3DMark06, pointless....

[LukeBird]

Just thought I'd have a quick look into one of the many suggestions that people give when a 3DMark score is not where expected (that is to put a quad in the system! 😀).
With the amount of 'I'm not scoring what I expected' threads, I thought I'd see what a difference a quad made in a 3D test.
My new work machine being the ideal candidate (CPU stronger than my home machine, but video card not even close! 😀).. My work machine - Q9300, 3870, 4Gb RAM, XP Pro SP2 32-bit.
My home machine (as per sig, with XP Pro SP2 32-bit) will score 10.8k with one GTX and 13.1k with 2 GTX's.
My work system with no optimising whatsoever and a fair few processes as well (NAS back-up that sort of thing, unlike my 'clean' home machine) scored 11337!!
In my mind, makes 3DMark completely pointless as a 3D test!! I mean, why so much weighting on a CPU?!
Which brings me to my point, people, if you want to go ****-waving, get a quad for 3DMark (a generalisation, yes, but a lot of people don't need a quad!), but for gaming, it really still isn't needed!
Especially as my PC at home (which barely scores higher) absolutely rapes my work system in game I'm sure!

 

[monst0r]

I agree somewhat, but sometimes it can help point out a large bottleneck (not all of the e-peen threads, the one's that are very far off from expected results). A high clocked dual core processor will actually give you higher SM scores than a lower clocked quad. This shows that the quad's clock can't keep up with the card in some scenarios. The CPU score is what gave you that 11000.

I also agree that the tests are too CPU weighted, but what are you gonna do, it's just one benchmark.

 

[bobwya]

Just thought I'd have a quick look into one of the many suggestions that people give when a 3DMark score is not where expected (that is to put a quad in the system! 😀).
With the amount of 'I'm not scoring what I expected' threads, I thought I'd see what a difference a quad made in a 3D test.
My new work machine being the ideal candidate (CPU stronger than my home machine, but video card not even close! 😀).. My work machine - Q9300, 3870, 4Gb RAM, XP Pro SP2 32-bit.
My home machine (as per sig, with XP Pro SP2 32-bit) will score 10.8k with one GTX and 13.1k with 2 GTX's.
My work system with no optimising whatsoever and a fair few processes as well (NAS back-up that sort of thing, unlike my 'clean' home machine) scored 11337!!
In my mind, makes 3DMark completely pointless as a 3D test!! I mean, why so much weighting on a CPU?!
Which brings me to my point, people, if you want to go ****-waving, get a quad for 3DMark (a generalisation, yes, but a lot of people don't need a quad!), but for gaming, it really still isn't needed!
Especially as my PC at home (which barely scores higher) absolutely rapes my work system in game I'm sure!

Hmmm,

Just thought I would weigh in here. Don't forget you are comparing apples and oranges. You are facing off a Core 2 architecture quad with a dual-core Athlon 64. Academic but it had to be stated.The Core 2 architecture is damn efficient when compared to outdated Athlon 64 architecture and the Q9300 has a monster cache to boot. The 3D 2006 looks to the future and games that offload more physics to the CPU and run more complex AI perhaps (??). NB of course it was written pre-Nvidia's purchase of Ageia which tore that road-map up quite a bit!!

I tend to ignore folks 3D Mark scores unless the 3 full scores are quoted (overall, CPU, and GPU) because it is meaningless as a single score. No body is going to argue that SLI GTX's can be beaten by a single 3870 (or even in crossfire)!!

I do agree though there is too much weighting on the CPU score in the final result!!

Bob

 

[Grimmy]

Noooo... say it aint soooo.

3dmarks is everything to me. I run it everyday to make sure my system is in tip top shape. :lol:. o O (NOT!!)

Although... I've noticed 3dmarks 2006 actually ends faster using my Q6600, simply because of the CPU test. 😀

 

[night_wolf_in]

Noooo... say it aint soooo.

3dmarks is everything to me. I run it everyday to make sure my system is in tip top shape. :lol:. o O (NOT!!)

Although... I've noticed 3dmarks 2006 actually ends faster using my Q6600, simply because of the CPU test. 😀

yeah, i run 3dmark first thing in morning. n one time before sleep. LOL

 

[LukeBird]

Hmmm,

Just thought I would weigh in here. Don't forget you are comparing apples and oranges. You are facing off a Core 2 architecture quad with a dual-core Athlon 64. Academic but it had to be stated.The Core 2 architecture is damn efficient when compared to outdated Athlon 64 architecture and the Q9300 has a monster cache to boot. The 3D 2006 looks to the future and games that offload more physics to the CPU and run more complex AI perhaps (??). NB of course it was written pre-Nvidia's purchase of Ageia which tore that road-map up quite a bit!!

