X-bit labs

Started by: Skywalker | Date 10/19/07

Comments: 28 | Last Comment:  11/23/07

[1-11]

1. you used catlyst 7.9...

good work.

not.

Still same old habit happen

2. Shouldn't HL2: Episode Two be superceding the HL2: Episode One benchmark suite by now? :-P

3. You have one of the most complete testing procedures on the net.

As I always consider heat radiation, not just noise (ideally metered at 1m) in graphics card purchase decisions, why not include some form of heat measurement, similar to Guru of 3D, or Hardwarezone?

Thanks for nice words.

Heat measurements of what?

4. "Low price." comparing to what? 8800GTS.

its too expensive DDR3 version and 86GT can be had almost twice as low, in easter europe.

5. I always appreciate your level of expertise and thoroughness, but there is one very big flaw in this article: you tested this card in circumstances in which nobody would use it!
This is a somewhat overclocked version of a budget card (under $150). Its aimed at casual and gamers on a tight budget. The former doesn't care about all of the eye candy, and the latter cares about smooth gammeplay more than anything else. Yet you tested the card with 4xAA wherever you could, and in resolutions up to 1920x1200 or 1920x1440 and you ended up with charts like the one for Neverwinter Nights 2, in which the card can't get 10fps on average even in the lowest tested resolution. In the end, you've got 20 pages repeating what was clear from the start, that this card won't be able to play latest games with all the eye candy in highest possible resolution, which is something anyone could have guessed for a budget card (if it weren't so why would there be the performance and high end segments?!), and most people buying these cards probably don't have such monitors anyway.
So I wish you concentrated on what this card can do instead. You mentioned yourselves, this card has decent computational power, yet low texturing speed makes it almost useless with AA on. Well turn it off, then. And skip the 1920 resolution. That way you'd show your readers what this card can deliver, instead of what it can't, and, I believe, give them much more useful information than you did.

Agree.. Xbitlabs should compare what the card can deliver so we can see the obvious result for budget segment. No body will buy budget card to play high resolution with high eyecandy and one more thing please make lastest driver for comparison.

We do not usually decrease image quality settings in games as different settings affect performance of various GPUs differently. Hence, for correct comparison we have to keep the highest possible settings all the time.

Full-scene antialiasing dramatically improves image quality. Given that in the last two years we tested a number of performance-mainstream graphics cards with 256-bit memory bus (ATI Radeon X800 GT, X800 GTO, X1800 GTO, X1900 GT, X1950 Pro, Nvidia GeForce 6800 GS, 7900 GS) and decent performance with FSAA 4x, we have no idea why we have to step back to no-FSAA mode ahead of the ATI RV670 and Nvidia G8M launch, both of which are projected to feature 256-bit memory bus. Moreover, enabling FSAA does not affect relative performance of different products, e.g. if the GeForce 8600 GTS is generally faster than the Radeon HD 2600 XT, then it will remain faster with FSAA activated.

I appreciate your taking time to reply to me, as well as the points you've made, especially the one about the necessity to have a uniform suite of tests.

Still, I wish you took a slightly more flexible approach, and "traded" the 1920x1200 tests for tests in 1280x1024 or 1600x1200 without AA (all other settings unchanged). That way you (and we) would be able to compare the card with the upcoming cards once they're out (save for 1920 resolution which isn't an option with this card anyway) and get a slightly better insight in what this card will give buyers under the circumstances they'd use the card when gaming. Especially since (that is the one thing I disagree with you about) that does change relative performance in this case, take the Far Cry Pier charts for a drastic example (compare the results with AA with those with HDR), or Splinter Cell: Double Agent for another one.

I thank you once again for your time, keep up the truly good work.

Hahahaha,

i will certainly laugh when you test something like 2400 and 8500 and lower
you guys will be benching on 4*AA and 16/8X AF???
because you guys are too f**king lazy to change youre benchmark suit?

6. All of these card manufactures really need to hire better artist instead of trying to sell their cards based on female models with
huge boobies as well as naked men in spandex like
on those Sapphire HD2900XTs box art.

I dont mind if the box is ugly fugly but the deliver of performance is the most important thing. By the way I agree what you said make more attractive and tempting box.

I agree that performance is more important than art, but sometimes people just like to
"judge a book by its cover" like the flames on the HD 2900XT. :)

7. this benchmarking was done using 7.9 ATI Catalyst drivers. They should have done it with 7.10 which are available for a few days now, and they promise much better performance.

Something tells me that this graphics card would benefit from 7.10 drivers...

dont even bother

all the results from the other cards came from catalys 7.6....
They are a bit retarded in this place

Would you mind stop spreading what I would call incorrect information here? Thanks.

you said it yourself in the previous reviews.....

do i need to look it up?
you said all reference scores from the other products used older drivers
so you guys are lying and lying again to the readers.....

Reread what I said once again.

you said results came from older driver version because newer drivers didnt change performance.....
Kinda strange,since any driver changes performance a little bit

but hell, i dont care anymore about the drivers,it f**ked up, and it will stay this way...

If you change those hilarous settings i will be really happy
benching mid range cards with 4*AA ist just plain stupid........

And dont give me the rant about the older 256bit cards....
really,i am a hardware enthusiast and i really know all about it
problem is,these are 128bit cards,start realizing it...

And change the benchmark suit....

8. Hi. I am wondering why the benchmark result showed is less than what i got. The testbed is high end one but mine is the mid-end one. This is my system:

- AMD 64 X2 5600+ @2.8 GHz (Stock)
- MSI K9N Neo V3 Socket AM2
- MSI 8600GT DDR3 OC Version (Core: 580 MHz, Mem: 800 MHz)
- Kingston Value RAM ( 3 x 1 GB )
- Sound Blaster Audigy 2
- Seagate 250GB SATA 2

My 3D Mark results are:
3D Mark 05: 11 117
3D Mark 06: 6536

So, i want to know, what is the factors that gives different benchmark result since the result for mid end system is higher than high end one?

This may sound like a dumb thing to say, but there is something very wrong with your scores. Across the web, 8600GTs, even the overclocked versions like yours, are scarcely able to get near the 5000 points mark in 3DMark06. How you managed to get through 6500, I have no idea, but you'd better check if everything is on it's default values in 3DMark and your drivers, and perhaps even to check what card really have you got, because it seems like a most extraordinary 8600GT.

All the details is on default condition. And i'd ran the 3D Mark twice and give slightly different result only. Maybe your system is not good or compatible enough. I also checked with 3D mark score statistic on Futuremark website, there are many people got more than 6500 mark on 3D mark 06 with geforce 8600GT. The highest score is 9+++ on 3d mark 06 with 8600GT.

I googled 8600GT 3DMark06, and all the scores I found were in line with those in the article, save for one guy who voltmodded his card and OCed it to 825MHz GPU and 1000/2000MHz memory to get a 67xx score, so were I not a really naive guy, I'd think you're just pulling my leg. ;)

the 7.10's were out on the same day as the NV drivers used, so no need to use the slower 7.9..
and if i look at the 3Dmark '06 score of the 1950pro, i find it very strange i 'm getting ~5500 (on different cat. versions) on a weaker system then used in the artikel (Skt939 opteron180 1gb 400ddr)
oh well think i have to find another unbiased review site, shame really for the artikels are usually very good
except for the neverending streams of cooler reviews...

9. 2600 2400 sucks qua performance real bad !

10. The 3DMarks seem a little low my 3700+@2830 and 7900GTX beats thoughs scores.

11. Since nwn2 is now at patch 1.10, maybe you could bring your reviews of cards with this game up to date.

[1-11]