Posts by Roo

Re: ooooouch

"Texas Instruments has been making ICs longer than Intel. They know much better than Intel how to compete on the low end high volume thin margin range as well as some of their competitors."

Not sure where you are going with TI, but last I heard they had shitcanned the OMAP roadmap - they didn't do very well at low end or high end on that front. Releasing fatally wounded silicon late might have had something to do with it, or perhaps they weren't able to adapt to the shift of manufacturing from the West+Japan -> Korea/China.

Re: ooooouch

Skaugen is probably right, but the Wintel partnership should not be underestimated. True, it lost the mobile war but seems pretty impregnable on the desktop, even in 2013.

"Content will always need to be generated and generation happens on the desktop, not the phablet. More phablets will demand more content which will require more desktops IMO."

Your opinion is drawn from incomplete data. While you may well be correct that more desktops will be needed to make more content, currently the vast majority of Desktop PCs are used to consume content - and it's this vast majority that is getting shit-canned.

Some people at Microsoft had the right idea with Win Mobile & the tablet PC efforts - but the execution was intentionally bolloxed - probably so it didn't eat into the desktop margins. Whatever else you say about Jobs he wasn't afraid of the new gear shredding the legacy box shifting effort, and that is why Apple are sitting on a huge pile of cash.

Meanwhile Microsoft flounder around as their ever loyal Excel fanbois yell about how they are too stupid to learn how to use alternative tools. The world is passing them all by, even in finance high frequency trading is now responsible for the majority of trades made, people slotting Excel into the critical path would lose a lot of money.

Re: Wintel irrelevance == x86 irrelevance

"IMHO it was very sad that DEC came up with one of the best CPU architectures ever (Alpha) just at the point when they were going bust for other reasons. Intel should have bought the IP and the designers from DEC and run with that but by that time they were far too heavily invested in x86 both financially and intellectually."

Well Intel *did* buy the IP & designers from DEC as part of a settlement of an IP infringement lawsuit, however their aim was to shut up the lawyers and bury one of the other 64bit architectures (Alpha) while they hacked away at IA64 (aka Itanic). At the time Intel planned to keep x86 32bit and relegate it to legacy/low-end stuff while everyone else was migrated to IA64. Thankfully the world was spared from a monopoly shoving an under performing freak-show ISA down it's throat by 'AMD64' and the Opteron. Thank you AMD.

A shortish while ago the Reg published an article claiming that a Chinese uni had built an Alpha ISA chip that yielded some very respectable FLOPS/W figures - I have had a hunt for more info - but I have struggled to find much beyond the initial press release. I think I'll have to learn Mandarin to find out more. :P

Re: There are a couple of lines that could work for *specific* applications

"So you need to design a number system that allows numeric operations without carry between digits."

Numeric operations that fit that category would vectorized nicely - oh look we have vector processors, Cray built a few of those, folks like Intel, AMD & Nvidia build millions of them a year.

"4) Transistors have a fan out and fan in of (typically) 10. Human neurons can go 1:10 000. Bigger is better, but at least matching a transistor is a start."

There are no 'limits' as such to a transistors (I suspect you mean gates) fan out and fan in. In the case of fan-out you simply sacrifice switching speed or burn more juice driving all those inputs (inputs typically look like capacitors - it takes time to (dis)charge a cap).

I guess there are practical limits in terms of how you wire up such a beast though, routing 10K lines across a die strikes me as tricky - maybe the tools make it trivial now. :)

Re: Grow the other baby.

"I would happily take a computer that was twice the size, and used twice the power but was ten times as fast."

I think they could probably whip up one that was 4 times the size, used 20 times the power and was 10 times a fast and 100 times more expensive using what they've already got if you have the money. Just make sure you keep the application's working set small - say <64Kwords of Data & <64Kwords of Instructions - so you don't get hit by memory access penalities too often... You may get away with making the words very wide - but making good use of that in software is tricky. That's kinda where GPUs are good at anyway. :)

ECL would be slower simply because you can't achieve the density - and you pay latency penalties as a result.

My guess is people will achieve some amazing results with alternative technologies, but I believe these guys when they tell me CMOS is nigh on tapped out (as far as *cost effective* chips go - specialist apps may well tolerate higher costs of course).

Re: DRAMA!!!!

"Some of us started with electronics too.

