Who Cares About 3DMARK?

[Machine350]

Any ideas?

 

[Cawchy87]

Because If you are getting an abnormal score in 3dmark, chances are there is something wrong with your system or something you could tweak a little more.

 

[Corporate Thug]

Originally posted by: Cawchy87
Because If you are getting an abnormal score in 3dmark, chances are there is something wrong with your system or something you could tweak a little more.

well put

 

[Rage187]

I love that game!

 

[Bar81]

Except that's not what he's asking. Besides comparing scores, 3dmark is useless. The argument put forth above would have a point except it would be much more valid to check your scores in the games you plan to play to make sure you're set up correctly.

imo the whole 3DMark thing is a small penis issue and a way for people who spent way too much to feel good about themselves. Its irrelevant.

 

[SonicIce]

its a quick way of determining performace, rather than running a whole bunch of games. actual programs that you are going to use make the best benchmarking tool.

 

[fstime]

psssssssssssssss haters.

 

[Torghn]

I like to use it to compare speed changes due to overclocking or new drivers.

 

[paladiin]

Originally posted by: Torghn
I like to use it to compare speed changes due to overclocking or new drivers.

Same. Though I am entertained by trying to score well on the ORB vs. similar spec'd systems. And I have to admit, seeing my 3DMark scores jump drastically helped justify upgrading my 5900u.

 

[zerocool84]

Originally posted by: Bar81
Except that's not what he's asking. Besides comparing scores, 3dmark is useless. The argument put forth above would have a point except it would be much more valid to check your scores in the games you plan to play to make sure you're set up correctly.

imo the whole 3DMark thing is a small penis issue and a way for people who spent way too much to feel good about themselves. Its irrelevant.

Duh its bragging rights. like having a faster car, you cant drive it that fast anywhere but you can say you have it

 

[DAPUNISHER]

Originally posted by: Cawchy87
Because If you are getting an abnormal score in 3dmark, chances are there is something wrong with your system or something you could tweak a little more.

Sometimes, but it can also mean you just still have AA/AF on that you use in the games you actually play Part of the tweaking process for higher 3DM scores entails looks like shat IQ settings and res. Thus tweaking for a higher 3DM score is often completely irrelevant to why you purchased that $300+ card, to play@high res with AA/AF. Most who proudly display the score in their sigs are participating in a HUGE pissing match, plain and simple.

You can evaluate every subsystem, the stability, and the performance without ever installing 3DMock, and better yet produce real world relevant results. Or better yet, just read the reviews of others who have tested everything already, Zebo's ram testing is a good example, and there are many, many more of course. 3DM scores also vary, sometimes radically, from 1 driver to the next, sometimes because of specific opts for it that do not bring better performance to most your games, or borks the media playback, ect. and yet some will switch to them just to get that extra 500 3DM points :thumbsdown:

Look, used properly 3DM is a good tool and can provide you with a good baseline for what each price range card performance will be with newer games graphics demands. But, if I understand the motivation for the OP's question, the answer to that is: Many who care about the score are in a pissing match/e-penis contest.

 

[DJQuanta]

It creates a common ground, something that allows you to compare the performance of your rig to detect if something is wrong (by comparing with someone who has a similar rig).

It's more complicated to do that with real games, because everyone runs with different settings and some of those affect performance.

Imagine a post like this:

"Guys, I just put together my new rig, I ran a HL2 benchmark at 1024x768, Medium textures, High model, reflect world, 4x AA, no AF, medium shadows, high shader, v-sync disabled..... and get an average of 75 FPS.........you think is ok ??", and HL2 is a game that doesn't have that many different settings. Need for Speed U2 has like 15 graphics settings.

It's easier running 3Dmark2003 at 10x7 and getting 5800

I agree that 3Dmark won't give you actual gaming performance, but it will let you to a quick check of the performance of your rig.

At least for me...

 

[hans030390]

because 3dmark 05 looks prettier then any other game around.

 

[DAPUNISHER]

Originally posted by: hans030390
because 3dmark 05 looks prettier then any other game around.

Except its not a game, and doesn't produce the AI or physics engine load a real game does. And most run it on low res no AA/AF so it looks like shat and produces that score they so proudly put in their sig :roll:

 

[hans030390]

i still think it looks pretty

i'm not even talking about the score...i find it useful to compare cards, and compare an OC on your card to stock speeds...and check for artifacts

as for comparing with others....who cares? but 3dmark does give what some games cant, or it gives it easier

 

[nebuchanezzar]

I use them alot. 01 I loop nature for 25 runs to test my GPU when I oc it and I keep rerunning 03 to generate some serious heat to look for artifacts. As far as scorekeeping..my gaming rig isn't online so my scores don't even upload to futuremark. I could care less but to some yes it's a p trip thing.

