UT07x01: Proving the Performance of Solidigm SSDs at StorageReview – Utilizing Tech

Analysts and press spend a lot of time talking about specs and performance numbers. So it's always a treat when we get to talk to people who are testing and using these products in the real world. And when I say in the real world, I literally mean out there in the world.

As you're gonna hear this time, this episode of Utilizing Tech is focused on AI data infrastructure and features Jordan Rono from Storage Review and is co-hosted by myself and a Stryker from our sponsor. Soy. Welcome to Utilizing Tech, the podcast about emerging technology from Tech Field Day, part of the Futureum Group.

This season is presented by soy and focuses on AI data infrastructure. I'm your host, Steven FoST, organizer of Tech Field Day, and joining me today as my co-host is Ace Stryker of Soy. Ace.

Welcome back. Hey, Steven, thank you so much. I'm very excited to be here and looking forward to this season of the podcast with you.

Absolutely. It's gonna be a lot of fun. Um, we're gonna be able to bring in a whole bunch of folks all season long from different customers, different practical applications, and that's kinda what we're talking about today, right?

Ace, I mean, you know, we hear a lot about speeds and feeds. We see a lot of numbers, we see a lot of bragging from vendors, but really nothing matters until this stuff is out there in the field and being used, right? Yeah, I'm really excited about, uh, the guests we've got lined up today.

You know, in a lot of these AI conversations, what we sort of hear is the industry talking to itself, right? And we hear various, uh, solution providers and hardware and software vendors sort of hawking their wares and, uh, extolling the virtues of, of, of their products. But where it gets really exciting for me, uh, is to be able to hear from folks, uh, who are out in the field, who are, who are, um, using these things in the real world, uh, where the rubber meets the road and, and producing really cool results.

And so, uh, in that spirit, um, I'm, I'm excited about the conversation today. Yeah, totally. And, um, and although they're not, I guess strictly speaking, an end user, um, one of the sites that I love, one of the, you know, the companies that, that has this incredible video content that I enjoy is storage review.

I love running to see you guys at, uh, at the show, uh, Jordan, um, you guys are always playing with the coolest toys. Welcome to the show. Yeah, thanks for having me on.

Uh, I'm Jordan, I'm from Storage Review. I, uh, more officially, my title is the Advanced Workload Specialist. So anything kind of advanced, uh, whether it's AI or HPC, it's my job to take all the cool toys that we have in our lab, put 'em all together and make 'em do fun stuff.

And that's really what we're talking about here. So this season of, um, utilizing Tech is focused on AI data infrastructure. You're the guy that's out there taking these things, trying 'em out, trying to see what these, uh, systems will do in terms of AI data infrastructure.

So I guess, um, give us a, a a little bit of background in terms of what level of, um, of cool, of coolness have you put together with, uh, solid I Mrs. D? Yeah.

So Soys been a, uh, longtime friend of storage review. Uh, we got together with them not too long ago and had a fun idea for some as, uh, ACE alluded to where a rubber meets the road, almost literally in this scenario. Uh, plan to take some of these high capacity, pretty quick SSDs, uh, and stick 'em out in the field and do some real field AI work with them.

Um, we came up with a pretty neat concept around astrophotography, which is one of my personal favorite hobbies. And we filled up a Dell XR 76 20 with four of the, uh, solid Im, uh, QC SSDs and took it out to the frozen wilderness and shot some pretty incredible space pictures with it, and we're able to take that data and bring it back to our lab and do some AI work with them. So, Jordan, what's the, uh, what's the connection there?

Where does AI play into this? You know, I've, I've been out there and I've, I've done a little stargazing. I've got a little telescope that the kids and I, you know, look through at home.

Obviously the images, uh, you're producing are much higher quality. Is it, is it the AI that enables, uh, that level of quality? Or how, how are you using AI models to kind of develop your outputs?

Yep. So there's a couple different ways that we were taking advantage of it. Uh, the first aspect, we're using the high capacity SSDs to capture all of the data.

