Slashdot Mirror

IBM had also planned on releasing multi-power5 processor computers, although I haven't heard anything as of late

Never heard anything as of late? Where have you been listening? IBM has many models with multiple power 5 processors and have for a while. All listed on their web site.

... not every processor is documented. In fact, the documentation for a lot of high volume processors isn't available. I'd like PSP docs, or docs on the processor in the ipod, or even the processor in the disposable digital cameras or the VMU game system -- but none of these docs are publicly available. The hardware manual for the G5 PowerPC was finally published last November, over a year after apple started selling the hardware.

So, no, the documentation doesn't always get released.

For me, the crux of the usefulness and eventual adoption and finally complete embracing of AJAX lies in the article's paragraph:

Some of the buzz surrounding AJAX has been generated by Web designers as well as programmers. AJAX?s flexibility is invigorating for Web designers because JavaScript can control any aspect of any images or type on a page. Fonts can grow or shrink. Tables can add or lose lines. Colors can change. Although none of these capabilities are new to programmers accustomed to building client applications -- or, for that matter, Java applets -- they are novelties to Web designers who would otherwise be forced to rely on Macromedia (Profile, Products, Articles) Flash.

I've seen what Google has done with AJAX (e.g., Google suggest), and it's stuff I never imagined could be so repsonsive in a web context. For me it starts to make programming fun again, and web programming an acceptable form of application development.

When browsers and web first emerged I could see the writing on the wall, but I wasn't happy about it. Browser application writing from the programming perspective was probably the single most giant leap backwards in technology for me (not including technologies introduced by Microsoft)....: you mean, all the years I've spent honing skills writing applications no longer apply? You mean I no longer have "state" as a tool for maintaining sanity in my application???? Hwaahhh??? I have to do what to change the web page???

While there have been some technologies (ASP, JSP, etc) to help with these issues, none have addressed the responsiveness issue with the web page round trip message loop. AJAX comes close. Now all I have to do is learn it.

For a great example of the responsive nature of this (I've referenced this before), go to Google Personal Home, set up your own home page, and play... Configure your modules by dragging them around... open and close your g-mail previews. This all starts looking alot like programs actually running locally on your own machine. (I'm assuming all are familiar with and have played similarly with Google Maps.)

Additionally, here are some very good resources to learn more about AJAX:

• Very Dynamic Web Interfaces
• XMLHttpRequest Introduction
• An example
• Using the XML HTTP Request object
• Dynamic HTML and XML
• XMLHttpRequest API madness
• Sarissa
• JavaScript: The World's Most Misunderstood Programming Language
• What kind of language is XSLT?

That's it, I'm done.

Currently all of Intel's stuff runs hotter

My dual processor G5 is watercooled. And very loud if the CPU load is high for long periods. The maximum power envelope at 2.5Ghz is 100W, which I believe is not a million miles away from Prescott.

No, really... that's what it says.

With all the advances in KM over the years, maybe a more interesting approach would be to have the system aggregate/rate responses over a period of time and respond with the top 3 or so. As planned, I would think the system described is less than useful, it would be downright obnoxious once it hit critical mass (go read every first post on /. for a preview).

MYTH #2: "New, innovative companies won't start up overseas."

Really? What do you think these laid-off chip designers are gonna do when they get back to Chennai? Sell trinkets to tourists?

MYTH #3: "R&D jobs don't go overseas. Hell, they don't even leave the US east and west coast, for the most part."

REALITY:

• GE Corporate Research in Bangalore and Shanghai
• HP Opens New Research Center in Singapore
• HP Bangalore Research
• IBM India Research Center
• IBM China Research Center
• Microsoft Research Beijing

Per nasscom.org, "A recent study on the biotech market by business intelligence firm, Ernst & Young, has shown that India has the potential to become a leading hub of biotech projects. Indian companies have the capability to enter segments such as manufacturing biogenerics, contract research services, clinical trials and even areas such as bio-informatics."

MYTH #5: "Ultimately, what xenophobes need to realize is that writing shitty code doesn't make anyone "high-tech." You're no more entitled to an inflated salary than the auto workers who saw their work moved overseas - if someone with no education can do your job cheaper, you don't deserve your job."

