Hello and welcome to the NVIDIA AI Podcast.
I'm your host, Noah Kravitz. For 81 years, the name Hewlett Packard has been synonymous with computers.
The HP Garage in Palo Alto, California, where the company was founded in 1939, is now designated an official California historical landmark and is marked with a plaque calling it the birthplace of Silicon Valley.
So it's no surprise that today, July 2020, HP is making waves in AI.
Joining us today is Jared Dame. Jared is Director of AI and Data Science at Z by HP.
And he's here to talk about what HP is doing with workstations, with AI, and what that Z means.
Jared, thanks so much for joining the NVIDIA AI Podcast.
Thanks for having me on. So let's start at the beginning.
What's your role at HP? What's HP doing with AI?
And what's the Z all about? sure so why don't we start with z so z by hp is a sub brand of hp um and we're in my opinion we're the best part of hp So we're the workstations.
We have over 40 years of experience. So it's more of a name brand that we're trying to delineate in order to to call attention to some of the new products that we have coming out, especially around data science, AI, honestly, the new form of collaboration that our world's kind of pushing into.
So that's kind of our focus. Okay. My role at HP is to kind of direct the future of where we go with with data science, what strategic partners, not only with NVIDIA, but also with ISV partners, to provide the best experiences, creative experiences around data science, creative experiences around AI, that we can handle and deliver to basically the world at this point.
Yeah, we get to get to a lot of fun, get to play with a lot of brand new technology, also very well established existing technologies.
So it's never a dull moment. So let's dig for a second more into the Z name brand before talking more specifically about AI.
How old is the Z brand and does this encompass all workstations or is it kind of a more professional and even kind of hardcore computing?
How does, how does that delineate? Yeah.
So, uh, that's a really good question actually.
So that then the name Z we branded, uh, about two years ago, maybe two and a half years ago at that point. and what we wanted to do is separate our product lines that any of the lines that actually have ZBook or the Z2, Z1, Z, all the way to the powerhouse Z8, indicate the professional line of compute.
These are compute devices that have the most configurable settings from the memory component to the CPU to the GPU components that give you the most power.
They also have the largest level of customer support and the longest standing support.
So you get the three-year basically white glove support.
The machines work for well over three years.
I actually still have a Z840 at my house. that's about seven and a half years old now and still work amazingly well and actually still has a fairly up-to-date GV100 from NVIDIA that I use on a regular basis.
Z is about performance. Got it. Okay. And it's laptop and desktop form factors.
Correct. And so how does data science and folks working with everything under the AI umbrella work?
How does Z work with them? And maybe just tell us a little bit about the efforts that the z brand is undertaking to work with folks who are uh you know working on and using ai tools sure uh so We start with listening to the customer, the voice of the customer.
We gather a tremendous amount of insights.
In fact, we spend a majority of our time prior to the design of a lot of our newest additions to the lineup, which I'll be talking to you a little bit about that on the mobile lineup.
And these designs aren't the fruition of just HP thinking inside their own locker of, you know, this is what we think you need as the customer.
The customers actually tell us what they're wanting.
We go to design it. We pull up plans. We go back and talk to them, make sure that We're hitting the mark and these products are specifically designed to enhance a person's experience in their workforce or in their workflow.
And we're trying to specifically take away pain points that they've indicated.
You know, oh God, this takes me too much time to load or my traditional data science plan takes me 15 days to crunch numbers.
Do you guys have anything else? So they're starting to, they look to us for solutions and it's more of that consultative selling that HP does, but it's more of a, it's, bridged from consultative selling, which was a thing I think actually HP started in the past.
Don't correct me if I'm wrong. It's come more to a collaborative partnership with our customers.
So it's not about us selling a product, it's about us designing a product they need and filling those gaps.
And then as we get that product to them, they come up with more needs.
And it's just a matter of listening with the intent to learn. and listening to hear, not listening to respond with our customers.
Right. I would presume that HP sells to literally every sector you can think of.
