AMD Launches New Mobile APU Lineup, Kabini Gets Tested 102
An anonymous reader writes "While everyone was glued to the Xbox One announcement, Nvidia GeForce GTX 780 launch, and Intel's pre-Haswell frenzy, it seems that AMD's launch was overlooked. On Wednesday, AMD launched its latest line of mobile APUs, codenamed Temash, Kabini, and Richland. Temash is targeted towards smaller touchscreen-based devices such as tablets and the various Windows 8 hybrid devices, and comes in dual-core A4 and A6 flavors. Kabini chips are intended for the low-end notebook market, and come in quad-core A4 and A6 models along with a dual-core E2. Richland includes quad-core A8 and A10 models, and is meant for higher-end notebooks — MSI is already on-board for the A10-5750M in their GX series of gaming notebooks. All three new APUs feature AMD HD 8000-series graphics. Tom's Hardware got a prototype notebook featuring the new quad-core A4-5000 with Radeon HD 8300 graphics, and benchmarked it versus a Pentium B960-based Acer Aspire V3 and a Core-i3-based HP Pavillion Sleekbook 15. While Kabini proves more efficient, and features more powerful graphics than the Pentium, it comes up short in CPU-heavy tasks. What's more, the Core-i3 matches the A4-5000 in power efficiency while its HD 4000 graphics completely outpace the APU."
Re:AMD should move into other areas (Score:4, Insightful)
Re:hUMA (Score:5, Insightful)
But even more puzzling to me is why both MSFT and Sony picked the absolute WEAKEST CHIP that AMD sells for their flagships...what the fuck?
Because of exactly what parent said:
AMD can provide unified memory (hUMA) with a decent GPU and a decent CPU on the same die. Intel cannot, nvidia cannot.
hUMA will not make your PC faster in general, but it will provide you with a feature, even a PC with 20 Geforce Titans does not have: Latency free data exchange between CPU and GPU.
It will make GPU processing more feasible especially on a small scale. I can't give you an example from gaming, but I can give you an example from my own expertise. When we simulate big proteins, we do it on a GPU. However, for small proteins, the latency overhead simply kills us. Processing on the GPU would be faster, but we need to copy back and forth all the time. We don't need faster GPUs, we need faster transfers. With hUMA: no problem.
Re:matches power consumption? (Score:3, Insightful)
None of the benchmarks have made an apples to apples comparision. Either they compare a 35W Pentium to the 15W Kabini, or it's an expensive Core i3/i5.
Core i3-3217U only appears in laptops costing more than $500. Kabini replaces Brazos which typically appears in cheap (sub $400) laptops.
Re:Heh (Score:2, Insightful)
"But we found notebooks with this exact CPU selling for as little as $360 on Newegg."
They found one notebook, which is a $650 model, on a temporary sale for $360. The cheapest i3 notebook with the CPU they are comparing, not on sale, is $525, and it's a shitty one.
The cheapest B960 laptop is also $400, which makes it quite a bit above the $300-$350 atom models that this will be competing with. Maybe they should have compared it with the standard $300 laptop and it's shitty 1.1ghz celeron.
Re:Heh (Score:4, Insightful)
You missread his post.
Your failure to understand the argument does not constitute a failure on my part to comprehend his comment.
He was asserting that single threaded performance is what matters on these laptops, because no one is going to use them for big number crunching tasks that can actually use multiple cores effectively. He's correct.
No, he's complextely wrong. What do you imagine that typical users need single-thread performance for? Most users need this only for games, and poorly-written ones to boot. PC games which require single-thread processing power are now vanishingly rare thanks at least in part to the influence of the tri-core Xbox 360, and the overwhelming tide of console to PC ports. Everything else the user typically does which requires very much CPU is already multithreaded. Most things the user does require virtually no CPU.
Running a GUI, editing files, I literally did these things on machines with single-digit MHz speeds which, when they were less responsive than using applications of today, were only so because of disk access times. And these tasks are today multithreaded, because they are based upon multithreaded libraries. Take a look at the programs running on a typical windows machine today, virtually all of them have a crapload of threads. Windows makes thread creation cheap in the way that Unix makes process creation cheap... not least because Windows is heavily multithreaded itself. And we are talking about what the majority of users will do with this hardware, which means running windows, playing the occasional game, watching cat videos on youtube.
Aside from games, the only times that most users use much CPU is during video encoding or possibly decoding, both of which are aggressively multithreaded and often even GPU accelerated, or while using graphics or video editing applications which are also typically heavily multithreaded, and have been for years. In short, practically no typical user actually needs serious single thread performance any more — what they need is good multithreaded performance, so that their computer can do a million pointless things behind the scenes without causing their cat videos to skip.
The Pentium beats the Brazos at single threaded performance, therefore, is a better chip for this kind of task.
The Pentium is only better than the new AMD cores we're talking about at the kind of task that people who buy APUs don't do. Thus, while your statement is factual, it is also irrelevant.