"Accenture in India has also been moving into front office work such as doing clinical data management for its pharma clients. Accenture's pharma team here, which consists of doctors, dentists and biologists, analyses data from tests and helps its pharma client to gain `time-to-market' advantage. "Normally, for a BPO, back office activities are the target, but we are beginning to spot opportunities in front office activities as well," Cole said."

MYTH #2: "New, innovative companies won't start up overseas."

Really? What do you think these laid-off chip designers are gonna do when they get back to Chennai? Sell trinkets to tourists?

MYTH #3: "R&D jobs don't go overseas. Hell, they don't even leave the US east and west coast, for the most part."

REALITY:

• GE Corporate Research in Bangalore and Shanghai
• HP Opens New Research Center in Singapore
• HP Bangalore Research
• IBM India Research Center
• IBM China Research Center
• Microsoft Research Beijing

Per nasscom.org, "A recent study on the biotech market by business intelligence firm, Ernst & Young, has shown that India has the potential to become a leading hub of biotech projects. Indian companies have the capability to enter segments such as manufacturing biogenerics, contract research services, clinical trials and even areas such as bio-informatics."

MYTH #5: "Ultimately, what xenophobes need to realize is that writing shitty code doesn't make anyone "high-tech." You're no more entitled to an inflated salary than the auto workers who saw their work moved overseas - if someone with no education can do your job cheaper, you don't deserve your job."

"Accenture in India has also been moving into front office work such as doing clinical data management for its pharma clients. Accenture's pharma team here, which consists of doctors, dentists and biologists, analyses data from tests and helps its pharma client to gain `time-to-market' advantage. "Normally, for a BPO, back office activities are the target, but we are beginning to spot opportunities in front office activities as well," Cole said."

Universities that sign up can let students download WebSphere Studio, DB2 Universal Database, WebSphere Application Server, Rational XDE and Lotus Domino. You don't even get the boxed product. It's IBM's answer to MSDN, with a big tilt towards web-oriented middleware.

This is not "computer science". This is vocational training. This is material IBM used to teach new hires in-house. Now they're dumping their product-specific training requirements on universities.

And then they whine that they're not getting "the best and the brightest".

... or in any other general-purpose operating system on a general-purpose computer. PCs are fundamentally insecure. There are a dozen ways to spy on cryptographic operations done in them, ranging from trojans, to hardware side-channel attacks, and dozens more to get copies of keys that they store. This is just one particular attack that may permit an attacker who can't get a trojan running with sufficient privileges to spy on operations directly to obtain some key bits. But if the attacker can't do that, there are lots of other ways to get the keys. General-purpose computers are simply not trustworthy.

If security is important, you do your crypto in a secure crypto module, like the FIPS 140-2 Level 4 IBM 4758 or the Level 3 Luna SA. Or, you use a general-purpose computer with special-purpose, very simple software and then provide strict physical access control to the machine and very limited network access -- often through a serial link using a custom protocol rather than via a real network. Or you could theoretically use a general-purpose machine with a TCPA chip with a regular, general-purpose operating system that has been modified to make use of the TCPA chip and with keys tightly bound to a well-defined system software configuration. But only if you have good physical security. In many situations it's still better to use a FIPS 140-2 Level 3 or Level 4 device.

IMO, the existence of weaknesses like this in Linux, and the fact that they're widely known, is a *good* thing, because it helps convince people not to trust that which is inherently untrustworthy. We need more publicity of similar problems in Windows (and there are lots of them).

But the team didn't just take an existing core like the PowerPC 970FX and build an SoC around it. The core for Cell is new and appears to have been designed before the clock-frequency-is-dead era. The core was designed to reach certain power and die-size goals and is designed to be able to run at clock frequencies in the 4+GHz range. The engineering theam did simply some of the core design (for example, it's an in-order design and only a dual-issue superscalar) and used some dynamic logic in the design in certain critical timing areas.

The core complies with the PowerPC instruction-set architecture version 2.02 (and the 2.01 public version of teh specification). The core was designed with a particular balance of die size, clock speed and architectural efficiency that is different from that of PowerPC 970. This instantiation of the Power Architecture still has a relatively long pipeline, much like the Power 4 and PowerPC 970, but the Cell design does not have a very wide issue pipeline or out-of-order execution, nor does it have as many functional units.The Cell Power core has hardware fine grain multi-threading.