Our images are about 62 megapixels each raw, and that's before you add color and chroma data to 'em. Um, from there we were able to use that and fine tune man manually go through the data and comb through it and get a really good subset of images, uh, combine that with some actual Hubble legacy data as well, and run them through a novel convolutional neural network to create a more advanced, uh, modern, uh, CNN based de-noise and sharpening algorithm that outperforms the traditional kind of Richard Richardson Lucy algorithms that you see out there. Because of the, uh, amount of data and the amount of time that we were able to spend out in the field, it helped us really kind of drive that model forward, which we can then bring back out into the field and do edge inferencing on the images real time to see if we're having some other issue that may be, um, we're slightly outta focus or there's a little too much do on the lens, or there's a vibration happening from a car drive in by, and we need to know about that so we can make an adjustment if there's something, you know, stray lights, uh, being able to kind of, this is a, a, a very hyper-specific task where we focused in and we're able to take that time to actionable result and shrink that down by using, uh, a neural network in order to help make decisions in real time out in the field.

To add a little bit of additional color to that, um, traditionally, you know, you would, uh, take a, take a photo and you could spend hours processing through it and, and working through the, the sharpening or the, uh, the processing and the stacking, and only to find out that the entire weekend that you went out on, on, on your camping trip and took all of your photos, uh, it just didn't work out for some reason because you were slightly out of focus. We're helping to drive methods that can bring, like I said, that actionable, uh, real time feedback into the, the scenario. And this has a lot of applications beyond just the astro photography side.

Uh, any sort of imaging is actually compatible with the, uh, the network that we put together. It was just specifically that we trained it on the space photos because that was something that was near and dear to my heart and something that I'm fairly decent at, or at least I like to think to, um, in order to, uh, prove out the idea and flesh it out. It sounds to me like you've basically created sort of a, an edge, uh, computing workload or, or test case there because, you know, you've got, uh, poor connectivity, uh, bad environmentals maybe, and, um, and yet, uh, some high, high throughput, uh, ai, uh, applications that are running there.

So, I mean, if, if you, if you sub sub out telescope and, and stars, you know, you could easily think that this similar situation could apply in energy, uh, medical, uh, military, all sorts of different ca areas, and yet you might not be able to talk about those use cases because they're sensitive. Whereas what you're doing is something that you can try out and experiment with and play with in a way that, that is wide open, that you really can have that conversation, right? Right.

The stars aren't going anywhere, and if we mess up the ai, it's not gonna turn off, you know, Andromeda on us. But, um, yeah, the, the couple of the big things that come to mind, like you said, the other industries, self-driving cars, providing more real time clar image clarification, uh, to the cameras, uh, that are needed on a self-driving car, for instance, or exploration. Uh, oceanic exploration is another one where you're, uh, capturing a lot of data and you're may need to move quickly.

And having the ability to go back and clean it up in a meaningful way is really important because of the value and the amount of costs associated with that data acquisition. Jordan, a lot of what we're exploring, uh, in this, uh, season of the podcast is around data infrastructure and what are the, um, kind of requirements, uh, needed to support, uh, use cases like, like the one you're talking about. So we talked a little bit about, um, what's happening at the edge and the inferencing that's going on there.

Can you speak a little bit to, uh, the upfront work of developing the model and sort of what your infrastructure, uh, requirements look like there, um, uh, maybe particularly around the, the storage and, and kind of take us through that process of, of developing the model that you ended up using? Yeah. So without getting too nitty gritty into the math of it, there was a lot of backend work that happened in our lab.

Initially. We were doing some training on some super micro, uh, blade systems. We were doing some distributed training, and that involved putting together some really fast SSDs to be able to work in the data in and out of the GPUs as fast as possible.

Our dataset wasn't exceptionally large, but we ended up requiring large amounts of VAM due to the nature of, uh, the image rec, uh, the image processing that was actually happening. We were, I think at, uh, over 2000 layers for the neural network at one point, which was able to eat up enough VAM to span four H one hundreds. So getting that data in and out of the GPUs, specifically the model check pointing and being able to take all of those different checkpoints and save them out so we can go back after the fact and assess the performance was one of the keys that we needed.

