Why do people use 3dmark06 over Vantage?

Why are there still so many benchmarking 06? It is like 5 years old now, and was meant for Geforce 7 series and lower.

Why not use vantage which gives you a much more accurate score?

 

then why do you have 3dmark06 in your sig?

 

I have Vantage as well, and at least i posted my res for 3dmark06. If u want u can run 06 in 320x200 and post ur score and win the internet, no one know what res u tested it in unless u post it.

Vantage has a standard res that everyone needs to use, so you dont have that problem.

 

off topic: what is the "standard' resolution that 3dmark is set at during the test?

for example, on notebookcheck, what resolution do they use? in fact.. what computer setup do they use? doesnt cpu affect these scores somewhat? Notebookcheck: Mobile Graphics Cards - Benchmark List

 

3dmark06 is dx9 centric. A lot of ports of game consoles are still written in dx9. 3dmarkvantage is dx10. It's good to know the results of both benchmarks to gauge overall performance.

 

What resolution should someone with a laptop that has a 1600x900 max res use on 3d mark vantage to get a semi accurate comparison to the 1280x1024 res of a desktop?

 

If u would bother to explore the site you can find all the info you are seeking. The standard is 1280x1024

 

Standard is 1280x1024 for both. If your laptop LCD can't handle that res, then hook up to an external monitor that supports that resolution if you want true results to compare with others. Vantage won't give you a score unless you run at that resolution. The problem with Vantage is that while you can get one free run, it's annoying to keep re-registering and it costs you money. Granted not much, but still to pay for a benchmark that has advertising in it too is a bit ridiculous.

 

Because a lot of lower end GPUs can't even run vantage, so 06 is a good gauge to see how much your GPU is destroying the lower end GPUs.

 

because it can be cheated and make a crap gpu look faster than the other one by using optimization tweaks specific for 3dmark 06

 

+1 to that .... why would I test my performance in DX10 while 80-90% of the games are still DX9

 

end up the one who really lose out is yourself, be true to yourself and be proud

 

Is it true that 3DMark06 is CPU centric? From what I have read, Vantage scores are much more proportionate to GPU's raw power.

 

The CPU can skew the results a little bit, but it's not as bad as people make it out to be.

 

This. PC gaming market is still filled with console ports, which don't even support DX10, let alone DX11. As long as the current generation console lives before the next generation console appears and takes over, 3DMARK06 will be used over and over. There is little point running a DX10 or DX11 benchmark when most of the games you play are based on DX9.

 

Because '06 has been in use longer and using it I get a longer set (historical) of comparisons to how current GPU's are performing compared to old ones. DX9 is still alive and well even after all this time and sadly, for the consumer, the laptop GPU industry hasn't advanced very much. I like both scores really, but '06 lets me go to any old list that could be who knows how old and get a comparison with a machine that might not be in most current tester's test benches.

 

3dmark06 is good for notebooks when you don't have a monitor with a res of 1280x1024 - at least that's why I use it (w/ the resolution used)

 

I can't use vantage on my Asus so I use 3dmark06, as simple as that....

 

O.K. Here is the lowdown.

3DMark06 is old now. Very old it has become way too easy to render for any modern high and even mid ranged gpu's as a result the system becomes very easily cpu bottle-necked, and the benchmark becomes more or less a cpu test.

Also due to that nature and how rendering with multi core works out a fast dual core cpu does much better than a stronger slower clocked quad core cpu.

The last problem with 3dmark06 is that there is no standardized resolution nor do most people tell you there resolution tested at, so often people compare there lower resolution tests to somebody elses higher resolution tests as an equal comparison and its not.

3dmark06 by all means is now just a cpu test or a "just because" benchmark it has absolutely no real bearing on performance of a system with modern systems or modern games.

3dmark09 Vantage has changed a few things.

First it is much more demanding again so it puts a adequate load on the gpu so that you do not find a cpu bottleneck giving falsified results. Also it does a great job of giving accurate cpu & gpu scores individually so you can see how strong each part of your system is.

