In a quest to deliver better application performance and greater efficiency, many software and hardware vendors are turning to custom silicon to achieve their goals.
Apple’s A-and M-series processors are prime examples. Meanwhile in the cloud, Amazon has spent the last few years developing custom CPUs, AI accelerators, and smartNICs, to drive down their costs. Now Intel’s making a big play for the custom silicon market.
Not every independent software vendor (ISV) or hardware vendor has the resources, expertise, and deep pockets necessary to blaze their own silicon trail. In fact, most software vendors aren’t all that great at optimizing their apps for the hardware they’re already using, Intel’s Jeremy Rader tells The Register.
Rader, Intel’s GM of enterprise strategy and solutions under its Data Center & AI Group (DCAI), argues that the direction forward for Intel isn’t just to sell a better, faster chip year after year – though that’s still part of it. Instead, it’s to engineer processors better optimized for customers’ workloads, and to make it easier for them to bake support for those features into their software.
“If you don’t understand the end-customer challenge, then all you’re going to be doing is showing up with a box,” he said of the company’s biggest competitors.
Here, Rader says Intel’s more than 20,000 software engineers can help their ISV partners get more out of their silicon, and ultimately deliver better performance and efficiency to their end customers.
“There are so many software companies entering the market, and they’re all looking for guidance on how they should design and take advantage of not just CPU – CPU is the workhorse, it’s the foundation – but they want to understand where does the FPGA play? If they’re not in the cloud, how does the IPU play into this as you look at trying to offload critical tasks or overhead tasks from the CPU,” he said.
Rader’s argument essentially boils to down this: software vendors aren’t optimizing their platforms to take full advantage of the silicon, let alone the litany of specialized accelerators now flooding the market.
- Why the end of Optane is bad news for all IT
- China’s 7nm chip surprise reveals more than Beijing might like
- Intel experiences another kind of meltdown
- Too little, too late: Intel’s legacy is eroding
“We’ve worked with software vendors that weren’t taking advantage of some very simple things that are part of our CPU architecture,” he said. “The types of companies that we work with – these are the big, top-tier software vendors, but also the startups and disruptors – they may have great software expertise, but they tend not to have hardware expertise.”
Here, Rader says Intel has a chance to help, but in order to do that, it first needs to make it simpler to take advantage of the features baked into its silicon, and second, it has to make it easier for them to understand where each piece of Intel’s portfolio fits into that puzzle.
Done right, “instead of it costing X to your end customer, you can bring that down to Y, because you’re running more efficiently,” he said.
Intel’s acquisition of cloud optimization startup Granulate earlier this year is key to this strategy. The software platform is pitched as a means to optimize workloads for modern CPU core architectures without requiring changes to the code base. The silicon stalwart says the software has helped customers reduce infrastructure costs by 40-60 percent.
Of course, Rader admits there’s nothing stopping the chipmaker’s competitors from embracing a similar philosophy. In fact, several, including Apple and Nvidia, arguably have a head start.
Nvidia in particular has not only expanded its hardware portfolio to include GPUs, CPUs, and networking, the company has built out an expansive software library with the express purpose of making it easier for customers to employ its tech in a variety of HPC, data visualization, graphics, and AI/ML workloads.
Rader opines that Intel, as one of the largest chipmakers on the market, is better positioned to get this right.
“The difference is, we’re bringing some talent in some sort of mass and scale,” he said. “It really gives us a chance to home in on where we can bring value and do it as quickly as possible.” ®
In a quest to deliver better application performance and greater efficiency, many software and hardware vendors are turning to custom silicon to achieve their goals.
Apple’s A-and M-series processors are prime examples. Meanwhile in the cloud, Amazon has spent the last few years developing custom CPUs, AI accelerators, and smartNICs, to drive down their costs. Now Intel’s making a big play for the custom silicon market.
Not every independent software vendor (ISV) or hardware vendor has the resources, expertise, and deep pockets necessary to blaze their own silicon trail. In fact, most software vendors aren’t all that great at optimizing their apps for the hardware they’re already using, Intel’s Jeremy Rader tells The Register.
Rader, Intel’s GM of enterprise strategy and solutions under its Data Center & AI Group (DCAI), argues that the direction forward for Intel isn’t just to sell a better, faster chip year after year – though that’s still part of it. Instead, it’s to engineer processors better optimized for customers’ workloads, and to make it easier for them to bake support for those features into their software.
“If you don’t understand the end-customer challenge, then all you’re going to be doing is showing up with a box,” he said of the company’s biggest competitors.
Here, Rader says Intel’s more than 20,000 software engineers can help their ISV partners get more out of their silicon, and ultimately deliver better performance and efficiency to their end customers.
