I Tried to Break a Million Dollar Computer - IBM Tour

[Draeconix]

Very cool video. I work in the mainframe world as a Systems Programmer specializing in Automation. I have been doing that for the last 9 years now. I was a Mainframe Operator for 6 years before that. There is very little in the server space that can compete with a Z series mainframe in terms of raw performance, compute power and redundancy. It is impressive technology. Most people don't realize it but almost every banking transaction hits a mainframe at some point in its lifetime and the up time is real. There are places that don't IPL (Initial Program Load - think "reboot") their mainframe for months and even years. It just sits in the data center calmly doing what it does.

 

I was at Poughkeepsie, NY for some training and got a tour of the factory as well. We got to see how they put them together from the ground up. One thing the video didn't show is how heavy those drawers can get when fully loaded. In the assembly area instead of lifting the drawers into place they raise and lower the entire rack to the correct height so the technician assembling the machine never has to climb a ladder or raise a heavy object over their head. The rack will actually lower into the floor so the technician can reach the top.

 

By the way, if anyone is looking for a good paying job in IT, definitely look into the mainframe world. Because people think that the mainframe is a "dinosaur" and "on its way out" the industry is somewhat hurting for people. I assure you the mainframe is going to be around for quite some time. The company I work for is definitely looking for people (PM me for more info) and will train you (they have a whole academy).

[MichaelMouton]

8 hours ago, faketruth said:

good job, ibm is trying to promote and make their image looks positive after all of the age discrimination lawsuit that have been plaguing them. and using LTT is their way to do it.

 

and for anyone who doesnt know about it, just search using keyword "ibm age discrimination lawsuit".

 

wonder if LTT just dont do enough research about the controversy of a company that they are covering/sponsored with or just blatantly ignore it  straight away at this point and just apologize later.

 

Oh wow, this seems to actually be a real thing!  See here...

https://techmonitor.ai/leadership/workforce/ibm-age-discrimination-lawsuit-dinobabies

 

I'm not sure how the TechLinked staff go about finding new articles (I know they look here on the forums for some), but I can't imagine that they missed something like this.  Granted, I don't watch the show too often nowadays.

 

As an innovative company I absolutely love IBM, but they have been involved in some dark sh!t.  Here is one small example...

https://www.wired.com/2001/02/did-ibm-help-nazis-in-wwii/

[maartendc]

No offense, but this type of clickbait title is exactly why I stopped watching LTT (disclaimer: I didnt even watch the video, the content might be fine).

 

To me, this is in the vain of the type of content like Mr. Beast and "shock value" Youtube channels:

"I GIFTED my Friend $1 million DOLLARS IN PENNIES OMG LOL"

"We BROKE his PRIVATE JET OMG"

 

This is so off-putting to me and makes me NOT want to watch the video honestly. But I guess it works for some people, that is probably what the metrics and algorithms tell us.

 

 

[NemesisPrime_691]

15 minutes ago, maartendc said:

No offense, but this type of clickbait title is exactly why I stopped watching LTT (disclaimer: I didnt even watch the video, the content might be fine).

 

To me, this is in the vain of the type of content like Mr. Beast and "shock value" Youtube channels:

"I GIFTED my Friend $1 million DOLLARS IN PENNIES OMG LOL"

"We BROKE his PRIVATE JET OMG"

 

This is so off-putting to me and makes me NOT want to watch the video honestly. But I guess it works for some people, that is probably what the metrics and algorithms tell us.

 

 

Yeah. I was waiting for him to try and break the thing, but that moment just never came. The only breaking he did was in the first 10 seconds. Technically Linus can just say that he did show trying to break this and hence its not a clickbait. But without explaining anything about that, what the screen showed, what did he break was sort of cheating.

 

The rest of the video was cool info about the part of the industry that people like me who just write C# (or any other language really) code on a small laptop screen for a living would never have got the opportunity to see, learn about. The titles are getting a bit off-track these days though and that needs to be brought back on track. LTT is big now and I believe they can be just as good as the titles from someone like Gamers Nexus.