And, uh, we had selected some of the faster SSDs that were maybe a little smaller on the capacity side 'cause we're only working with, uh, 320 gigabytes worth of VAM that we had to, uh, then condense out and flush out for the checkpoints. But the, the key there was the speed in getting the GPUs, 'cause they, we only had 'em for a short amount of time and getting the data back out of the GPUs and getting those checkpoints saved. So what's, what's next for you in the, in the world of astrophotography?

Is this a project that, that continues to live on here? I've seen the images on, on the storage view site. They're gorgeous.

And, uh, and I I I, I highly encourage our listeners to go check those out. But I'm curious about, uh, are there, are there bigger challenges you're looking to tackle, uh, in the astrophotography realm specifically, as well as, uh, are there ways for others with similar interests to leverage some of the work that you've done recently in the space? Yeah, so the next, the next goal immediately on the horizon is to finish up the paper with, uh, my partner in crime on this model and, and get that pushed across the line.

And then actually open source this model out for folks to be able to use in some of the, uh, existing software that's out there right now. And really put it in the hands of the developers who can help make those, uh, implementations and some of the open source capture software happen and help give the realtime feedback to the users. The whole idea here is to help improve the hobby as a whole and not to, um, you know, not get too hyper-focused on one specific thing, but just help everybody out with the, uh, with the network.

'cause at the end of the day, it is actually relatively small to run. And being able to take all this information and all this time and data that we spent capturing it and condense it into a really small, really efficient model, I think that's one of the more, um, rewarding aspects of doing this project. So What did you find, um, in terms of, uh, you know, what were some of the nuts and bolts, nitty gritty kind of the fun, the fun things that you found when you were experimenting with this setup?

Yeah, so we had, uh, I'll touch back on my Edge server that we used for this. Uh, Dell sent over a XR 76 20, which is, there's one of the ruggedized platforms. And we had it in a big, uh, uh, ruggedized road case that was impact resistant and weatherproof and shock resistant.

And when we were looking at kit it out and kind of building everything up, obviously A GPU was on the list. We ended up with, uh, Nvidia L four in there. Uh, they ate a Lovelace, uh, it's a 70 watt card, or 75 watt card, I believe.

But it's a 24 gigabytes of V RAM to be able to work with. So we were able to do some kind of realtime work at the edge. Uh, but as far as, as far as fine tuning goes, but when we started looking at storage solution, what's generally provided, we found to not be quick enough for what we were trying to do.

Traditionally, when you're going out to capture this stuff, you're gonna be, you know, a guy with a laptop and a camera and a USB cable connected to your scope. You don't need all this overkill stuff. But when we start getting into, we need to collect a lot of data, we need to look at it in real time, we need to be able to make decisions on is this, are we doing what we're doing properly?

So we are capturing the best data to provide the best training data. All those factors combined led us to needing a lot more higher throughput from both the CPU perspective as well as our storage perspective. So we did go with four of the udot two NVME SSDs.

Uh, we had two flavors of the, um, of the, uh, P 53 30 sixes. 68 terabyte, uh, drives in there. Um, both of which performed great.

The main reason why we ended up going with these, especially for the high density, was actually the old Sneakernet story. Um, getting the data back to the data center for larger processing later. It was more efficient for me to fill up one of these 60 terabyte drives and drop it in a FedEx mailer and have it two day back to the lab where I could then interact with it on, uh, bigger big iron, so to speak, as far as the H one hundreds and the bigger AI machines go then to try and upload it over limited bandwidth.

'cause we actually ended up doing this out in the middle of nowhere on, uh, the, the banks of the Great Lakes. Uh, you're someone who, um, uh, from, from my point of view, very much has their finger on, on the pulse of what's going on, uh, in the AI world by virtue of, of your work and, and, and your personal kind of, uh, uh, pursuits as well. I'm curious about this general trend of, uh, more and more AI work, you know, moving closer to the data, moving closer to the edge.

You mentioned in your case, you know, there's a benefit of having a better sense of how you're doing while you're out in the field, right? And sort of what your outputs are gonna look like so you don't, uh, find out down the road that you've ended up with, with nothing usable. Can you talk about maybe other ways you're seeing that trend manifest across, across the, the AI landscape today?

