CIPWTTKT&GC v0X12 (v18): Makol can't Computer Very Good - Hardware & Software

[QUOTE=Dr. Deeps;35323936][B]*WOOSH*[/B] That's the sound of everything you guys are talking about flying right over my head.[/QUOTE] Quantum computers, extremely high clock rates but slow memory and innefective processors that can't have conductive-enough materials without being colder than -150°C That's what I gathered. Correct me if I'm wrong.

[QUOTE=Wooops: 404;35323809]I know but I still think it would be easier find a way to cool something in space that make a super conducter that works at 300k.[/QUOTE] The 300K superconductors was necessary for *us* to have one. The NSA report says, and I quote, "Refrigeration is not expected to be an issue". They figure, if they run into any problems, they can just go ask NASA or Fermilab or Los Alamos "hey, how do you get a really tiny thing really really fucking cold?" [QUOTE=Wooops: 404;35323903]The way I understand it superconductivity works because their is little vibration from heat to disrupt it(correct me if I'm wrong), so creating a high heat super conductor would be next to impossible. Cooling something in space, yes, fucking hard and expensive, but still easier than breaking the laws of physics.[/QUOTE] Not quite. Superconductors work by allowing electrons to pair up, which means they act like bosons instead of fermions. At high temperatures, thermal energy is able to overcome the tiny quantum energy that binds them. However, a significant role is played by the protons of the superconductor, so the upper limit is pretty much unknown - scientists have pretty much given up on saying how cold you absolutely need to be for it to work, after originally thinking you'd never get a superconductor above 10K, then above 30K, then above 70K. There's even suggestions that hydrogen, compressed to a few hundred gigapascals, would be a superconductor at room temperature. [QUOTE=wingless;35323924]I think I found one of the main problems with that RSFQ processor. Large latency in memory and deeper pipelines in the chip and only a small chip area reachable in a single cycle, so you'll be running multiple cycles for the whole chip.[/QUOTE] Latency is actually the main thing they deal with in the paper. But a 50-stage pipeline wouldn't be *terrible* - it would still have far less latency (at 50GHz) than a 15-stage pipeline (at 4GHz). [editline]27th March 2012[/editline] [QUOTE=wingless;35324142]I should really re-enable data on my phone plan. Last time I managed to rack up $100 on my phone bill, but I realized I shouldn't go to Facepunch where people don't exactly resize things, they just thumb 5MB png's. All I really want it for now is SSH into my VPS.[/QUOTE] I just always keep data access disabled unless I need it. I'll sometimes turn it on at Taco Bell so I can check Twitter while I eat.

[QUOTE=wingless;35324142]I should really re-enable data on my phone plan. Last time I managed to rack up $100 on my phone bill, but I realized I shouldn't go to Facepunch where people don't exactly resize things, they just thumb 5MB png's. [/QUOTE] Download an application that tracks the data usage and setup alerts. I use Onavo on my Android phone.

[QUOTE=gman003-main;35323342]Also: I started reading an [url=http://www.nitrd.gov/pubs/nsa/sta.pdf]old, 2005-era report[/url] from the NSA, researching the feasibility of [url=http://en.wikipedia.org/wiki/RSFQ]Rapid Single-Flux Quantum Computing[/url]. [b][u][i]HOLY FUCK WHY ARE WE NOT DOING THIS?[/i][/u][/b] * 50GHz expected by first usable model, [b]expected to reach 100GHz+[/b] * Basic RSFQ circuits have been [u]tested[/u] at 770GHz * 100,000x [i]less[/i] power usage than CMOS * Can be made with existing technology and methodology * Isn't a massive change from the programmer's view, unlike qubit computers - [b]you could run C on it[/b] * RSFQ-based SRAM would be able to match clock speeds of the processor * Expected price tag to invent the entire field: [b]$20M and five years[/b] Only downsides: * Because a signal at 100GHz would only go 2mm in one clock cycle, they expect processor pipelines about 50 stages long (compare: 31 stages for Prescott, 20 stages for early Pentium IV, 14 for SB, 10 for PIII, etc) * Oh, and you need to cool the thing with liquid nitrogen. Not because it makes a shit-ton of heat, but because you have to make it superconducting and the highest-temperature superconductor stops superconducting at 133K OK, so barring a sudden discovery of a 300K superconductor, they're not going to get it into a laptop. Or an affordable desktop. And probably not commodity servers. But dude, a 0.05THz processor? For $20M? We have missiles that cost more than that, per shot. We can afford $4M a year to INVENT NEW MICROPROCESSORS.[/QUOTE] I read a few years ago about a superconductor that operates at room temperature but it seems to be one of those things that get a big report and then people stop caring for 15 years