[maartendc]

17 minutes ago, NemesisPrime_691 said:

Yeah. I was waiting for him to try and break the thing, but that moment just never came. The only breaking he did was in the first 10 seconds. Technically Linus can just say that he did show trying to break this and hence its not a clickbait. But without explaining anything about that, what the screen showed, what did he break was sort of cheating.

 

The rest of the video was cool info about the part of the industry that people like me who just write C# (or any other language really) code on a small laptop screen for a living would never have got the opportunity to see, learn about. The titles are getting a bit off-track these days though and that needs to be brought back on track. LTT is big now and I believe they can be just as good as the titles from someone like Gamers Nexus.

Linus has always been a smart businessman if nothing else.

 

They do extensive research on video titles, video thumbnails, etc. and I am sure this type of title and thumbnail gets the most clicks. It just turns off a certain percentage of your audience.

 

I much prefer the no-nonsense approach of channels like Paul's hardware or Hardware unboxed (although all of them do some "eye catching thumbnails" to some extent.)

[Apple_Pie]

I still have 2 T-shirt from IBM (because my mom work for them) and it doesn't fit on me anymore I'm sad its a good quality too!

[Nystemy]

10 hours ago, yakovlev said:

You're right, but I'm pretty sure Linus didn't quite get the facts straight on the details of ECC vs. RAIM in these systems.  To do RAID you have to have some way of detecting errors on individual drives, and I would expect the same is true on the individual modules in a RAIM scheme.  IBM Z memory architecture is somewhat unique, so I suspect that may be part of the reason for the unusual DIMM configuration.

Well, each RAM module has an extra DRAM chip on either side of the board, so that is likely doing more regular ECC on the module itself.

 

And a CRC checksum can usually correct single bit errors without major issue, and with the dedicated controller on the module it likely continuously checks for errors, making the risk of an uncorrectable error rather slim.

And for module to module correction, well there one can take the regular RAID approach of considering all the working units to be correct and not consider the dead ones. _{(Though, taking a quick look at RAID, then a RAID 5 is technically not even error detecting when it has lost a drive, and RAID 6 meanwhile can't fix it since it can't tell who is wrong. Now, there is many ways to skin the error detection/correction cat.)}

But, all of this error correction and RAIDing of modules likely has a pretty hefty amount of overhead associated with it, and likely makes access latency worse. _{(one other reason for why caching on a completely different CPU is potentially viable.)} But mainframes are all about that reliability and uptime, so slowing down access times a bit isn't the end of the world. _{(though, even regular x86/ARM servers do also tend to focus on uptime and reliability, and why the performance/$ tends to be worse compared to regular desktops. All though usually regular servers focus on this to a lesser degree, since it is typically a lot more cost effective to rely on multiple systems where an individual failure of one isn't particularly impactful, but a more clustered approach has its own huge slew of downsides. (There is however a lot more nuance to this topic. And what is best between single system vs multiple servers vs mainframe is debatable depending on the application.))}

[dogwitch]

30 minutes ago, Nystemy said:

Well, each RAM module has an extra DRAM chip on either side of the board, so that is likely doing more regular ECC on the module itself.

 

And a CRC checksum can usually correct single bit errors without major issue, and with the dedicated controller on the module it likely continuously checks for errors, making the risk of an uncorrectable error rather slim.

And for module to module correction, well there one can take the regular RAID approach of considering all the working units to be correct and not consider the dead ones. _{(Though, taking a quick look at RAID, then a RAID 5 is technically not even error detecting when it has lost a drive, and RAID 6 meanwhile can't fix it since it can't tell who is wrong. Now, there is many ways to skin the error detection/correction cat.)}

But, all of this error correction and RAIDing of modules likely has a pretty hefty amount of overhead associated with it, and likely makes access latency worse. _{(one other reason for why caching on a completely different CPU is potentially viable.)} But mainframes are all about that reliability and uptime, so slowing down access times a bit isn't the end of the world. _{(though, even regular x86/ARM servers do also tend to focus on uptime and reliability, and why the performance/$ tends to be worse compared to regular desktops. All though usually regular servers focus on this to a lesser degree, since it is typically a lot more cost effective to rely on multiple systems where an individual failure of one isn't particularly impactful, but a more clustered approach has its own huge slew of downsides. (There is however a lot more nuance to this topic. And what is best between single system vs multiple servers vs mainframe is debatable depending on the application.))}

oddly what you talking about here. wendell talk about trying to recover data from a 1.2? tb nand drive.

