Puzzling PassMark scores 7400 vs X1400 vs 7600

Hi all,

I just ran PassMark ( http://www.passmark.com/download/pt_download.htm) against 3 laptops and I got some puzzling results for my 3D test scores. I have the following:

Sony Vaio VGN-AR130G with NVidia GeForce 7400 (256MB)
HP Pavillion dv8330us with NVidia GeForce 7600 (256MB)
Fujitsu Lifebook N6410 with ATI Radeon X1400 (256MB)

I expected the 7600 to be top, and the X1400 and 7400 to be pretty close... Instead I got a score of around 310 for the Radeon, a score of 245 for the 7600 and a score of only 87 for the 7400!! The Sticky thread on this forum suggests that the 7600 should win out over the X1400 pretty easily... I know the systems are different, but all the software is identical (Windows MCE on all systems, out of the box, and same test software), and the supporting hardware (CPU, bus, memory, etc) can't make *that* much of a difference (they are all similarly spec'd Core2Duo systems with similar FSB, resolutions, etc)...

Now, either something is very wrong (all three machines were out-of-the-box configurations), PassMark (or Windows) works better with ATI GPUs, or maybe a driver/BIOS issue?

Any comments?

Thanks!

S.

 

PassMark?! What is this?! Check the stickies or download 3dMarks and test it again, there's no way X1400 outperforms GF7600.

 

who cares for those tests.
even 3dMark is not so reliable, there're so many reasons for that...

I'm sure that you can do something better on these GPUs instead of doing tests..

 

Imperical evidence is the key to knowing what you have. Reliable tests are important to define the differences between chipsets. Passmark does a series of standard tests which use various hardware acceleration functions and allows you to do repeatable tests on various platforms. It has received a lot of acclaim for being simple and repeatable, unlike some other benchmarking programs which are bloated and have pretty useless figures.

Manufacturers charge a lot for these graphics cards so it's good to test the hardware to ensure you are getting what you paid for. Just saying "Dont bother testing" isn't exactly constructive!

Thanks!

S.

 

yeea

doing questionable tests is the base of the constructivism...

how do you decide which test is reliable?
(you count the people that like it or the people that got what they'd wanted...or when it's simple and repeatable so it's good)

Empirical data are valuable, then and only then when they are made under equal other conditions.

 

Hi all,

Dreamer - Everything was virtually identical - the only major difference with respects to the graphical system was the GPU. Playing a game is each platform is not going to tell me which card is hugely better unless I was comparing, say, a 7600 with a 7900GT. Doing subsystem benchmarks is the only way to compare similarly spec'd GPUs. I don't see how this was a "questionable" test.

Anyway, I did the 3DMark test on the 7400 and got a score of 290 for SM2.0 and 252 for HDR/SM3.0. Can anyone with an X1400 please post their 3DMark scores (I don't have my X1400 available right now).

I'll post the 7600 scores when I can run the test.

Thanks!

S.

 

Ok,
but just it's good to know that manufacturers usaully clock the GPUs in different way, that's way the comparison is a bit questionable...

and the gaming performance is something a bit different from benchmarks, there're so many examples for that...

PassMark - ? what's that... sorry

Otherwise, nothing against benchmarks but just don't trust them implicitly

Anyway, here is the full Mobile Graphics Cards Benchmark List - you can find the information that you're looking for..

http://www.notebookcheck.net/Mobile-Graphics-Cards-Benchmark-List.844.0.html

 

Errr, yeah, never heard of Passmark.....use 3Dmark instead.

 

what parts of the gpu does passmark stress?, all the above cores you mention are vastly different from eachother, especially going from ati to nvidia, ati has a edge in the advanced shader department, where nvidia cards have an edge by having more pipelines, hence a higher theoretical fill rate, it really helps to know what aspects of the cards your benchmarking

 

See this is why I hate and dont trust benchmarks. Most of them are optimized for certain systems. There was a deal with 3dmark and nvidia card scores being inflated a while back; plus when you open 3dmark05 it says right on the opening screen "runs best on alienware". Wow I guess I wont score so the best since my pc is an : insert non alienware brand PC:

I do 3dmark after driver installs to see if there is a difference, But I by no means compare my benchmarks with other x1600s or 7600 anymore cause there are way too many factors and issues.

 

Exactly. Benchmarks give a good overall picture (My X300 benchmarks much lower than the 6600GT in my desktop, and that's very much so a reflector of actual performance in other applicatons), but when you're talking about things like what features are supported and benchmarked, things get murky. Such as the FX series of NVIDIA cards having crappy Shader Model 2 support, even though they're decent cards overall. The parts that Passmark tests may just show up really well on ATI and not so good on NVIDIA. It only uses Direct3D tests, which are typically well-represented by ATI. If it used just OpenGL tests, or even a mix of the two, I'd guarantee you that the NVIDIA cards would wipe the floor with the ATI ones. So it's just a matter of WHAT is tested. The 7600 is a faster card overall, though.

As an aside, why are you running a Cisco router, liquidxit2? You have a reason for that kinda functionality, or is it just for the halibut?

 

If you dont like benchmark programs, dont freaking use them. I could care less what you think about em, no one asked, because we allready know what your going to say.

Emmy, not sure what your looking at. I ran passmark and I get 2 sets of marks, one for 2d, one for 3d. My x1300 is scoring 332 for 2d, and 79 for 3d. What numbers are you looking at?

 

Agreed to the nth degree. See how running passmark gave you an ati biased? Well run 3dmark with two identical cards one ati and one nvidia; and the nvidia one will win by a good margin. Now the gameplay and real world difference is nil. You could compare 3dmark scores with others that have the same card to be more on the same level, but then again there will be too many factors to influence a difference (e.g. clocks, cpu, ram, etc.).

This is why I only use benchmarks for comparisons between different tweaks and drivers I have installed, thats it. When I first got my notebook right out of the box I scored just under 2800, now Im hovering around the 3000 mark. The difference in gameplay you say? well it hasnt changed a bit.

Just in fun, we had to make banners for our routers in my cisco class a while ago to warn people not to mess with our routers; and mine was kinda funny but a litle obscene (spelling?). SO I censored it. But yeah I deal with cisco equipment from time to time at work and consulting jobs, fun stuff. And yes Im a command line junkie....

 

no need for hostility, I have no problem with benchmarking programs, what I do have a problem with are the uneducated folk out there that dont understand the full picture of the benchmarking results, and how they relate to real world performance, there are a lot of folks out there that live and die by their 3d marks, I dont know about you, but I personally like to play games, not benchmarks, and its sad that the hardware makers take steps, often times to the detriment of image quality to make synthetic test results look more appealing, I use 3d mark to test new systems and overclocks, and I find it very useful for that, when shopping for a new card, I dont even bother to look at the 3dmark scores, I compare the results of the current games and applications I intend to run, and base my decision from those results

 

can someone tell me which laptop to take of...

TOSHIBA (Go 7600) (256) All the other specs remain the same for all the 3
DELL (ATI X1400) (256)
HP (Go 7400) (256)

the principle purpose would be playing games like AOE3 or something that needs some real graphics...

please suggest me on which to take people...

All the 3 r principally coming in almost the same range exept for dell being a bit less...

cant afford Go 7900 else would have taken that...

thanks...
purav.

 

yea 3dmark is more accurate... and if you have somethign like css or hl2 do the video stress test and check fo ryour average fps rate... fear also can do this for you make sure they are alls et to same screen size and every thing else in the game must be the same.

 

well part of it depends on the other things in the computer(cpu ram) but the 7600 woudl be way better for gameing than the x1400 or 7400