Widescreen Gaming Forum

Here's the first pass:

http://www.widescreengamingforum.com/wiki/index.php/WS_and_WTH_Benchmarking_%28Core_i7_920_and_9800_GX2%29

Gonna add a few more, then drop in the GTX 295 I have sitting here.

Thanks a lot for all the time you put into this, on top of the rest. I love reading benchmark charts :)

The i7 platform looks very interesting for gaming in high res's with multiple GPUs, be it multi-GPU boards or Crossfire/SLI. It seems to be much better than any nforce chipset to date. 2.7 Gig by core is pretty modest however for such an expensive upgrade, and WiC looks clearly CPU-bound to me - but it's impressive to see what the very same graphics card can achieve on that new platform. :shock:

It'd be great to see how the i920 with two cores disabled compares to the E6700, in order to focus a bit more on pure chipset performance. Quad-cores are still often vastly underused in gaming, but dual-cores tend to show their limits - especially at lower frequencies and coupled with multi-GPU solutions. Well, I know you can't test everything... ;)

Racing title : I'd suggest Trackmania Nations Forever for its built-in benchmark feature. Pretty rare in a racing title afaik, not to mention its native support of ws and TH. If you prefer to conduct long runs in actual gameplay conditions, check out Grid - cockpit view on Shibuya tracks for instance (wet streets in the night = a lot more demanding than the rest of the game...)

Can't wait to see the GTX295 (hopefully) blow those results away :D

Yeah the GTX is just the same as SLI GTX260's (plus a little) so you'd hope so! Still when are you getting quad-SLI! that'd be interesting LOL

I'll stick with SLI GTX280, the memory bandwidth is so much more important at high res...it is still all that holds us back at 5040 (esp with AA).

The core i7 is so much better than core2 but for gaming? Not really that relevant except in a few games (WiC as mentioned) so I am happy with my core2 with a backup of my core quad if some games that I want come out that really need 4 cores.

I'd also be interested in the 295 with heavily OC'd memory, that would be good....

I bought GRID. I'll see what I can do to load it up and run a track. Hopefully I won't suck too bad. I think I'm going to upgrade my power supply as well to a Corsair 1k. Then I could try GTX 285 SLI. I know the bandwidth is going to be an issue. If it's that much better, I would return the GTX 295 to the retail I got it at.

Well the 285 SLI will perform similar to mine, I would imagine. It is only a small clock jump cos of die shrink. My 280's are factory o'clocked at 670/2.5 and I have bumped it to 700. The stock 285 only 650/2.5 ish.
Maybe it will OC much better though, but it'll certainly run cooler and (hopefully) quieter.

Great writeup but I echo the Quad Core vs Dual Core request to see if it really is the new platform or extra core that is giving the increases.

Also, you should overclock the CPU a bit to see how much further improvement can be realized, especially at the higher resolutions.

Looking forward to any updates!

I've now posted some benchmarks for the i7 920 and the GTX 295

http://www.widescreengamingforum.com/wiki/index.php/WS_and_WTH_Benchmarking_%28Core_i7_920_and_GTX_295%29

Great review, as always. I never seem to get around to doing that sort of stuff. I still have heaps of games I want to finish!

I have order WiC so I will run that on my system to see just how CPU core limited it is, it shouldn't be GPU limited on mine! However I have the dual core in it 'cos so far I have seen no advantage to having a slower quad in games. Be interesting compared to yours.

It takes a lot of time to do them. I forget how much time. I'm torn with doing them. I need to do them to help generate traffic for the site, but I don't enjoy them. I feel like I've accomplished something when I'm done, but I'd rather be playing the stacks of games that I have.

Why do you have to do them to generate traffic? Just because more people search for 'GTX295 review' than wide screen help?
I know nuffink about web design/running

Great Work...


I have a question ...on the GTX 295 what OS were you using for the Benchmarks??

I have a Gateway FX6800-01e i7-920 running Vista 64 and I could not get 5040x1050 on my GTX 295 on Drivers versions 181.20 / 181.22 beta /185.20 beta. Custom Mode Fails

I was just wondering if you were using Vista 64 and how you achieved 5040x1050 and also if the results are different under different OS's.

On a side note. Have you tried running PhysX on a second card??

(I am running two video cards my GTX295 and my second I'm using GTX280 for Physx. )

Posting benchmarks and reviews gets hits from web searches, it also keeps fresh content to give people a reason to come back.

I am running XP 32-bit. Knowing of the 200-series problems in Vista-64, I didn't even try. I saw the beta driver that Mach1.9, and I'm going to try that. I have PhysX enabled, but I don't think I have any real PhysX games. That is why I wanted to try Mirror's Edge.

I want to try GFX and PhysX on the GTX 295, then GFX on 295 and PhysX on something like a 9600. Then try different combinations of 285/9600, 285-SLI and 285-SLI/9600. I'd like to see how Mirror's Edge benches in all of them.