You know, in the, in the course of your work and your interactions with, with lots of folks across the industry. Uh, how, how prominent is this trend and, and how quickly, uh, are you seeing, uh, AI work moving from kind of the core data center where a lot of it has lived, uh, historically, uh, to where more intensive, uh, work is happening at the edge? Yeah, so that, I mean, your, your question almost is kind of the holy grail in what everybody's searching for right now.

The feedback that I'm hearing going to the shows and interacting with the viewers and, and the people who ask questions about these products coming to me for answers are, I've got my data center full of GPUs, and I think I have an AI model, and now I needed to deploy it to 1500 retail locations. Or actually just the other day, this was kind of a fun one. I went through a McDonald's drive-through that had an AI ordering for the first time.

That was kind of a wild experience actually. But the, the name of the game for everybody is, we've got all this data, we've got it right now, very centralized, and, and a lot of folks are taking the, the kind of the similar steps that we did with this project, which was, okay, let's keep that data out at the edge, let's inference on it out there. And then feedback just the key metrics into either the data center or into the, um, you know, management interfaces, real-time dashboards, that sort of stuff.

And just getting it down to that real granular level really quick. Traditionally, with the big data stuff working in, you know, large databases, a lot of those things have to deal with ETL jobs. They take a long time.

Whereas if you can process that stuff and save that important data out at the edge, even though you might not be able to send it all the way back to the data center right away at the time of capture, having that real time performance, um, at the edge in order to be able to make a decision on it as a business is where things are going. And I think it's what a lot of people already see is, this is what we need to do with ai and this is how AI can actually help the business. We don't have to move truckloads of data.

We can move metadata or, or a few values or, or a few model outputs, um, for, for analysis and then worry about moving the heavy, the heavy stuff later. Yeah, that's something I really wanna zoom in on there, Jordan, because you're completely in line with what we heard when we did. Uh, so previous seasons of utilizing tech, we focused on ai, uh, we talked about industrial OT and, and industrial vision applications.

We talked about media production. Um, we, and, and, you know, all of these industrial iot applications, that's exactly what they're looking for. They're looking for a way to move processing to the edge, not just data collection, but move processing to the edge to use AI as a way to collect, not just, um, you know, not just to process data, but to collect more data, to process more data and to get more value at the edge and then ship back the anomalies or the interesting aspects, the interesting elements.

We're seeing that more and more, I think in a lot of these edge use cases. I mean, think about, you know, pretty much any kind of, uh, IOT vision environment, they're collecting all sorts of camera and sensor data all the time. The last thing they want to do is just be writing all that data and then taking it back home.

Instead, what they wanna do is they wanna have intelligent processing of that data and then take the good bits back home. And similarly, a lot of the things that you're describing to me, it sounds like, uh, a lot of the same constraints that people are facing at the edge, um, you know, you're, you're, you're talking about ruggedized servers, you're talking about power. I think that's another aspect here too.

If, if you're gonna have a fairly high powered GPU, uh, you need to think about cooling, you need to think about power requirements, you need to think about, um, adverse conditions that these things may be installed in back of a pickup truck. Pretty adverse, but actually not that wild when it comes to things like, uh, energy exploration or military applications or things like that. Uh, that's actually a pretty nice environment compared to what the military faces.

And, and also, you know, if, if you've got something like SSDs that have large capacity, large performance, and a low power footprint, that really helps. I mean, I know that your stuff, for example, is, is battery powered, right? And, um, and, and so you're, you know, having the ability to reduce that power envelope gives you hours more, um, processing time, right?

Yeah. I mean, the difference of 25 or 50 watts, uh, that you would have to spread out across spinning rust, uh, versus consolidating it into a single large capacity SSD at the edge, certainly that can mean the difference of hours depending on your system, uh, especially if you're loading up the gpu. You brought up something quite quite interesting there, though.

We did a, a, a demo with soy actually back in, uh, at FMS in, uh, 2023 that we did just that, where we had a, a camera up and we were capturing all 10 80 P 60 FPS of the camera and saving it down to large capacity hard drives that could then be yanked out, shipped back to your data center. But we are doing real time inferencing on it. It's a lot cheaper in more than one way, not just financially to send a single line of text back 60 times a second than a, than a whole video frame or a whole 4K frame even.

