What Benchmarks do you guys trust?

I figure the folks on this discussion group know more about laptop/notebook graphics realworld performance than anyone else.
Years ago I worked in workstation development at two diff companies and in those days we were dissatisified with commonly used and published benchmarks. none really got at the issues of what real people doing real things would experience, and some benchmarks were deliberately 'tilted' to favor/bias an outcome.

so fast forward to today: I was reading on another forum and somebody suggested this site as the "final word" on making purchase decision for a rig that does good job with graphics [games were in the discusssion, though not exclusively..] the site:

Mobile Processors - Benchmarklist - Notebookcheck.net Tech

Right off I have two issues with the page/display: no reference to age/generation/currency: when was which one introduced? which ones are newer? "Core" 1st gen or "Core" 2nd? No common graphics naming labels: i.e. "radeon 6xxx" so one has to go run down the description to see what's marked.

then, it just makes me suspicious when you sort this list a couple of diff ways, and for the most part, all the Intel Core series dominate no matter what. is that realworld?

So, what benchmarks do YOU trust as at least an indicator of platforms that should be considered viable for gaming?

 

I trust 3DMark Vantage GPU the most. I've almost never seen it's % of difference not transfer into real world performance.

 

The reason why the Core series absolutely dominates is because AMD has never released a direct competitor to the Core series ever since Arrandale, because they just cannot keep up in terms of architecture, performance and efficiency. The high-end AMD chips can only match up to perhaps the mid-range Sandy Bridge processors. As such, the Intel chips utterly wipe out any competition AMD may have. However, AMD recently pulled out of the BapCo SYSMark consortium due to what they believed as biased benchmarking against AMD in favour of Intel chips. Whether you believe them is up to you, but regardless of this, Intel chips are the best for now.

The standard GPU benchmark for now is 3DMark06 since it's the most widely used. 3DMark Vantage is used to test DirectX 10 performance, and the latest 3DMark11 to test DirectX 11. Don't bother trying 3DMark11 at all unless you're running the desktop GTX590 or HD6990 because it lags so bad on anything else.

 

No one uses or trusts '06 anymore. It's dead.

 

excellent feedback.
so, is the problem with 2nd gen Core i processing in the marriage with gpu du jour? It doesn't appear anyone take Intel HD3000 seriously, and nvidia seem racked with driver issues - maybe they have a specific driver all tidy for benchmarking but not so hot for real people?
the anecdotal feedback seems to run in Radeon favor

 

what's your "Go To" mark?

 

I think Unigine Heaven benchmark works pretty good.
It does Dx9 to 11 and also the latest OpenGL.

But i am curious to see if AMD Bulldozer sees a September release.

 

Exactly, it became obsolete when DirectX 10 was introduced, and that was like 5 years ago.

 

For me 3DMark Vantage and Unigine are very good, buuuutttt you should check what are the configurations those benchmarks are running on, otherwise the benchamarks would be quite off.

 

The Intel HD3000, while impressive for a discrete GPU, is by no means even close to average standards for today gaming. As for Nvidia or AMD, its your choice on GPU. Both their drivers work fine.

 

uni heaven kinda favours cards with better tessellation ie nvidia.

3dmark06 is no longer valid due new dx and the fact new cpus are soo much higher score in the cpu tests.

vantage and 11 are fine

 

For games, game benchmarks are available, and they are undoubtably far and away the best indicators of absolute and relative performance.

If youre interested in performance differences with one application, there is no reason to compare performance in a different application if the results of the former are available. Games and benchmark apps are so varied in performance that there really is no single benchmark that delivers the full picture. Take a collection of real use benchmarks instead.

 

Cinebench 11.5 from the makers of Cinema 4D graphic 3D modeling and animation application.
Tests CPU and OpenGL in "real life" examples.

Also no need to install, just download and run.

MAXON: CINEBENCH 11.5

 

I trust none of them. As the guy from the notebook suggests, you need to get the bigger picture. Try to get results on the games you want to play, or games that use same/similar engine to get a feel for it.

