XDMA: Improving Crossfire

Over the past year or so a lot of noise has been made over AMD’s Crossfire scaling capabilities, and for good reason. With the evolution of frame capture tools such as FCAT it finally became possible to easily and objectively measure frame delivery patterns.  The results of course weren’t pretty for AMD, showcasing that Crossfire may have been generating plenty of frames, but in most cases it was doing a very poor job of delivering them.

AMD for their part doubled down on the situation and began rolling out improvements in a plan that would see Crossfire improved in multiple phases. Phase 1, deployed in August, saw a revised Crossfire frame pacing scheme implemented for single monitor resolutions (2560x1600 and below) which generally resolved AMD’s frame pacing in those scenarios. Phase 2, which is scheduled for next month, will address multi-monitor and high resolution scaling, which faces a different set of problems and requires a different set of fixes than what went into phase 1.

The fact that there’s even a phase 2 brings us to our next topic of discussion, which is a new hardware DMA engine in GCN 1.1 parts called XDMA. Being first utilized on Hawaii, XDMA is the final solution to AMD’s frame pacing woes, and in doing so it is redefining how Crossfire is implemented on 290X and future cards. Specifically, AMD is forgoing the Crossfire Bridge Interconnect (CFBI) entirely and moving all inter-GPU communication over the PCIe bus, with XDMA being the hardware engine that makes this both practical and efficient.

But before we get too far ahead of ourselves, it would be best to put the current Crossfire situation in context before discussing how XDMA deviates from it.

In AMD’s current CFBI implementation, which itself dates back to the X1900 generation, a CFBI link directly connects two GPUs and has 900MB/sec of bandwidth. In this setup the purpose of the CFBI link is to transfer completed frames to the master GPU for display purposes, and to so in a direct GPU-to-GPU manner to complete the job as quickly and efficiently as possible.

For single monitor configurations and today’s common resolutions the CFBI excels at its task. AMD’s software frame pacing algorithms aside, the CFBI has enough bandwidth to pass around complete 2560x1600 frames at over 60Hz, allowing the CFBI to handle the scenarios laid out in AMD’s phase 1 frame pacing fix.

The issue with the CFBI is that while it’s an efficient GPU-to-GPU link, it hasn’t been updated to keep up with the greater bandwidth demands generated by Eyefinity, and more recently 4K monitors. For a 3x1080p setup frames are now just shy of 20MB/each, and for a 4K setup frames are larger still at almost 24MB/each. With frames this large CFBI doesn’t have enough bandwidth to transfer them at high framerates – realistically you’d top out at 30Hz or so for 4K – requiring that AMD go over the PCIe bus for their existing cards.

Going over the PCIe bus is not in and of itself inherently a problem, but pre-GCN 1.1 hardware lacks any specialized hardware to help with the task. Without an efficient way to move frames, and specifically a way to DMA transfer frames directly between the cards without involving CPU time, AMD has to resort to much uglier methods of moving frames between the cards, which are in part responsible for the poor frame pacing we see today on Eyefinity/4K setups.

CFBI Crossfire At 4K: Still Dropping Frames

For GCN 1.1 and Hawaii in particular, AMD has chosen to solve this problem by continuing to use the PCIe bus, but by doing so with hardware dedicated to the task. Dubbed the XDMA engine, the purpose of this hardware is to allow CPU-free DMA based frame transfers between the GPUs, thereby allowing AMD to transfer frames over the PCIe bus without the ugliness and performance costs of doing so on pre-GCN 1.1 cards.

With that in mind, the specific role of the XDMA engine is relatively simple. Located within the display controller block (the final destination for all completed frames) the XDMA engine allows the display controllers within each Hawaii GPU to directly talk to each other and their associated memory ranges, bypassing the CPU and large chunks of the GPU entirely. Within that context the purpose of the XDMA engine is to be a dedicated DMA engine for the display controllers and nothing more. Frame transfers and frame presentations are still directed by the display controllers as before – which in turn are directed by the algorithms loaded up by AMD’s drivers – so the XDMA engine is not strictly speaking a standalone device, nor is it a hardware frame pacing device (which is something of a misnomer anyhow). Meanwhile this setup also allows AMD to implement their existing Crossfire frame pacing algorithms on the new hardware rather than starting from scratch, and of course to continue iterating on those algorithms as time goes on.

Of course by relying solely on the PCIe bus to transfer frames there are tradeoffs to be made, both for the better and for the worse. The benefits are of course the vast increase in memory bandwidth (PCIe 3.0 x16 has 16GB/sec available versus .9GB/sec for CFBI) not to mention allowing Crossfire to be implemented without those pesky Crossfire bridges. The downside to relying on the PCIe bus is that it’s not a dedicated, point-to-point connection between GPUs, and for that reason there will bandwidth contention, and the latency for using the PCIe bus will be higher than the CFBI. How much worse depends on the configuration; PCIe bridge chips for example can both improve and worsen latency depending on where in the chain the bridges and the GPUs are located, not to mention the generation and width of the PCIe link. But, as AMD tells us, any latency can be overcome by measuring it and thereby planning frame transfers around it to take the impact of latency into account.

Ultimately AMD’s goal with the XDMA engine is to make PCIe based Crossfire just as efficient, performant, and compatible as CFBI based Crossfire, and despite the initial concerns we had over the use of the PCIe bus, based on our test results AMD appears to have delivered on their promises.