Third it has introduced resolution standardization. Each test in vantage (High, Performance, Extreme) has a fixed resolution for the test to run, you are not able to select your own custom resolution for the test so if one person is comparing there Performance benchmark results to another persons you know they ran at the same resolution and there is new skewing of the results.

Vantage also properly takes advantage of a quad core in the cpu portions of its testing.

So in summary if you want to benchmark your system and see how well it performs and compare it to other systems and get a real life idea of the performance difference and gaming ability of different systems use Vantage.

06 is done, it has no benchmarking use anymore its just for show/fun.

 

This goes for me as well, especially considering I'm on old hardware.

 

thank u for this, its what i wanted to say all a long

 

The problem though is I think the CPU *SHOULD* be part of the benchmarking equation. It's your whole system that needs to be benchmarked not just the graphics card. Plus by saying Vantage takes advantage of quad core shows that it is also CPU-centric in some respects.

I think 3DMark06 is still valid in the world of notebooks because the technology lags behind a considerable amount from the desktop counterparts. Maybe not your high end notebooks like the G73 but surely for most other mid level notebooks with HD 5650 or nVidia 240M cards or slower which comprise a large portion of the dedicated GPU notebook market.

 

Dude ..... don't forget the guys with even weaker GPUs like the ATI HD4500 series or the low end NV 8xxxs, 9xxxs

 

No, not forgotten.. that's why I put in the caveat "or slower", as in 5650 or 240m are max for mid-grade. Tons of machines with GMA 4500 too, granted if you want to game, just not a wise choice.

 

I think one of the main reasons is that with the free version of 3Dmark06 you can run it as many times as you like. Comparing various drivers against each other as well as different gpu overclocks. I was under the impression that with the free download of vantage you can only run it once, thereby eliminating any tweaking and comparisons of different runs.

 

You are correct. You can keep requesting new keys I believe, and think you have to reinstall Vantage every time. Although you can buy the basic for $6.95 and Advanced for $19.95. That's my biggest peeve is that they charge for a benchmark. They should at least make the basic test free, with online only score viewing like 3DMark06, and pay for offline scoring, viewing, and configuration. Hopefully the next version will be like that.

 

Completely agree with this. Although most PC games today are limited by GPU, CPU still plays a big part in the whole play, many games still require a decent dual core CPU. High end CPUs can effectively up the end score by a few hundred, probably even up to a thousand if you got a really powerful CPU like i7-940M, but nowhere to the extent it becomes a CPU benchmark.

3DMark06 was never meant to be a GPU benchmark, it's a GAMER'S benchmark weighing the overall gaming capabilities of your system.

 

Uh. I wasn't involved in benchmarking when the transition from 3dmark05 to 3dmark06 happened to show an example, but I'm pretty sure the reason is because Vantage isn't free, since thats the reason I haven't used it yet. In fact, I only use 3dmark06 for testing system stability and to make sure the score is good compared to what I would expect from the system specs. If I want to test system performance, I simply play some games and see for myself.

 

The point of a benchmark is for analysis. A truly analytical person wouldn't be looking to get the "highest number" to show off. If they are, then the benchmark isn't for them.

 

It does, Vantage gives you individual scores for both cpu & gpu and then a total score.

So you get accurate results with vantage, have you a strong/weak cpu/gpu it does not matter, its accurate and not skewed due to any bottlenecks.

with 06 its not the same scenario due to how easy it is to render basically there is no gpu score in the equation it becomes a mhz race on the cpu.

if you do not understand go look up a benchmark called aquamark 3, back in the day when 06 was new, aquamark 3 was the one getting phased out and it was the same deal people used to have a huge goal to break 100k on aquamark 3, they would overclock there gpu & cpu like mad to reach it, by the time 06 was standard you could get 100k with integrated graphics just as long as you had a core 2 quad and overclocked it past 4ghz.

But when aquamark 3 was standard even if you had a 5ghz i7 by traveling back in time you were not going to break 100k without a heavily overclocked gpu because the gpu was the limitation on the benchmark.

