AMD Introduces New APUs for System Builder and DIY Market

[btarunr]

AMD today introduced the new AMD A10-7800 Accelerated Processing Unit (APU) to the component channel. This 4th generation A-Series APU with 12 Compute Cores (4CPU + 8 GPU) unlocks the APU potential with Heterogeneous System Architecture (HSA) features, and boasts powerful AMD Radeon R7 Series graphics for outstanding performance across applications. Combined with AMD's acclaimed Mantle API, the AMD A10-7800 APU can enable accelerated performance across select AMD Gaming Evolved partner titles.

"The AMD A-Series APUs bring a superior level of gaming and compute experiences to the desktop PC," said Bernd Lienhard, corporate vice president and general manager, Client Business Unit, AMD. "With support for AMD's acclaimed Mantle API that simplifies game optimizations for programmers and developers to unlock unprecedented levels of gaming performance transforming the world of game development to help bring better, faster games to the PC."

Revolutionary Architecture

• Up to 12 Compute Cores (4 CPU + 8 GPU) deliver amazing performance and responsiveness, that leverages the capability of the APU on various workloads, applications and top gaming titles
• The breakthrough new Heterogeneous System Architecture (HSA) allows the CPU and GPU to work together in harmony by quickly dividing and directing tasks to the appropriate cores for new levels of performance and efficiency for desktop and notebook PCs, enabling a new era of compute capabilities with compute cores
• Award-winning Graphics Core Next Architecture with AMD Radeon R7 series graphics and compute capability makes it a top choice for gamers who expect the best
• Support for UltraHD (4K) resolutions and new video post-processing enhancements that make 1080p videos look amazing when upscaled on an UltraHD-enabled monitor or TV

Equipped with AMD Quick Stream, AMD Steady Video and AMD TrueAudio technologies, the AMD A-Series APUs combine to deliver lifelike audio and video4. Utilizing a configurable thermal design power option (cTDP), enthusiasts are now able to fine-tune for efficiency. In addition, AMD also announced the introduction of the AMD A8-7600 and AMD A6-7400K APUs, ideal for consumers looking to upgrade their application and work experience on the PC. With the unifying FM2+ infrastructure for AMD APUs, users are enabled to build smaller form factors for gaming and home theatre PC (HTPC) systems.

In addition with AMD Radeon memory, users can benefit from AMD AMP technology to help boost their gaming experience with the AMD Radeon R9 2400 Gamer Series memory which has been tested and certified for AMD A10 APUs.The AMD A-Series APUs are now available worldwide from select e-tailers, retailers and system integrators.



View at TechPowerUp Main Site

 

[newtekie1]

So there is nothing really new here, just a slightly underclocked and locked 7850K and 7700K. I guess the dual-core Kaveri is something, but meh...

 

[Andrew LB]

Notice how careful they're being about not giving us the 8-core? I bet there are agreements with Microsoft and Sony where they exclusively get those chips for at least a year.

 

[Sempron Guy]

finally! the 45w APUs

 

[brian111]

Notice how careful they're being about not giving us the 8-core? I bet there are agreements with Microsoft and Sony where they exclusively get those chips for at least a year.

This is a different architecture. Regardless, I don't believe the current motherboards are designed to handle that many cpu cores.

 

[TheinsanegamerN]

This is a different architecture. Regardless, I don't believe the current motherboards are designed to handle that many cpu cores.

How does that work, exactly? It can't handle eight cpu cores, but it CAN handle 4 cores and 512 graphic shaders, 32 texture units, and 16 ROPs? that sure makes sense......

 

[NC37]

As with any new idea...never buy the first gen. If the CPU can only handle 4 threads then what the heck is the point with giving it GPU cores to compute with? This is just stupid. I like the concept but the implementation just isn't there.

Wonder how long it will take before AMD gets this working the way it should? Probably not by 2nd gen. Maybe 3rd gen.

 

[Blín D'ñero]

[...]
If the CPU can only handle 4 threads then what the heck is the point [...]
.

4 CORES, not threads. 4 CPU cores + 8 GPU cores = 12 Compute Cores. CORES, not threads.

 

[TRWOV]

As with any new idea...never buy the first gen. If the CPU can only handle 4 threads then what the heck is the point with giving it GPU cores to compute with? This is just stupid. I like the concept but the implementation just isn't there.

Wonder how long it will take before AMD gets this working the way it should? Probably not by 2nd gen. Maybe 3rd gen.

The technology works right now, the softwate hasn't catched up.

 

[Constantine Yevseyev]

The technology works right now, the softwate hasn't catched up.

"The software hasn't catched up"? And what exactly do you want developers to do, huh? Windows handles the CPU cores as it should, and the GPU is used for GPU-related stuff only. There's no "progression" past this point.
Look, we only use GPU "cores" if we need to do something that is:

• Repetitive;
• Doesn't return any sort of information before the full compute cycle is complete;
• Can be effectively executed on many computing devices simultaneously.

See? There's no way you can make an average business app benefit from GPU. And I'm not willing to debug GPU-specific code library written in C++ (which is the only language you can implement your compute operations with) every time I have to make a string search while I'm developing a typical Windows app with C#. Yeah, it would be 12% faster if it's done with "8 GPU cores", but it's just not worth it. Even if you done everything perfectly, which is not always the case, of course.
Same goes to their legendary "not ready for FX-8000" thing. For gods' sake, we DO NOT specify in any part of our code the MAGICAL NUMBER OF CORES we're going to use, we leave this to the compiler. And it will use as many as you give; so will the runtime. If I was designing my app to be ran specifically on a supercomputer with 2880 logical CPU's, I would've implemented more of parallel stuff (like PLINQ), sure, but not when I expect the target system to have 8 of them. It's average, we don't need to handle it in a specific way.

 

[TRWOV]

I'm aware that not every kind of software can be parallelized, my comment was in regard to the "Wonder how long it will take before AMD gets this working the way it should?" comment. The technology works right now but there's only a handful of software that uses it. I'm not stupid enough to think that ALL the software should/could use HSA.

 

[Constantine Yevseyev]

there's only a handful of software that uses it.

And that's absolutely fine! In fact, that's how it should be done. There IS only a handful of software that can possibly be created with a support for GPU computing. Support, not even a guaranteed performance boost!
HSA is just a myth, it's a sweet concept for those who have never tried to write an actual GPU-accelerated software. You know, like when somebody calls "AVX 2.0" a technology, we know that it's just a set of instructions, right? That's exactly the case with HSA, "GPU Cores", whatever they come up with tomorrow. It's all about marketing and not real technology.
The way that HSA practically works is simple. There's a plenty of things that you can use to manipulate data inside your application ("Sort", "Subtract", etc), and AMD just provides you with their take on them (like, "Sort: AMD Edition"). That's it. It simply uses GPU instead of CPU for executing them. No one says that it would be faster, or more flexible, or anything like that. More then that, you can only use these "accelerated" things when they are allowed, and these cases are roughly 1%, really.