This is quite insteresting. Unlike general purpose processors, which are often optimized for a set of specific benchmarks, the processor for a game console is actually designed to optimize the performance for a specific set of applications, i.e., 3D games. The most demanding applications driving the performance of high end PCs today also happen top be 3D games. I wonder if we are going to see a transition to back to simplified cores with higher clock speeds soon. Given the current trend to integrate multiple cores on a single processor die, a multi-core design with a large number of simple, high speed processors would be an interesting design trend.

The multi-threading feature of the Cell core may be ported over from the Power5 design as a way to deal with memory latency at high clock speeds.

I think it would be pretty safe to assume that the PowerPC core in the Xbox360 chip is very similar, if not the same design. Here is an IBM paper that shows, at least in the lab, they were able to run the cell processor above 4GHz.

But the team didn't just take an existing core like the PowerPC 970FX and build an SoC around it. The core for Cell is new and appears to have been designed before the clock-frequency-is-dead era. The core was designed to reach certain power and die-size goals and is designed to be able to run at clock frequencies in the 4+GHz range. The engineering theam did simply some of the core design (for example, it's an in-order design and only a dual-issue superscalar) and used some dynamic logic in the design in certain critical timing areas.

The core complies with the PowerPC instruction-set architecture version 2.02 (and the 2.01 public version of teh specification). The core was designed with a particular balance of die size, clock speed and architectural efficiency that is different from that of PowerPC 970. This instantiation of the Power Architecture still has a relatively long pipeline, much like the Power 4 and PowerPC 970, but the Cell design does not have a very wide issue pipeline or out-of-order execution, nor does it have as many functional units.The Cell Power core has hardware fine grain multi-threading.

This is quite insteresting. Unlike general purpose processors, which are often optimized for a set of specific benchmarks, the processor for a game console is actually designed to optimize the performance for a specific set of applications, i.e., 3D games. The most demanding applications driving the performance of high end PCs today also happen top be 3D games. I wonder if we are going to see a transition to back to simplified cores with higher clock speeds soon. Given the current trend to integrate multiple cores on a single processor die, a multi-core design with a large number of simple, high speed processors would be an interesting design trend.

The multi-threading feature of the Cell core may be ported over from the Power5 design as a way to deal with memory latency at high clock speeds.

I think it would be pretty safe to assume that the PowerPC core in the Xbox360 chip is very similar, if not the same design. Here is an IBM paper that shows, at least in the lab, they were able to run the cell processor above 4GHz.

Although, on the proprietary front, I've heard, IBM's JVMs and Java SDKs outperform Sun's.

I think these people may disagree with you there.[1]

[1] This is annoying/depressing. I can't think of another non-x86 processor useful enough to act as a server, and still in production. Via, Transmeta, AMD64 might as well be x86 since they're all hamstrung by backwards-compatibility requirements, ARM is a toy (but kicks arse in embedded apps), Itanium is dead, as is Alpha. OK, there are probably a few Crays and mainframes kicking about that it would be pointless to replace, and yes, people do run mailservers on Commodore 64s and the like, but realistically there are only three usable architectures out there: x86, Power and Sparc.
Cell had better be worth the hype, because life will get awfully boring otherwise.

Seriously. Their Tivoli brand of software is aimed toward enterprise-class systems management...and the apps all run on Linux. Tivoli Security Compliance Manager lets you build profiles of what each system should have installed and how it should be configured (even the BIOS) and periodically scan them to make sure they match. And when any systems don't match your profile for how they should look, use Tivoli Configuration Manager to push out the changes that will bring it back into compliance. Also useful for pushing out patches, updates, etc, etc. Did I mention that it runs on Linux? http://www-306.ibm.com/software/tivoli/products/co nfig-mgr/

Seriously. Their Tivoli brand of software is aimed toward enterprise-class systems management...and the apps all run on Linux. Tivoli Security Compliance Manager lets you build profiles of what each system should have installed and how it should be configured (even the BIOS) and periodically scan them to make sure they match. And when any systems don't match your profile for how they should look, use Tivoli Configuration Manager to push out the changes that will bring it back into compliance. Also useful for pushing out patches, updates, etc, etc. Did I mention that it runs on Linux? http://www-306.ibm.com/software/tivoli/products/co nfig-mgr/