Are there specific sectors that you've been working with recently that you know, either seeing a bigger push into needing AI specific solutions or maybe, you know, even more interestingly, ones that have kind of you've been working with that on something particularly exciting or just kind of you've noticed, yeah, this sector is really kind of pushing the envelope when it comes to you know, using the workstations in AI workflows.
Yeah, absolutely. What's amazing, HP, Z by HP is selling literally literally everywhere.
So every vertical market does data science.
Every vertical market is adopting various types of artificial intelligence, deep learning, into their workflows from anything from autonomous robotics to autonomous vehicles.
I mean, so it's everywhere. Specific industries that we're seeing the greatest kind of turn up in the last six months, Believe it or not, with all these trying times with the world going through its pandemic,
Oil and gas has taken a pretty hard hit, but we've actually had a number of customers come in to look at how do we augment what we're doing in various workflows from artificial lift to reserves and calculating reserves and estimation and production, mostly so that they can remain profitable and afloat until we get some semblance of the new norm back.
So they're looking to adopt a greater AI strategy and they're even doing something which I tout as data modernization from not only the data architecture, but all the way through.
Obviously, biopharmaceuticals, you can imagine right now.
Yeah. heavy adoption. But we've also seen a massive amount, and this has never really changed, the federal The federal government uses a ton of this.
State and local government use a ton of this type of technology. and they consume a lot of our Z by HP workstations, both mobile and desktop.
And yeah, I mean, it's all, I could go into every single vertical, but those three are the ones that are knocking my door down the most right now.
Yeah, no, it makes sense. When we talk about the workstation and kind of the AI workflow, we use that word a few times already, Can you maybe paint a picture in sort of broad generic terms or a specific example, if you like, of the role that the workstation itself plays in the AI pipeline?
Is it a matter of now it's like, oh, I've got the you know, the top of the line Z by HP workstation and this is it.
This is start to finish my whole project. or does it kind of fit into sort of a larger workflow?
How are the workstations positioned by the end users?
Sure. So as the joke would say, answers vary.
Right. But in general, the way I like to look at this and this is what I've been, I guess. evangelizing ever since I actually worked at SAS prior to coming to HP, and it was about finding a balance. of compute and there's a There's a number of meso trends out there that the IDC Gartner report on that talk about the explosion of data and the amount of data that we have exceeds the bandwidth and the latency. issues to be able to move data from one point to the other at will.
In fact, there's a data set I'm literally downloading onto my workstation right now that has been going for a little over an hour and a half. at this point and it just takes forever to move this data back and forth now.
So there's a prevalence with this. Workstations are uniquely positioned to do end-to-end, You can do absolutely everything on the workstation from your training to the inference engine from there.
And some customers in various verticals are doing that.
In others, they're balancing it out with their data centers, their cloud, instances and HP wants and we have designed these systems to work seamlessly within those environments.
We have a product called ZCentral, which actually enables a workstation to be not even at your desk.
It can be in a data center, it can be in a coat closet, and these things are designed to cool themselves so it does not need a massive chiller room in order to do that.
You can connect securely and work from pretty much anywhere from any device on these.
And what we talk about is having the right balance add the compute where it needs to do it because ultimately, if we go back to data science and AI, it's about determining what the answer is and then being able to have an actionable answer.
Something that a business can make a quick turn on if you know the answer, but actually can't. implement anything with it, there's no value that comes from knowing.
Sure. So one of the themes that's kind of emerged over the the past couple of years of doing this podcast and ranging from, you know, sort of DIY types who built something cool with AI, like in their own coat closet, so to speak to, you know, folks like you who are working at, uh, these larger organizations that are kind of setting the tone for the hardware or the software or the applications of AI kind of on a broader scale. everybody kind of speaks to both the increases or the acceleration of hardware power over the past, let's say five years And then a little bit more recently, the acceleration of the power and the availability of AI specific tools. as these two forces that have really helped fuel this AI heyday that we're kind of in.
And I'm wondering, and specifically the thing you said about being able to access the workstation.
It doesn't have to be where you are. It can be in the coat closet or wherever it is.