 

[Nystemy]

21 minutes ago, dogwitch said:

oddly what you talking about here. wendell talk about trying to recover data from a 1.2? tb nand drive.

 

Well, a lot of stuff in computer science is intermingled.

 

Even something simple like caching is done for similar reasons in most places where it is used.

 

For an example, caching website data is practical since it ensures that one don't have to download the same data over and over and over, saving in on bandwidth by a decent amount. It is the same story for websites using a Content Distribution Network, just viewed from the other side of the ISP's pipework.

 

And in a CPU, cache does the exact same thing. A lot of people think cache is for reducing latency, but it really isn't. We can look at a typical x86 core that decodes at least 1 instruction per cycle, every cycle, and an x86 instruction is 1-15 bytes _{(ARM meanwhile is fixed 4 bytes)}, but 75% of all x86 instructions are equal or less than 4 bytes. But roughly speaking, a 3.5 GHz CPU core is going to need 14 GB/s of bandwidth of memory bandwidth for just instructions. Then there is data on top of that, and the fact that we have more than 1 core in a CPU, where each typically decodes multiple instructions per cycle. So going into the high 100's of GB/s of bandwidth isn't particularly odd for a humble quad core CPU. It is simply not possible for a pair of DDR4 channels can keep up.

 

But with a few kB of L1 cache all the short repetitive program loops more or less don't bother memory, add a few hundred kB of L2 and the longer loops and some often used program data also become marginal from a memory bandwidth standpoint, by the time we have a few MB of L3 even a fair bit of program data gets easy to work with. But we need to first check L1 before we go to L2 since if we sent all calls to L2 directly, it wouldn't be able to keep up, and this trend of lowering amounts of available bandwidth continuous up our stack, so each layer of cache adds a bit of latency. In the end, having a 3 teir cache as is typical in most CPUs today can add a good few tens of cycles to our access time. But a few extra ns of latency doesn't matter, since without cache we would bottleneck on memory and not get any use out of our cores. _{(though, some workloads handle data sufficiently randomly and have a large enough dataset to dwarf cache, making cache "pointless" for the application data, so memory bandwidth easily becomes the bottleneck if the work being done is simple enough for each piece of data computed. (It gets even worse if the application reads the dataset sequentially, since then cache does nothing if it even is a single byte smaller than the dataset. One reason Dwarf's Fortress loves memory bandwidth.))}

Though, there is situations where access latency matters, especially in multi node systems. But such calls usually also aim for system wide coherency of the applicable data. Here I know that some systems will literally just skip checking cache and go straight to the source. Since it doesn't matter if the cache were technically faster if the data it provided were out of date. _{(Now, one can speculatively start to compute based on cache while also in parallel fetching the data from the source, and then abandon the "cached" version if it turns out to be out of date, this will technically be faster but less power efficient.)}

 

_{(I feel like I could have written this shorter, after all, most of this post is just me rambling about cache.... (So I won't bring up how Inter Process Communication is more or less similar on both the Internet wide level and the Core to Core level, as well as the application to OS service level, or how IPC is rather central to a lot of mainframe and HPC applications.))}

[dogwitch]

which i wont dis agree with you on that.

 

side bar note.

i notice when  i had a raid array fail. the terminology for some parts of raid have complete change now...

which for a bit was total confusing me. which am now seeing else where in the computer space.

 

 above comment side bar

i had to both learn raid(new terms) also on how it writes and cache data(modern).