I tend to ignore folks 3D Mark scores unless the 3 full scores are quoted (overall, CPU, and GPU) because it is meaningless as a single score. No body is going to argue that SLI GTX's can be beaten by a single 3870 (or even in crossfire)!!

I do agree though there is too much weighting on the CPU score in the final result!!

Bob

Very true, what you said regarding the old 90nm K8 dual to a spanking new Core 2 45nm quad, but as a 3D test, you would expect the CPU weighting to only make a big difference in the CPU test (surprisingly enough! 😉).
I don't feel I am ever bottlenecked in game with my 6000+, and my video cards are obviously enough for pretty much anything at any resolution
Although that isn't going to stop me going to a 9750/9850.
Will make a nice difference in CS3 (a decent reason for getting a quad) and I have no aversion to shiny new things (not such a good reason for a quad!). Will probably go to Vista 64 at the same time
Now if I cared, that'd be a 3DMark-eater!

 

[sailer]

I've used 3Dmark for years, from back when it was FutureMark and I've found it to be very useful. I could care less about how my machine's score compares to anyone else's score. What I use it for is figuring out what helps my computer and what is a waste of time, that is, various overclocks, different video cards, CPUs, etc. I could care less about whether or not the CPU score is too heavily weighted. I look at all the individual scores. In this manner, I figure out what helps my computer the most and that is where I see the value of 3DMark.

As to some guy setting a record by overclocking his CPU to 5.7 ghz and using four video cards, and blah, blah, blah, well, good for him. I'm glad he's entertained and has the money to replaced the various parts that meet an early end of their lives. I have better things to do and ways to spend my money.

Oh yes, since I only use 3DMark when experimenting with overclocking or different graphics cards, I guess I probably don't keep my computer as finely tuned as Grimmy and Nightwolf's computers, but I can live with that.

 

[Grimmy]

Heh... was I being too harsh?

I mean 3dmarks will help indicate or show for compared stock system if there is a problem.

But then again, I usually see people show scores (prolly cause of that other post for 2001 3dmarks to see who could score the highest) to show their pride and joy... most of the time, rather then trouble shoot.

 

[sailer]

Heh... was I being too harsh?

I mean 3dmarks will help indicate or show for compared stock system if there is a problem.

But then again, I usually see people show scores (prolly cause of that other post for 2001 3dmarks to see who could score the highest) to show their pride and joy... most of the time, rather then trouble shoot.

Nah, I was joking about you. If i wanted to be harsh, well, I rarely do that to anyone.

Yeah, I remember some of the posts of 3D points on Futuremark01. Sillyness, etc.

 

[jimmysmitty]

I think that 3DMark06 is just for bragging right. I mean its nothing to be amazed at really.

I got a 40" Samsung 550 the other day and decided to try TF2 on it. I had everything set to the highest settings, I also found that there are higher settings through the console and activated them, and ran it at 1920x1080.

So I was plaing TF2 in HD and not sure if my single HD2900Pro 1GB would handle it very well. But guess what. It ran so freakin smooth even when there was a lot of people on and heavy fighting it was like I was running it at 1280x1024.

Of course my 3DMark06 score wouldn't show that I could do that as easily.

 

[LukeBird]

Of course my 3DMark06 score wouldn't show that I could do that as easily.

No mine neither...
I'm more than happy with the way it runs in game though!
Just need a new monitor to take advantage of all that graphical goodness! 😀

 

[jimmysmitty]

No mine neither...
I'm more than happy with the way it runs in game though!
Just need a new monitor to take advantage of all that graphical goodness! 😀

I was amazed that a single HD2900Pro 1GB could run TF2 at max settings ( including the hidden higher settings) and at 1920x1080 on a full server. I tested it again yesterday just to make sure it wasn't a freak accident and Played for about 2 hours with it still smooth.

Oh and that TV I have rocks.

 

[Andrius]

My 3D Mark scores vary by time of day I run it (~12.500 early daytime, ~12.100 nighttime). Also 2 days of idle XP desktop makes it drop around 1k marks.

And the best thing.
It overheats my otherwise cool running passive cooled 9600GT to 99°C.
3+ hour Supreme Commander games 1920x1200, high fidelity presets, low shadows, 2xAA takes it to 75°C. Yay. Another new 3D Mark only feature.
The previous one is a whining sound during mode switches that only happens during 3D Mark.

Don't you just love random software features ?