3870X2 in the house, any1 else has it ? youre experience here!

[suedenim]

Crysis only comparison of the X2 and GTX

http://www.3dgameman.com/content/view/11998/103/

 

Good news for those that sold a kidney for the GTX as its still got legs.

[Angry_Games]

this pretty much shows EXACTLY what I have been saying...3dMark = useless compared to real games...

 

 

vs

 

 

 

Now of course this is just one game, but this is a trend I've seen too much to ever rely on 3dMark for anything but stability testing.

 

Now we all know the ATI drivers are still crappy while the Nvidia 8800 drivers are finally mature...but it's the fact that the 3870x2 can destroy a single 8800GTX in something like 3dMark and yet still can't match it (keep in mind...DUAL gpu's vs a single older gpu...) in REAL games...

[Guest ElAguila]

Unfortunatly you are correct about these reviews. There are some games that showed my G92 GTS as good as the X2 but in reality my x2 is much better and it goes the other way as well. It is almost as if some of the review sites are becoming like the media. You can usually tell which media channel is liberal and which is conservative. It is getting so that you can tell which review sites are pro ATI and which are pro nVidia. The best thing I have found is to look at forums with informed and experienced users that can give you some real performance numbers.

[ShinyG]

I totally agree with Angry on the 3dmark issue.

Still, I don't consider Crysis to be a fair benchmark either. The game is obviously favorable towards nVidia, which actually contributed to the game's development.

Using the same reasoning, I wouldn't consider HL2 a good benchmark either, as ATI is still actively involved with Valve and Softimage (they do XSI, the 3d program used the most by Valve.

Best thing is to compare games that don't have crappy logo's like "nVidia, the way it's meant to be played" or "best experienced on ATI graphics".

I hate seeing corporate logos in games...

[Praz]

Most of the gaming results posted are no better then 3DMark. Not only are a good portion of the games optimized for either nVidia or ATI but the built-in benchmarks are also optimized and are not indicative of actual play. Of the few sites that base the benchmarks on actual game play a majority of those results can't be relied on either. For high-end cards they use wrong settings/resolutions and don't match the processor speed to the card.

[Angry_Games]

I don't know there is such a thing as "matching cpu speed to the card" as that doesn't really make much sense to me.

 

As for using in-game benchmarks, I'm completely opposed to that because of the fact that you are running a pre-scripted routine that really doesn't vary from run to run.

 

The ONLY way to get a fair and accurate measure of a game by benchmarking is using a utility like FRAPS and play a certain level or area within a real game and do this for every card. While most games are pretty scripted and don't change a lot, a lot of games DO change things up here and there and can give you slightly different variations.

 

One example would be STALKER...STALKER doesn't follow some scripted routine...it's a pretty open-ended game that varies each and every time you play it, even the same areas.

 

The proper way would be to play for say 10 minutes while doing a mission (Defend Bandit Camp @ Garbage for example) while running FRAPS as a benchmark, and play the EXACT same mission with each different video card or rig. Each time you defend this camp, it plays out differently, but the results are valid because you are actually playing a real game, and it is getting random results (ie one time you play 10 enemies might rush your position and you fire off many magazines from your AK-97 while the next time you might only get rushed by 5 enemies but you throw 3 frag grenades, etc).

 

Multiple runs over different sections within the game and then averaging them is the proper way to do it, but it takes time (but...hey...you get to play instead of watching a scripted benchmark!).

 

The other thing I see that turns me off to 'standard benchmarking' is that they rarely give you details on exactly how the game's video settings are set. They generally give you a vague "using Medium graphics settings" which is Ok I guess, but I don't know of a single gamer around here that just sets their vid options to one be-all/end-all setting and then plays. Most of us love to turn shadows off or down to low because it can boost framerates by 25% without really impacting the graphics quality experience. I like to turn textures up to max and model quality up to max to make everything much sharper.

 

But I have healthy respect for reviewers who DO take the time to play the actual segments of the game and then average out the scores. I don't buy into any "cpu speed doesn't match the card" because as far as I'm concerned, that's purely bogus.