What have you seen as far as the impact of both the hardware but then also the surrounding tools? in really kind of fueling the advance of AI and related applications in the past couple of years since you've been working with Zee.
Sure. Actually, it's been a great time for being in this type of field and working in these types of with the types of hardware we do and with the software acceleration that's come out.
I mean, Rapids having come out last year has taken my own personal data science and absolutely launched it and accelerated it.
We just recently did a project with NASA Goddard, and you were talking about the implications of this.
So they're not only using a highly configured Z8 running an RTX 8000 Quadro, And they took a project, their business impact on literally detecting using AI to detect solar flares for basically what impact that was going to have on our atmosphere and potential health health implications of that.
They took a process that took them 90 days prior to using this advanced flow. using Rapids and these RTX cards down to 15 hours.
Wow. So if you consider that, I mean, that's just phenomenal that NASA is able to do that. if you take that and i have more time than we have on this podcast to go through um how many companies that can impact.
If you know something that your competitors will know in four days, but you know it now, What does that ROI look like to your company?
And talk about trading engines and whatnot.
They're all moving to GPU acceleration. any advantage they can get.
Can I make a trade nanosecond faster than the competitor? in which case like commodity markets, you're going to make a lot more money because you're going to be able to jump on something the minute you know you want to buy it.
There's just so many different advantages companies can leverage with this technology that's come out.
And the beauty is the data is rapidly just continuously expanding and growing and we're learning new ways to capture data.
And the tools that are coming out, the new startup companies that have come out and actually become I'd say they're actually standing on two legs now.
Not the fragility of the original one to two year phase, but partnerships that we have with OmniSci doing extremely well at taking billions of rows and records that you know, back in my earlier days, that was a dream, right?
I've never done something like that. I have to take small little slices of the pie And then extrapolate from there.
And it would still take weeks and months to get an answer that I can now get in a couple of mouse clicks with this acceleration.
It's just phenomenal. time to be in this industry.
Our guest today is Jared Dame. Jared is Director of AI and Data Science at Z by HP.
HP's high-performance computing sub-brand that was launched a couple of years ago.
We've been talking about the role of workstations in the AI workflow and all of the advances in hardware, obviously, but then the surrounding tools that have kind of been fueling all the cool AI stuff we talked about. on the podcast.
Jared, you mentioned, I think you used the phrase your early days, and we were talking before we hit record a little bit about your background, but Let's get into it for the audience, if you don't mind.
How did you get started working in data science and then AI?
Tell us a little bit about those early days.
So actually, to start with that, it's interesting.
So I went to actually a junior college for the first two years of my education, Cerro Coso Community College in Ridgecrest, California.
So it's a shout out to that original one. which was amazing because there's a really fantastic military base there.
And I actually got very interested in data when it came to organic chemistry, which was my original major.
Dr. John Stingersmith got me very, very interested in it and I was actually using the PC in the lab. to do data analytics, qualitative analytics, and he actually got me working in NMR. work on the base doing polymer studies, which was fantastic.
And this is junior college, right? So amazing opportunity for somebody in a two-year program.
Eventually transferred to Cal Lutheran University and Dr. Pong ran that program and it was a small data I wouldn't say data science program is a traditional computer science program.
But what I was more interested in was what data told us.
Then every job that I got from there was essentially interpreting various types of data from different types of signals.
I can remember when cell phones first came out and doing triangulation work with with the original cell phone companies and essentially did all of that and eventually landed a job at the SAS Institute a number of years ago.
And I call them the 800-pound gorilla in data science.
I found it in 1976, phenomenal products. and actually it was there that I started doing my first AI projects, feature identification. and it was actually on animals that we did that, and that was a side project, and I worked a lot in oil and gas,
Is there, not to stop you, but that first AI project, do you have a time?
Remember how long ago that was? That was 2013.
Okay. So yeah, and just started and actually fell in love with it from there.
And I've been, I dabble in it on the side and I obviously I
I found a great opportunity at the HP to come over here and completely immerse myself in hundreds of companies, adventures in AI from the starting crawl to the, you know, tentative steps to a couple of jogs.
