AMD's Radeon HD 4870 X2 - Testing the Multi-GPU Waters
by Anand Lal Shimpi & Derek Wilson on August 12, 2008 12:00 AM EST- Posted in
- GPUs
These Aren't the Sideports You're Looking For
Remember this diagram from the Radeon HD 4850/4870 review?
I do. It was one of the last block diagrams I drew for that article, and I did it at the very last minute and wasn't really happy with the final outcome. But it was necessary because of that little red box labeled CrossFire Sideport.
AMD made a huge deal out of making sure we knew about the CrossFire Sideport, promising that it meant something special for single-card, multi-GPU configurations. It also made sense that AMD would do something like this, after all the whole point of AMD's small-die strategy is to exploit the benefits of pairing multiple small GPUs. It's supposed to be more efficient than designing a single large GPU and if you're going to build your entire GPU strategy around it, you had better design your chips from the start to be used in multi-GPU environments - even more so than your competitors.
AMD wouldn't tell us much initially about the CrossFire Sideport other than it meant some very special things for CrossFire performance. We were intrigued but before we could ever get excited AMD let us know that its beloved Sideport didn't work. Here's how it would work if it were enabled:
The CrossFire Sideport is simply another high bandwidth link between the GPUs. Data can be sent between them via a PCIe switch on the board, or via the Sideport. The two aren't mutually exclusive, using the Sideport doubles the amount of GPU-to-GPU bandwidth on a single Radeon HD 4870 X2. So why disable it?
According to AMD the performance impact is negligible, while average frame rates don't see a gain every now and then you'll see a boost in minimum frame rates. There's also an issue where power consumption could go up enough that you'd run out of power on the two PCIe power connectors on the board. Board manufacturers also have to lay out the additional lanes on the graphics card connecting the two GPUs, which does increase board costs (although ever so slightly).
AMD decided that since there's relatively no performance increase yet there's an increase in power consumption and board costs that it would make more sense to leave the feature disabled.
The reference 4870 X2 design includes hardware support for the CrossFire Sideport, assuming AMD would ever want to enable it via a software update. However, there's no hardware requirement that the GPU-to-GPU connection is included on partner designs. My concern is that in an effort to reduce costs we'll see some X2s ship without the Sideport traces laid out on the PCB, and then if AMD happens to enable the feature in its drivers later on some X2 users will be left in the dark.
I pushed AMD for a firm commitment on how it was going to handle future support for Sideport and honestly, right now, it's looking like the feature will never get enabled. AMD should have never mentioned that it ever existed, especially if there was a good chance that it wouldn't be enabled. AMD (or more specifically ATI) does have a history of making a big deal of GPU features that never get used (Truform anyone?), so it's not too unexpected but still annoying.
The lack of anything special on the 4870 X2 to make the two GPUs work better together is bothersome. You would expect a company who has built its GPU philosophy on going after the high end market with multi-GPU configurations to have done something more than NVIDIA when it comes to actually shipping a multi-GPU card. AMD insists that a unified frame buffer is coming, it just needs to make economic sense first. The concern here is that NVIDIA could just as easily adopt AMD's small-die strategy going forward if AMD isn't investing more R&D dollars into enabling multi-GPU specific features than NVIDIA.
The lack of CrossFire Sideport support or any other AMD-only multi-GPU specific features reaffirms what we said in our Radeon HD 4800 launch article: AMD and NVIDIA don't really have different GPU strategies, they simply target different markets with their baseline GPU designs. NVIDIA aims at the $400 - $600 market while AMD shoots for the $200 - $300 market. And both companies have similar multi-GPU strategies, AMD simply needs to rely on its more.
Facebook
Twitter
RSS
93 Comments
View All Comments
random2 - Wednesday, August 13, 2008 - link
I didn't really notice much in the line of bias, however I'm beginning to wonder if this card didn't quite meet their expectations....or maybe after years of running similar types of benchmarks on video cards (which are all basically the same), and having to spend countless hours trying to jam a review together in time for deadlines, isn't quite as much fun as it used to be. We on the other hand can hardly keep our bodily fluids in place when we hear there is a new card review almost ready for release. Now I'm not sure just who the crazy ones are here...Is it Anand and Derek for wasting years of there lives trying to appease us unappeasables, or is it us, because even at my advanced age, I get a woody just thinking about a dual GPU ATI card....Mmmmm....Maybe I should talk to someone about this...Anyhow, thank you Anand and Derek for your efforts and more late nights no doubt.