Seriously. Their Tivoli brand of software is aimed toward enterprise-class systems management...and the apps all run on Linux. Tivoli Security Compliance Manager lets you build profiles of what each system should have installed and how it should be configured (even the BIOS) and periodically scan them to make sure they match. And when any systems don't match your profile for how they should look, use Tivoli Configuration Manager to push out the changes that will bring it back into compliance. Also useful for pushing out patches, updates, etc, etc. Did I mention that it runs on Linux? http://www-306.ibm.com/software/tivoli/products/co nfig-mgr/

Are you sure you know what you're talking about in regard to Security Certifications

Darwin still sucks in lmbench, sorry.
(http://www-106.ibm.com/developerworks/lib rary/l-y dlg5.html)

Which is obviously a microbenchmark that doesn't measure real-life application performance. But having a small thing that gets done quite often (context switches) take 4x more time or so does translate into real-life performance losses too (obviously not 4x, but still measurable).

You might be able to "hide" all the low-level performance losses by optimizing elsewhere, but it still doesn't mean you're doing the best possible thing.

There's absolutely no reason to carry around a dinosaur like Mach around. Portability isn't a reason (see number of platforms supported by Linux and NetBSD vs. Darwin). Scalability isn't one either.

Auto-boxing changes nothing in terms of performance. If you pass a value type as a parameter to a method that expects an object it will generate code that is precisely the same as if you were wrapping it yourself by hand.

It doesn't even really alter 'strictness' since it basically just promotes value types to their object wrappers, and only where that actually makes sense.

While your statement is true when talking about cases where you intend object conversion, the problem is when you don't want conversion to occur. IBM notes the problem in this article, for the same reason I do. If the compiler uses objects without telling you, it's a problem. Having done a bit of real-time image processing with Java, it is clear that Java is an adequate high performance platform, and that primitives are significantly faster than objects - as long as you can count on them remaining primitive.

Everybody is focusing on those two guys smiling together, instead of looking at why they called the press release together, and why what they announced is considered important enough to warrant a Ballmer/McNealy co-presentation!

The reason why this is news, is that both companies, along with a ton of other groups of all sorts of sizes and purposes, have been working on creation of standards that will allow web authentication on the internet to cross boundaries of OS platform, browser platform, and development platform. The Metadata Exchange and Interop protocols are just two of a whole HOST of protocols that are going to link everything up.

Some of you will say - who cares? But the technology they are working on now will be used in the future by most people, on most platforms, to access protected web content.

That's pretty big. This little niche of the industry is set to explode into mainstream consciousness, just wait and see...

If you want to be ahead of the curve:

Check out the Fact Sheet from the MS-Sun announcement.

Check out the WS-* White Paper

Check out Microsoft's Vision For an Identity Metasystem

Check out the Liberty Alliance Technology Review

And if prefer blogs to White Papers, check out Kim Cameron's Blog. That's really the happening place in Identity Management right now...

Pixie

Are you surrounded by IBM-compatible machines or Lenovo-compatible machines?

When we were evaluating technologies over a year ago, KHTML and KJS stood out. Not only were they the basis of an excellent modern and standards compliant web browser, they were also less than 140,000 lines of code. The size of your code and ease of development within that code made it a better choice for us than other open source projects. Your clean design was also a plus. And the small size of your code is a significant reason for our winning startup performance...

Quit trolling: that webpage renders the same in MSIE v6.0 and FF v1.04.

IBM doesn't market PCs any more.

Just remember there are many PowerPC processors and cores that IBM makes. This is a "custom" fabbed chip for Microsoft with a (3) VMX unit(s) attached. This is not a Power4, Power5, or 970/FX/GX or like chips. Also, has pricing for XBox live has not been announced, but $300USD seems right.

But it sure would be nice to have >3GHz G5's or dual cores....or dual/dual...or ...

Just remember there are many PowerPC processors and cores that IBM makes. This is a "custom" fabbed chip for Microsoft with a (3) VMX unit(s) attached. This is not a Power4, Power5, or 970/FX/GX or like chips. Also, has pricing for XBox live has not been announced, but $300USD seems right.

But it sure would be nice to have >3GHz G5's or dual cores....or dual/dual...or ...

According to IBM's site http://www.ibm.com/ibm/us/ they currently have 329,000 employees. With an organization that large you don't simply install a new application and hope for the best!

