Conclusion
Conclusion
Going into this review, we really didn’t think there would be all that much to get excited about and those predictions proved correct. The A10-7800 is certainly a capable APU in its own right with Steamroller bringing in some much-needed performance improvements over previous generation parts. However, the competition at this end of the CPU market is cutthroat and AMD’s latest addition does tend to fall flat in some key areas.
One major benefit of Kaveri APUs is of course their overall cost in relation to the alternatives. Intel’s Haswell architecture simply doesn’t have a quad core offering in this price range, a fact that makes the A10-7800 jump out as a value front-runner despite a continual presence near the bottom of our performance charts. It is the clear winner in onboard graphics processing and pulls significantly ahead when OpenCL comes into play. With that being said, all of you budding gamers out there should remember many dual core Haswell chips will walk all over this APU when either solution is paired up with a capable dedicated GPU.
While it may provide an excellent option for anyone buying a prebuilt system, the A10-7800 is something of a dead end for DIYers. It may be about $20 less expensive than the A10-7850K but many of us would be much better off jumping onto the unlocked multiplier bandwagon since it provides some peace of mind without spending money to upgrade components in the future. As we said previously, the A10-7800 is perfect for system builders which is why AMD will likely see the vast majority of this APU’s volume going towards the likes HP and ASUS rather than the processor-in-a-box market. That’s good news considering a quick glance at the available options at big box retailers shows
a distinct lack of Kaveri offerings, even after half a year of availability.
In terms of overall platform costs the A10-7800 and FM2+ couldn’t be better positioned. Technologies like Dual Graphics, Mantle, TrueAudio and minor revisions to the platform’s storage capabilities have brought it up to date in most situations. The only stumbling block for some buyers may be the lack of more advanced features like SATA Express and M.2 SSD support but with every next generation addition comes associated costs. For the time being AMD’s ecosystem doesn’t need those capabilities since it is primarily focused on delivering good performance at an outstanding value rather than being a lab rat for expensive new hardware testing.
We can talk about costs till we’re blue in the face but there are some stark pieces of reality that need to be discussed as well. Here we are, seven months after Kaveri’s initial unveiling and more than a year after AMD’s Heterogeneous System Architecture’s key components (hQ and hUMA) were announced and thus far there aren't any applications that support them. Kaveri actually launched with those features disabled and they remain in that exact same locked down state today. For early adopters who were counting on this technology to give APU’s a leg up on the competition (and it can….supposedly) but have yet to see any glimmer of hope, that’s a bitter pill to swallow. In the past we chalked up this lack of back-end application support to the relative newness of the underlying feature sets and gave Kaveri a pass in the hope that some forwards progress would be realized.
At this point, the time for playing a wait and see game is over: AMD needs to have HSA up and working with a full stable of developer engagement before they can properly market those same technologies in upcoming architectures. Even marketing current APUs like the A10-7800 is becoming challenging since potentially defining elements remain nothing more than talking points. Some help is on the horizon though. As OpenCL 2.0 begins rolling out later this year, both Corel AfterShot 2 and Libre Office will boast support for hUMA and hQ. That is one small step towards broader acceptance but one giant leap for HSA's founding principles.
There’s a reason why AMD desperately needs all the help they can get. Despite their graphics superiority and even with Steamroller’s notable IPC improvements over Piledriver and Bulldozer, Kaveri’s general processing performance in everyday applications is still in a decrepit state. If we rolled back the clock to 2011 and told you then that AMD’s 2014 lineup would barely outperform a three year old X4 980 in games and many productivity-based benchmarks, pitchfork-wielding mobs would have turned up at our doors. However, that’s exactly what is happening.
HSA was meant to be the saving grace of the APU dream but after three successive generations we have yet to see anything concrete materialize. That’s worrying but according to AMD change is on the way. Simply put, they can throw as much GPU horsepower as they want at their upcoming APUs but x86 performance will remain a key factor for years to come. Better OpenCL GPGPU support can somewhat mitigate these shortfalls but it will never be able to completely make up for them.
The A10-7800 is a well-placed addition to the 2014 APU lineup even though it won’t make any massive waves in the industry. Due to a relatively low price and an abundance of capabilities it should become a darling for large system builders but its appeal within the retail market may be limited when compared against AMD’s growing stable of unlocked parts. The real concern here is that until HSA features can be leveraged for better processing efficiency, AMD’s massive graphics strengths are currently the only thing keeping the A10-7800’s head above water. With Intel investing so heavily in substantial graphics improvements of their own for Broadwell and future architectures, there’s a desperate need to find something else that can help APUs distinguish themselves now and in the future.