The core i7 is so much better than core2 but for gaming? Not really that relevant except in a few games (WiC as mentioned) so I am happy with my core2 with a backup of my core quad if some games that I want come out that really need 4 cores.

The I7 SMASHES the Core2duo when more than 1 gpu is used in gaming. Sometimes almost double the performance. Guru3d has a great article on it.

Slightly better, sure but not that much, and lots more expensive.

Doubling the performance looks like a pretty big difference to me. Sure it doesn't double performance at every rez, but this is HUGE for multi-gpu machines. Not slightly.
http://www.guru3d.com/article/core-i7-multigpu-sli-crossfire-game-performance-review/7

Doubling the performance looks like a pretty big difference to me. Sure it doesn't double performance at every rez, but this is HUGE for multi-gpu machines. Not slightly.
http://www.guru3d.com/article/core-i7-multigpu-sli-crossfire-game-performance-review/7

But look at the drop off as it approaches 2560 res. Once you get to 5040 res the gain will be negligible. I doubt you'd see any real advantage at 5040 on my system if I swapped to a 920 or even a 965. I certainly aren't going to pay a few thousand dollars (mobo, CPU and watercooling) for the privilege of a few FPS!

Using 280 SLI @ 2560:
BIAHH: +50%..very good
CoD4: +2% ..very poor
Far Cry 2: +5% ..very poor
FEAR: -3% ..arrgh!
Crysis WH: +8% poor
So not that good, accept at BIAHH, and the shape of the graph shows the difference rapidly shrinking as resolution increases. So at the res I want to play (an extra 1.3MP over 2560- almost 1/3 more) I will see virtually no gain, or none at all.

If they detailed settings more, I'd be inclined to trust their results, right now though, they lack too much of it.

For starters they don't seem to use FSAA, which as CPU for GPU use, is one of the biggest customers.

As for you (xetura), so far you seem to me like a fanboy, and therefore not worth taking seriously, period.

Know what I mostly thought of from that reviews? how people used to claim great improvements benchmarking quake3 when it went from 300fps to 400fps.

[quote]Doubling the performance looks like a pretty big difference to me. Sure it doesn't double performance at every rez, but this is HUGE for multi-gpu machines. Not slightly.
http://www.guru3d.com/article/core-i7-multigpu-sli-crossfire-game-performance-review/7

But look at the drop off as it approaches 2560 res. Once you get to 5040 res the gain will be negligible. I doubt you'd see any real advantage at 5040 on my system if I swapped to a 920 or even a 965. I certainly aren't going to pay a few thousand dollars (mobo, CPU and watercooling) for the privilege of a few FPS!

Using 280 SLI @ 2560:
BIAHH: +50%..very good
CoD4: +2% ..very poor
Far Cry 2: +5% ..very poor
FEAR: -3% ..arrgh!
Crysis WH: +8% poor
So not that good, accept at BIAHH, and the shape of the graph shows the difference rapidly shrinking as resolution increases. So at the res I want to play (an extra 1.3MP over 2560- almost 1/3 more) I will see virtually no gain, or none at all.

Well sure, I didn't say it was better all the time, I said at certain resolutions in my second post. Obviously the resolution you're running at isn't going to see a bump at all. But for the majority of people out there with one monitor, they will see big jumps. And thank you for being mature about this, Mach1.9pants, as apposed to others in here.

If they detailed settings more, I'd be inclined to trust their results, right now though, they lack too much of it.

For starters they don't seem to use FSAA, which as CPU for GPU use, is one of the biggest customers.

As for you (xetura), so far you seem to me like a fanboy, and therefore not worth taking seriously, period.

Know what I mostly thought of from that reviews? how people used to claim great improvements benchmarking quake3 when it went from 300fps to 400fps.

Maybe before you start calling names, you should do a bit more research and back up your claims with proof. That wasn't the only site I've seen those kinds of gains. There's a few sites out there that have similiar results as well. But if you want to ignore the numbers and act like a little kid calling names, that's fine. Not my problem.
Like I said before though, single gpu results for the I7 are meh. But more than one gpu can see big gains. Sure, if you're running at obscene resolutions like 5040x1050, it won't matter much. But for those(like me) that run 3840x800, or 3840x1024 and lower, there are gains. Is it worth it? Well, that's up to the user to decide.

Without FSAA the gain fails at 2560x1600.

With FSAA it'd likely drop before that, probably well before it would help you gain anything whatsoever on a 3840x800 or 3840x1024 setup.

And if you want to be treated with respect, you should post with some. It's a two way street.

Your last reply only confirms what I said "look like" which if not true wouldn't matter to you at all.

Fanboys plague the net like humans do the planet. If you're not one of them, learn not to behave like one.

I would research it more, however, I have no time to waste on reviews that don't bother with detailing their testing methods.

Now crawl back under a rock and let the mature ones discuss.