You are correct, it is just a web browser, but when Jack and Jill in accounting cannot figure out where their 'Favorites' list went to and call the help desk, help desk staff need to know the answers. Granted, that is a simple question to answer, but think of the 100 other things that might be different between Firefox and IE (common message text for browser warning dialogs, etc), with 329K employees, even if only a small fraction of them run into questions it can mean time wasted by both the employee asking the question and the help desk person trying to find the answer. That time adds up quickly.

In any case, I see this as a huge success for Firefox!

Maybe they'll stop optimising their employment page for IE 6 then. Or maybe they think you're only supposed to start using Firefox AFTER you get hired on at the company...

Indeed. I see the question as being "when you can't cope with your expanding J2EE app server needs yourself, who do you want to call in, IBM or JBoss ?"

Note that's jboss.com... IBM isn't the only for-profit company involved in open-source J2EE app server products here. Maybe that's what you're missing. Or maybe you're just missing the concept that someone might look to an outside company for help with their app server, rather than hiring more guys like you and I.

Sorry, the name "Boson" was already named... ;P

And there are no computrons.

Though I guess having the first quantum-spelled name (IBM in xenon atoms with nickel base) is pretty cool.

Gotta hand it to IBM. At first they were the ones to bring the computer industry down, but now they're bringing it up, up and up. I think they really understand what our computer and tech culture is becoming.

• 6502 - September 1975, 8 bits
  
• Z-80 - July 1976, 8 bits
  
• 8086 - 1978
  
• VAX - 1978
  
• 68000 - 1979, 32 bits
  
• ARM2 - 1986, 32 bits
  

Robocode is designed to use a game-development environment to teach programming. Check it out.

Secondly it is antifuse-based one-time programmable ROM. It is NOT a flash which can be re-written 100,000 times. So it is more useful for storing application code but not for data storage etc.

Antifuse base memories are diode like and can be much smaller than regular FLASH memories. But these are inherently slower and also don't have any gain element (like transistor). This requires careful design to achieve good signal-to-noise ration for memory read operation

More aggressive 3D technology was demonstrated by IBM last year where they have circuits in 3D.

A startup R-cube logic is also designing 3D microprocessor where memory is put on top of the logic core to reduce latency.

Xanoptics is more into hybrid design (mixed analog, RF, optics) on a single footprint.

I'd like to argue that Perl is very maintainable. Many studies show a correlation between LOC and number of defects. So the important thing is reducing the LOC written. The differences accountable to language syntax is negligable when you can reduce the amount of code you write by order(s) of magnitude. With Perl, you have CPAN (http://search.cpan.org/), a repository of pre-written, tested code, the vast majority of which are classes ready to use in object-oriented fashion. Now, I'm not going to say that Perl's OO syntax is anything but inelegant, but it works. What is elegant is the way CPAN works. I won't go into that here, but it does allow you to decrease by orders of magnitude the amount of code you have to write. There are great solutions for all kinds of things, such as object-relational mapping (point Class::DBI at your database, and you instantly have get/set methods to update your tables) and powerful templating (http://template-toolkit.org/), just to name a couple. There are 4000+ other examples there.

As an example of reducing lines of code, here's an article on writing a database-driven web application in 18 lines of code, and a similar article here.

But this doesn't apply to just database or web frameworks. Thanks to CPAN, this applies to anything you set out to do. CPAN modules (in general; there are exceptions) are encapsulated best practices, well tested, and a major productivity booster. And, they drastically reduce the amount of code you need to write, making your code more maintainable.

First, look at the freaking date before submitting something as "News". The PDF is dated May 5, 2004 (not 2005).

Here are some observations:

- The Windows version tested was 2k3 pre-release. To be fair, VeriTest should have tested RHEL 3 and RHL 9. Otherwise, it should have tested against Windows 2k server.

- The IIS version tested was 6.0. To be fair, VeriTest should have tested with the current Apache releases at the time (1.3.29 and 2.0.48/9) and the current Tux release included in RHEL 3.

Other notes:

- The Linux kernel rev for RHEL 2.1 is 2.4.9-e3. IBM benchmarked web serving performance for 2.6 vs. 2.4 and the results are huge performance gains by using a 2.6 kernel.

- The filesystem used was ext3. ReiserFS is a faster filesystem for filesystem, although it uses more CPU.

- There was minimal performance tuning to Linux or Apache. TUX was performance tuned. The testers attempted to reverse-engineer settings for Apache.