 

 

[Sir Asvald]

On 4/5/2022 at 10:16 PM, HenrySalayne said:

Congratulations, you found an oversight. Watch the news, it might give an idea why they did not want to have this symbol in their video.

Do you have a link.

[HenrySalayne]

47 minutes ago, Sir Asvald said:

Do you have a link.

Here you go!

 

Or try this one.

[Sir Asvald]

44 minutes ago, HenrySalayne said:

Here you go!

 

Or try this one.

[admiralq]

Very cool to see Linus at my workplace here in Poughkeepsie.  I worked on the raised floor years ago when I was in Server Ops.  Z systems are beasts and the Z16 is something else.  

[faketruth]

On 4/6/2022 at 6:58 AM, MichaelMouton said:

 

Oh wow, this seems to actually be a real thing!  See here...

https://techmonitor.ai/leadership/workforce/ibm-age-discrimination-lawsuit-dinobabies

 

I'm not sure how the TechLinked staff go about finding new articles (I know they look here on the forums for some), but I can't imagine that they missed something like this.  Granted, I don't watch the show too often nowadays.

 

As an innovative company I absolutely love IBM, but they have been involved in some dark sh!t.  Here is one small example...

https://www.wired.com/2001/02/did-ibm-help-nazis-in-wwii/

the best thing is that after leaving this for a while, nobody from the staff even reply to address this, just like what happen with RING back then lol.

 

https://www.theregister.com/2022/04/14/ibm_discovery_discrimination/

 

just look at what ibm do lol

 

id say ltt research team are just ignoring how bad ibm image atm with all of the lawsuit for the age discrimination and doesnt even consider how ltt coverage can influence ppl to show that ibm is "cool" and have "awesome" technology when they are getting the ibm invite for this coverage.

[Quackers101]

I break millions, IBM linus version.

Kinda sus, would need a small loan for that. :V

[Jtalk4456]

On 4/6/2022 at 2:58 AM, MichaelMouton said:

 

Oh wow, this seems to actually be a real thing!  See here...

https://techmonitor.ai/leadership/workforce/ibm-age-discrimination-lawsuit-dinobabies

 

I'm not sure how the TechLinked staff go about finding new articles (I know they look here on the forums for some), but I can't imagine that they missed something like this.  Granted, I don't watch the show too often nowadays.

 

As an innovative company I absolutely love IBM, but they have been involved in some dark sh!t.  Here is one small example...

https://www.wired.com/2001/02/did-ibm-help-nazis-in-wwii/

so I'm gonna start with the second link. Yes IBM had some participation making machines used by the government. It was the germany division of the company. there was a dictator killing anyone who didn't help him, I wonder if they might have been doing what they were told to stay alive like many germans who didn't believe in what was happening but were part of it anyways. I'm not saying we excuse anyone participating in what happened, but lets not forever condemn a country for making punch card systems for the dictatorship they were under. IBM is not a big nazi corporation pretending to be normal, they're a normal company who had a german division during the time germany went bad. 

 

As far as the discrimination, reading thru that link talks about 2 higher ups using a rude term for old people. This is an ongiong investigation into company culture, but if that's all the info we've got right now, I wouldn't miss the chance for a mainframe factory tour just because 2 douchebags in management were caught saying something we've seen pretty much every other company in the industry act on. I'm not saying there's not problems, but with the info given here, I'm tempering my anger to a reasonable level

[chrome_wheels]

What a cool video. I'm surprised too that IBM is still this big.

[faketruth]

On 4/15/2022 at 10:20 PM, Jtalk4456 said:

so I'm gonna start with the second link. Yes IBM had some participation making machines used by the government. It was the germany division of the company. there was a dictator killing anyone who didn't help him, I wonder if they might have been doing what they were told to stay alive like many germans who didn't believe in what was happening but were part of it anyways. I'm not saying we excuse anyone participating in what happened, but lets not forever condemn a country for making punch card systems for the dictatorship they were under. IBM is not a big nazi corporation pretending to be normal, they're a normal company who had a german division during the time germany went bad. 

