On 19 Feb 2012 at 18:02, William Donzelli wrote: CDC? SDS? XDS? NCR? RCA? --Chuck

Wouldn't Amdahl break the meme?

So can I gather that there is no clear definition and Jim's is probably the nearest thing that comes to defining a mainframe these days...for those people that use the term? Terry (tez)

On 02/19/2012 10:09 PM, Terry Stewart wrote: There's nothing in there that jumps out at me as being obviously wrong. I'd say it's summed up there pretty well. -Dave -- Dave McGuire, AK4HZ New Kensington, PA

On Mon, 20 Feb 2012, Terry Stewart wrote: That is CORRECT. There has never been a definitive definition, but MANY by marketing people who do NOT understand. It's certainly a good EXAMPLE. But, it is NOT a definition. The whole world is not 370. The whole world is not Unix. The whole world is not Windoze. The best that we've come up with have been differentiation examples. You can pick up and carry a microcomputer A minicomputer has casters, or you will need a really solid handtruck. A mainframe requires a forklift and a union moving crew. Howzbout: How do you get it up or down stairs? You can lose a screw in a microcomputer. You can lose a screwdriver in a minicomputer. You can lose a scope in a mainframe. (It is an exaggeration to say that a person could get lost in one. I think.) A microcomputer requires a mains outlet. A minicomputer requires a mains circuit. A mainframe requires a mains account. Your neighbors find your microcomputers amusing. Your neighbors find your minicomputers worrisome. Your neighbors find your mainframes terrifying. Your microcomputer won't have a significant effect on your power bill. Your minicomputer will have a significant effect on your power bill. Your mainframe will attract the attention of the DEA (hydroponics?) and HSA.

On Sun, Feb 19, 2012 at 08:08:39PM -0800, Chuck Guzis wrote: Well, these days people are building 19" racks for high density datacenters with builtin water cooling at the rack level. And when you look at the power density of e.g. a fully loaded blade system has it makes a lot of sense. Wouldn't call those setups mainframes, though ;-) Kind regards, Alex. -- "Opportunity is missed by most people because it is dressed in overalls and looks like work." -- Thomas A. Edison

On Thu, Feb 23, 2012 at 05:38:03PM -0800, Chuck Guzis wrote: I haven't seen those setups in person, but the water cooled racks $VENDOR was pitching to us in a previous job had direct hookups for cold (and return hot) water. They were (IIRC) aiming to get 40+ KW of heat out of the rack. Kind regards, Alex. -- "Opportunity is missed by most people because it is dressed in overalls and looks like work." -- Thomas A. Edison

Water actually is not very conductive if you keep it clean - chemically clean, that is, free of dissolved minerals and salts and such; mechanical contaminants don't matter unless they (a) are conductive in their own right or (b) dissolve into the water and render it conductive. Of course, even if you start with pure water, keeping it pure is going to be a bit of a fight if you're using it to cool "normal" computer hardware, at least until the hardware gets washed clean. And, while it's not a large effect, metallic corrosion may be enough to be a problem on a timescale of months or years - I'd have to try it, or dig up more data, to say anything definite. True, that. /~\ The ASCII Mouse \ / Ribbon Campaign X Against HTML mouse at rodents-montreal.org / \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B

