Update: May 18, 2015:
HBM represents the revolutionary step that has been so badly needed in the evolution of memory standards. The first generation of HBM promises to deliver 4.5X the bandwidth of GDDR5 and a staggering 16 times the bandwidth of DDR3.
Nvidia will be using 3D stacked memory on its GPUs starting with Pascal in 2016, a full year after AMD. Volta was Nvidia’s scheduled GPU for 2015 that was supposed to use the Hybrid Memory Cube or HMC for short which was s 3D stacked memory technology competing with HBM. However after HMC’s development fell behind the roadmap it put a lot of pressure on Nvidia to look for an alternative technology, which they did. What’s painfully ironic however is that Nvidia ended up using the AMD and SK Hynix co-developed HBM with Pascal. Which the company confirmed earlier this year. Nvidia very rarely uses a technology developed by their competition if ever.
AMD bet on HBM early on, while Nvidia bet on the competing HMC standard. It seems AMD bet on the right technology and they stand to gain an entire year’s worth of exclusive access to the technology. Nvidia will simply have to make due with GDDR5 for 2015.
UPDATE: May 7, 2015:
AMD analyst day in 3 years, provide details of x86 “ZEN” core with 40% IPC per core, that is a big architecture improvement. Giving their lower power consumption borrowed/learned from ARM64, if the 16-cores rumor is correct, AMD should come back to top after 10 years leading the x86_64 debut. AMD made the correct decision to get rid of SeaMicro and refocus on its strength (interconnect, server, GPU). For the past 2 years, AMD’s every step is correct which increases our trust to management team.
AMD representing the advantage of both x86 and ARM64, represent the idealistic architecture and design. If US is to continue its dominance on CPU and future computing, AMD will be the future and Wall-street and US gov should provide meaningful support, unless we want other country to use ARM64 to democratize CPU industry of the future.
It makes great sense for Google, Qcom, Amazon, HPQ, Apple to buy AMD, to strength and accelerate its R&D activities. Thus position US in another decade of dominance in CPUs.
April 30, 2015, TopCool Investment Report.
A lot of our progress as a human specie will depend on computation power of tomorrow, which should be hundreds of times faster. Who will win the long match will make your chip investment boom or bust, as there will be dramatic fight ahead. as dimension reach a limit and only if people choose the optimal design will eventually win.
You might have wonder why ARM A57 powered phone is very fast(Samsung S6), but both Extremetech and U of University of Wisconsin – Madison had try to prove there is no difference between ARM and x86 architecture. Moreover, on April 23rd, 2015, ARM is claiming A72 is going to be 70% faster than A57 (20~60% faster IPC (instructions-per-clock) while using 18~30% less power). Mike Filippo, ARM’s chief architect for Cortex A72, said: “we are certainly sure that, running at full tilt, we’re going to use 75 percent less power [than Broadwell] or “less than 1/3 of the power for equivalent performance”(see wording in the following ARM slide):
16nm finfet A72 Comparing to Intel mobile processor of similar finfet at 14nm:
Whom you are going to believe? who will win the ISA war? who (AMCC, AMD, Nvidia, Cavium, BRCM, QCOM vs. Intel) will eventually win in server space if ARM conclusion is true?
Our research report will address the following issue:
1. Is the above ARM A72 claim credible? how much discount on ARM A72?
2. Intel’s path in the next 3~6 years.
3. Is AMD’s dump of SeaMicro a good idea or not? why?
4. what is the difference between server and mobile CPU design, pricing, performance and power consumption.
5. who (AMCC, AMD, Nvidia, Cavium, BRCM, QCOM vs. Intel) will eventually win in server (and CPU space) if ARM conclusion is true?
6. Will the GPGPU (CPU+GPU) computing change the CPU performance landscape?
7. what will 4K HD impact the computing space?
There are details which none of those public articles or company spoken person talk about, but will playing a determine role to long term failure and success. Remember industry will have huge lead than academics and a lot of secret is unspoken and misleading.
If you have or plan to invest in CPU design space, this report is a “must” read report as it is from the frontier-line first-hand of research lab knowledge on power, performance, x86 and ARM64, server vs. Mobile architecture and ISA. Thing is not what U of WISC or ExtremeTech said. (e.g. answer to one of the question is: x86 has architecture/ISA disadvantage and only one of them will win in server given long enough time frame; AMD dump SeaMicro is a good idea).
For detail report and pricing, contact email@example.com. Our author is long time Silicon valley architect who most recently spent 6 years in Intel doing AMD, Intel, ARM performance and competitive research and analysis; and 1 year in Cavium 48-core-Thunder-ARM64. With deep knowledge of all the above competing players.
Add to favorites