 

As far as the discrimination, reading thru that link talks about 2 higher ups using a rude term for old people. This is an ongiong investigation into company culture, but if that's all the info we've got right now, I wouldn't miss the chance for a mainframe factory tour just because 2 douchebags in management were caught saying something we've seen pretty much every other company in the industry act on. I'm not saying there's not problems, but with the info given here, I'm tempering my anger to a reasonable level

amazing, i wish i can learn from you that JUST from reading from 2 article given by other ppl, you decided to reach a conclusion that IBM worth the benefit of the doubt. simply just amazing.

 

oh since this is the case with you, here another link https://www.theregister.com/2022/06/27/ibm_age_discrimination_emails/

 

IBM totally dont do wrong - i get you

[Jtalk4456]

1 hour ago, faketruth said:

amazing, i wish i can learn from you that JUST from reading from 2 article given by other ppl, you decided to reach a conclusion that IBM worth the benefit of the doubt. simply just amazing.

Amazing, I wish I had the ability to completely ignore facts for the sake of my arguments. That would certainly make things easier. So you completely skipped responding to the nazi part of my post, presumably because there's no good argument for that. Lots of companies in Germany contributed to the bad government many years ago, but because there's an ongoing investigation of bad business practice, we're gonna draw lines and say the company was intentionally evil in germany and is still intentionally evil now.
As for your link on the actual issue at hand, it seems to be recent, like today recent. So IDK, maybe it's a stretch, but it SEEMS like you had nothing to argue with my post for the last 2 months when I said:

On 4/15/2022 at 6:20 PM, Jtalk4456 said:

This is an ongiong investigation into company culture, but if that's all the info we've got right now ... I'm not saying there's not problems, but with the info given here, I'm tempering my anger to a reasonable level

You know reasonably waiting to find out more information. And now that there is more information, you are trying to take pleasure in reopening a months old discussion to tell a rando on the internet they were wrong for saying they would wait for info because months later, wow there's info now...
Point is, for the lack of info we had at the time of my post, I still stand by my reluctance to call a company evil. I also pointed out that, while not an excuse (stated in my post), IBM is not the only company to be caught with bad business practices of even this kind. Many in the tech field lament constant ageism from within most of the industry. So if more info has come out against these managers, ok. They have now been proved to be as bad as we know most every other tech company is. My early skepticm still stands and I maintain a reasonable amount of anger.

1 hour ago, faketruth said:

IBM totally dont do wrong - i get you

Do you really get me? Funny cuz I literally didn't ever say IBM was guiltless. I kinda said they did bad things if you read what I wrote. But given your other arguments, I'm not expecting that.

[faketruth]

On 6/28/2022 at 2:31 AM, Jtalk4456 said:

Amazing, I wish I had the ability to completely ignore facts for the sake of my arguments. That would certainly make things easier. So you completely skipped responding to the nazi part of my post, presumably because there's no good argument for that. Lots of companies in Germany contributed to the bad government many years ago, but because there's an ongoing investigation of bad business practice, we're gonna draw lines and say the company was intentionally evil in germany and is still intentionally evil now.
As for your link on the actual issue at hand, it seems to be recent, like today recent. So IDK, maybe it's a stretch, but it SEEMS like you had nothing to argue with my post for the last 2 months when I said:

You know reasonably waiting to find out more information. And now that there is more information, you are trying to take pleasure in reopening a months old discussion to tell a rando on the internet they were wrong for saying they would wait for info because months later, wow there's info now...
Point is, for the lack of info we had at the time of my post, I still stand by my reluctance to call a company evil. I also pointed out that, while not an excuse (stated in my post), IBM is not the only company to be caught with bad business practices of even this kind. Many in the tech field lament constant ageism from within most of the industry. So if more info has come out against these managers, ok. They have now been proved to be as bad as we know most every other tech company is. My early skepticm still stands and I maintain a reasonable amount of anger.