The term 'mainframe' helped to categorize the physical aspect of machines from the 1950s through 1980s , but since then it has diminished as a differentiator. The mainframe architecture which emphasized dedicated processors to handle IO, front-end processors, support for large numbers of diverse peripheral devices, OSes which included batch and interactive services, also appears to help categorize machines into the 1980s. After that these combination of features started to appear in new systems or were added as extensions to so-called midrange systems (not only IBM), increasingly on lower cost systems. The distinctive mainframe 'style' of computing: the way of interacting with the operating system, OS design, communication protocols, legacy support - captures systems onwards beyond the 1980s till present day. This includes desktop (PC), small deskside cabinet implementations of mainframes like IBM XT/370, P/390, Unisys SCAMP (A-series, Unisys System 80), but also the newer (current day) technology implementations like the z-Series and the Unisys Libra. This criteria helps to answer the question: "What type of computing is a P/390 or Hercules Simulator showing? Answer: a mainframe style of computing, it is not a physical mainframe, but the manifestation shown is a mainframe.'. The marketing folks latched onto the mainframe term as a way to garner some of the halo effect of big, powerful, expensive ("THIS product is just as good as a mainframe but cheaper" etc), meet-the-needs-of-any-business mantra which sold them so well through the early decades. I think at various times marketeers tied themselves into knots trying to reconcile the categories. Unisys advertised the A1 as a "small-frame" computer in the late 1980s. Therefore mainframe only exists today as a identifiable category because examples of the legacy architecture (although significantly updated in technology and implementation) continue to exist. If the mainframe products of IBM and Unisys [*] disappeared tomorrow, the idea of mainframe (as we collectively know them) would fade too. [*] I expect there are other manufacturers, I am using IBM and Unisys as just two examples.

On Sun, 19 Feb 2012, Chuck Guzis wrote: Perhaps if you define computers in number of refrigerators it's equivalent in size to. A mini fridge up to one or two standard fridges is a mini. More than two fridges makes it a mainframe. -- David Griffith dgriffi at cs.csubak.edu A: Because it fouls the order in which people normally read text. Q: Why is top-posting such a bad thing? A: Top-posting. Q: What is the most annoying thing in e-mail?

OK, WHAT IS A "WORKSTATION"? One day, one of our newer "fresh out of UC" (SCHEME) instructors came running into the CIS lab (50 PCs, 6 Macs) shouting, "We're getting SUN COMPUTERS! We're getting SUN COMPUTERS!" It turned out that she had just gotten word that there were three "Multimedia Workstations" for us on the loading dock! A week later, two of them arrived (an administrator had snagged one because it was shinier than what they had, although their old stuff would always disappear rather than end up with us) Each one was in a three foot by four foot by one foot box. "Some assembly required" Yep. A "Multimedia Workstation" was a shiny computer desk on wheels. -- Grumpy Ol' Fred cisin at xenosoft.com

With the decline of the desktop, and rise of the content consumption appliances, I would argue the desktops remaining will be the new 'workstations'. The low end of desktop land is likely to be eroded away leaving the more powerful Windows and Mac systems dominant. And like historic workstations, the hardware is likely to be *relatively* high end. The OS platforms will be mildly diluted and skinned server software. And they will be targeted at specialist content creation. Seems analogous to the workstation of old. -- Colin Cameron Kaiser <spectre at floodgap.com> wrote: I dunno. I'd claim the true workstation is dead. Probably the last was the HP C8000. -- _____________________________________________ personal: http://www.cameronkaiser.com/ -- Cameron Kaiser * Floodgap Systems * www.floodgap.com * ckaiser at floodgap.com -- There's an old proverb that says just about whatever you want it to. -------

m?n 2012-02-20 klockan 21:29 +0000 skrev Tony Duell: How about this: the capacity to in the current publishing sw display an easily readable A4 or folio format page of copy and being able to jump between pages without latency. Think about Frame/(that other big DTP package) on an Sun 1+ or Apollo ws. Current day manual writers/instruction booklet for your BMW or Scania truck, what do they use ? But this becomes mostly an question of sw architecture. The latency issue is a problem for current Word.

On 02/20/2012 10:43 AM, William Donzelli wrote: VERY well put!! -Dave -- Dave McGuire, AK4HZ New Kensington, PA

ILLIAC IV was not a mainframe. It was a lab animal. - Will

It was thus said that the Great Chuck Guzis once stated: Back in the late 90s I worked at a company that had a Stratus. If you rebooted the machine, within twenty minutes, the Status company would call up asking if there were any issues we were experiencing and should a technician be sent out? There's also the apocraphyl story of a new fan unit showing up at the office---the Stratus noticed one of the existing fan units was marginal and had ordered itself a replacement. So perhaps it is the service level that makes a mainframe a mainframe. -spc (The software the we wrote for the Stratus also ran under Windows, but cost significanly more ... )

