X-bit labs

Started by: fastpunk | Date 01/25/08

Comments: 32 | Last Comment:  02/26/08

[1-20 | 21-25]

1. All the tech details and the thorough benchmarking of both products makes this one of the best reviews I've read in this field. Good job! For me, the 3850 is looking sweet, it really strikes a balance between price and performance, it's all I need from my graphics card. These are the type of products we want to see more often. Now I can only hope that x-bit will do equally great reviews for the Radeon HD3870 X2 and the GeForce 9800 GX2. Peace!

2. What I do not understand is why the fucking game companies release titles that can only be run normally on a few fucking expensive graphics cards...

I love the 3850 though...

Because by deliberately pushing things to their limits, it gets people buying new hardware...As a result, the software market drives the hardware market.

Its why I don't bother playing the latest games anymore. (I'm amazed at how much money I've saved in 2007 when I didn't spend a dime on hardware!)

What's the alternative? Make games whose 'highest' graphics look like crap so you can run it on your Voodoo 3? That's stupid.

It makes far more sense to make games that push the hardware's limits. That way you have a more marketable and interesting product.

People who don't want to buy better equipment can drop the quality settings down, or wait for a few years, and people who care enough to buy powerful hardware get to see amazing things. It's win-win.

3. This review's only half a year late, its all cool.

But to be fair, this has to be the most comprehensive and informative review on the net. So many games benched, wow. GJ xbitlabs.

4. good review, a lit tardy though.
how about get working on that GTS 512

5. Where are the no AA and no AF tests??? Everyone knows that ATI's ROP has problems with AA.

Ha. No shrimp. ATI has no problems with AA, its just the Nvidia's 4xAA is very efficeint and ATI's isn't. Once you have no AA or 2xAA or 8xAA you can see that the gap between ATI and Nvidia shrinks.

6. Excellent test, but I'd like to echo lucassp's comment: why no tests without AA? I personally haven't found it necessary to play any game on 1600x1200 or 1680x1050 with AA turned on, and even if I did, 2xAA would have been good enough.

Also, please, pretty please, do some passive cooling tests for both the 38xx and 88xx cards, using the Accelero S1 and Thermalright HR-03 coolers, along with measuring GPU, VRM and memory temperatures. Some of us are aiming for truly inaudible computers, with passively cooled CPUs and GPUs.

It's no problem to use Accelero S1 with 8800GT. For 3850 it will be more than sufficient cooler.

Accelero S1 + 12 mm fan @ 800rpm = 50°C in full load! :o And it is absolutely silent solution. I recommend it to everyone. Even without any fan is Accelero more efficient than stock cooler.

7. plz make the test with the latest ati driver??are you a pro geforce site....???

8. Aaaaargh!!! Will you guys please stop spewing out nonsensical guesses as to why a particular card wins in a particular game? You aren't using any scientific method and are just throwing out random conclusions with no basis. I like your reviews but this practice is exceedingly annoying and unprofessional!

No, you are wrong. It is indeed very "scientific".

The problem is that it's too scientific at times. Just benchmark that game in 3 resolutions and then provide a benchmark chart. That's it?

An ideal thing would be to include a best playable setting for each video card in a given game. There is one popular site that does that but that site does not include "scientific" benchmarks. I would love for there to be a website that includes BOTH.

Of course, I would prefer that the "best playable setting" chart comes AFTER the scientific maximum quality/3 resolutions chart. Also, at this point it would be ideal to always have both DX9 and DX10 benchmarks in games that support DX10, since more of us still use WinXP, and in some games the DX10 benefits are barely noticeable or even worth the 40-50% performance hit in games that really shine in DX10 mode.

9. I think your comments about 8800 GT 512 availability issue is way dated as both 8800 GT 512 & 256 can be found pretty easily nowadays, and at the date of starting this article both HD 38xx had their own supply issues once the launch date inventory depleted. That's a worldwide talking not only US driven.
Taking the price difference between 8800 GT 512 vs. HD 3870 & 8800 GT 256 vs. 3850 is kinda misleading as the performance difference is higher than the price difference at current prices.
Both HD 38xx have some advantages though, they are both quiter and have onboard sound.
All in all this article could have made a much more impact if it's done earlier, now the 9600 GT is around the corner and it will push both 8800 GT 512 & 256 prices down which changes the situation completely, at least me think so !

