Nvidia promises once again to let Arm keep its Switzerland-of-chips biz model – and even license some Nv GPU tech
(2020/10/07)
- Reference: 1602059592
- News link: https://www.theregister.co.uk/2020/10/07/nvidia_arm_plans/
- Source link:
Arm DevSummit Nvidia and Arm are confident their $40bn [1]proposed merger , set to be the largest deal of its kind in the semiconductor world, will be approved by regulators, and the duo look forward to a long and happy marriage.
Nv's big cheese Jensen Huang and Arm’s chief executive Simon Segars put on a united front for a virtual "fireside" chat on Monday, kicking off Arm's virtual [2]DevSummit conference.
Huang said he expected Nvidia's acquisition of Arm would take more than a year to gain watchdog approval. “As soon as we explain the rationale of the transaction, the regulators will realize that we are two complimentary companies,” he said.
“We’ll explain what we do, and where we belong in the ecosystem. We will create innovations that are good for the market, and good for customers.”
Huang insisted he'll let Arm keep its neutral, Switzerland-like position in the industry – licensing any chip for any customer, more or less, without picking favorites or sidelining rivals – and even wants Arm to license some of Nvidia’s own GPU technology.
Time for a virtual love affair: ESXi-Arm Fling flung onto the web for peeps to test drive with Raspberry Pi 4, other kit [3]READ MORE
“The genius of Arm is its invention of energy efficient CPU architecture combined with its business model of licensing and having the knowledge to create vast computing systems," Huang said.
"We have every intention of protecting it, nurturing it, and growing the business model. We value a network of partners that can take difficult designs and turn them into soft IP products. It’s something that I wish [Nvidia] could always do.”
Indeed, Huang announced he will make at least some of Nvidia's GPU and [4]DPU designs, and related software, available to customers via Arm's licensing platform. That will include at least some of Nvidia's deep-learning frameworks and algorithms for imaging and graphics, as well as its high-performance computing CUDA tech that taps into specialist hardware to accelerate computationally intensive simulations and such workloads.
Huang made the same points in his personal keynote for Nvidia’s virtual [5]GPU Technology Conference , which is coincidentally taking place at the same time as Arm DevSummit this week. Here’s his speech, just in case you missed it:
[6]Youtube Video
We note Nvidia already offers a deep-learning accelerator – from software down to the hardware design – [7]on GitHub .
Nvidia and Arm also talked up their partnership with Fujitsu, which is building what could be a record-breaking [8]supercomputer that uses Fujitsu's homegrown A64FX processors that are packed with 48 64-bit Armv8 cores. The pair also gave a hat tip to Ampere, a hyperscale cloud computing startup that designs data-center-level Arm-compatible processors, and Marvell, a semiconductor giant that uses Arm designs to build server-grade silicon.
“The [acquisition] is about expanding,” Segars said. “It’s about creating new technology and putting it in the hands of people that will do really cool stuff with it. I’m sure it’s going to take a while. The right thing is to scrutinize it, to make sure that process is done thoroughly. We will get approval.”
Last month, an [9]open letter directed at UK Prime Minister Boris Johnson was circulated by Arm co-founder Hermann Hauser, who raised concerns about Arm losing its ability to remain neutral if new owner Nvidia puts the squeeze on its customers.
Also at DevSummit, Arm launched [10]SystemReady , a certification program and standards-setting effort for Arm-based hardware. ®
Get our [11]Tech Resources
[1] https://www.theregister.com/2020/09/14/nvida_arm_acquisition/
[2] https://devsummit.arm.com/
[3] https://www.theregister.com/2020/10/07/vmware_esxi_arm/
[4] https://www.theregister.com/2020/09/25/smartnic_dpu/
[5] https://www.nvidia.com/en-us/gtc/
[6] https://www.youtube.com/watch?v=pzbhU4ttSvM
[7] https://github.com/nvdla
[8] https://www.theregister.com/2019/11/13/cray_fujitsu_arm_supercomputer/
[9] https://www.theregister.com/2020/09/14/arm_nvidia_deal_details/
[10] https://community.arm.com/developer/ip-products/processors/b/processors-ip-blog/posts/arm-systemready-where-software-just-works-across-a-diverse-ecosystem
[11] https://whitepapers.theregister.com/
Nv's big cheese Jensen Huang and Arm’s chief executive Simon Segars put on a united front for a virtual "fireside" chat on Monday, kicking off Arm's virtual [2]DevSummit conference.