So, when we look at that, that was kind of a, our proof of concepts of this idea that everyone's been talking about, that everyone's been saying, this is what AI can do for you. This is how you can use it. You can do that edge inferencing and take back the data, but you don't have to throw away.

That's where the, the, the larger storage stuff comes in. You don't have to throw away that raw data because that's valuable. At the end of the day, everybody's data, they wanna save every bit and bite of it, because who knows what the next evolution of AI is gonna look like.

Oh, you can just point it at a drive full of stuff and it'll figure it all out for you, maybe, but without saving that, if you were just throwing it away because you didn't have enough storage or your storage wasn't fast enough to keep up, right? That's where you kind of get the double-edged benefit of things like these, these huge QLC SSDs at the edge. And you mentioned the rugged thing.

I've got a little fun anecdote. Uh, I I, there there's a video floating around on the internet somewhere of me actually running these things in a blizzard. And I had, I had sent it to, uh, my PR friend at soy, and I think the first question was, what's the temperature?

And the response was, I'm pretty sure we're not rated for that. But they were fine. They were, they were rugged enough they could handle it.

Um, you know, that's not something where things with moving parts would necessarily be able to, uh, to handle it, but we, we had to go through it as part of our capture when a storm ran through and the, the rugged hardware, uh, chugged right through it. Jordan, I wanna, I wanna pivot a little bit here, um, because one question that a lot of folks are asking, uh, we heard it at, uh, GTC earlier this year, uh, we've heard it in various forums and conversations with various customers, is how do you evaluate, uh, storage performance for ai? Right?

And I know this is an area where you've done a lot of work, so I'd love to pick your brain here a little bit. Um, you know, in, in the, where I go, you know, as a, as a PC guy is if I, if I want to sort of measure CPU performance, i, I run syn a bench, right? Uh, and I can make an apples to apples comparison between a couple of processors, or I can run PC mark on a system level.

Uh, and there's a lot of interest in understanding how that's done, uh, for, um, you know, storage for AI in the data center specifically. Um, so what's, what's your view of kind of the state of things in that space now? Are there emerging tools that are designed to, um, uh, serve that purpose?

How well do they work? What are you, what are you learning there? And what do you see, uh, as, as recommendations for folks with an interest in kind of understanding that a little better?

I thought we talked about no loaded questions. Um, so there, there's, there's a lot to unfold there, right? So if we think through our, our phases of AI training, right?

You've got data ingest, data prep, uh, the actual training that goes into it, checkpointing, and then out to inferencing once you're done with your model. So when we look at those five different phases, there's different needs for each of those phases that when you take that into consideration, you don't wanna have to go, maybe necessarily buy out five different SKUs of an SSD to fill up your data center, but you need to thoughtfully design all the way through the, the platform that you need for your business, right? Every, every, every AI is not gonna be the next LAMA or chat GPT or, or Dolly, you know, it's every, everything's gonna be unique.

And so what we do right now in our lab is we're taking a look at a lot of different types of storage, uh, Q-L-C-T-L-C, um, throwing in cash layers, looking at using, utilizing the CPU and the DRAM, going GPU direct and kind of profiling these different workloads that we're seeing coming out, uh, whether it's synthetic benchmarks or open source projects, looking at how those are actually impacting the system, how they're treating the desk, how they're, um, working with the system memory with the CPU, and then mimicking that using some open source tools like G-D-S-I-O, for example, is really powerful if you know what kind of AI you're going to use. 'cause there's 15,000, or I'm sorry, 16, no 17,000, no 18,000 now kinds of ai, it's always changing. But if you have a general idea of what you're gonna be doing, you can actually go out and profile and set up, you know, these, these tests.