On 02/20/2012 06:24 PM, Dave wrote: That's scary. I have an MP3000 here that I'm trying to get running. Do you know a bunch about them? I'd definitely call it a mainframe; it uses the same processor chipset as an S/390 G5. (for others reading this, ZERO relation to PowerPC G5) The z9 has massive, massive I/O capacity that is most definitely NOT emulated. -Dave -- Dave McGuire, AK4HZ New Kensington, PA

Yes, the Stratus machine at the Milwaukee School of Engineering did that while I was there, but it was a processor board. Many Stratus sites have similar stories. Very nice machines, if a bit sluggish. -- Will

On 2/20/12 6:56 PM, Fred Cisin wrote: yes, it's called a service contract.

On Tue, 21 Feb 2012, Sean Conner wrote: A properly implemented defestration peripheral will defenestrate admninistrators, with no alteration to the installation nor configuration of the computer. Mainframes, with the possible exception of microcomputer based mainframes such as AT/370, are not subject to the need for inverse defenestration - (throwing windows out of the computer). -- Grumpy Ol' Fred cisin at xenosoft.com

On Sun, 26 Feb 2012, Stefan Skoglund wrote: I certainly can't say that I know Prague, never having been there. :-( During the famous defenestration, there were no fatalities! Those loyal to the defenestrees claimed divine intervention. Proponents of the defenestrators pointed out that landing in a maure pile makes a less than completely positive divine statement. -- Grumpy Ol' Fred cisin at xenosoft.com

On 02/20/2012 08:00 PM, William Donzelli wrote: I remember wandering around Weta Digital circa 2001, and the same was true of their SGI systems (in my case it was hard disks that I was told about) - I'm not sure if the "phone home" was an SGI feature, or something that they'd cooked up themselves. It seemed like a useful feature for any kind of environment where uptime was crucial. cheers Jules

On 02/22/2012 12:15 AM, emu at e-bbes.com wrote: :-) It's the only time I've personally seen the "call back to base and order new parts" feature - plenty of setups where the machine would inform local admins of potential faults so they could decide what to do, of course, but not taking it that extra step. cheers J.

Chuck Guzis wrote: ILLIAC IV is perhaps better classed as a supercomputer. Though that probably starts another argument about how to distinguish mainframes There were certainly a lot of supercomputers that didn't have RAS as defined above.

Early Crays and Cyber were anything but RAS. That is why they came with a little technician that lived inside the cabinets. -- Will

On 2012 Feb 19, at 7:21 PM, Fred Cisin wrote: Whirlwind & SAGE

On Sun, 19 Feb 2012, Jim Brain wrote: It's a good EXAMPLE, but . . . 1) It is NOT a definition 2) There exist mainframes that do NOT execute 370 assembler. (IBM shops may deny that, but it IS true.) 3) There exist NON-mainframes that DO execute 370 assembler. (add-on boards for 5160, do NOT change an XT into a mainframe.) If a microcomputer FALLS on you, you may be injured. If a minicomputer falls on you, you may be killed. If a mainframe falls on you, they may not find the body. are also examples (of differentiation) -- Grumpy Ol' Fred cisin at xenosoft.com

Well, don't tell Unisys that. Clearpath Libra runs the Burroughs architecture, and Clearpath Dorado runs the Univac architecture. Both are still in production, albeit in tiny numbers. -- Will

