference between you and Keys that many share. You are a hardware enthusiast, Keys is a gamer. Sure, you may cross into each others territory, but honestly I will take a moderately educated gamers perspective over the top engineer in the world that doesn't game- the gamer's viewpoint means a LOT more to me, FAR more then a benchmark chart could ever hope to show.

This is one of the reasons I have found myself using [ H ] for reviews far more then AT's when it comes time to reccommend or buy something. I don't like Kyle's approach as far as not reporting all benchmark numbers(particularly in this day and age- how many options I can play with at 19x12 don't do squat for me at 16x10) however I tend to know far more about how cards perform in game reading one paragraph of Kyle's reviews then I do reading a series from AT.

 

[Tempered81]

I like the maxishine video reviews on youtube. They let you get a "feel" for how it plays, versus just reading benchmark numbers.

Good thread here.

 

[apoppin]

Originally posted by: MarcVenice

Originally posted by: keysplayr2003

Originally posted by: soccerballtux

Originally posted by: keysplayr2003

Originally posted by: CP5670
More information is always better, but I would much rather know the minimum framerate alone than the average framerate alone. It gives a better indication of what is happening during the big fights, when you need the performance the most.

Agree.

More info is always better. In this particular case, minimum framerate alone would tell the tale, as you say. Actually, minimum framerate, and a report on how the game "feels". As in smooth, choppy, hitching, etc.... I do appreciate those kind of comments when reading a review article. Much like N7 did with his 4870X2 GTX280 commentary/comparison.

I prefer to stay away from the "feel" comments, they hold no weight. Completely at the mercy of the reviewer's bias.

If they're biased, what business do they have benching and reviewing for web sites in the first place? How do you know they don't falsify numbers? By your logic, there should exist zero benchmarks because of undetectability of bias.

There is something called science. They have established secure ways of emperically determining something, and through the use of statistics the are able to say that they are sure about something for 90/95% of the time, depending on the alfa they use.

Feel can not be measured, and is thus always inaccurate, non reproducable and thus not objective. AVG, MIN or MAX FPS can be measured, and remeasured, it can be averaged, the worst and the best run can be substracted. Hence they are meaningfull. Feel should have no place in a review, save for an editorial comment, that everyone should look at with a lot of scepcis.

Btw, I do agree about the MIN FPS, they should have been included. On of the downfalls of many scientists is their obscure use of statistics, you can make anything look interesting, but if you dig a little deeper you often notice there's more going on then meets the eye.

What do you mean "feel" should have no place in a review?
:Q

Your own reviews of games are 100% "feel" and subjective

When i review HW, i tell you how the game "feels" on different HW

for example, in FC2, the 4870 should have played it smoothly according to the Min/Max/Av FPS .. but it didn't ..
- it lurched and chugged from frame to frame .. horribly annoying without the framerate cap

should that be left out and just the "scientific" methods be used?

i think not

 

[MarcVenice]

Originally posted by: apoppin

Originally posted by: MarcVenice

Originally posted by: keysplayr2003

Originally posted by: soccerballtux

Originally posted by: keysplayr2003

Originally posted by: CP5670
More information is always better, but I would much rather know the minimum framerate alone than the average framerate alone. It gives a better indication of what is happening during the big fights, when you need the performance the most.

Agree.

More info is always better. In this particular case, minimum framerate alone would tell the tale, as you say. Actually, minimum framerate, and a report on how the game "feels". As in smooth, choppy, hitching, etc.... I do appreciate those kind of comments when reading a review article. Much like N7 did with his 4870X2 GTX280 commentary/comparison.

I prefer to stay away from the "feel" comments, they hold no weight. Completely at the mercy of the reviewer's bias.

If they're biased, what business do they have benching and reviewing for web sites in the first place? How do you know they don't falsify numbers? By your logic, there should exist zero benchmarks because of undetectability of bias.