 

Not everyone has a QX6800 + 8800GTX SLI + 4GB DDR3-1667

 

I'm interested in more realistic results (if this is what you are talking about Praz then my apologies for not understanding until now haha) like a real, normal user would come up with using realistic equipment like:

 

X2 3800+ @ stock (2Ghz) or Core2 E2180

2GB DDR2-800 / 2GB DDR-400

8800GT 512MB / ATI X3850 256MB

 

None of the reviews I've read lately are even close to what my own (high end!) machines are, and not even in the same universe as the majority of friends, customers, and users here have (again, most of us have 3800+ X2 or 4000+ single core or E6400 or E2180 or E4500 etc, 2GB of random RAM, and video cards that mostly range from X1900/7800GTX to 8800GTS/GT/ATI 38xx series).

[Praz]

QX6800 + 8800GTX SLI + 4GB DDR3-1667

 

X2 3800+ @ stock (2Ghz) or Core2 E2180

2GB DDR2-800 / 2GB DDR-400

8800GT 512MB / ATI X3850 256MB

Both of the above are what I would consider matched for a review. But all too often reviewers will pair up something like the X2 3800+ @ stock with 8800GTX SLI at a low resolution and wonder why there is no performance increase. Or like some recently did with the 3870 x2 run at 1280x1024 and show little improvement over a single card.

 

The 3870 x2 is a good example of testing methodology. All someone has to do is google reviews for it. Same games, different sites. Some will have the X2 leading in almost all games while other will have it trailing the 8800GTS or GTX. So as an across the board comparison most all of these reviews are useless for the average consumer.

[Angry_Games]

Both of the above are what I would consider matched for a review. But all too often reviewers will pair up something like the X2 3800+ @ stock with 8800GTX SLI at a low resolution and wonder why there is no performance increase. Or like some recently did with the 3870 x2 run at 1280x1024 and show little improvement over a single card.

 

The 3870 x2 is a good example of testing methodology. All someone has to do is google reviews for it. Same games, different sites. Some will have the X2 leading in almost all games while other will have it trailing the 8800GTS or GTX. So as an across the board comparison most all of these reviews are useless for the average consumer.

 

haha all right then i DID get what you meant

 

just making sure...I wholly agree...why bother with E8500 @ 4GHz + 2x 8800 Ultras in SLI but only 1280x1024 or lower resolution...?

[ebdoradz]

real man hav e alower end PC and a bigger screen,

 

why!?!?!? becuse we usualy watch more porn on the bigger screen then gaming

 

so real test/review would be computer like angry and Praz mention above

 

no need for ther 2x Quad Xeon with 16gb of ram and 2 8800Ultra in SLI if you game in 1024*768 on a 24inch LCD WTH !?&"?

 

almost all reviewers out there are doing marketking for either ati or nvidia, same thing goes to cpu test

 

im just reading hardwarecanucks review, and thats all, ...

 

my 2cent

[uneedav8]

I'm not too happy with mine right now, but troubleshooting issues.

[Guest ElAguila]

I have had my card for a couple of weeks now. I have been satisfied with it so far. My oc's while decent aren't anything to yell about. I am running both cores at 860 and both vrams at 975. If I modded the bios I could probably get a little more out of the cores. I have seen the prices go down on these and now you can pick one up for around $420. For that price it is not a bad card. I have been reading a lot of the reviews and the more I read the post by Angry and others I really don't know if there are many reviews out there for this or other cards that I can really go by anymore. I have this card in my system and I can run any of my games(it is understood that crysis is the exception) at my native 1920x1200 resolution with eyecandy turned up and I rarely go below 60fps. But others with a similar system may not get the same results. If I was running XP 32 bit I might get better or worse results. It just seems to me that there are so many variables out now that buying cards based on reviews may not be the best thing. Some reviews can give us a general idea but when you are paying $300+ plus for video cards you want more than that. Unfortunately I think it comes down to trial and error by the consumer. Buy your card from a place that will let you return it for a refund. Wait to buy it until the drivers have matured. Then if it doesn't perform well on your setup with your games or other needs, return it and get a different one.