I don't think anybody that I'm working with right now is at a full run outside of NVIDIA.
But yeah, it's just It's great. I love working on tons of different projects.
So that's kind of the journey. So to kind of get a little bit meta here, How, either in your own day-to-day role or, you know, you can kind of speak to other places in the company, but how is HP using AI internally these days?
Okay, so this actually is one of the differentiators of Z by HP.
Inside our machines and that computer you're looking at right now and using, there are hundreds, thousands of sensors. and both machine learning and artificial intelligence is being used to ensure the life expectancy of that machine.
We have something called DAS, which is Devices of Service.
Essentially what this is, service offering, but it's around our product, the Z product where we fill out your uh it with with every type of compute needs you have and the device as a service is actually built off of ai which will actually tell you a month or two before a drive goes bad.
Instead of you having to deal with loss of potential data, downtime of work, a drive will just show up on your IT manager's doorstep saying, So you plan maintenance.
Hey, why don't you go surf for the next four hours?
We're going to fix your PC for you. But a lot of that stuff from thermal sensors to all of that.
And then internally from manufacturing, we have... and I can't go into this super detailed, but we're drinking our own Kool-Aid, we use our own machines. and we're using the latest and greatest technologies from CUDA, TensorFlow, to the traditional programming languages to actually design AI accelerated for manufacturing, product quality control, and material science internally.
And so we're really adopting it fast. I mean, in some cases, we're able to go to customers that are doing very similar, let's say discrete manufacturing, and we're able to give them pointers on, this is where we failed and this is how we figured out how to do it right. we can save you 100 steps of trying to figure this out by showing you how we did it.
And then in other cases, we go to a customer and they've solved a problem that we haven't solved yet.
So, you know, it's all about relationships, right?
Absolutely. You've hinted, or no, I shouldn't say hinted, but it's come up a few times earlier this in our conversation talking about being able to access these workstations from anywhere.
And then just now you were mentioning the device as a service initiative.
And it makes me wonder, and both in the context of the pandemic and lots of people working from home and kind of you know, still to be determined how this is all gonna shake out and what offices will look like, you know, a year or two years, five years from now.
But then also just kind of more broadly, how are developers sharing access to workstation resources?
And I'm thinking, you know, particularly of the higher end workstations, that maybe folks can't afford to own on their own everybody who wants access to one, but might be able to access them in ways that they couldn't five, 10 years ago. but not necessarily limited to those.
I'm just wondering, are there any kind of interesting or surprising ways that devs are able to share HP Workstation resources these days?
Absolutely. It's a very good question. So going back to that zCentral, Zcentral concepts.
The Zcentral is, if anyone was familiar prior to remote graphics system, RGS, Zcentral is a real graphic system on steroids.
We've got an expanded platform. We've got a broker that allows you to do scheduling and asset allocation.
But essentially, if I'm talking with my hands, of course, no one can see.
I'm doing the same. I like the whiteboard, so no one can see that either.
Right. If you consider the part I was talking about earlier, the trend around massive data, being created and you need to compute close to that.
Zcentral is uniquely positioned for that because you can actually put the workstation alongside the data so you don't have to worry about long traffic times trying to get that data to and from.
What Z Central allows you to do is you can be on your ZBook 360, which is actually what I use, And it's a touch screen and it's a lot of fun.
And I can connect to this in DC. I can connect to this in Taipei.
And what it's doing is it's transferring just pixels back and forth.
So none of my data or my customer data Nothing that would pose potentially a security issue is passed back and forth.
So number one, hyper-secure. It's also built to where you can encapsulate your own IT, cybersecurity parameters around it, encapsulated in your own VPN, for added security and company compliance.
But in the pandemic, what we have found is a lot of customers weren't what I call COVID-19 compliant.
A lot of workers had to go home, but the work still needs to go on I mean, the company needs to maintain, and in companies where they are manufacturing medications and servicing the healthcare and the first responders, they need to be able to still connect to this stuff.