“There are so many software companies entering the market, and they’re all looking for guidance on how they should design and take advantage of not just CPU – CPU is the workhorse, it’s the foundation – but they want to understand where does the FPGA play? If they’re not in the cloud, how does the IPU play into this as you look at trying to offload critical tasks or overhead tasks from the CPU,” he said.
Rader’s argument essentially boils to down this: software vendors aren’t optimizing their platforms to take full advantage of the silicon, let alone the litany of specialized accelerators now flooding the market.
- Why the end of Optane is bad news for all IT
- China’s 7nm chip surprise reveals more than Beijing might like
- Intel experiences another kind of meltdown
- Too little, too late: Intel’s legacy is eroding
“We’ve worked with software vendors that weren’t taking advantage of some very simple things that are part of our CPU architecture,” he said. “The types of companies that we work with – these are the big, top-tier software vendors, but also the startups and disruptors – they may have great software expertise, but they tend not to have hardware expertise.”
Here, Rader says Intel has a chance to help, but in order to do that, it first needs to make it simpler to take advantage of the features baked into its silicon, and second, it has to make it easier for them to understand where each piece of Intel’s portfolio fits into that puzzle.
Done right, “instead of it costing X to your end customer, you can bring that down to Y, because you’re running more efficiently,” he said.
Intel’s acquisition of cloud optimization startup Granulate earlier this year is key to this strategy. The software platform is pitched as a means to optimize workloads for modern CPU core architectures without requiring changes to the code base. The silicon stalwart says the software has helped customers reduce infrastructure costs by 40-60 percent.
Of course, Rader admits there’s nothing stopping the chipmaker’s competitors from embracing a similar philosophy. In fact, several, including Apple and Nvidia, arguably have a head start.
Nvidia in particular has not only expanded its hardware portfolio to include GPUs, CPUs, and networking, the company has built out an expansive software library with the express purpose of making it easier for customers to employ its tech in a variety of HPC, data visualization, graphics, and AI/ML workloads.
Rader opines that Intel, as one of the largest chipmakers on the market, is better positioned to get this right.
“The difference is, we’re bringing some talent in some sort of mass and scale,” he said. “It really gives us a chance to home in on where we can bring value and do it as quickly as possible.” ®
In a quest to deliver better application performance and greater efficiency, many software and hardware vendors are turning to custom silicon to achieve their goals.
Apple’s A-and M-series processors are prime examples. Meanwhile in the cloud, Amazon has spent the last few years developing custom CPUs, AI accelerators, and smartNICs, to drive down their costs. Now Intel’s making a big play for the custom silicon market.
Not every independent software vendor (ISV) or hardware vendor has the resources, expertise, and deep pockets necessary to blaze their own silicon trail. In fact, most software vendors aren’t all that great at optimizing their apps for the hardware they’re already using, Intel’s Jeremy Rader tells The Register.
Rader, Intel’s GM of enterprise strategy and solutions under its Data Center & AI Group (DCAI), argues that the direction forward for Intel isn’t just to sell a better, faster chip year after year – though that’s still part of it. Instead, it’s to engineer processors better optimized for customers’ workloads, and to make it easier for them to bake support for those features into their software.
“If you don’t understand the end-customer challenge, then all you’re going to be doing is showing up with a box,” he said of the company’s biggest competitors.
Here, Rader says Intel’s more than 20,000 software engineers can help their ISV partners get more out of their silicon, and ultimately deliver better performance and efficiency to their end customers.
“There are so many software companies entering the market, and they’re all looking for guidance on how they should design and take advantage of not just CPU – CPU is the workhorse, it’s the foundation – but they want to understand where does the FPGA play? If they’re not in the cloud, how does the IPU play into this as you look at trying to offload critical tasks or overhead tasks from the CPU,” he said.
Rader’s argument essentially boils to down this: software vendors aren’t optimizing their platforms to take full advantage of the silicon, let alone the litany of specialized accelerators now flooding the market.
- Why the end of Optane is bad news for all IT
- China’s 7nm chip surprise reveals more than Beijing might like
- Intel experiences another kind of meltdown
- Too little, too late: Intel’s legacy is eroding
“We’ve worked with software vendors that weren’t taking advantage of some very simple things that are part of our CPU architecture,” he said. “The types of companies that we work with – these are the big, top-tier software vendors, but also the startups and disruptors – they may have great software expertise, but they tend not to have hardware expertise.”
Here, Rader says Intel has a chance to help, but in order to do that, it first needs to make it simpler to take advantage of the features baked into its silicon, and second, it has to make it easier for them to understand where each piece of Intel’s portfolio fits into that puzzle.
