Earlier this week, we discussed new test results for Ashes of the Singularity that showed GPUs from AMD and Nvidia running the gameside-by-side. Unfortunately, Oxide has made the decision not to make this build of the game public immediately. What we dohave, courtesy of Oxide, is video evidence of the ultimate Frankenstein rig running both AMD and Nvidia GPUs.
Note that in this case, the hardware obviously isn’t configured for any kind of SLI passthrough — AMD doesn’t use physical bridges the way Nvidia still does, and this kind of hardware wouldn’t be able to take advantage of it, since the two company’s use very different pinouts. Nevertheless, the rig works and works well.
Oxide’s benchmark test pairs a GTX 980 with a Fury Nano and notes that scaling is extremely good, though not quite perfect. While the video isn’t quite as gameplay-focused as we might like, the footage that exists runs without apparent problems. As we’ve noted, there’s very little that any ODM can do to prevent this mode from working. AMD and Nvidia can certainly try to encourage developers to prefer LDA (Linked Display Adapter), which mimics DX11 functionality, as opposed to the MDA mode shown here, which allows GPUs from multiple vendors to work together. But AMD and Nvidia can’t prevent it outright.
Some readers have asked whether or not PCI-E has enough bandwidth to make MDA plausible in all scenarios. This is a valid question, and we’ll need more data from other games before we have a guaranteed answer, but given that AMD stopped using bridge cables back in 2013, we suspect PCIe 3.0 has all the bandwidth necessary to feed even 4K multi-GPU configurations.
With PCI-Express 4.0 now not expected to debut until 2017 or 2018 due to the difficulty of achieving the standard’s 16GT/s transfer rate, it’ll be important to balance any new capabilities against what PCI-E 3.0 can provide. Then again, Ashes shows extremely strong scaling in the Nano + GTX 980 test, and a still-impressive 1.62x when the higher-end Fury X and GTX 980 Ti were used. Neither AMD nor Nvidia have fallen over themselves to publicly congratulate Oxide, so we’ll do it for them — gentlemen, we appreciate your efforts in making gaming more accessible to consumers who prefer to buy based on price/performance ratios rather than sticking to any single company.
Ashes of the Singularity is currently in Early Access on Steam. As with all EA titles, be aware that the game is still in early stages and that current gameplay or unit balance may not reflect the final retail product.
Yamaha’s new humanoid robot, Motobot, just learned to ride a motorbike and already it’s talking trash. At the recent Tokyo Motor Show where Motobot was unveiled, it announced that the reason it was created was to simply “surpass you.” Presumably, that means piloting a 1000cc Yamaha R1M around a racetrack at over 200kmph, and beating MotoGP champion Valentino Rossi.
Blathering on in a voice font crafted to ambiguous perfection, Motobot further inquires of us: “There has to be something only I am capable of?” Indeed there is one, and perhaps only one edge that Motobot now has on the track — it can’t die. While that advantage is a power that should never be underestimated in human-robot interaction, it’s going to take a lot more than courage to best a real rider around the track.
Without an impressive flagship robot, many Japanese companies might be devoid of vision. For nearly a decade, engineers everywhere have been inspired to greatness by the dance moves sported by Honda’s Asimo robot. Toyota’s equally impressive violin-playing robot delicately displays just how good predictive servo loops fed by 1,000+ encoder ticks-per-rev can be. However, actual sport usually entails a bit more than just cuing up a pre-programmed sequence or rant.
Although in theory you could program in the perfect track run as easily as you would plot a course in Pac-Man, the realty is that in any interesting endeavor, there will always remain enough unpredictability to wreak havoc. In other words, as the “undisputed truth” himself (boxer Mike Tyson) was often fond of saying — “every robot has a plan till it gets punched in the mouth.’ Yamaha’s approach of using a humanoid robot to autonomously pilot a largely unmodified motorbike probably isn’t the easiest way to do it, but there are certainly many merits to that conception.
