Sign in to follow this  
StoneC0ld

why 3Dmark03 is no good as a benchmark......

Recommended Posts

....As related to helping us benchmark our machines to improve FS performance.FS2k2 is highly CPU dependant (AA and AF aside) and as at least with 3Dmark01 the CPU made about as much differance as the video card and you could use it to "fine tune" etc.Well 3Dmark03 has moved heavily towards the video card, take a gander:http://www.gamepc.com/labs/view_content.as...=axp3000&page=7As has been stated if you want a true benchmark of your machine as to how it wil run your favorite game....benchmark it using you favorite game. :-roll :-lol

Share this post


Link to post
Share on other sites
Help AVSIM continue to serve you!
Please donate today!

Well, my Radeon 7200 will bearly run 3dmark 2003 anyway LOL. It can run only 1 game test out of 4 and like 2 normal tests. All the Radeon 7xxx's are like this too, they are all DX7 based cards, meaning only 1 game test (the airplane one), will run on it. Oh well, my fps in FS2002 is pretty good anyway, don't need a 3dmark to tell me that LOL. As long as its 20fps and up, im happy (and its usually there anyway) LOL.

Share this post


Link to post
Share on other sites

If you want to read a rant on why 3Dmark03 is not a valid benchmark for determining how you system will handle just about any game, head over to HardOCP. Kyle posted a little blurb about it the other day.

Share this post


Link to post
Share on other sites

Hi,I think the title of this thread is a bit misleading... Indeed, 3DMark03 may not be a good GAMING benchark, since it doesn't yet (if ever - too early to tell) reflect actual game designs.But that doesn't make 3DMark03 itself is a poor benchmarking program. It can be extremely useful in many situations to compare systems with each other in very specific ways. General games are not good at doing specific bench comparisions - they are general game system benchmarks by design.3DMark03 most certainly has its place in benchmarking: just not as a "Gamers Benchark" today, as they market it.Take care,Elrond

Share this post


Link to post
Share on other sites

Hi,>Hi, >>I think the title of this thread is a bit misleading... >Indeed, 3DMark03 may not be a good GAMING benchark, >since it doesn't yet (if ever - too early to tell) reflect >actual game designs. "As related to helping us benchmark our machines to improve FS performance.">It can be extremely useful in many situations to >compare systems with each other in very specific ways. >General games are not good at doing specific bench >comparisions - they are general game system >benchmarks by design. >>3DMark03 most certainly has its place in benchmarking: just >not as a "Gamers Benchark" today, as they market it. Sorry, cant agree at all with that as the link showed the difference between a P4-3.08 at 4800 score and the P4-2.2 was less than 150 points difference! 3Dmark2003 is as I said video card mark plain and simple.Go back and check the link comparing ten different CPUs running on the two different motherboards the difference between all of them is just 300 points! The only thing they all had in common was the video card.So lets see: Going from an xp2200 to xp3000 or p4-2.2 to p4-3.08 comparing the results in 3Dmark2003 that was an improvement of less than 6%! is that even close to what we will see in any real game or aplication that you can think of with that upgrade? I think the post was not misleading at all, 3dmark2003 is a waist of a benchmark for anyone intrested in how sell system X compares to system X as it might relate to running FS or anyother game for that matter.Maybe I'm wrong but I dont think less than 7% difference between all those CPUs is anything close to reality.JMHO. :)

Share this post


Link to post
Share on other sites

Isn't 3DMark advertised as a "video card" benchmark, and not necessarily a video game benchmark?It does exactly what it advertises to do. Does graphically intensive benchmark tests on the video card. It does not advertise (based on what I've seen) that it accurately simulates a true video game environment.So, here's an idea for madonion (or whoever is making 3dmark now):Include an option in the benchmark to throw a high load of nonsense calculations to simulate a real game environment while performing the video benchmark. Must be able to enable or disable it, though. Either include this as an option in the next release, or create 2 seperate product lines, the "video card" benchmark without the cpu load and the "video game" benchmark with the cpu load.Although, including a high cpu load as an option may become an option available only to registered users.........An alternate idea, if this works.... Dunno, since I'm just making this up as I type.....Run FS2002. Go into the options and disable the "Pause during task switch". Go on a long flight and cruise using autopilot. Minimize FS2002. If this works, FS2002 will continue flying and hogging mega cpu cycles. Run your 3DMark tests. Unless the FS2002 locks the video card from running other directx applications, this should work and reflect the video performance under a high cpu load. :)