What I really find interesting is the bit of backlash we are starting to see against Nvidia. Not that I am a fanboy of anything...ok I've looked at a few girls...but When I think of Nvidias pricing strategies over the years while competition was in the realm of little to none from ATI, I cannot help but feel much the same as a choir boy in the preachers shower stall being asked to bend over and grab that soap. I know, no one says I have to, but if I want to play the latest games and become the lead singer it might help.
I like hardware.....I like it functional, fast, reliable and affordable. Till now Nvidia has been kind of sticking it to us so to speak. Providing only the basics at more reasonable pricing and forcing us people of average incomes to sell drugs or our bodies in order to afford their high end products.
Thank you AMD/ATI for saving me from a life on the street, for giving me the opportunity to turn away from a life of corruption, evil and self loathing. Never again will I feel the degradation and the compromising of my moral values in the search of higher frame rates. Bless you...
Down with Nvidia! which I am now sure have become a key player in the evil axis...somewhere...or is that axis of evil?
I wonder if I have time to stop into NCIX tomorrow before therapy?
Denny Crane
DarthAgitated - Tuesday, August 12, 2008 - link
I am bothered by the fact that the test system specs are not shown which seems rather silly.I am even bothered more however by the ridiculous reactions to the article.
Please look over the previous reviews with regards to the 4850 and 4870. Stating that both cards are the best cards at their price points. Or even the 9800GTX review where they praise the 4850 over the 9800GTX.
Also take into account that the GX260 and 280 reviews happen when ATI had no new product out there and here was this brand new architecture being released.
In this review he points out that the Nvidia price drop is due to AMD’s influence. As they did in the 4850 and 4870 review.. “You can either look at it as AMD giving you a bargain or NVIDIA charging too much, either way it's healthy competition in the graphics industry once again (after far too long of a hiatus).”
So please stop with your “this site is pro-“whatever doesn’t fit with what I like” bullshit. Try actually reading the articles instead of doing what most people on the internet do and skim through everything. Here you have a very fast video card, which is shown to win in a bunch of benchmarks. And when he points out the issues with crossfire and SLI implementation and support for game titles, he is speaking a valid truth. And like the majority of x2 cards, it’s hard to get excited over something that can be achieved by simply plugging in a second video card.
Please grasp the concept of this card being a niche product, a product that is not reaching its full potential if your not pushing a 30” lcd (or an hdtv I suppose) and 800+ watt power supply and playing specific games that are properly supported by the crossfire drivers and you happen to have $559.00 (OMG you didn’t automatically account for shipping and my local sales tax when posting that!!!) dollars burning a hole in your wallet.
If you want your pro-whatever review then do a search on website that contains the name of the company you want to win and find their review and enjoy a, what I’m sure is a unbiased “this is fastest evarrr!” review.
I for one appreciate this site and fact that it contains well thought out and technical articles on some new stuff like new cpu or gpu architecture. But don’t expect them to go off on a product that’s an updated version of an existing product that contains two other existing products and offers nothing new.
Please put up the test system specs though please.
Thanks.
Zaitsev - Wednesday, August 13, 2008 - link
I couldn't agree more, DarthAgitated. In the article, I thought credit was given when it was deserved and nothing more. Someone's panties did seem to be mildly bunched over the Sideport issue, however. ;)DRSoul - Wednesday, August 13, 2008 - link
What's happening?? This article is not up to the usual Anandtech standard. I am writing this because I have over the past 4 - 5 years never read such a sober review on this site on such a high performing card. Might just be that they are getting so much new stuff that its becoming boring...AnnoyedGrunt - Wednesday, August 13, 2008 - link
I thought the review was very good. I didn't detect bias, just a good mixture of praise for the performance but concern that the CrossFire support could not always be counted on.In general the feeling (at least on Anantech) towards SLI and CF solutions at this time seems to be similar to the original Single Core Hyper Threaded Intel Chips. Sometime, Hyper Threading helped and sometimes it didn't. It wasn't until true dual core processors arrived (along with multi-threaded software) that the dual CPU's really became good competition for higher clocked single core CPU's.
I agree with the concern that both SLI and CF solutions are not ready for "prime-time" and are instead best for the early-adopters and enthusiasts that like to tinker and don't mind getting frequent driver updates and browsing forums for optimal driver configurations.
I liked that the article provided some factual commentary regarding the tradeoffs between single and multi-GPU setups.
Aside from that, I didn't notice a discussion regarding fan noise. Did I miss it somewhere, or is it not included in the article?
-D'oh!