 

But you can always turn tessellation off.

While I do agree that game benchmarks are more significative, if the performance of certain cards is what the OP wants to compare, he should get a bunch of game benchmarks in order to look at the big picture. The problem is that some game engines run better on Nvidia or AMD architectures (you can see such discussion in many tech websites such as anandtech when reviewing GPUs).

 

Playing the games.

 

It does under Linux.

 

hi i would recomend 3dmark vantage, 3dmark 11 ,furmark,
furmark to see cooling probs and stability the other 2 to see performance.
this is what i use when running Oc and test new drivers to see perf dif in ex new G drivers and also i run the games i like with fraps running since i played them so much i know the prev fps in a certain part of the map and so on and so forth
for the hole setup i run Pcmark 7 and Pcmark Vantage comes from the same company that makes the my primary gpu bench soft
if you want to compare to gpu's from amd and nvidia remeber to turn of phyxs on the nvidia card to get more reliant scores (im nvidia fan)

 

I ran Unigine Heaven Benchmark as did a Sager user with his Geforce GTX 485m.
My AMD Radeon HD 6970m scored nearly the same as he did.

 

give me a d link and i will run it with the h ware in my sing and give the result of course at stock settings

 

Heaven DX11 Benchmark | Unigine - advanced real-time 3D engine (for games and virtual reality systems)

There are multiple settings and render modes to choose from.
This way you can sorta test out how OpenGL performs on linux vs windows ect.

 

what is the stadard setting every one is using just to give a ref to the OP

 

most games don't leverage physx. even fewer support GPU physx, iirc.

 

3DMark Vantage, 3DMark 11 and Unigine Heaven.


Good stuff

 

Either 720p or 1080p. DX11 with Normal tesselation setting. 4x Anisotropy. High shaders, high textures. No Ambient Occlusion, no Volumetric shadows and no Refraction.

 

i wasent saying it did but what i was saying was if he is comparing amd to nvidia i these soft then he needs to turn it of on the nvidia cards to get a acurate reading even that im a fan boy i like acurate readings in bench progs
and as you say physx is a nvidia thing but let me say as of now at max for me is in Vantage is 28979 gpu points not confirmed as of at that time no wifi thats with physx of and at realy high gpu settings and a bumpt V in Bios thanks to NiBiTor from MVKtech.net or mavke as his handle is

 

Im running Performance on them havent lookt at specifics that much

 

benchmarking like this has limited real world usage.

this is akin to being involved in tuner cars.

 

for me its just to know if i get low fps in one game i can play around with this and this and zap zarap its running with ecepted fps

 

this makes good sense for the knowledgeable, but for someone trying to sort the current/new generation laptops for something that is at least Capable of running a variety of high-end games, we end up relying on some 'expert analysis' with its benchmarks.
But, and maybe supportive of your point, once a shortlist is arrived at somehow, it boils down to "user experience" which can be awfully subjective.
I've seen a number of AMD-equipped models earn high user-ratings, and a number of 'Core' models with nvidia gpu's that don't score so high with real users. also seen the reverse condition... Specs & marks can only tell a little about what SHOUD be the capability of a model. execution is everything... keyboard, case, build quality, placement of devices, hinges, diplays that actually look appealing....weight, heat, battery life.

 

one more question for you folks:
- do most of you use SSD's?
- and, I assume if you are, it ain't because of great benchmarks, but rather fast load times and cooler operation?

 

IM using a 160gb intel
x25 m some thing for me its fast and works great no bugs and i can recomend it to run your OS and favorite games from ssd then the rest from hdd
i like to use a ssd since it cut down start up and shut down time for the sys not only for games i have a compleat sys image on a 160gb standard if some thing would hapend to the ssd just in case

 

I have an SSD, and not because of games. Mainly for vastly improved responsiveness, durability, and battery life.

 

I want an SSD badly, but I'm still holding out for them to have a better GB per dollar ratio. When a 256GB model goes for $150, I'll jump in happily.