The result of a very bored pepper, 3 mass effect you tube videos and 10 minutes of dicking about in audacity. [url]http://dl.dropbox.com/u/5831868/Mass%20Effect%203%20-%203%20in%20one%20mix%20thing%20rev3.mp3[/url] LAME only likes exporting at 128KBPS

[QUOTE=kukiric;35324162]Quantum computers, extremely high clock rates but slow memory and innefective processors that can't have conductive-enough materials without being colder than -150°C That's what I gathered. Correct me if I'm wrong.[/QUOTE] It's more of a latency issue. It would be a hell of a processor for anything that fits in cache. Like cracking encryption.

[QUOTE=gman003-main;35324290]I just always keep data access disabled unless I need it. I'll sometimes turn it on at Taco Bell so I can check Twitter while I eat.[/QUOTE] I have data disabled from my telco. Serious. $100 phone bill from data usage, fuck that. [quote]Latency is actually the main thing they deal with in the paper. But a 50-stage pipeline wouldn't be *terrible* - it would still have far less latency (at 50GHz) than a 15-stage pipeline (at 4GHz).[/quote] But if you're going to want this thing for supercomputing, you're going to want to realistically think bigger than 50Ghz, We have Supercomputing clusters measured in Thz and for cheaper than the development costs we have now. Then again, most supercomputing projects are just funded by a blank cheque and you do have the advantage of cheaper running costs with much lower power requirements. Still, there has to be something holding us back from using this. I really think it's people holding out for Graphene. Some recent tests got IBM to get the shit resonating at 100Ghz (Not actual clock speed, though. But fucking impressive in comparison)

[QUOTE=ButtsexV3;35324326]I read a few years ago about a superconductor that operates at room temperature but it seems to be one of those things that get a big report and then people stop caring for 15 years[/QUOTE] "Room temperature superconductor" is about on par with "room temperature fusion", as far as journalism goes. Don't believe it until you see a *real* scientist say something about it.

[QUOTE=ButtsexV3;35324326]I read a few years ago about a superconductor that operates at room temperature but it [b]seems to be one of those things that get a big report and then people stop caring for 15 years[/b][/QUOTE] Like cures for cancer? [editline]28th March 2012[/editline] [QUOTE=gman003-main;35324388]"Room temperature superconductor" is about on par with "room temperature fusion", as far as journalism goes. Don't believe it until you see a *real* scientist say something about it.[/QUOTE] Modern journalism is just sensationalism.

[QUOTE=wingless;35324385]I really think it's people holding out for Graphene. Some recent tests got IBM to get the shit resonating at 100Ghz (Not actual clock speed, though. But fucking impressive in comparison)[/QUOTE] Somehow I haven't heard about this graphene computer shit. Please, do tell.

[QUOTE=wingless;35324398]Like cures for cancer? [editline]28th March 2012[/editline] Modern journalism is just sensationalism.[/QUOTE] [url]http://facepunch.com/threads/1173155[/url]

[QUOTE=Wooops: 404;35324431][url]http://facepunch.com/threads/1173155[/url][/QUOTE] Oh look. That's like, what, the 15th in the past 2 years?