On 2/19/2012 10:12 PM, Fred Cisin wrote: :-) The XT/370 is not the last word in such units (http://en.wikipedia.org/wiki/PC-based_IBM-compatible_mainframes). I've heard them referred to as "legacy environments", which is ambiguous since people call the 370 architecture "legacy" as well. The only terms I've heard folks use are the manufacturer or model: I've folks talk about the "Burroughs", and the "UNIVAC System". I won't argue that a complete definition needs to include these other machines, but the term has been largely "redefined" to be 370-based systems at this point. I will still stand by my statement that the term refers to 370-compatible systems, but I did find a few notes (http://www.eweek.com/c/a/IT-Infrastructure/Unisys-Unveils-New-ClearPath-Mai…) referring to Unisys' machines as mainframes. I'll concede the point, but I think one would be hard pressed to find many in IT that even know about Unisys' big-iron solutions. Most people do know them in relation to their Tandem machines. Maybe, the best contemporary definition is "a current computing system that can natively execute code written in a machine architecture commercially available before 1970" (or some arbitrary date that covers the various machines one typically identifies with a mainframe moniker). Note that this would preclude such Hitachi systems that actually run on Xeon CPUs and emulate the z Architecture, but I'm sure some wordsmithing could fix that while keeping things like Hercules emulator from falling into the definition's space.

On Feb 20, 2012, at 12:25 AM, David Griffith <dgriffi at cs.csubak.edu> wrote: PCs! I'll bet several people got Intel-branded yachts for making that move. -Dave -- Dave McGuire, AK4HZ New Kensington, PA

POWER. There might not be Power Macs anymore, but every major console is PowerPC, and IBM makes buck$$$ on POWER servers (says the man who dropped $9K on a used POWER6). Plus PowerPC is crazy common in embedded environments and there are still PPC Amigas. </powerbigot> -- ------------------------------------ personal: http://www.cameronkaiser.com/ -- Cameron Kaiser * Floodgap Systems * www.floodgap.com * ckaiser at floodgap.com -- There's always free cheddar in a mousetrap, baby. -- Tom Waits -------------

Zitat von David Griffith <dgriffi at cs.csubak.edu>: Don't write off MIPS that fast. A lot of new embedded Designs are MIPS based, a lot of research (china!) goes into the MIPS architecture. Seems that Itanium is not there to stay either. So "just" an Intel shop ... ARM still has to deliver the performance there. Nice for power saving tablets, but for anything else ... Cheers

On 20/02/12 1:26 PM, Liam Proven wrote: Tilera* http://www.tilera.com/products/processors What's the status of Sun's CoolThreads/Niagara? Fujitsu? http://www.fujitsu.com/global/services/computing/server/sparcenterprise/tec… What happened to Transmeta's technology? And is XMOS going anywhere? http://www.xmos.com/technology/architecture http://www.xmos.com/technology/xcore --Toby

On Mon, 20 Feb 2012, Alexey Toptygin wrote: I dont think XMOS CPUs are in any way targeted at general purpose computing, more at the FPGA/DSP/real time embedded control space... Peter Wallace

On 20/02/12 4:07 PM, Alexey Toptygin wrote: What if you could map Erlang processes to it? There are many industrial and general purpose Erlang programs now. It thrives on low-latency message passing between as many cores as possible. --Toby

On Mon, 20 Feb 2012, Alexey Toptygin wrote: Much much faster and lower power for multiple low latency coupled tasks X86 is simply not a player here (too hot, and too slow to switch tasks) Peter Wallace

On Tue, 21 Feb 2012, Alexey Toptygin wrote: It all has to do with real time and latency. X86s (and their assosciated chipsets) have great throughput but dreadful latency (this is a common tradeoff) especially for hardware replacement type tasks. The XMOS chips have low latency (ns) and independent I/O available on multiple cores, the X86 cores share a (high latency/buffered) common data path As I said before, the XMOS chips are designed for multiple coupled tasks with (somewhat in FPGA space if the signal frequencies are not above perhaps a few 10 of MHz). It is designed to simulate hardware in software. The x86 cannot do this very well, so as I said simply does not play in the same space at all. Peter Wallace

On Tue, 21 Feb 2012, Alexey Toptygin wrote: I just fits in the continuum of computing devices ranked by latency/parallism you might also choose a GA or XMOS device for ease of programming relative to a FPGA, I think most people would say that XMOS and GA devices are easier than FPGAs to program (or at least use more familiar tools) FPGAs,etc lowest latency greatest parallelism, core local I/O XMOS/GA medium latency medium parallelism, core local I/O DSPs medium/high latency, little/medium parallelism, shared I/O path GP CPUs high latency/ medium parallelism, shared I/O path Peter Wallace

