X-bit labs

Started by: Mr. BonBon | Date 02/18/08

Comments: 36 | Last Comment:  04/20/08

[1-19]

1. That's weird because other sites show solid gains with the 3870X2 over 3870 in games like COH, COD4, ET:QW and etc....

Maybe something is wrong with your setup.

Or maybe something is different in their testbed that causes the change in framerate.

Is it "wrong" or different? I wouldn't tell them something was wrong unless you specified how so.

We cannot compare results obtained on different test beds with different hardware, software and settings.

The drivers for the card are still work in progress. Some flaws were discovered right before the planned launch date, which is why the card was postponed until a better driver was made, but since it was a mere matter of days and it happened after some cards were sent to the reviewers, so apparently not everyone got the same drivers. With the right ones, which will soon be officially released, the card performs admirably. A review of three-way SLI which I read yesterday showed (though that was not the focus and the author made no comment about it) the X2 beating a single 8800Ultra almost across the board (CoD 4, ET:QW, UT3 and HL2:E2, with Crysis being a mixed bag).

2. And the multistuttering problem is mentioned where exactly?

3. Good job! nice article - quad crossfire in month should be even more interesting. 3 way sli vs 2 slot quad is the high light of my year. lets not forget mr nvidia says dual gpu solutions are not as goos as single.

That's right. My 9-month old 8800 Ultra still bests AMD's best - without any OC'ing too.
Yeah, I'd say Mr.NVIDIA knows what he's talking about.

Yea and how much is a 8800 Ultra, almost double of what the 3870 X2 is.

my my we love mr nvidia dont we? mr nvidia is the dude that said that double gpu's where not as good as single not the company!
would you still be paying $700 not $450 if there was no ati? No you be paying a lot more!

compare it to 8800GT/GTS in SLI 400-500$.
ATI can't impress since 19xx.
The only thing they good for, is price wars.
Exactly as they owner AMD since core 2 release.

4. Just as in my comments regarding CPU reviews, I would begin by saying that I trust xbit labs more than any other wanna-bees out there, just because I'm reading those articles for years now and they never used to lie before, at least I believe so.

Of course, a number of trusted sources, like anandtech, posted a different picture of the X2, so I must confess I'm a bit puzzled, too. Not that I would ever go and by a X2 by myself - I'm pretty damn paranoid about the bad consequences of such a made-up design.

To sum up, this is a cowardly solution to me, which, as noted in the review, is intended just to give a little more time for the AMD engineers to come up with something more elegant. For now, I don't see any reasons for buying an ambiguous transitional product as the X2. The past has shown that such sandwiches get blown away by a next generation single-chip designs pretty soon.

But I would buy an ordinary 3870. It's such a bless for non-demanding users, with such a low power envelope... Nice :)

i really think ati is pushing for new tech based on old tech, 2 gpu's in one card and 1 slot with a water cooler = 4 gpu's with 2 slots. with the right drivers this could be a great product. we see with time?

The keyword by all multi-stuff (CPUs, GPUs) is scalability - or how much do you gain by multiplying your raw processing powe. In the case of SLI/Crossfire, the speedup ratio often falls below 1, which is absolutely scary...

Would you buy a 4-core CPU that SOMETIMES is slower than a 2-core? Or even than a single core? All 3-SLI or Quad-Crossfire setups I've seen being benchmarked so far suffer from such pitfalls. See that link, it's a total disgrace:

http://www.anandtech.com/video/showdoc.aspx?i=3183&p=3

The X2 review showed that there is a decent amount of games which don't profit from the second GPU onboard, I don't see any purpose in adding another pair of GPUs to the soup... Your games that used to run fast will run even faster, and those who didn't won't see any profit at all.

Now that's a goog idea, kudos to AMD/nVidia for providing their customers with the opportunity to waste 4 digit sums on dead-born technology... But you're right, having the X2 to perform even that high is a miracle for itself.

Call me a prophet for what I wrote two days ago, now we have a 4-GPU setup benchmarked, and it confirms my point: CrossFireX with 4 GPUs sucks exactly as 3-way SLI did, check it for yourself:

http://www.anandtech.com/video/showdoc.aspx?i=3232&p=3

And these are results with games hand-picked by AMD.
I wonder if you would ever want to buy a multi-core CPU if it scaled that nice :) Come on, nVidia and ATI must be kidding, no idiot in the world would buy such crap. Now that's why I feel very curious about what Intel has got to add to the topic.