Share this post


Link to post
Share on other sites

Hi again,"As related to helping us benchmark our machines to improve FS performance."That is still incorrect. See below."Sorry, cant agree at all with that as the link showed the difference between a P4-3.08 at 4800 score and the P4-2.2 was less than 150 points difference!"Well of course it did. The final "3DMark" score is based upon the results of the one, three or four games only (depending on graphic card used) and none of the other tests 3DMark03 performs. You do realize that isn't all 3DMark03 tests don't you? I can't see how you might miss that it was made with a wide variety of tests, including:- video performance- CPU performance- fill-rate- shader performance- audio sub-system performance- IQ (image quality) tests"3Dmark2003 is as I said video card mark plain and simple."That too is incorrect. See above."3dmark2003 is a waist of a benchmark for anyone intrested in how sell system X compares to system X as it might relate to running FS or anyother game for that matter."Well, if you only focus on one set of numbers and tests that the program gives you - sure. But, why would you do that? It doesn't make sense.3DMark03 is indeed an absolutely viable synthetic benchmark - just as Sandra, SYSMark, SPEC ViewPerf and many other synthetic benchmarks are. Does that mean it should be used alone without "Real Games" as a reliable source of information (or even worse, ignore whole swaths of its own information)? No, of course not. For FS specific system comparisons: CPU, Audio and IQ tests might be the most relevant.Take care,Elrond

Share this post


Link to post
Share on other sites

One could argue (and many are) that these synthetic benchmarks are not so good at deteremining system performance as they are at determining which graphics card manufacturer can release drivers that are the most optimized for these particular programs. This optimization may include the elimination of entire render cycles for certain tests. Thus, artificially boosting their numbers.I'll agree if the playing field is level, these programs can give insight into relative performance capabilities. Unfortunately, it seems that unsavory practices by graphics card manufacturers are clouding the results.

Share this post


Link to post
Share on other sites

Elrond,Thanks for the corections I think. :) Are you saying according to the link that I provided that the publisher of those benchmarks ignored these other test?Doesn't make sense to me.???BTW, all test on all systems were done with ATI9700-pro using latest catylist drivers and DX9.

Share this post


Link to post
Share on other sites

Sorry about the somewhat terse response I gave you above... I am in the middle of a difficult project that was frustrating the heck out of me last night. Unfortunately, I let that color my response to some degree. Please excuse me.Yes, GamePC completely ignored the CPU Score (a combined score made up of two separate tests, actually) and only focused on the video heavy 3DMark Score that '03 generates. Why they would choose to publish the video mark scores and not the CPU mark scores when what they were testing was CPU's is beyond me. It wasn't very professional of them in my opinion.To be precise, there are 16 different results that 3DMark03 provides (plus a staggering variety of exact IQ comparisons). Only four of those tests make up the actual "3DMark Score". 10 of the others contribute nothing to the "3DMark Score"... They are specifically designed to test other aspects of a system beyond the video card. Specifically:CPU Score (This is what GamePC should have published considering what they were testing) - CPU Test 1 - CPU Test 2Feature Tests - Fill Rate (Single-Texturing) - Fill Rate (Multi-Texturing) - Vertex Shader - Pixel Shader 2.0 (DirectX 9+ cards only) - Ragtroll (tests heavy video + heavy CPU combined performance)Sound Tests - No Sounds - 24 Sounds - 60 Sounds (when supported by sound cards eg: Audigy2)I agree with you... What GamePC published doesn't make sense to me either.Let me be clear here: I'm certainly not advocating one way or the other the video portions of 3DMark03 due to some of the questions that have been raised (some seem to have merit, some seem like PR spin control from NVidia and nothing more). We'll have to wait and see what types of DX9 games get released and what types of internal shader paths those games use before we'll know if the video portions of 3DMark03 were accurate (lets remember they are trying to look into the future - never easy to do). But, and this is what I've been talking about in this thread, 3DMark03 provides more than the questioned video tests. Taken as a whole, that makes 3DMark03 still a very valid synthetic benchmark tool.Take care,Elrond

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this