It was a test of both cpu & gpu but because it became outdated the gpu portion became too easy and the cpu became a huge bottleneck and thus the benchmark became useless, it stuck around for a long time, people posted results even though they did not mean anything, and now time has repeated itself and people are doing the same with 3dmark06.

Anything past a integrated graphics these days is too strong for 06, and even if you had integrated there is no way to use 06 to compare it to other modern systems, the benchmark is useless.


If anybody has trouble understanding the relationship between cpu & gpu and how it forms a system bottleneck on these benchmarks I offer you this analogy.

Games are rendered in FPS (frames per second) pretend that each frame is a page in a book and you are flipping through the pages.

The difference is that the gpu has a job of drawing each picture on each page in real time. You cant view the pictures on each page if you do not turn the page. This is the cpu's job it turns the pages.

The cpu & gpu work together by having the gpu render the page (image) and the cpu has to process that data thus turning the page.

Its basically impossible to not have a bottleneck and have a perfectly balanced system where the cpu is using 100% of its power to turn the pages and the gpu using 100% of its power to render the pages. There is always one faster than the other, and thus the slower of the two creates what we call a "bottleneck" aka a limitation on your system.

Having a bottleneck is not necessarily bad, if you were maxing out the gpu at 100% but getting 60+ frames per second in the game your good to go, even if your cpu is only at 50% use.

The higher the graphics settings are the harder the images is to draw, this puts more load on the gpu. If you were in the above situation where the gpu was 100% and your game was at low settings and you turned them up to high your frames per second would go down because it would be harder for the gpu to draw the pages and it was already at its limit.

However if say the game was on ultra high settings and at 60fps and you turned it down to medium suddenly your frames per second would go up, lets say 100fps now.

As a result with more pages being drawn the cpu load would go up from 50% to lets say 80% so your getting more fps as you lessened the load on the gpu so it can draw faster, and now the cpu is working harder to turn the extra pages.

Now say you turn the game settings down to low, you may find that the cpu is now at 100% load and the gpu only at 80% load.

This reversed the bottleneck you had before where the gpu was the limit, now the gpu is drawing those pages so fast that the cpu cant keep up and reached 100% load. Because the cpu cant keep up the gpu cant draw any more pages and thus is now not at 100% load.


If you understand that, now you understand why 3dmark06 is a problem.

The graphical portion is so easy for any modern gpu to draw that it easily has the ability to do say 300fps if it had as much cpu power as it needed to turn those pages that fast, but no modern cpu has that power behind it so you end up with a very severe cpu bottleneck and a score that is only reflecting the power of your cpu.

3dmark06 gives both a cpu & gpu score but the cpu score while accurate (for dual core/single core only) the gpu score is wrong because the gpu was not able to give 100% due to the lack of cpu power to back it.


Now 3dmark09 Vantage has separated the cpu & gpu tests to give you accurate results no matter the power of your cpu & gpu because it puts each component at 100% load individually in each test. In the cpu test it will max out the cpu @ 100% and also can use multi core cpu's like quad cores to reflect the power of modern systems.

In the gpu portion of the tests, it will max out the gpu easily even without a lot of power from the cpu because the pages are incredibly difficult to draw so the cpu must not work hard.

It then gives you the individual scores and a total score.



_____________________________________________________________


If my words alone are not enough please test for yourself as I have done many times in the past.

Take 3dmark06 and run it at stock settings on a desktop even with a old card like a 7000 series Geforce because thats the card I have done these tests with.

Write down your score.

Now Overclock your CPU and test again, if you see the cpu score rise but not the gpu score you know the gpu was 100% loaded. But if you see both scores go up you will know the cpu was the limit for the system even on the gpu testing portions of the benchmark.

To make it even more clear, test again at stock speeds on your cpu and UNDERCLOCK your gpu some, the score will stay the same for the test as it did at stock settings even for the graphics portion of the testing because the cpu was a bottleneck to the system.

The cpu is a bottleneck because what I explained above, the "pages" of the test are too easy to draw for modern gpus and the cpu has a very hard time keeping up it cant turn the pages fast enough.

many people experience this phenomena when they go back to play some of there old favorite games, like say half life 2. They ask why is my new computer at 100% load on the cpu when I am playing an old game? They think its a bug/error because an old game should be easier to play/render than it was on there old computer.