5. Any chances of getting power readings included in this review too?

+1

We will at some point in future. Sorry.

6. Thank you for a good review, this is why I always wait for XBit to publish their reviews before I make any judgment about a new product.

I would however prefer that the following be mentioned:

The problems with Crossfire and Sli's micro stuttering, I have an SLi system myself and I am not very happy with SLI and a card build around Sli and Crossfire with the micro stutter issue is a real deal breaker for me, it makes gaming an irritation instead of a pleasure. Not to mention that Sli does not work in the one game that I spend most of my time in Lord of the Rings Online.

The X2 is not a mature product by any means, I do not think that it can be crowned the fasted card on the market due to its low average scoring. Winning in Synthetic benchmarks is not a champion. Though it did show potential in some games the minimum frames where always lower than the Nvidia parts.

I am however glad that Ati found some sort of bragging rights for the next month or two so that their image is lifted a bit. As mention by some other comments we need strong competition in the market and I am glad that Ati still has some fight left in them.

Again thank you for a very thorough article!

Thanks for nice words.

Are you referring to AFR lags? What do you mean by "stuttering"?

7. I am so confused... i thought this card was a dual RV680 gpu effort, with pci-ex 2.0 and gddr4... or am i simply going mad? where is the RV680 if i might be so bold as to noob myself here :)

You've been living in a dream world vampmonk.

R670 is the 3870 512 with GDDR4 and PCI-e 2.0

R680 is the 3870X2 1GB with GDDR3 and PCI-e 1.1

The slower ram doesn't make that big of a difference in performance, and PCI-e 2 isn't mainstream yet.

eish, then why does the review say dual R670's? and ive found this (Asus EAH3870X2/G/HTDi/1G , dual HD3870 chipset , HDMi ready , support crossfire-X ,pci-e 2.0 x16) thats the asus card saying it sports pci-e 2.0. so are these ppl lying to me? ARG so hard to figure out a decent card to buy. so would it be better to get 2x 3870's or a single 3870X2?

8. Just like I've been saying from my experience with the X2; inconsistent, unpredictable and not a new champ.

9. Great article. Which is really too bad for AMD and the consumer. Reality is a harsh mistress.
It does show promise however.

10. Why didn't they use the Catalyst 8.2 drivers? They were released Feb. 13

PLX PCIe Switch Unleashes Maximum Throughput on ATI Radeon HD 3870 X2 Graphics Cards

Maybe because they spent more than week on the review? Anyhow, the drivers they did use were pre-release 8.2's, so the performance should be the same (since ATI's monthly release cycle means the month prior to release is by and large frozen in terms of performance and features, with the main focus on stability). Finally, they are the same drivers everyone else used when reviewing this card.

11. Yet another total waste of space unless all you do is bench

12. i've read the reviews from anandtech.
similar system, very different result.
what gives?

13. most review sites showed 3870 X2 faster than 8800 GTS and GTX in most scenarios... Is this a driver issue?

14. I think that the multi-GPU approach to boost performance is cheaper in the short term and long term. Since the majority of costs are in the refinement stages where the very last bit of performance is tweaked from the GPU at the micro-code level. From the ATi perspective, why create a +$500 video card that no-one buys?

I'd rather be able to afford a video card than have some ego trip about how the GPU on my video card is made by the same company as the GPU on the fastest video card on the planet.

15. want a part 2 of the review with new catalyst drivers.

16. Useless...
why using 2 differnt systems ?? an old Ati580 chipset, and the latest (working ) Nvidia chipset ??? ever heard of scaling ??
what is wrong with a neutral intel chipset to test both cards ???
to get atleast some fair comparing

or are you eyes here on Xbit gone completely NVgreen ??

stick with cpu coolers testing in the future ,
to avoid becoming the joke of the web..(god knows you do enough every week)

17. I always wondered why you guys always tested Oblivion without AA. You just have to set HDR in game and force AA from the control panel, most people who are spending hundred of bucks on video cards want to play with AA enabled.

18. any one who wants a more honest less bias review go to toms hardware

19. "We cannot compare results obtained on different test beds with different hardware, software and settings."

Keep saying that. If you see such dramatic differences between this review and other reviews, wouldn't you figure out why your numbers are so low in for example call of duty 4?

Instead you keep using these weird results in every review where the 3870X2 is used.

@ Kieran, or go to http://www.anandtech.com/video/showdoc.aspx?i=3232&p=3

My personal experience looks much more like that from the link above than what is accomplished here.

[1-19]