There is something called science. They have established secure ways of emperically determining something, and through the use of statistics the are able to say that they are sure about something for 90/95% of the time, depending on the alfa they use.

Feel can not be measured, and is thus always inaccurate, non reproducable and thus not objective. AVG, MIN or MAX FPS can be measured, and remeasured, it can be averaged, the worst and the best run can be substracted. Hence they are meaningfull. Feel should have no place in a review, save for an editorial comment, that everyone should look at with a lot of scepcis.

Btw, I do agree about the MIN FPS, they should have been included. On of the downfalls of many scientists is their obscure use of statistics, you can make anything look interesting, but if you dig a little deeper you often notice there's more going on then meets the eye.

What do you mean "feel" should have no place in a review?
:Q

Your own reviews of games are 100% "feel" and subjective

When i review HW, i tell you how the game "feels" on different HW

for example, in FC2, the 4870 should have played it smoothly according to the Min/Max/Av FPS .. but it didn't ..
- it lurched and chugged from frame to frame .. horribly annoying without the framerate cap

should that be left out and just the "scientific" methods be used?

i think not

I never said it should, did I not say you could use an editorial comment for that? It has to be looked at with scepsis though, because lurching or whatever doesn't mean much. I can't visualize it, where as I know that when a game dips below 25-20fps, it will start stuttering. Basicaly I'm saying that reviews for videocards should mostly be about numbers, min/max/avg fps. Not feel. Like I said, some people notice microstuttering, some don't, some people get motionsickness from a certain FOV, some don't.

Also Ben, your statement is pretty ridiculous. I review games, I'm a gamer first and foremost. I've got dozens of games, and only 1 videocard, maybe 2 in the near future

Also apoppin, gaming reviews and videocard reviews shouldn't be compared. I tend to 'trust' a certain website when it comes to videocard reviews, and I'm going to have to belief that the numbers they posted are real. When it comes to gaming reviews, it's so biased, you'll have to read several reviews from an author and see if you and him are on the same level, then he can start earning your trust. Games also tend to have demo's, etc etc. So I think it's an apples and oranges comparison ...

 

[apoppin]

Originally posted by: MarcVenice

Originally posted by: apoppin

Originally posted by: MarcVenice

Originally posted by: keysplayr2003

Originally posted by: soccerballtux

Originally posted by: keysplayr2003

Originally posted by: CP5670
More information is always better, but I would much rather know the minimum framerate alone than the average framerate alone. It gives a better indication of what is happening during the big fights, when you need the performance the most.

Agree.

More info is always better. In this particular case, minimum framerate alone would tell the tale, as you say. Actually, minimum framerate, and a report on how the game "feels". As in smooth, choppy, hitching, etc.... I do appreciate those kind of comments when reading a review article. Much like N7 did with his 4870X2 GTX280 commentary/comparison.

I prefer to stay away from the "feel" comments, they hold no weight. Completely at the mercy of the reviewer's bias.

If they're biased, what business do they have benching and reviewing for web sites in the first place? How do you know they don't falsify numbers? By your logic, there should exist zero benchmarks because of undetectability of bias.

There is something called science. They have established secure ways of emperically determining something, and through the use of statistics the are able to say that they are sure about something for 90/95% of the time, depending on the alfa they use.

Feel can not be measured, and is thus always inaccurate, non reproducable and thus not objective. AVG, MIN or MAX FPS can be measured, and remeasured, it can be averaged, the worst and the best run can be substracted. Hence they are meaningfull. Feel should have no place in a review, save for an editorial comment, that everyone should look at with a lot of scepcis.

Btw, I do agree about the MIN FPS, they should have been included. On of the downfalls of many scientists is their obscure use of statistics, you can make anything look interesting, but if you dig a little deeper you often notice there's more going on then meets the eye.