The XDMA engine alone can’t eliminate the variation in frame times, but in its first implementation it’s already as good as CFBI in single monitor setups, and being free of the Eyefinity/4K frame pacing issues that still plague CFBI, is nothing short of a massive improvement over CFBI in those scenarios. True to their promises, AMD has delivered a PCie based Crossfire implementation that incurs no performance penalty versus CFBI, and on the whole fully and sufficiently resolves AMD’s outstanding frame pacing issues. The downside of course is that XDMA won’t help the 280X or other pre-GCN 1.1 cards, but at the very least going forward AMD finally has demonstrated that they have frame pacing fully under control.

On a side note, looking at our results it’s interesting to see that despite the general reuse of frame pacing algorithms, the XDMA Crossfire implementation doesn’t exhibit any of the distinct frame time plateaus that the CFBI implementation does. The plateaus were more an interesting artifact than a problem, but it does mean that AMD’s XDMA Crossfire implementation is much more “organic” like NVIDIA’s, rather than strictly enforcing a minimum frame time as appeared to be the case with CFBI.

Hawaii: Tahiti Refined PowerTune: Improved Flexibility & Fan Speed Throttling
POST A COMMENT

396 Comments

View All Comments

  • SolMiester - Monday, October 28, 2013 - link

    So you can OC a 780 on stock, but not the 290x to sustain the OC, which means 780 wins!, especially after the price drop to $500!, oh dear AMD 290x just went from hero to zero... Reply
  • TheJian - Friday, October 25, 2013 - link

    I gave links and named the games previously...See my post. At 1080p 780 trades blows depending on the games. Considering 98.75% of us are 1920x1200 or less, that is important and you get 3 AAA games with 780, on top of the fact that it's using far less watts, less noise and less heat. A simple drop in price of $50-100 and 780 seems like a no brainer to me (disregarding the 780TI which should keep the same price as now I'd guess). Granted Titan needs a dunk in price now too, which I'm sure will come or they'll just replace it with a full SMX up-clocked titan to keep that price. I'm guessing old titan just died as 780TI will likely beat it in nearly everything if the rumored clock speed and extra smx are true. They will have to release a new titan ULTRA or something with another smx or up the mhz to 1ghz or something. OR hopefully BOTH.

    I'm guessing it's easier to just up the 100mhz or put it to 1ghz as surely manufacturing has gotten them to where all will do this now, more than having all SMX's defect free. Then again if you have a bad SMX just turn a few more off and it's a 780TI anyway. They've had 8 months to either pile up cherry picked ones, or just improve totally anyway so more can do this easily. Clearly 780ti was just waiting in the wings already. They were just waiting to see 290x perf and estimates.
    Reply
  • eddieveenstra - Sunday, October 27, 2013 - link

    Titan died when 780gtx entered the room at 600 Euro. I'm betting Nvidia only brings a 780gtx ti and that's it. Titan goes EOL. Reply
  • anubis44 - Thursday, October 24, 2013 - link

    This is the reference card. It's not loud unless you set it to 'Uber' mode, and even then, HardOCP thought the max fan speed should be set to 100% rather than 55%. Imagine how quiet an Asus Direct CUIII or Gigabyte Windforce or Sapphire Toxic custom cooled R9 290x will be.

    Crossfire and frame pacing all working, and R9 290X crushes Titan in 4K gaming (read HardOCP's review of this 4K section), all while costing $100 less than GTX780, and the R9 280X (7970) is priced at $299, and the R9 270X (7870) is now going for $180, and now Mantle API could be the next 3dfx Glide, and boost all 7000-series cards and higher dramatically for free...

    It's like AMD just pulled out a light sabre and cut nVidia right in half while Jsen Hsun just stares dumbly at them in disbelief. He should have merged nVidia with AMD when he had the chance. Could be too late now.
    Reply
  • Shark321 - Thursday, October 24, 2013 - link

    There will be no custom cooling solution for the time being. It's the loudest card ever released. Twice as loud as 780/Titan in BF3 after 10 minutes of playing. Also Nvidia will bringt the 780Ti in 3 weeks, a faster cart at a comparable price, but quiet. AMD releases the 290x one year after NVidia, 2 years after NVidias tipeout. Nvidia will be able to counter this with a wink. Reply
  • just4U - Thursday, October 24, 2013 - link

    Shark Writes: "It's the loudest card ever released."

    Guess you weren't around for the Geforce5...
    Reply
  • HisDivineOrder - Thursday, October 24, 2013 - link

    The FX5800 is not ever dead. Not if we remember the shrill sound of its fans...

    ...or if the sound burned itself into our brains for all time.
    Reply
  • Samus - Friday, October 25, 2013 - link

    I think the 65nm GeForce 280 takes the cake for loudest card ever made. It was the first card with a blower. Reply
  • ninjaquick - Thursday, October 24, 2013 - link

    lol, the Ti can only do so much, there is no smaller node for either company to jump to, not until March for enough shipments to have stock for sales. The 290X just proves AMD's GCN design is a keeper. It is getting massively throttled by heat and still manages to pull a slight lead over the titan, at sometimes 15% lower clocks than reference. AMD needed a brand for this release season, and they have it.

    Both Nvidia and AMD are jumping to the next node in 2014. Nvidia will not release Maxwell on the current node. And there is no other node they would invest in going to.
    Reply
  • HisDivineOrder - Thursday, October 24, 2013 - link

    The Ti could theoretically open up all the disabled parts of the current GK110 part. Doing that, who knows what might happen? We've yet to see a fully enabled GK110. I suspect that might eat away some of the Titan's efficiency advantage, though. Reply

Log in

Don't have an account? Sign up now