On Tue, 21 Feb 2012, Sean Conner wrote: You are confusing XCore and SMP. They are very different - XCore cores do not share memory, but communicate using IO channels. Alexey

Zitat von Toby Thain <toby at telegraphics.com.au>: If you didn't see it yet ... ... you're not using 8 cores at work ;-)

On 21/02/12 8:47 AM, emu at e-bbes.com wrote: You are perfectly right. I barely use one core, most of the time. Bad wording on my part. :) --Toby

On 21/02/12 12:24 PM, Sean Conner wrote: Where do you get 21x from? I see a wall clock ratio of about 12x (still amazing; can you explain why it's superlinear?) --Toby

On Feb 21, 2012, at 1:30 PM, Toby Thain wrote: It's superlinear because of the constraints of I/O (reading the file off the disk) vs. CPU (compiling the file). Actually compiling the file (depending on language, of course) can be miniscule compared to the overhead of reading off the disk. I typically use a number of jobs equal to 1.5 the number of processors in the system; I have a 4-core i7 which masquerades as 8 due to hyperthreading (which they actually got right in the i7, as opposed to the P4). I don't see a *lot* of difference between 9 and 12 jobs. If I were to guess why it would be superlinear in the general case, I would imagine it's because the header files then wind up in the file cache while you're using them. I've never profiled it. - Dave

On Feb 21, 2012, at 3:49 PM, Dave McGuire wrote: Well, after the P4 mess, they just dropped it from the original Core architecture (which was a complete reboot from the mess that was the P4). They added it back in on the i7 family, and it does help a bit; I get something like a 30% speedup on my simple raytracer running 8 threads instead of 4, and that's basically CPU and memory bound (very, very little I/O and system call overhead; no thread locks except for getting the next pixel position, even). ISTR that hyperthreading generally imposed a performance *penalty* on the P4. - Dave

On 21/02/12 3:04 PM, David Riley wrote: (still amazing; can you explain why it's superlinear?) off the disk) vs. CPU (compiling the file). Actually compiling the file Wouldn't a large buffer cache pretty much make that disparity vanish? Most systems can buffer many times more than the total size of source. I still don't see a clear explanation, because my reasoning is: - assume a very small cache in relation to source - parallel job kicks off 1..N compiles, let's say N unique source files and 20*N common include files. So each compile process works through 21 files. - the problem is, if your cache is smaller than that overall job, ANY of these compiles is going to blow the cache. Running in parallel doesn't help. On the other hand, if ALL of the source is cached, again, parallel make doesn't win, because serial make gets exactly the same advantage. Did I miss something - exactly what mechanism is helping if it's not cache? --Toby (depending on language, of course) can be miniscule compared to the overhead of reading off the disk. I typically use a number of jobs equal to 1.5 the number of processors in the system; I have a 4-core i7 which masquerades as 8 due to hyperthreading (which they actually got right in the i7, as opposed to the P4). I don't see a *lot* of difference between 9 and 12 jobs. would imagine it's because the header files then wind up in the file cache while you're using them. I've never profiled it.

On 21/02/12 9:56 PM, Mouse wrote: Right, if certain tasks are cpu bound and other tasks are i/o bound then perhaps we run into the more subtle analysis hinted at by Sean, where accidental overlaps might be helping (since nothing is done to *deliberately* optimise this situation). But if everything is I/O bound then I don't see how parallelising it would help. --Toby

On 02/21/2012 08:30 AM, Toby Thain wrote: I've *only* ever seen a superlinear speedup on parallel builds. Are your Makefiles ok? Actually there's one exception: running under SunOS 4. That was ages ago of course. It did SMP, but only one CPU can be in the kernel at once (one big lock!) so system calls were a big bottleneck. -Dave -- Dave McGuire, AK4HZ New Kensington, PA

It was thus said that the Great Toby Thain once stated: If a "make clean ; make -j" breaks, then the Makefile isn't correct 8-) The directory I posted figures for has 389 files to compile (C++). It wouldn't surprise me that there are a few files that take a while to compile, while in the same time, other cores could compile a multiple number of files. I'm sure there's an answer in queueing theory (it's faster for one line of customers to N tellers than one line for one teller). In another section of the codebase, there's a subsection that takes around 1:30 (one minute, 30 seconds) to compile serially, and 1:15 parallel, but that's because there's one file that's a few megabytes in size [1]. -spc (I also avoid recursive makes) [1] I can't be sure of certain data files being install where the program runs, so it's easier to just embed said data files into the executable. This is purely for testing purposes, not release.