Do you really get me? Funny cuz I literally didn't ever say IBM was guiltless. I kinda said they did bad things if you read what I wrote. But given your other arguments, I'm not expecting that.

more like i didnt really check this place much unless there is a controversy related content made by LTT,  not because im waiting for more information. If you really did do your research, the details about age discrimination lawsuit and stuff are hefty enough and the last link i gave you is literally just the latest update which prompt me to check if there is any update on the thread.

 

only checked this thread again after seeing update on the ibm discrimination lawsuit to see if there is any discussion further, and only after that i saw your post.

 

On 4/15/2022 at 10:20 PM, Jtalk4456 said:

so I'm gonna start with the second link. Yes IBM had some participation making machines used by the government. It was the germany division of the company. there was a dictator killing anyone who didn't help him, I wonder if they might have been doing what they were told to stay alive like many germans who didn't believe in what was happening but were part of it anyways. I'm not saying we excuse anyone participating in what happened, but lets not forever condemn a country for making punch card systems for the dictatorship they were under. IBM is not a big nazi corporation pretending to be normal, they're a normal company who had a german division during the time germany went bad. 

 

As far as the discrimination, reading thru that link talks about 2 higher ups using a rude term for old people. This is an ongiong investigation into company culture, but if that's all the info we've got right now, I wouldn't miss the chance for a mainframe factory tour just because 2 douchebags in management were caught saying something we've seen pretty much every other company in the industry act on. I'm not saying there's not problems, but with the info given here, I'm tempering my anger to a reasonable level

you literally said that after reading only 1 link that "higher ups being rude", this is literally downplaying the issue. you said "ongoing investigation" and "all the info we got rn" - maybe if at that time - you can try to see if you can find more details yourself? i even try to search into google news with "ibm discrimination" and can even find more details easily back from apr even.

 

wow, you are right, there is more info now and back then there is nothing about ibm discrimination details at all.

[Jtalk4456]

14 hours ago, faketruth said:

1. more like i didnt really check this place much unless there is a controversy related content made by LTT,  not because im waiting for more information.

2. If you really did do your research, the details about age discrimination lawsuit and stuff are hefty enough and the last link i gave you is literally just the latest update which prompt me to check if there is any update on the thread.

3. you literally said that after reading only 1 link that "higher ups being rude", this is literally downplaying the issue. you said "ongoing investigation" and "all the info we got rn" - maybe if at that time - you can try to see if you can find more details yourself? i even try to search into google news with "ibm discrimination" and can even find more details easily back from apr even.

4. wow, you are right, there is more info now and back then there is nothing about ibm discrimination details at all.

1. Why would LTT controversy mean anything in this discussion? That's completely irrelevant to me saying I wanted to wait and see what surfaced in the trial.

2. Really? If I just read some articles I'd have heaps and heaps of info about them and not have to question anything I read with any ounce of critical thinking skills? What a world we live in where being skeptical in absence of proven fact is a bad thing...

3. Let's read one of the other links, then. hmmm...

Quote

 

IBM has disclosed that the median age of its staff in 2020 was 48, unchanged from 2010. This means its workforce is significantly older than some of its rivals in IT outsourcing such as consultancies like Accenture or Deloitte, says Josh Bersin, founder of Bersin Associates, a research company focused on enterprise learning and talent management.

“The disclosures were pretty surprising,” he adds. “IBM has a long history of reskilling and continuously developing its workforce, and also a history of offering early retirement programs. If these statements are true it’s a sobering story about IBM leadership which may be a wake-up call.”

 

So what another article says is while the industry is used to this practice, they're not as used to it coming from IBM, who has a statistically older workforce than many companies. Thanks for suggesting me to do more reading that you didn't do apparently.

4. So me saying waiting for more is equivalent to me saying there's no info. You have a very strange grasp on the use of the english language... It's almost like you're trying to use hyperbole to make it seem like I said something completely different than what I said to argue that your point is somehow emboldened from it. If you stop putting words in peoples mouth and actually read what they said, you might have an easier time understanding what was said.