What do you mean "feel" should have no place in a review?
:Q

Your own reviews of games are 100% "feel" and subjective

When i review HW, i tell you how the game "feels" on different HW

for example, in FC2, the 4870 should have played it smoothly according to the Min/Max/Av FPS .. but it didn't ..
- it lurched and chugged from frame to frame .. horribly annoying without the framerate cap

should that be left out and just the "scientific" methods be used?

i think not

I never said it should, did I not say you could use an editorial comment for that? It has to be looked at with scepsis though, because lurching or whatever doesn't mean much. I can't visualize it, where as I know that when a game dips below 25-20fps, it will start stuttering. Basicaly I'm saying that reviews for videocards should mostly be about numbers, min/max/avg fps. Not feel. Like I said, some people notice microstuttering, some don't, some people get motionsickness from a certain FOV, some don't.

Also Ben, your statement is pretty ridiculous. I review games, I'm a gamer first and foremost. I've got dozens of games, and only 1 videocard, maybe 2 in the near future

Also apoppin, gaming reviews and videocard reviews shouldn't be compared. I tend to 'trust' a certain website when it comes to videocard reviews, and I'm going to have to belief that the numbers they posted are real. When it comes to gaming reviews, it's so biased, you'll have to read several reviews from an author and see if you and him are on the same level, then he can start earning your trust. Games also tend to have demo's, etc etc. So I think it's an apples and oranges comparison ...

well, Marc, for me the "feel" of HW is as important as the "feel" of driving one car compared to another.

You can compare all the specs you want, but they do not tell you the driving experience.

Just like with ver1 of FarCry2 and the first buggy hotfix drivers - 4870 looked awesome on paper but played FC2 like crap; 280GTX looked about the same on paper and in the charts but the gaming experience was not comparable - 280 blew away the 4870

also in reviewing AUDIO equipment, the same is true .. you have equipment with identical specs but one simply "sounds" better

keep your scientific methods - by themselves - they are completely worthless
- it takes a skilled human reviewer to relate the benchmarks to actual game play

that is why KyleB is successful with awful benchmarks - it still paints a picture - albeit biased

 

[nRollo]

Originally posted by: BenSkywalker

Feel can not be measured, and is thus always inaccurate, non reproducable and thus not objective.

And I think this points to a much larger difference between you and Keys that many share. You are a hardware enthusiast, Keys is a gamer. Sure, you may cross into each others territory, but honestly I will take a moderately educated gamers perspective over the top engineer in the world that doesn't game- the gamer's viewpoint means a LOT more to me, FAR more then a benchmark chart could ever hope to show.

This is one of the reasons I have found myself using [ H ] for reviews far more then AT's when it comes time to reccommend or buy something. I don't like Kyle's approach as far as not reporting all benchmark numbers(particularly in this day and age- how many options I can play with at 19x12 don't do squat for me at 16x10) however I tend to know far more about how cards perform in game reading one paragraph of Kyle's reviews then I do reading a series from AT.

I agree with this.

I think a review site that offered 3 minute FRAPs graphs at 16X10 4X16X, 19X12 4X16X, and 25X16 4X16X could do well.

 

[BFG10K]

I don?t see any ?slapping? going on there; I see the GTX280 winning in two games and the 4830 CF winning the other two. Even if 4830 CF won all four games I?d still pick the GTX280 because of the inherent advantages it enjoys due to being a single card. I have 90 games under active play rotation, not four. Testing four games doesn?t tell me a damn thing about the bulk of my library.

On another note, I agree that a subjective analysis is also important along with raw benchmark figures.

I also agree that minimums can be important but they have to be done right. Quite often spikes in minimums aren?t accurate because of benchmarking ?noise?, especially right at the start or end of benchmark runs. In actual gaming such noise will not be there.

As an example, I remember UT2003 always posted excessive minimum and maximum scores but when actually playing the game you?d never see those kinds of figures.

I also think [ H ]?s graphs are useful but I don?t agree with their manual Fraps run methodology because it means every time they run a benchmark, it?s different to what they ran before. If they?re scared of drivers being optimized for ?canned? benchmark runs (a legitimate concern), they should be creating custom timedemos.