Forget my last paragraph, 9600 GT (G94) turned out to have lower specs. than 8800 GT (G92) yet they followed the same trend of using a higher number for a card still belongs to the current generation of cores. I think they are out of 8800 suffixes (Ultra, GTX, GT, GTS & GS) and they already used 8600 and 8500 monikeres, still they could have used 8800 S (but that would be way too much 8800 suffixes) or maybe 8700 (whatever) and that would be logical as it's higher than 8600 and lower than 8800 (both new and old ones), for it's aimed directly againest the 3850.
Once this one is out the 8800 GT 256 will have no point at all though !
Out of the topic but I had to correct myself.

10. Why review a graphics card and not prove anything about graphics quality?

We need to see real pictures compared, how colors and details are rendered in games, and, in movies too. A slower graphics card may render detail much better. This is the manufacturer cheating ground.

Standard list of questions to be answered:

A. Price
B. Speed
C. Image quality (!)
D. Noise
E. Power draw
F. Compatibility
G. Extras

11. Overall a good review, but as usual a few issues:

- What exactly does testing Crysis on Very High detail tells me, the casual customer/gamer?
that i shouldn't buy/play that game? or i shouldn't buy a graphics card?

- The above is also true for Call of Juarez, Lost Planet and World in Conflict. I just don't get it why are xbitlabs so stubborn. Why can't they (at least for those few games) provide a playable benchmark as well. Come on look at Anand, Tom's, Vr-zone etc.. all of them are smart enough to understand that some games are to be tested in PLAYABLE settings!

- The end table (sufficient/insufficient performance) is very deceiving, because a person may conclude that Radeon 3870 has absolutely insufficient performance in some of the games (Call of Juarez, Crysis, Lost Planet) and turn to its competitor (nVidia) and buy a card that has the SAME INSUFFICIENT performance in the same games!!

- Its abit queer to conclude (and despite it's correct) that Radeon 3850 is better than Geforce 8600 when the latter hasn't been in any of the benchmarks.

Well as usual same issues repeat itself. I'm not saying you're doing a bad job, not at all, i'm just saying that you just stubborn and either don't learn anything or completely ignore these feedbacks.

12. HM...
nice ...
sweet ...
I'm using 8800GT 512 with DELL 2407WFP-HC (1920x1200).
I'm a casual gamer, and mostly prefer FPS where the requirement usually lower and fps are higher. But the main problem is Vista. (in my case 64 bit due to 4 GB of ram). There is a significant drop in performance when you switch to Vista. ETQW became almost unplayable @1920x1200. (funny, it is perfectly playable on same rig, with same setting with "other" (linux) OS).
Anyway ...
The price difference between 8800 512 vs (8800 256 and ATI's 38xx) so low in the real world, that it's definitely 8800 512 to choose.
Much better card. And can easily be overclocked with official NVIDIA's nTune to 700/1600/2000 (core/shader/mem) and that will give the performance of 8800 GTX/ ultra.

And i like the "excuse" tone of compares in the article. "Almost like NVIDIA's". Especially when compared to old (and year is old for entry level of gamers cards). This AMD/ATI's "almost" compared to Intel/NVIDIA in last year is sad because leaders don't make their top product better/cheaper.

13. after the graphs of Company of Heroes: Opposing Fronts it says
"The ATI Radeon HD 3870 has a higher minimum of speed than its predecessor and almost equals the result of the Nvidia GeForce 8800 GTS 512MB. You can try to play at 1280x1024 with FSAA turned on although there will be little playing comfort then. So, we don’t recommend to enable 4x FSAA in this game on such graphics cards as Nvidia GeForce 8800 GT and ATI Radeon HD 3870."

1. there is no such card as 8800 GTS 512MB in the review.
Im assuming they meant 8800 GT 512MB, because there is no point in comparing it to the 320MB 8800 GTS
2. 8800 GT 512 MB is more than 50% faster than the radeon HD3850

oh they were talking about the minimum fps. then at the highest resoution the 8800gt has a ~43% higher fps, at medium it has ~38% and at the lowest 17%. Still wouldnt call those numbers "almost the same."