Bottom line: This survey isn't much better than the Mindcraft survey done several years ago. It didn't tune Linux at all, and received tuning help and funding from Microsoft.

Another good one, same site. It is a four part article that does a decent job of showing how it works.

This definitely isn't a news story, though.

Great article on countering buffer overflows.

The Computing-Tabulating-Recording Company was renamed to "International Business Machines Corporation" in 1924.

The roots of the company go back to the 1880s - see IBM's website.

You are wrong...

From IBM:

Although IBM was incorporated in the state of New York on June 15, 1911 as the Computing- Tabulating- Recording Company (C-T-R), its origins can be traced back to developments at the close of the 19th century. For example, the first dial recorder was invented by Dr. Alexander Dey in 1888, and Dey's business became one of the building blocks of C-T-R. Similarly, the Bundy Manufacturing Company was incorporated in 1889 as the first time recording company in the world, and it, too, later became a key component of C-T-R.

I would like to rat out various folks who offered me counterfeit copies of your "windows" product: these, and these, and these, and these, and .....

Stick it to 'em. And good luck!

Certain once-large organizations have shed a lot of (and will soon shed all ) folks around here, and others are threatening to do the same. Not surprising, since RTP has been so telecom-centric, and unlike Silicon Valley, concentrates its employment base in a handful of large companies (vs. gajillions of startups).

The upshot is that there are a lot of unemployed techies around here who need re-training. Enter TechEngage. The proposition is simple: if you're unemployed, you get to attend a certain number of classes for free (or close to it), and in return, you donate 40 hours of your time to the cause.

I really wish the public sector would wake up to this effort. Oh, sorry, that would be socialist. Can't have that. Unless of course you're an unemployed textile worker in the western part of NC, without even a high-school diploma. Then our state legislature bends over for you, even though you could never contribute as much to the tax base once employed (grumble)..

QBIC is part of IBM's DB2 content manager. It has been available for at least 5 years now, and is now part of a DB2 extender. You can check it out here:

http://wwwqbic.almaden.ibm.com/

Their 1350 Clusters all run Linux and use Infiniband, Myrinet, or GigE as the high speed interconnect and can bring you anywhere from 4 to 1024 nodes. You actually can order more than 1024 nodes, it's just not listed on the website. IBM 1350 recently built a cluster for a customer that was ~8000 nodes.

ThinkPad now. ThinkPad always.

There's some standarization going on.

http://www.iata.org/whatwedo/fuel/datastandards.ht m

http://www-306.ibm.com/software/ebusiness/jstart/c asestudies/faa.shtml

They could probably find room for one somewhere in the Power5 CPU.

You're describing ClearCase. And it's a horrific piece of garbage.

The blog is a bit misleading: "Details are scarce about the STELLA supercomputer, built by IBM using some of its Blue Gene/L technology."

Details are plenty since what IBM gave to the project is a couple of racks of BlueGene so everything applies, scaled proportionately.

Here are some details: http://www.research.ibm.com/bluegene/ and http://www.llnl.gov/asci/platforms/bluegenel/

And your system won't work anymore, either. The order is important, and if you just add a "&" to the end of the commands, you lose control of that. Conventional init satisfies dependencies by prefixing the scripts with numbers indicating when they are to be launched - S10network, S12syslog, etc - and doing so sequentially. But this is overly conservative; the whole system doesn't start until a single service does, even if nothing depends on it. It also means you have to maintain a total order instead of a DAG, which sucks.

NetBSD 1.5 replaced init, too. See The Design and Implementation of the NetBSD rc.d system for details.

There was also an IBM article called Boot Linux faster on replacing initscripts with make. (This is less weird than it seems. Correctly resolving parallel dependencies is what make was designed to do.)

So Apple's not the only one to see a problem here and replace it. I don't know why the major Linux distributions haven't yet; the init system really sucks. Hopefully they'll take a look at launchd, which I believe is open source.

Wasn't this paper co-authored with Joe Darcy .. now the Java floating point czar working on Tiger (Java 1.5)?

Much shorter version of the paper is here, and a good java floating point paper is also over here

oh .. and if you think that nobody at sun will admit java's weaknesses .. you gotta stop talking to the sales and marketing drones, and spend some time in targeted discussions with the engineers ..