At this point, it would seem that any new vehicle technology worth its salt would have at least some minimal capability for a default autonomous recovery mode in the event of human failure. As an example of this, consider that the mandatory retirement of pilots beyond a certain age, even those flying with co-pilots, is precisely scaled to the incidence of heart attack. If self-driving car technology is to be more than a fiction, then it might be fair to ask where is that minimal program and hardware that safely powers down a school bus when the heart of its aging veteran driver suddenly falters?
Instead of simple self-driving trains, planes, or boats, often the first vehicles to offer society at large some measure of autonomous control are those for which failure could be most catastrophic — and perhaps even likely. In other words, those cars with the most insane acceleration capabilities, even “ludicrous mode” powerplants, also tend to be the ones with the most capacity for self-governance. By contrast, we have $2 billion military blimps busting free from their moorings left and right without so much as the guidance capability of your smartphone on tap to bring it home. One needn’t know how to derive the equation for the volume of a blimp in order to realize that the textbook procedure for bringing its rampage to a halt — namely, shooting it full of holes — wastes ungodly amounts of strategic helium.
Yamaha isn’t the only one working on robobikes; Google and even private DIYers have all put forth different designs. No doubt in time a winning combination will be hit upon. And yet, the potential market for a self-driving motorcycle might appear rather grim. Nonetheless, the appeal of having all the brains for an autonomous control in a human footprint is hard to ignore. Not only could the same form factor potentially drive any number of different vehicles with only slight adjustments to its program, but the vehicles themselves would be dual-use ready right out of the box. The line between ‘light single engine craft’ and ‘drone’ instantly becomes blurred. By the same token, there is need to buy an expensive aftermarket retrofit for hands-free trawling on your Boston Whaler — just bring along your universal Motoboat buddy.
We’re being a bit hard on self-driving technology here, but there are several indications that the field is due for a dose of tough love. Much of the recent hype over how to imbue autonomous vehicles with ethics contains within itself many fallacies. For example, it sounds straightforward enough to ask whether a self-driving vehicle, whether car or bike, should imperil its single occupant to avoid 10 looming pedestrians. However, if your autonomous vehicle suddenly finds itself on an “unavoidable” collision course with such a crowd, it should and will be more than just the car that is surprised. The main falsehood herein is that a self-driving vehicle could ever properly determine the “greater good” — the problem simply dwarfs the problem of self-driving.
There is no foreseeable computational solution to a moral problem in infinite time, let alone the 150ms that Google says is how long their vehicles need to respond. The larger fiction inherent in this latency issue is that doing what we do 99% of the time using perhaps 1% of our ‘computational’ power means that autonomous vehicles, when called upon, can also do what we routinely must only by using all our power.
Software behemoth Microsoft is once again warning system admins that the end-of-life (EoL) date for its MS SQL Server 2005 is now just six months away.
The company sent out yesterday a reminder to customers that its ten-year-old database server edition will officially be considered out of support on April 12, 2016.
This simply means that SQL Server 2005 systems will no longer be able to receive free security updates from Microsoft after that date.
While extended support can be had at a specific cost, Microsoft would rather customers simply update from SQL Server 2005 to its most recent MS SQL Server 2014.
In sounding off the six month alarm, Microsoft made its sales pitch for upgrading to SQL Server 2014. Nothing new there.
Among the promised benefits of migrating from 2005 to 2014 are improved performance, better uptime stability, and better integration with Excel and Power BI.
Additionally, and this is where it really counts for the software giant, it also means nice new license revenues for Microsoft. That's several millions of new dollars in newly minted cash.
"In short, MS SQL Server 2014 delivers a high-performing, advanced data platform that keeps up with modern business needs," wrote Microsoft corporate vice president of enterprise marketing Takeshi Numoto.
Microsoft has long been warning enterprise customers of the looming end-of-life for SQL Server 2005, and trying to sell them on an upgrade to SQL Server 2014 in the process.
Back in April of this year when the one-year notice was posted, Microsoft made a similar announcement touting the virtues of the new 2014 edition.
Earlier this summer, Microsoft moved the upcoming SQL Server 2016 into its public preview trials. Its new SQL Server 2016 touts improved data security and this time, better integration with Azure, the company claims.