On 02/21/2012 02:00 PM, Toby Thain wrote: I did not mean to suggest otherwise. I disagree with some of your viewpoints, but I know that you know your stuff. Dave Riley addressed this in a later email when I wasn't looking. :) I agree with his assessment based on my own experiences. -Dave -- Dave McGuire, AK4HZ New Kensington, PA

On 02/21/2012 02:22 PM, Cameron Kaiser wrote: Yes, SunOS 4 supports SMP. I ran dozens of multiprocessor machines running SunOS 4. Solbourne scaled it beyond four processors and improved things a bit. -Dave -- Dave McGuire, AK4HZ New Kensington, PA

Also, didn't Solbourne do it earlier? Or is my wetware storage dropping bits again? /~\ The ASCII Mouse \ / Ribbon Campaign X Against HTML mouse at rodents-montreal.org / \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B

On 02/21/2012 02:57 PM, Cameron Kaiser wrote: Yes. I had a six-processor Solbourne system for awhile. I never could get it running because I didn't have OS media. It was a very well-built piece of hardware! -Dave -- Dave McGuire, AK4HZ New Kensington, PA

On 02/21/2012 03:44 PM, Cameron Kaiser wrote: More wacky than a Voyager? That's saying something! Pics? Mine was a 5/600, a deskside box somewhat resemblant of an AlphaServer 2100 but not quite as deep. I'd love a copy of that, if you can manage it. -Dave -- Dave McGuire, AK4HZ New Kensington, PA

What versions? -- ------------------------------------ personal: http://www.cameronkaiser.com/ -- Cameron Kaiser * Floodgap Systems * www.floodgap.com * ckaiser at floodgap.com -- "I [still] adore my Commodore 64" ------------------------------------------

On 2/22/12 12:57 PM, Dave McGuire wrote: 4.1c and patches along with the release notes have been uploaded to bitsavers.

http://www.floodgap.com/iv/715 http://www.floodgap.com/iv/716 This is a better picture of the profile: http://danjuliodesigns.com/resume/page12.html I think it's packed in with one of the pizza boxes. I'll put it up on the Gopher server when I dig it out. -- ------------------------------------ personal: http://www.cameronkaiser.com/ -- Cameron Kaiser * Floodgap Systems * www.floodgap.com * ckaiser at floodgap.com -- Greek tailor shop: "Euripedes?" "Yes -- Eumenides?" ------------------------

I believe so. The sick SGI Indy is occupying the bay where I plan to put it. It's in its zipper case in the closet waiting for a slot on the workbench. The display is very crisp and you just don't see those kinds of garish gas plasma displays anymore. :) Dworkin helped me out with my first S4100 which I needed to root. It booted that first box eventually let the magic smoke out of its hard disk, so the only one I have in fully operational order is the S3000. I have another pizza box as a donor unit, but I haven't merged the two, and the S3000 is much more convenient with its built-in display. -- ------------------------------------ personal: http://www.cameronkaiser.com/ -- Cameron Kaiser * Floodgap Systems * www.floodgap.com * ckaiser at floodgap.com -- Math according to Pentium: 2 / 2 = 1.037587439439485486372112039523781385 ...