And that's what we're aiming to do with kind of a, a, a script that we've been working on to look at storage from a, a, a total perspective and kind of Gantt chart it out and say, okay, this type of device or this type of, um, you know, this specific MVME drive is really good if you're doing this type of ai, but if you're doing this kind, you need to look at a storage appliance like this. Or if you are doing, um, you know, another kind of AI shoving your GPU server full of as much MVME as you can get in there, is the way to go and then let the CPU worry about offloading those checkpoints over your network at a later, at a later date. When we see stuff like, um, Nvidia just came out with the, uh, and I've got a video on this there, 800 gigabit networking, we're starting to see that be less and less of a bottleneck, but then the storage servers and the storage appliances are gonna have to start to catch up to that too.

Um, we were doing a test recently, I can saturate, uh, 200 gigabit Ether Ether or 200 gigabit in finna band rather of very easily with a single A 100 GPU. And then we start talking about H one hundreds and now the Blackwell stuff's, um, on its way, uh, you're, you're gonna need those bigger, faster, stronger stuff. And that's where Gen five gets really exciting, uh, as well as gen six.

Uh, I know gen five, right? When we look at the speed on there, the name of the game at the end of the day is to keep the GPU working as fast or as fast as possible, as well as as nonstop as possible. Everybody knows that selecting your storage infrastructure and what you're gonna be doing around that is something that's getting more and more focused pretty much every day right Now that we need a way to, if we're doing really read heavy stuff because we're doing ETLs, um, but we need to, we're using, uh, a NEMO the NEMO framework that does the real time augmentation of the data.

So we're not, you know, over overfitting our model or something like that. Um, we need a lot of random 4K read performance, um, you know, and so there's, there's layers to it, right? And that's kind of what we're aiming to do at storage review at least, is provide the, maybe not the textbook, but give out the, the playbook for folks to be able to profile their storage or make a, make a smart decision on either their existing infrastructure or some infrastructure that they might be looking at and say, this is where this fits, and this is why it's in important for that.

You need the density above all for this reason, or you need the gen five speed above all for this reason. And, and providing that is, I think that's kind of what everybody's looking for right now. So it's, yeah, it's a loaded question.

I think we're, I think we're onto it as far as getting these GSIO stuff up and going, getting our own, uh, FIO scripts, because you can mimic a lot of these workloads in FIO. There's multiple ways to skin this, uh, and, and, and to, to make it happen. Um, but I think we're on the right track.

It's, it's interesting you mentioned this, Jordan, because I've actually, uh, found that it is actually, this equipment is getting so fast and so good that it's actually kind of difficult to, uh, to push it, you know? Mm-hmm. I mean, how do you test a 60 terabyte SSD that can push, you know, gigabyte gigabytes per second of throughput?

How do you test a processor that has, uh, you know, over a hundred cores? How do you test, you know, terabytes of memory? I, I, you don't find that that's a challenge.

I, you're, you're, you're bringing back memories to win the A MD uh, 96 Core Gen Noahs came out, and, and that was actually one of my first testing tests that I had when I joined storage review as, Hey, go test these 96 core cpu, by the way, here's a terabyte and a half of DRAM and a bunch of 30 terabyte SSDs in it. And it's like, uh, okay. Um, I, I like the approach.

I like the approach. If we look at it as each individual piece of the system and profile that, right? So when we looked at our CPUs, we decided to do something crazy with the CPUs where we run the traditional benchmarks that could scale, that could handle that level of thread and that level of core.

Um, and, and then pit those tests down both the product stack currently and historically to be able to show that scaling. We started talking about SSDs. We gotta look at the, you know, we, we look at the total performance of 'em by absolutely hammering 'em with rights.

So we took one of our more traditional tests, right? So for these big SSDs and big CPUs, we took white cruncher, and then we needed to test SSDs with it. There's a swap partition in there.

We just did 105 trillion digits of pie with that set, that world record, and then we just did 202 trillion digits of pie on the 60 terabyte drives and took that world record. But that was all basically in the name of, let's put these things up there. And Soddy was willing to send 'em to us and say, our SSDs can survive this level of abuse.

We put absolute petabytes through these things when we were doing it. Um, and it, and it was just that total system test. But then I can go out and when people approach me at shows and say, Hey, are, is that QLC good enough for my workload?