What we actually did is in our zCentral group, they were getting phone calls from our long-term HP customers asking, okay how do we leverage this in our infrastructure and we were giving out free licenses across the board if you have the Z by HP you already get an RGS license and it's a small nominal fee if you use a different platform.
Despite their infrastructure, maybe it's a mixed model of our competitor, plus us, we were more about, okay, we want to ensure that you can get your jobs done and to make sure your companies can move forward.
Zcentral actually enables you to put the workstation wherever it needs to be.
It enables you to have that experience that you're on the workstation.
You can watch a 4K video, over a cell phone connection from your laptop to cell phone connection to a Z8 which means you can do 3D rendering, you can do architecture work, you can do AI, image identification.
You can do anything you want over it and it doesn't need to be, you know, this fiber optic high speed internet that you only find in the major cities.
Yeah. That's, that's phenomenal. Speaking of the ZBook you mentioned that you're using right now, I think there is a new addition to the lineup that we wanted to touch on before we wrap up here.
Can you tell us about that? Yes, so this is coming in the fall.
It's the Z book studio and it's going to be equipped with quadro graphics and this data science laptops that will have Linux capabilities and compliance as well as obviously Windows 10. compliance as well.
32 gigs of memory, i9 and RTX 5000. A little bit of power.
A little bit of power, and this is a form factor.
It's small. It's light. And so you can carry it with you to the coffee shop or from the basement to the dining room, whatever.
Exactly. Whatever the times dictate. And you said that's slated to come out fall of 2020?
Correct. Very cool. And what's the screen size on that?
Or is it a range? It's 15 inch. Okay. Very cool.
So before we wrap up here, and there's a ton here. to talk about but but i think um you know without leaning too hard into massive changes that the pandemic or kind of has, have already wrought on us and, you know, going forward, kind of who knows when you think about the two or so years since you've, been in this role and then kind of look ahead to the next two or five years beyond that,
Obviously, we've touched about the ability to access these powerful workstations from wherever you are, and that seems like it's a theme that's gonna continue.
But what other things are on your radar going forward, whether specific to Z by HP or kind of broader to the industry? that you might want to leave the audience to think about as we ponder the next couple of years.
Sure. So obviously there's a lot of things I could say, but I think the most pertinent would be with the rise of data and data being classified as the new oil or the most renewable energy source that we have.
The need for compute to actually be at that edge of data generation is going to be what I feel will dominate the next five years and probably beyond.
But let's just crystal ball it out for five.
I think that HP is uniquely positioned with the current offering as well as things that we are going to unveil very soon around Edge compute and platforms to enable that.
I think you're going to see a lot more of a push in that because there needs to be, and I said this earlier, there needs to be a balance.
You're going to need to know where to use your workstation.
You're going to need to know where to use Z, going to need to know how to best optimize your cloud, best optimize your data storage. and when I say Edge, I do not mean just IoT.
IoT is part of that, but Edge is a more encompassing term to basically state where the data is being generated and those endpoint devices are going to have to be accelerated.
They're going to have to have a tremendous amount of compute in order to ensure businesses and corporations and education firms are able to get the most out of their data so that they can make the best decisions possible for whatever industry they do.
Well, I can't imagine that the world's output of data is going to slow down anytime soon.
So we'll wrap up this conversation and leave you to... you know, creating the machines that will allow us to keep up with the data flow.
But Jared, this has been a great conversation.
Thank you again for taking the time For folks who want to learn more about Z by HP in particular, and other things going on with AI at HP, Where on the internet should they go to start checking this out?
So you can find us on hp.com and also connect with us on social media.
So hp.com forward slash data science. And we're also on LinkedIn.
We have Twitter feeds and we're all over the social webs.
Excellent. Well, Jared Dame, again, thank you for the time.
All the best to you. Stay safe and healthy.
Good luck with the launch of the new ZBook and everything else you're doing.
Appreciate it. Stay safe out there. We'll see you soon.
Thank you. ¶¶ Thank you.
