AMD Opens New Global Design Center
AMD has opened a new design center as part of phase two of its transformation.
AMD said on Thursday that it has launched a new global design center in India. It features "world-class" lab facilities dedicated to furthering both software and hardware APU-focused innovations. The news arrives after AMD introduced a dual-core APU for fanless devices, the GX-210JA, with a maximum TDP of 6 watts and an expected average power usage of around 3 watts.
The company said on Thursday that its new design center is actually located at Raheja Mindspace, HITEC City, Madhapur, in the heart of Hyderabad's technology hub. Hyderabad is the capital and largest city of the southern Indian state of Andhra Pradesh.
"AMD is committed to providing our customers with innovative, tailored technology solutions that empower people and deliver exceptional experiences," said AMD president and CEO Rory Read. "Our Hyderabad Design Centre will play an important part in that mission as the team works in concert with our other design centers around the world to deliver AMD's next round of innovative products."
The new facility features 175,000 square feet of engineering labs, equipment and office space for hundreds of engineers. It joins the other design center AMD has established in Bangalore, and has sales offices located in New Delhi and Mumbai.
"Our design centers in both Hyderabad and Bangalore are key design and development hubs for our business," said Madhusudan Atre, corporate vice president, Design Engineering at AMD. "Like our talented engineering teams around the world, the engineers working in AMD's new Hyderabad Design Center are every bit as focused and committed to the sustained delivery of hardware and software innovations that can help drive the company's business forward."
Just weeks ago, Read said in the company's second quarter 2013 financial results that it has entered phase two of its "restructure, accelerate, and ultimately transform" realignment project. Now that the restructuring aspect is complete, AMD will be able to focus on accelerating its business in the second half of the year. The company may even return to profitability in the third quarter.
"Our focus on restructuring and transforming AMD resulted in improved financial results," Read said. "Our performance in the second quarter was driven by opportunities in our new high-growth and traditional PC businesses. Looking ahead, we will continue to deliver a strong value proposition to our established customers and also reach new customers as we diversify our business."
AMD said that for 3Q 2103, it expects to see revenue to increase 22 percent, plus or minus 3 percent, sequentially.
IPC improvements and GPU are not mutually exclusive.
Also the FX line doesn't have GPU.
How about focusing just on CPU's without built in GPU's that can compete against the core i7 in terms of ipc?
Because that would not be 'moving forward' (it would effectively mean focusing on an outdated premise).
The CPU in terms of single-threaded tasks is also falling out of relevance, and more and more tasks are being delegated to the GPU (not the CPU) due to the sheer computing power it has compared to the CPU.
Intel doesn't have a HSA/hUMA equivalent, and AMD made a smart move by focusing on that - still it focused on those features when software support was next to non-existent.
The only reason people focus on single-threaded tasks in the first place is because Intel dominates the market more or less and pays developers and manufacturers, but this is the last period of its usefulness (because everything is going multithreaded - which is where AMD is more than capable).
Also, people tried (via virtual machine) to switch CPU ID's, which effectively gave AMD (that registered as an Intel cpu) a pretty good performance advantage in games (WoW comes to mind). Intel under such circumstances performed lower (its results were very similar to AMD with only 10% to 20% difference going to Intel).
Point is, that it would appear software is optimized to take full advantage of Intel instructions while intentionally crippling AMD (to this day) which is what produces lower scores in benchmarks, various games (which seem to be more CPU bound), etc.
Bottom line is, until we have a properly written software that can fully take advantage of all instruction sets in both Intel and AMD, saying which one is 'noticeably better' or plainly 'superior' is pointless.
The market is saturated with bias - neutrality is effectively non-existent and it would appear that the more money company has, they can pay developers and manufacturers to shun the 'competition' and shift the market in their favor - and if this continues, AMD can probably make a CPU which is realistically 4x faster and better than Intel, but it would still register as slower if the software is not optimized to take advantage of the architecture or all of its instruction sets.
@6falchard Nice generalisation there, India has some the most qualified and highly skilled technical people in the world, AMD is not the first to tap into this. It's good enough for them but obviously not for top brass like yourself.
And it is not like the Brits I worked with impressed me with the quality of their work or professionalism, pot-kettle-black.
They still release CPUs...but their main focus is on APUs, which seems the right choice to me, especially since they can remain in the market and give Intel a run for their money at certain price points.
Right now the APU as AMD said it would is evolving take Llano to Richland every release has increased performance/watt and per core performance along with iGPU gains. APU's are not limited to DT silicon, mobile, low powered systems, servers and hybrids along with Custom silicon (ie: PS4, XBone) have far greater markets than traditional designs. If you have read anything AMD in the last 4 months you would realize that AMD are expanding markets which ultimately means a product line.
For the same reason Intel isn't really focusing on this area anymore, the money's elsewhere. Even when AMD was winning this race Intel simply used there monopoly status to freeze them out of the market. But working on a broader market they stand a much better chance of growing, unlike Intel which seems to be stuck in neutral at the moment.
Half the people in Intel are Indian, or Chinese. Same applies to AMD, Nvidia (their Tegra stuff), Microsoft, IBM, NASA, MIT, most of your Ivy League colleges, etc. Hotmail was made by an Indian. Lots of people in all of your tech companies are Indian. Kabini, Tamesh and Kaveri are/were designed in India.
Not that i really believe in what i'm going to say after this, but since you're choosing to be the stereotypical stupid American, i have to:
If Americans designed better chips, maybe AMD wouldn't have to outsource production?
/rant
Half the people in Intel are Indian, or Chinese. Same applies to AMD, Nvidia (their Tegra stuff), Microsoft, IBM, NASA, MIT, most of your Ivy League colleges, etc. Hotmail was made by an Indian. Lots of people in all of your tech companies are Indian. Kabini, Tamesh and Kaveri are/were designed in India.
Not that i really believe in what i'm going to say after this, but since you're choosing to be the stereotypical stupid American, i have to:
If Americans designed better chips, maybe AMD wouldn't have to outsource production?
/rant
it actually is really good and professional. india is actually one of the(if not THE) world leaders in nanotech research and engineering. india has many capable researchers that would rather stay at home rather than move to the US to pursue work in advanced engineering. salaries are also lower than in US. Im all for US companies keeping jobs in US, but for a company in AMDs situation(that of a company just getting by), saving on cost with as minimal effect on performance as possible is a necessary thing to do.
One of the reason may be Intel IPP (integrated Performance Primitives) library, which contains a lot of code optimized using SIMD instructions like SSE. By default, IPP works only on Intel processors. Still alternate processors like VIA or AMD do support those SSE instructions, and can run the IPP faster when it recognizes that processor supports SSE instructions. Many commercial software vendors use IPP in their applications, so this may be a reason of this benchmark effect.