[QUOTE=gman003-main;35323701]Sorry, meant to say 42U. And [url=http://www.cisco.com/en/US/products/ps5854/index.html]there are 2800 series routers[/url] - from the looks of it, we got two 2801s. [editline]27th March 2012[/editline] Well, the paper I read was commissioned by the NSA, back in 2005. It laid out a plan to have a usable 50GHz system by 2010. Maybe... Maybe they already *did*. They're the fucking NSA man, they can hide shit. Like the bodies. Of the people who found out. Oh shiiiiiiit[/QUOTE] AHH, those Cisco ISR's. I could've sworn they were discontinued. Proven wrong again. :P Anyways, the throughput on those will be GARBAAAAGE.

[QUOTE=wingless;35324435]Oh look. That's like, what, the 15th in the past 2 years?[/QUOTE] Yup.

[QUOTE=gman003-main;35324388]"Room temperature superconductor" is about on par with "room temperature fusion", as far as journalism goes. Don't believe it until you see a *real* scientist say something about it.[/QUOTE] it wasn't a news article, it was a published journal. I don't remember the name else I'd reference directly. It could operate as a superconductor up to 260 kelvin iirc which isn't strictly room temperature but it's very easy to sustain in most environments.

[QUOTE=pure.Joseph;35324441]AHH, those Cisco ISR's. I could've sworn they were discontinued. Proven wrong again. :P Anyways, the throughput on those will be GARBAAAAGE.[/QUOTE] Look, I think maybe *one* of our servers has a working gigabit ethernet interface. Throughput will *not* be an issue. [editline]27th March 2012[/editline] [QUOTE=ButtsexV3;35324487]it wasn't a news article, it was a published journal. I don't remember the name else I'd reference directly. It could operate as a superconductor up to 260 kelvin iirc which isn't strictly room temperature but it's very easy to sustain in most environments.[/QUOTE] That would probably have been palladium hydride. Quoth Wikipedia: "PdHx is a superconductor with a transition temperature Tc of about 9 K for x=1. (Pure palladium is not superconducting). Drops in resistivity vs. temperature curves were observed at higher temperatures (up to 273 K) in hydrogen-rich (x ~ 1), nonstoichiometric palladium hydride and interpreted as superconducting transitions.[4][5][6] These results have been questioned[7] and have not been confirmed thus far."

[QUOTE=gman003-main;35324417]Somehow I haven't heard about this graphene computer shit. Please, do tell.[/QUOTE] Graphene is probably as you know the strongest material known to man. It's also the material used for fastest transistors I know of and possibly known to man, 155Ghz per transistor I think is the record (You should probably know that doesn't translate to clock speed :v:). IBM are the ones leading in it, mostly funded by DARPA from what I can tell. In comparison to 12.6Ghz from silicon, if memory serves. But I'm probably wrong on the Silicon stuff. It's an interesting and fucking awesome field right now, look into it. The main advantage to graphene is it's size, one atom thick sheets and strong as fuck. [editline]28th March 2012[/editline] Interesting. It appears like the college blocks anything that's "Uncategorised". That means Facepunch is categorized and not blocked. So, which one of you works for DET in Au?

[QUOTE=gman003-main;35324493]Look, I think maybe *one* of our servers has a working gigabit ethernet interface. Throughput will *not* be an issue. [editline]27th March 2012[/editline] I got used to playing with gigabit line rate switches. My apologies. Anyways, I seem to be able to get any type of product from this one company. Too bad they don't make 10GbE SFP+ transceivers. ( THOSE THINGS ARE MORE EXPENSIVE THAN YOU THINK )

[QUOTE=pure.Joseph;35324580][QUOTE=gman003-main;35324493]Look, I think maybe *one* of our servers has a working gigabit ethernet interface. Throughput will *not* be an issue.[/quote] [editline]27th March 2012[/editline] I got used to playing with gigabit line rate switches. My apologies. Anyways, I seem to be able to get any type of product from this one company. Too bad they don't make 10GbE SFP+ transceivers. ( THOSE THINGS ARE MORE EXPENSIVE THAN YOU THINK )[/QUOTE] Get me a 20 port Gbit switch for my room.