 

[Mykl]

Who cares about people who care about 3DMark?

 

[homercles337]

Maybe someone here can answer this then. When my command rate is set to 2T i get ~5100, but my rig does bad with any memory benches. When i change the command rate to 1T, my memory benches improves significantly BUT my 3dmark drops ~1000 points! Explain that! Please...

 

[Bar81]

It's quite simple really, memory benchmarks are useless. In memory benchmarks 2T drops your score significantly but in real world usage it's practically negligible. Also, 3DMark is useless. Don't use synthetic benchmarks because as you are seeing they are worthless.

Compare 1T and 2T in the games/application you use and see that it makes no discernable difference.

 

[BFG10K]

Because If you are getting an abnormal score in 3dmark, chances are there is something wrong with your system or something you could tweak a little more.

The problem is that many people don't really do this. Instead they overclock cards to get 200 extra points or condemn drivers because they lost 150 points. That's just really silly.

 

[Insidious]

Originally posted by: homercles337
Maybe someone here can answer this then. When my command rate is set to 2T i get ~5100, but my rig does bad with any memory benches. When i change the command rate to 1T, my memory benches improves significantly BUT my 3dmark drops ~1000 points! Explain that! Please...

different error response methodologies. (counting bits vs. using them and having to re-read if incorrect)

Looking for the point where 3DMark score peaks is one of the good indicators of an OC heading south.... find that point, back off a bit and you have a very fast, stable OC in one regard.

 

[Mykl]

Originally posted by: Bar81
It's quite simple really, memory benchmarks are useless. In memory benchmarks 2T drops your score significantly but in real world usage it's practically negligible. Also, 3DMark is useless. Don't use synthetic benchmarks because as you are seeing they are worthless.

Compare 1T and 2T in the games/application you use and see that it makes no discernable difference.

Then why do people make such a huge deal about the 1T versus 2T thing? From the way some folks around here talk there's this HUGE performance difference that may as well be the end of the world.

Now, I understand this site is filled with hobbyists whose goal is to squeeze every last ounce out of their machine. But sometimes it can be difficult to differentiate between these people and the advice that the real world folks need to be listening to.


So basically, what's the real world difference between an inexpensive stick of memory and a top end expensive one?

 

[Bar81]

Originally posted by: Mykl

Originally posted by: Bar81
It's quite simple really, memory benchmarks are useless. In memory benchmarks 2T drops your score significantly but in real world usage it's practically negligible. Also, 3DMark is useless. Don't use synthetic benchmarks because as you are seeing they are worthless.

Compare 1T and 2T in the games/application you use and see that it makes no discernable difference.

Then why do people make such a huge deal about the 1T versus 2T thing? From the way some folks around here talk there's this HUGE performance difference that may as well be the end of the world.

Now, I understand this site is filled with hobbyists whose goal is to squeeze every last ounce out of their machine. But sometimes it can be difficult to differentiate between these people and the advice that the real world folks need to be listening to.


So basically, what's the real world difference between an inexpensive stick of memory and a top end expensive one?

It's just ignorance. For a long time on these boards there were dozens of ignorant people touting RAID0 as a way to improve performance even though it had long been disproven on storagereview.com It wasn't until Anand posted his results that the masses got a clue.

The whole 1T versus 2T has been blown up because of synthetic benchmarks. The guys who are posting there memory bandwidth to show how great their system is are the most clueless. For example take single channel versus dual channel memory on the A64. Your memory score when go dual channel goes through the roof but in real world benchmarking that huge memory bandwidth advantage gives you maybe 3% across the board, something you would *never* notice in game or application usage but you'll see people telling you how much better performance is on 939 versus 754.

To put your mind at ease do your own benchmarks on the games/apps *you* use to see what works best for you.

Oh, another example is DDR333 versus DDR400 on the Athlon64. People were going on and on about how usig DDR333 would cripple the perfomance of the A64. Well, I did my own tests and using the slower speed memory resulted in a 0-3% drop in framerates (notice the zero) so I realized that all the people making these claims were ignorant benchmark whores who never once actually tested real world performance.

 

[KoolDrew]

Originally posted by: Cawchy87
Because If you are getting an abnormal score in 3dmark, chances are there is something wrong with your system or something you could tweak a little more.

 

[Sparky19692]

Originally posted by: KoolDrew

Originally posted by: Cawchy87
Because If you are getting an abnormal score in 3dmark, chances are there is something wrong with your system or something you could tweak a little more.

For this reason I will always use these types of tools. When I first put this rig togather I was very happy with its speed until I started comparring 3dmark and Aqua. I had a Few Major changes to make, one being a poor PSU. I also leave my system set for normal usage and have Aqua set for my normal gaming 12x16 aa/af on. this make my life very easy when I try new bois or a new O/C setup. what easier way to do a fast test lap.