I can say, well, if you're gonna put 20 petabytes through it in a year, yeah, you'll be fine. It'll last you maybe 10 years. And that's, that's a pretty good, uh, that's a pretty good, you know, thing to be able to say, we did it.

Here's the raw data. Apply it for yourself, see how it works out. Yeah.

That, that's, that's so cool. And, uh, and I, I don't, I, that's a, it's a, it's a crazy, it's a crazy ability that you guys have over there at storage review. That's one reason that I'm a reader, and, uh, that, that I enjoy looking at what you're doing.

So thank you so much for joining us on here today. Um, ACE, before we go, um, how would you summarize this, uh, real world testing, real world benchmarking, uh, proving out AI data infrastructure with storage review? Well, it sounds like, uh, uh, like everything else in ai, uh, it's moving rapidly, right?

Uh, and, um, uh, I think what we're, what we're learning here is that, uh, uh, the possibilities are opening up in terms of what you can do at the edge, which is really exciting in a lot of fields, right? From, uh, medicine to manufacturing to energy, right? Uh, and so that's a trend that, uh, is going to be, uh, driving a lot of, uh, innovations and probably touching, you know, our daily lives in more and more ways, uh, going forward.

And so, uh, I'm, I'm thrilled to hear, um, you know, kind of some of the, the, the developments there and, and firsthand from, from you Jordan, about, uh, your own findings and, and some of your projects. Very, very exciting stuff. Um, glad you also found a way to plug the pie, uh, project in there as well.

I think that's, uh, that's super cool. That's become one of my kind of go to, uh, party facts. Hey, did you know the, the 105 trillion digit of pie is a six, and maybe that's why I don't get invited to too many parties.

But, uh, anyway, I, I find it super interesting, uh, and, uh, and appreciate your thoughts on, on the benchmarking piece as well. That's something that, uh, is not as easy as it sounds, right? Uh, there's, there's so many, uh, variables in there.

It depends on the use case, it depends on the architecture, uh, but, uh, I think it's of interest to a lot of folks in this space as we look forward, uh, that, that, uh, uh, you know, continues to mature and that, and that the industry finds a way to measure and communicate storage performance within an ai, um, uh, infrastructure in a way that enables folks to make easy, you know, A to B comparisons. Uh, and so we'll certainly, uh, uh, look forward to your continued work, uh, on, on that front as well. Yeah, it's been great to be able to go through and test all of these different kind of permutations and the stuff that everybody's talking about.

When you go to the trade shows and you see the keynotes and you see all the booth demos, it's really fun to see those and then actually to be able to take 'em out into the real world. I feel like I'm one of the luckiest people in the world. I have the best job ever.

I get to take this fun stuff, take it out, and actually, like you opened with Make the Rubber Meet the road, put it to the test, and it's been absolutely great. Uh, partnering with Soy has been, they're long time friends and, uh, great friends to work with in the industry. com for all the latest and greatest in data center hardware, tech news and reviews.

Yeah. Thank, thanks a lot, Jordan. Uh, ACE, before we go, uh, where can we continue this conversation with you?

Apart from listening to utilizing Tech every Monday? We'll be very busy this summer, uh, at, at events all over the place as well. So keep an eye out for us at, uh, you know, all the major OEM conferences and industry events going forward.

com/ai. And as for me, uh, you'll be seeing me at Tech Field Day events, uh, this month. And, uh, after the bit of a summer break here, uh, we're gonna be doing an awful lot of stuff.

Of course, you can also catch me here at, uh, utilizing Tech every, uh, Monday, uh, the Tech Field Day Podcast, podcast every Tuesday, and of course, our rundown of the week news every Wednesday at galt. Thank you for listening to this episode of, uh, utilizing AI data infrastructure, part of the Utilizing Tech podcast series. You can find this podcast in your favorite application.

Just look for utilizing tech as well as on YouTube. If you enjoyed this discussion, please do give us a rating or a review. Uh, we'd love to hear from you.

The podcast is brought to you by Tech Field Day, home of IT experts from across the enterprise now part of Futureum Group, as well as our friends at Solid ime. com or find us on x Twitter and ma it on at utilizing tech. Thanks for listening, and we will catch you next week.