[QUOTE=wingless;35324532]Graphene is probably as you know the strongest material known to man. It's also the material used for fastest transistors I know of and possibly known to man, 155Ghz per transistor I think is the record (You should probably know that doesn't translate to clock speed :v:). IBM are the ones leading in it, mostly funded by DARPA from what I can tell. In comparison to 12.6Ghz from silicon, if memory serves. But I'm probably wrong on the Silicon stuff. It's an interesting and fucking awesome field right now, look into it. The main advantage to graphene is it's size, one atom thick sheets and strong as fuck.[/QUOTE] Don't know what the limit on silicon is, but [url=http://www.highbeam.com/doc/1G1-157174290.html]here's an indium-based transistor running at 845GHz[/url].

[QUOTE=wingless;35324594]Get me a 20 port Gbit switch for my room.[/QUOTE] Would you want 4port SFP+ 10GbE to go with those 20 port RJ45 GbE? 10GbE = around 1.2GB/s, yes GIGABYTES, not gigabits.

[QUOTE=gman003-main;35324601]Don't know what the limit on silicon is, but [url=http://www.highbeam.com/doc/1G1-157174290.html]here's an indium-based transistor running at 845GHz[/url].[/QUOTE] FFFFascinating. [editline]28th March 2012[/editline] [QUOTE=pure.Joseph;35324611]Would you want 4port SFP+ 10GbE to go with those 20 port RJ45 GbE? 10GbE = around 1.2GB/s, yes GIGABYTES, not gigabits.[/QUOTE] I wouldn't complain (Unless it broke).

You guys make me feel dumb with your sophisticated 'puter talkin'.

[QUOTE=wingless;35324613]FFFFascinating. [editline]28th March 2012[/editline] I wouldn't complain (Unless it broke).[/QUOTE] Well, so. I was on WHT ( webhosting talk ) asking others opinions of those particular switches. Someone said D-LINK OEM'd for them, so after talking with the people at the company of those said switches, I now know that, it was the completely opposite. They build these switches in house, and OEM for the likes of D-Link. Nice Eh? The house I'm working on, should have a 12-24 port PoE switch for the security cameras, and another 12-24 for web connected devices. Such as Wi-Fi access points, Media Server, puters, and PS3 / TV's. Should be nice to play with. Maybe I'll just segregate the network onto one single server, and VLAN it out. But then if the server fails, there goes my Media VM, security VM, and even my PFSense setups :( Maybe I should document the whole build in glorious 1080P?

[QUOTE=pure.Joseph;35324671] Maybe I should document the whole build in glorious 1080P?[/QUOTE] Please do.

[QUOTE=SuperDuperScoot;35324645]You guys make me feel dumb with your sophisticated 'puter talkin'.[/QUOTE] It's not so hard. First, [url=http://en.wikipedia.org/wiki/Main_Page]read this.[/url] All of it. If big words confuse you, try [url=http://simple.wikipedia.org/wiki/Main_Page]this instead[/url]. Then start reading [url=http://slashdot.org/]this.[/url] And follow [url=https://twitter.com/#!/climagic]this[/url]. And read [url=http://thecodelesscode.com]this.[/url] And [url=http://www.catb.org/jargon/html/index.html]this[/url]. And maybe [url=http://www.theregister.co.uk/odds/bofh/]this[/url] too. Probably wouldn't hurt to start reading [url=http://www.gamasutra.com/]this[/url], although that's more gaming than computing (lots of technical shit though). And I still don't know why I read all of [url=http://tvtropes.org/pmwiki/pmwiki.php/Main/HomePage]this shit[/url], but why not add that to the list too. Once you're done with that, pick a side to fanboy for - pick a processor, an operating system, and a language. Me, I like SPARCs, OpenBSD and C/C++; Wings likes Itanic, Arch and... some language. Look hard enough, you'll find a guy who insists on using Common LISP on the OS/2 and cross-compiles stuff to the 6502.