On Feb 20, 2012, at 5:49 PM, Alexey Toptygin wrote: Maybe not outperform, but (as has been mentioned before) sometimes it's a matter of having to fit in a size/power/heat envelope. Sure, you could bring the hammer down with a Core i7 Xeon or 4 (that's what we did with our application; 32 cores of Nehalem goodness), but you'll burn almost a kilowatt doing what works in a hundred watts for a more purpose-built chip just for the sake of going with COTS hardware. Or, if you're going smaller, it means you can fit it on an AMC card (which has a power budget of 35W). It's also a question of application; there are a lot of things people use DSPs for these days when there's no reason they shouldn't be using an FPGA, but there's also lots of idiots (usually in management) who specify an FPGA because it "sounds cool" when they really should be using a CPU, particularly for networking. In cases like that, manycore CPUs are really handy, because you can just dispatch each incoming packet to a CPU. Probably not, but again, it's a matter of application constraints in a lot of cases. - Dave

On 02/20/2012 04:01 PM, David Riley wrote: I'm sad to report that I haven't even powered it up yet. I've recently gone from not having enough money to do anything to not having enough time to do anything. -Dave -- Dave McGuire, AK4HZ New Kensington, PA

On Mon, 20 Feb 2012, Dave McGuire wrote: OK, but why do your GSM modems need to be talking to 2+ interlinked cores rather than one core that is 2+ times as fast? What are you doing with them that makes controllling them with a single fast core impractical? Alexey

On 20/02/12 8:52 PM, Dave McGuire wrote: That sounds like the Transputer model -- there was quite a diversity of devices accessible via the standard serial links: I/O, storage, graphics, networks, etc. --Toby

On 20/02/12 10:43 PM, Dave McGuire wrote: My point was that Inmos sold a diversity of adapters interfaced through the links. It *wasn't* necessarily done with Transputers bit-banging (they wouldn't necessarily be fast enough anyway - e.g. video). --Toby

It was thus said that the Great Alexey Toptygin once stated: One core for handling traffic out, another one for handling traffic coming in. It makes the code easier to write. Handling traffic for both directions, without using threads, means you end up writing a cooperative multitasking system where you have to manage all the state information. It's not impossible, but it tends to turn the code inside out (as it's event driven) and (at least to me) harder to follow. If the system provides threads, then you still have overhead of task switching, so a single CPU running at 2X does not match two CPUs running at X. -spc (Has to deal with event driven code at work, and personally, I don't like it)

Hi I doubt they'd be doing much different today if it was the old company. It is all about calculations per watt per unit volume. It is no longer about the fastest processor design. That is for gamers with cooling systems designed to cool houses but used to a single graphics processor. What Cray did then was push the current technology to the limit. The limits have since been redefined. Dwight

All the 68000 based pizza boxes, Apple & Motorola on the PowerPC

From: Jim Brain Sent: Sunday, February 19, 2012 9:02 PM Whose Tandem machines??? Compaq bought Tandem Computers, which was then part of the HP/Compaq merger. HP still sells Non-Stop systems based on the Tandem system architecture. Rich Alderson Vintage Computing Sr. Systems Engineer Vulcan, Inc. 505 5th Avenue S, Suite 900 Seattle, WA 98104 mailto:RichA at vulcan.com mailto:RichA at LivingComputerMuseum.org http://www.LivingComputerMuseum.org/

On Mon, 20 Feb 2012, Chuck Guzis wrote: Easy to forget "midi" when memories of minis and micro-minis are plentiful, memorable, and sometimes quite explicit. I will gladly find a way to provide a 5100 to anybody who can transport me back half a century!

On 2/20/2012 12:38 AM, Mouse wrote: Yep, I did. It's a good distinction. What tripped me up is that the common vernacular is to call is 370 assembler code, not 370 machine code. I'm not sure why, but that's the term they use. Jim

On Sun, 19 Feb 2012, Fred Cisin wrote: I remember seeing magazine ads for this sort of thing. It showed a PC in a room once occupied by a mainframe of some sort with all the cables now going into the PC. -- David Griffith dgriffi at cs.csubak.edu A: Because it fouls the order in which people normally read text. Q: Why is top-posting such a bad thing? A: Top-posting. Q: What is the most annoying thing in e-mail?

Err, Disks, tape reels, golfball-type printheads.... [Could not resist] -tony