Well infact it is, they just do not understand the concept that now the game is at 200fps and the cpu is maxed out trying to keep up with the gpu, when back on there P4 system it was at 50fps and the gpu was doing all the work.

This is when things like v-sync come in handy

 

did the wall of text kill everybody?

 

As much as i liked your text, it was totally overkill

You just said everything pluss some more

 

im thorough when I need to be

 

Sometimes overkill is needed to prove people wrong that are only interested in showing off big numbers and their e-peen.

 

I think that's basically it. If you're not a professional reviewer, why would you pay for a benchmark? I suppose you can keep re-registering, but that's probably even worse. I don't think people care about the finer details of what exactly it's measuring, they just want something that is free, easy to use and vaguely accurate.

 

Hehe, as far as I can tell, if you got a notebook 4 years old or older, 2006 and later, then maybe 3DMark06 is good. Anything around 2007 and newer probably should use Vantage.

Either way for the current high end, gaming notebooks, Vantage definitely.

Anyways, hopefully we can expect a DX11 benchmark from 3DMark. With AMD's latest HD6xxx, I think DX11 gaming is finally here. AMD's latest HD6xxx and whenever Nvidia decides FERMI is a lost cause and introduces their next DX11, we should have hardware capable of DX11 gaming to it's fullest.

That said, I think I'll be saving to build a desktop rig for future DX11, fully AMD with CPU also, I don't mind the slower Phenom

Laptop gaming is fun, but I think DX11 is still a few years away for laptops. But DX11 gaming should be full on in 2011 I believe.

 

it killed me.. but the reason ppl use 3D Mark Vantage is that its much more accurate especially in CPU department.. also overall its much more accurate.. at stock G73 has 4 times as many points in 3D Mark 06 compared to my Dv5t however in vantage its 5 times as many... i guess its too do more with accuracy...

 

Curious about the CPU influence. Not that I don't agree or am debating, but look at these results from these netbooks/subnotebooks:

ASUS EeePC 1215N review -- Engadget



The specs for each are as follows:

Eee 1215N = Atom N550 / ION 2 (9400m?)
Eee 1201N = Atom 330 (desktop) / ION
Dell Inspiron M101Z = Athlon II Neo K325 (2 core - 1.3GHz) / ATI Radeon 4225
Acer Aspire One 721 = Athlon II Neo K125 (1 core - 1.7GHz) / ATI Radon 4225
Lenovo IdeaPad U160 = Core i7-640um / GeForce 305m
Asus UL50Vf = Core 2 Duo SU7300 / GeForce 210m
Alienware M11x = Core 2 Duo SU7300 / GT 335m

For example, the U160 CPU is more powerful than the M11x CPU, but the GPU is weaker. It scores about 20% of the M11x with 3DMark06 but it outscores it with Vantage. In the case of the Athlon II Neo the single core and dual core scores are about even with 3DMark06, but the dual core is about 50% faster in Vantage.

 

Those scores are total scores you need the individual scores to break down the performance properly when there is a bottleneck.

Looks like something is not right there since the UL50 and M11X share the same CPU but yet got totally different scores on the CPU testing. I assume they put the CPU/GPU score for 3dmark06 on there since its not a single score listed.

 

The UL50 has 210m GPU and M11x has 335m, both have the same CPU. Results in 3DMark06 make sense but Vantage do not.

Also the 3DMark06 with two scores shows integrated/discrete scores.

 

I always find it comical, when people who spend thousands on notebooks have an issue with spending $6.95 on Vantage.

 

It's the principle. Spend money for a benchmark? With advertising blatantly thrown in? No thanks. It's $20 to avoid all the web based crap and to configure it. I ended up winning an advanced copy, so I'm good, but I would have never paid for it. I don't even think it's that great of a benchmark.

Imagine if every benchmark charged $6.95. That'd be an issue. I hate to see Futuremark set a precedent.