[QUOTE=gman003-main;35324716]It's not so hard. Me, I like SPARCs, OpenBSD and C/C++; Wings likes Itanic, Arch and... some language. Look hard enough, you'll find a guy who insists on using Common LISP on the OS/2 and cross-compiles stuff to the 6502.[/QUOTE] [QUOTE=SuperDuperScoot;35324645]You guys make me feel dumb with your sophisticated 'puter talkin'.[/QUOTE] You're not the only one :(

[QUOTE=gman003-main;35324716]It's not so hard. First, [url=http://en.wikipedia.org/wiki/Main_Page]read this.[/url] All of it. If big words confuse you, try [url=http://simple.wikipedia.org/wiki/Main_Page]this instead[/url]. Then start reading [url=http://slashdot.org/]this.[/url] And follow [url=https://twitter.com/#!/climagic]this[/url]. And read [url=http://thecodelesscode.com]this.[/url] And [url=http://www.catb.org/jargon/html/index.html]this[/url]. And maybe [url=http://www.theregister.co.uk/odds/bofh/]this[/url] too. Probably wouldn't hurt to start reading [url=http://www.gamasutra.com/]this[/url], although that's more gaming than computing (lots of technical shit though). And I still don't know why I read all of [url=http://tvtropes.org/pmwiki/pmwiki.php/Main/HomePage]this shit[/url], but why not add that to the list too. Once you're done with that, pick a side to fanboy for - pick a processor, an operating system, and a language. Me, I like SPARCs, OpenBSD and C/C++; Wings likes Itanic, Arch and... some language. Look hard enough, you'll find a guy who insists on using Common LISP on the OS/2 and cross-compiles stuff to the 6502.[/QUOTE] Later, after sleep.

[QUOTE=gman003-main;35324716]It's not so hard. First, [url=http://en.wikipedia.org/wiki/Main_Page]read this.[/url] All of it. If big words confuse you, try [url=http://simple.wikipedia.org/wiki/Main_Page]this instead[/url]. Then start reading [url=http://slashdot.org/]this.[/url] And follow [url=https://twitter.com/#!/climagic]this[/url]. And read [url=http://thecodelesscode.com]this.[/url] And [url=http://www.catb.org/jargon/html/index.html]this[/url]. And maybe [url=http://www.theregister.co.uk/odds/bofh/]this[/url] too. Probably wouldn't hurt to start reading [url=http://www.gamasutra.com/]this[/url], although that's more gaming than computing (lots of technical shit though). And I still don't know why I read all of [url=http://tvtropes.org/pmwiki/pmwiki.php/Main/HomePage]this shit[/url], but why not add that to the list too. Once you're done with that, pick a side to fanboy for - pick a processor, an operating system, and a language. Me, I like SPARCs, OpenBSD and C/C++; Wings likes Itanic, Arch and... some language. Look hard enough, you'll find a guy who insists on using Common LISP on the OS/2 and cross-compiles stuff to the 6502.[/QUOTE] ... I... ........ I didn't even get to read any of what you linked and my brain is already broken.

[QUOTE=gman003-main;35324716]Once you're done with that, pick a side to fanboy for - pick a processor, an operating system, and a language. Me, I like SPARCs, OpenBSD and C/C++; Wings likes Itanic, Arch and... some language. Look hard enough, you'll find a guy who insists on using Common LISP on the OS/2 and cross-compiles stuff to the 6502.[/QUOTE] Itanium is amazing and you know it, sure it had a rough start but it's grown. SPARC has gone the other direction. It has history as a great arch but now it's as good as dirt. Maybe if MAJC was properly released, it would have easily been the best high end server processor ever. Oh and HP-UX over Arch anyday, I've been converted. It's fucking brilliant. Oh and I'm just a standard C# guy. It has it's problems, but it's like Java++, the way I look at it.