 

[nRollo]

Originally posted by: BFG10K
I don?t see any ?slapping? going on there; I see the GTX280 winning in two games and the 4830 CF winning the other two. Even if 4830 CF won all four games I?d still pick the GTX280 because of the inherent advantages it enjoys due to being a single card. I have 90 games under active play rotation, not four. Testing four games doesn?t tell me a damn thing about the bulk of my library.

On another note, I agree that a subjective analysis is also important along with raw benchmark figures.

I also agree that minimums can be important but they have to be done right. Quite often spikes in minimums aren?t accurate because of benchmarking ?noise?, especially right at the start or end of benchmark runs. In actual gaming such noise will not be there.

As an example, I remember UT2003 always posted excessive minimum and maximum scores but when actually playing the game you?d never see those kinds of figures.

I also think [ H ]?s graphs are useful but I don?t agree with their manual Fraps run methodology because it means every time they run a benchmark, it?s different to what they ran before. If they?re scared of drivers being optimized for ?canned? benchmark runs (a legitimate concern), they should be creating custom timedemos.

BFG-

I think the end of the world is at hand.

We're in agreement on every point in your post.

:beer:

 

[Insomniator]

I have 90 games under active play rotation, not four.

?!

 

[Ares202]

Regardless of the review the 4830 is a little faster than the 9800gt, and Crossfire usually scales better than SLI so it seems to me as it is a better option to go for the 4830XF than 9800gt's in SLI

but yeh, i agree that its probably better to go for the single GTX260 or 4870 1GB over 4830-crossfire because of lower power draw, more stable framerate and no scaling issues

 

[apoppin]

Originally posted by: nRollo

Originally posted by: BFG10K
I don?t see any ?slapping? going on there; I see the GTX280 winning in two games and the 4830 CF winning the other two. Even if 4830 CF won all four games I?d still pick the GTX280 because of the inherent advantages it enjoys due to being a single card. I have 90 games under active play rotation, not four. Testing four games doesn?t tell me a damn thing about the bulk of my library.

On another note, I agree that a subjective analysis is also important along with raw benchmark figures.

I also agree that minimums can be important but they have to be done right. Quite often spikes in minimums aren?t accurate because of benchmarking ?noise?, especially right at the start or end of benchmark runs. In actual gaming such noise will not be there.

As an example, I remember UT2003 always posted excessive minimum and maximum scores but when actually playing the game you?d never see those kinds of figures.

I also think [ H ]?s graphs are useful but I don?t agree with their manual Fraps run methodology because it means every time they run a benchmark, it?s different to what they ran before. If they?re scared of drivers being optimized for ?canned? benchmark runs (a legitimate concern), they should be creating custom timedemos.

BFG-

I think the end of the world is at hand.

We're in agreement on every point in your post.

:beer:

it is close

. . . and who is the "we" that are in agreement?
:Q

if you were predicting, i agree also; every point, except i don't have 90 games in active rotation - just about a quarter of that and my oldest game is '05 that is still installed on my PC; i tend to look at about 15 games and benches when i compare anything.

i would definitely take a 260+ - or a single 4870 - over CF-4830. i would hate to play on a single 4830 when CF does not scale; then you would wish you had something to slap IT with and would be thinking you were an idiot for only looking at a 4 game benchmark suite without mins, to make your buying decision.

 

[solofly]

Originally posted by: apoppin
i would definitely take a 260+ - or a single 4870 - over CF-4830. i would hate to play on a single 4830 when CF does not scale; then you would wish you had something to slap IT with and would be thinking you were an idiot for only looking at a 4 game benchmark suite without mins, to make your buying decision.

Let me rephrase that another way.... If you're into newer games you're pretty much safe with multi-gpu but if you're into old or older games then single card solution would be a better bet, providing it's powerful enough.