ZootyGray - Tuesday, August 12, 2008 - link
I can be very sensitive to some strange things sometimes.While reading this, I had to stop. I was sensing great rage. That's pretty dark. And it was taking over me! Like evil something. I don't read reviews to have experiences like that.
Anyone else get a feeling like that? Maybe just me.
I met a guy once who was about 8 months clean from crak coc. He was full of rage. I had to stop talking to him too. It was like he wasn't even there. This just reminded me of that. Which makes no sense and the whole thing was pretty insane. But I wonder why I sense such anger in a hardware review that's supposedly unbiased. I always thought I would enjoy such work - must be stressful. Passing stress to an audience is not right either.
I don't know much about why; but my feelings don't lie. Anyone else get a sense of pain or anger? Maybe I got this from a different source. Maybe I am getting stressed from all the flame games. Maybe time to just let this go. Time to move on.
anonymous x - Tuesday, August 12, 2008 - link
i don't see any bias in this article- you can clearly see the performance benchmark scores, hardly any improvement over the other cards.Dark Legion - Wednesday, August 13, 2008 - link
If you don't see any bias in the article, you clearly only looked at the pretty pictures and not the actual writing.Also, why not include the test setup? For all we know, there could have been another limiting factor, especially since we saw very little gains from having a pair of 4870 X2's at such a high resolution. I remember back when the GX2 was coming out and you reviewed it, you used an Intel Skulltrail board because you could use both Crossfire and SLI in it, and you weren't limited by anything else. THAT is unbiased, and you can truly compare the cards that you're reviewing. Now you review the best performing competing card, and you don't even tell us what setup you used. Oh, and you don't even include the GX2, which is Nvidia's best performing card, and also happens to be a multi-GPU single card solution (which takes up as much energy as the 4870 X2, so don't only put AMD down for that). Did it not stack up well enough compared to this card for you? This review was horrible. I have come to expect this from DT (*cough cough Jason *cough cough), but not so much from Anand.
glynor - Tuesday, August 12, 2008 - link
I find it very interesting that the one solitary example of a game that doesn't scale well on the 4870 X2 is Assassin's Creed with the patch. Also, I didn't see a detailed explanation of what settings and setup were used to test the different games (were these custom timedemos? FRAPS? built-in benches?).Either way... It is interesting to say the least. Assassin's Creed posted numbers well under the GTX 280 in the Anandtech results. However, if you look at Tech Report's numbers which were generated using the pre-patch (DirectX 10.1 supporting) version of the game from before they took Nvidia money, the picture is dramatically different. Instead of losing to the GTX 280 (not to mention the GTX 260 SLI), the 4870 X2 easily bests the GTX 280 SLI (x2) setup in both average FPS and in Median Low FPS.
Just seems awful fishy to me. Overall this is good information. I'm particularly interested in the info on Sideport being disabled, as I'm not seeing similar information reflected in other reviews out there. I'm sure this is simply a case of Anand asking the right questions of the right people, but it'd be nice to see some independent confirmation.
Overall, it seems to me that this review does seem to have a slight (not over the top, but slight) anti-dual-GPU bias. It feels like more stock is put in the "failing" shown by Assassin's Creed, which is dubious at best, and no other evidence is shown to back this up. Surely, any dual GPU product may suffer optimization problems with new games, but wouldn't this apply to SLI equally (if not more -- most results I've seen show Crossfire scaling better than SLI more often than not)? I guess I just feel that the conclusions are being drawn based on results "not in evidence".
Discounting the outlier and contradicted Assassin's Creed results, I fail to see how the GTX 280 is in the same league at all as the new ATI dual-GPU card.
JarredWalton - Tuesday, August 12, 2008 - link
The games tested impact the results. As someone who has been running CrossFire HD 3870 for the past year or so (well, maybe more like 9 months?), plus someone who ran X1900/X1950 CrossFire before that, I can attest to the fact that CF support for new games is terrible. Basically, you get support in all major titles, but it's usually about two months after a game comes out. I've taken to not rushing to purchase new games, but that's okay since I'm busy of late.As for Assassin's Creed, the lack of performance with 4870X2 is odd and indicates perhaps a remaining driver issue for the new architecture. The game is definitely demanding of your CPU, but it should be running much faster. Maybe forcing on 4xAA (the game doesn't support 4xAA above 1680x1050) made the results worse than you would normally expect.
Personally, I am very cautious about recommending dual-GPU configurations for gamers - they're much better for benchmarks. Or at least, I would only recommend them for gamers that don't immediately buy the latest games and want top performance. GRID required updated drivers for CF, as did Mass Effect, Assassin's Creed, and pretty much every game I recall purchasing in the last two years.