Done right, “instead of it costing X to your end customer, you can bring that down to Y, because you’re running more efficiently,” he said.
Intel’s acquisition of cloud optimization startup Granulate earlier this year is key to this strategy. The software platform is pitched as a means to optimize workloads for modern CPU core architectures without requiring changes to the code base. The silicon stalwart says the software has helped customers reduce infrastructure costs by 40-60 percent.
Of course, Rader admits there’s nothing stopping the chipmaker’s competitors from embracing a similar philosophy. In fact, several, including Apple and Nvidia, arguably have a head start.
Nvidia in particular has not only expanded its hardware portfolio to include GPUs, CPUs, and networking, the company has built out an expansive software library with the express purpose of making it easier for customers to employ its tech in a variety of HPC, data visualization, graphics, and AI/ML workloads.
Rader opines that Intel, as one of the largest chipmakers on the market, is better positioned to get this right.
“The difference is, we’re bringing some talent in some sort of mass and scale,” he said. “It really gives us a chance to home in on where we can bring value and do it as quickly as possible.” ®
In a quest to deliver better application performance and greater efficiency, many software and hardware vendors are turning to custom silicon to achieve their goals.
Apple’s A-and M-series processors are prime examples. Meanwhile in the cloud, Amazon has spent the last few years developing custom CPUs, AI accelerators, and smartNICs, to drive down their costs. Now Intel’s making a big play for the custom silicon market.
Not every independent software vendor (ISV) or hardware vendor has the resources, expertise, and deep pockets necessary to blaze their own silicon trail. In fact, most software vendors aren’t all that great at optimizing their apps for the hardware they’re already using, Intel’s Jeremy Rader tells The Register.
Rader, Intel’s GM of enterprise strategy and solutions under its Data Center & AI Group (DCAI), argues that the direction forward for Intel isn’t just to sell a better, faster chip year after year – though that’s still part of it. Instead, it’s to engineer processors better optimized for customers’ workloads, and to make it easier for them to bake support for those features into their software.
“If you don’t understand the end-customer challenge, then all you’re going to be doing is showing up with a box,” he said of the company’s biggest competitors.
Here, Rader says Intel’s more than 20,000 software engineers can help their ISV partners get more out of their silicon, and ultimately deliver better performance and efficiency to their end customers.
“There are so many software companies entering the market, and they’re all looking for guidance on how they should design and take advantage of not just CPU – CPU is the workhorse, it’s the foundation – but they want to understand where does the FPGA play? If they’re not in the cloud, how does the IPU play into this as you look at trying to offload critical tasks or overhead tasks from the CPU,” he said.
Rader’s argument essentially boils to down this: software vendors aren’t optimizing their platforms to take full advantage of the silicon, let alone the litany of specialized accelerators now flooding the market.
- Why the end of Optane is bad news for all IT
- China’s 7nm chip surprise reveals more than Beijing might like
- Intel experiences another kind of meltdown
- Too little, too late: Intel’s legacy is eroding
“We’ve worked with software vendors that weren’t taking advantage of some very simple things that are part of our CPU architecture,” he said. “The types of companies that we work with – these are the big, top-tier software vendors, but also the startups and disruptors – they may have great software expertise, but they tend not to have hardware expertise.”
Here, Rader says Intel has a chance to help, but in order to do that, it first needs to make it simpler to take advantage of the features baked into its silicon, and second, it has to make it easier for them to understand where each piece of Intel’s portfolio fits into that puzzle.
Done right, “instead of it costing X to your end customer, you can bring that down to Y, because you’re running more efficiently,” he said.
Intel’s acquisition of cloud optimization startup Granulate earlier this year is key to this strategy. The software platform is pitched as a means to optimize workloads for modern CPU core architectures without requiring changes to the code base. The silicon stalwart says the software has helped customers reduce infrastructure costs by 40-60 percent.
Of course, Rader admits there’s nothing stopping the chipmaker’s competitors from embracing a similar philosophy. In fact, several, including Apple and Nvidia, arguably have a head start.
Nvidia in particular has not only expanded its hardware portfolio to include GPUs, CPUs, and networking, the company has built out an expansive software library with the express purpose of making it easier for customers to employ its tech in a variety of HPC, data visualization, graphics, and AI/ML workloads.
Rader opines that Intel, as one of the largest chipmakers on the market, is better positioned to get this right.
“The difference is, we’re bringing some talent in some sort of mass and scale,” he said. “It really gives us a chance to home in on where we can bring value and do it as quickly as possible.” ®