Huang said he expected Nvidia's acquisition of Arm would take more than a year to gain watchdog approval. “As soon as we explain the rationale of the transaction, the regulators will realize that we are two complimentary companies,” he said.
“We’ll explain what we do, and where we belong in the ecosystem. We will create innovations that are good for the market, and good for customers.”
Huang insisted he'll let Arm keep its neutral, Switzerland-like position in the industry – licensing any chip for any customer, more or less, without picking favorites or sidelining rivals – and even wants Arm to license some of Nvidia’s own GPU technology.
Time for a virtual love affair: ESXi-Arm Fling flung onto the web for peeps to test drive with Raspberry Pi 4, other kit [3]READ MORE
“The genius of Arm is its invention of energy efficient CPU architecture combined with its business model of licensing and having the knowledge to create vast computing systems," Huang said.
"We have every intention of protecting it, nurturing it, and growing the business model. We value a network of partners that can take difficult designs and turn them into soft IP products. It’s something that I wish [Nvidia] could always do.”
Indeed, Huang announced he will make at least some of Nvidia's GPU and [4]DPU designs, and related software, available to customers via Arm's licensing platform. That will include at least some of Nvidia's deep-learning frameworks and algorithms for imaging and graphics, as well as its high-performance computing CUDA tech that taps into specialist hardware to accelerate computationally intensive simulations and such workloads.
Huang made the same points in his personal keynote for Nvidia’s virtual [5]GPU Technology Conference , which is coincidentally taking place at the same time as Arm DevSummit this week. Here’s his speech, just in case you missed it:
[6]Youtube Video
We note Nvidia already offers a deep-learning accelerator – from software down to the hardware design – [7]on GitHub .
Nvidia and Arm also talked up their partnership with Fujitsu, which is building what could be a record-breaking [8]supercomputer that uses Fujitsu's homegrown A64FX processors that are packed with 48 64-bit Armv8 cores. The pair also gave a hat tip to Ampere, a hyperscale cloud computing startup that designs data-center-level Arm-compatible processors, and Marvell, a semiconductor giant that uses Arm designs to build server-grade silicon.
“The [acquisition] is about expanding,” Segars said. “It’s about creating new technology and putting it in the hands of people that will do really cool stuff with it. I’m sure it’s going to take a while. The right thing is to scrutinize it, to make sure that process is done thoroughly. We will get approval.”
Last month, an [9]open letter directed at UK Prime Minister Boris Johnson was circulated by Arm co-founder Hermann Hauser, who raised concerns about Arm losing its ability to remain neutral if new owner Nvidia puts the squeeze on its customers.
Also at DevSummit, Arm launched [10]SystemReady , a certification program and standards-setting effort for Arm-based hardware. ®
Get our [11]Tech Resources
[1] https://www.theregister.com/2020/09/14/nvida_arm_acquisition/
[2] https://devsummit.arm.com/
[3] https://www.theregister.com/2020/10/07/vmware_esxi_arm/
[4] https://www.theregister.com/2020/09/25/smartnic_dpu/
[5] https://www.nvidia.com/en-us/gtc/
[6] https://www.youtube.com/watch?v=pzbhU4ttSvM
[7] https://github.com/nvdla
[8] https://www.theregister.com/2019/11/13/cray_fujitsu_arm_supercomputer/
[9] https://www.theregister.com/2020/09/14/arm_nvidia_deal_details/
[10] https://community.arm.com/developer/ip-products/processors/b/processors-ip-blog/posts/arm-systemready-where-software-just-works-across-a-diverse-ecosystem
[11] https://whitepapers.theregister.com/
What a difference a vowel makes...
navidier
They may be "complimentary companies", but sweet talking doesn't make a business model. Better they were complementary.
Simple...
IGotOut
...make it a watertight legally binding contract...including any name changes, mergers or future sales.
"We have every intention of protecting it, nurturing it, and growing the business model."
Even if present management has every such intention there's no guarantee that a future management will.
"We value a network of partners that can take difficult designs and turn them into soft IP products. It’s something that I wish [Nvidia] could always do.”
And what was stopping you?