14. Another question ...
why different mother boards ?
There is no SLI/CrossFire tests.

15. Wow, awesome sufficient/insufficient performance chart in the last page conclusion!

However, it should include DX9 and DX10 modes because it could be misleading for WinXP users that can only game at DX9 in the first place. Many still do not know exactly how much difference there is between DX9 and DX10 performance in each and every game that supports DX10.

16. A few talking points I'd like to drive home... whilst the GeForce 8600 GTS has proven itself a tad underpar, I have a 512MB G-DDR 3 one myself (which I've easily managed a 30% overclock with for over 6 months now, and it hasn't crashed once, not even for a window of a minute), and I find that it's not that slow. This, of course, my be biased on my part, as this is coming from someone who only had a stock 128MB DDR ATi Radeon 9500 Pro before last June, so after receiving a 4.5x-12x augmentation in frame-rates (around 6-8x, much of the time), it's arduous not to be chuffed in full. Not everyone has a GeForce 8800 GT, GTX, or GTS - and not everyone has the dosh to dish out to upgrade each generation.

I find that at select times, the card I have is only 50%-60% slower than a 320MB GeForce 8800 GTS (and, for that matter, the Radeon HD 3870 and 3850), or at least, comparing it the results seen here, which may be unfair, seeing as how it's Vista, and a noticeable performance drop. I do feel Vista should be the future, but by using it you're not showing your fellow readers the full capacity of these boards, which is belied by Vista. Maybe with the exception of DirectX 10, but it's rather convoluted, since DirectX10 is supposed to theoretically render things faster, but it's doing quite the antithesis. But, yes, I can be in agreeance that the 8600's hardly matter when you have a HD 3850 for around $150, or a tad higher. I happen to be stuck with one myself, but it works for me.

Another point I'd like to drive home... up until a few months ago, nVidia's drivers for the GeForce 8800 series were so deplorable that most couldn't even play Half-Life 2 without yellow rendering artifacts and cold lock-ups. I couldn't watch the 3DMark05 demo without my frame-rate being cut in half, and that's still an issue. Not to mention the texture evict issue, which still affects me in F.E.A.R. So while the 8800 series may exceed the HD 38xx/29xx series in a terms of pure fill rate, bandwidth,and raw power, I find ATi's drivers and price points to have superiority.

And yes, what with 64 stream processors and what have you, the 9600 GT is looking to be a LOT better than the 8600 GT/GTS. Now all we have to do is wait for some new hidden threat on ATi's part, but the Radeon HD 3870 x2 is looking good nontheless. Not in the budget sense of the word, but yeah. As for DirectX 10.1 / Shader Model 4.1? I don't see how this is that much of an advantage, with things like better FSAA sample patterns and better FP32 filtering... or anything of the such. It has been confirmed by everyone under the sun, to my recollection, that it's going to be an exiguous upgrade over SM4.0 at best. So was SM2.0a/2.0b over SM2.0 - the GeForce FX series (a series that had contorted architecture to begin with, so "gumption" told me to stay away from them) and the Radeon X8xx series back in "the day" hardly had an advantage over SM2.0 because of such.

Nevertheless, your review is good, I'm just posting some sentiments I have towards the entire affair.

17. "What I do not understand is why the fucking game companies release titles that can only be run normally on a few fucking expensive graphics cards..." quoted from 31415

and game consoles survive for years without an update.

18. You wrote the 3850 isnt a choice when playing games at 1920x1200.
I played Bioshock, TDU, LOTRO to name some at this resolution using a factory oced powercolor 3850 and it runs very well.
I don ot play much fps games but bioshock run very well too, crysis at high settings is killing every videocard today avilable.
I am very very pleased with the performance of this card as it filled the wide open gap between high end and "mainstream" (8600 is low end in my eyes but not for everyone).
The 38x0 is the right card to the right time!

19. Incredible article

very very detailed and really is a true insight into workings of a graphics card

the only thing missing is the glossary/explanation of some terms (like TMU/ALU and rest)

thanks

20. good job

[1-20 | 21-25]