Actually one way out of this mess is an ANALOGUE cell that can handle more than one bit - imagine 8 logic levels between one and 8 (0--5V) handled by a single switching element. 8 bit adder would be a snap..as would an 8 bit comparator"

Right - so you have big voltage swings -> more power loss driving signals, oh and a lower signal/noise ratio.

You might be better off sticking with an analogue computer - they exist. :)

Re: DRAMA!!!! @ stu 4

Speaking a developer & very occasional software engineer, Amen to that. Have an upvote !

Re: DRAMA!!!!

"We'll have to build chips which implement better internal logic and design, rather than taking the easy route of shrinking everything..

The various chip design journals & papers are worth a read, you should read them so you don't have to take my next statement on trust. The chip design bods have zillions of neat tricks above & beyond sitting on their duffs waiting for the process guys to pull the next rabbit out of the hat.

Now back to HPC...

Essentially for 'real' HPC applications time is money - which is why they throw a lot of money at making stuff go faster. With that in mind...

1) When you start running programs on physically large systems (multi-core, multi-processor, multi-rack, multi-site etc), you will eventually want to aggregate all those results of those sub-calculations together, and that is where latency kicks you in the balls.

2) At present the lower bound on latency is set at the speed of light (C). So far we haven't found a way to increase the value of C.

3) So if we want to improve latency, but we can't increase the speed of the signals tweak the speed of light, we are left with reducing the distance travelled by the signals (ie: shrinking stuff).

The CMOS guys have actually done a very good job. Even Intel seem to have got to the point where the only real tweak they have left is upping the cache size, and at this point they seem to have run into the latency wall there too as far as CMOS goes (increasing cache size increases latency of the cache).

Re: The CLOUD brought down by a single NIC card?

"One would have thought that the people building the CLOUD would have designed in such failure detection from the beginning. What effect would failure of component X have on the system-wide performance etc."

Detecting sub-optimal performance can be tricky. In the example given the NIC appears to still be passing traffic, so it hasn't failed as such - it's just slow. Perhaps the sink for the data isn't keeping up so flow-control is throttling the data rate, or perhaps the auto-negotiation is picking the wrong value, or maybe segment congestion is killing the throughput, it could be starved of memory bandwidth etc.

If you choose to apply a simple threshold, what value do you pick for the threshold ? How do you account for averaging effect of legitimate idle periods or segment congestion on the measured throughput ?

Then if you decide to blacklist that component instead of tolerating it's degraded performance what will happen when you redirect that traffic via another set of components ? Sometimes (actually quite often in practice) fail-over can cause components to degrade or fail because they are suddenly deluged with extra work.

Some times fail-over is a very costly process in itself (state transfer, sync etc) so the time & space resources expended during the fail-over can actually outweigh the potential savings from blacklisting a degraded component.

It's nice to see folks trying to quantify this stuff, I just hope they're doing something new rather than repeating the distributed systems research done in the 50s/60s/70s/80s.

I suspect they are simply repeating research because they think the word 'Cloud' somehow changes all the rules of distributed computing... Either way I'm sure they'll be rewarded for a new buzzword that will give warm fuzzies to ignorant salesmen, fanbois and execs.

"He *claimed* that, but his actual observation is simply that they aren't talking as much as they used to."

Remember, for that kind of role you need absolutely zero education or competence at stuff like gathering evidence, assessing it and drawing conclusions from it. In fact the folks who pay no attention to inconvenient stuff like facts and reality tend to do better at gaining those positions because they are able to tell people what they want to hear.

Ray McGovern has written about this lack of care and attention to reality on the part of intelligence officials and their overseers over in Leftpondia for several years now. The same seems to apply in Rightpondia as well judging by the rubbish we see the state owned media (aka BBC) parroting at the moment,

Re: Only got themselves to blame..

"I'm struggling to feel any sympathy to his co-workers, but is anyone really surprised that even in the NSA people don't quite get the importance of credentials being private?"

The officials who provide oversight over this whole mass surveillance shebang really need to be investigated themselves. Snowden did all this stuff on their watch, and it has shown that their oversight of these operations has been opaque, inadequate and ineffective.

If they had any sense of duty to the nation they would be working hard to fix the lack of oversight that led to Snowden gaining access to this stuff just by asking people for their credentials, instead of wasting valuable time vilifying Snowden.

I do wonder if any of those cretins have actually considered the possibility that if Snowden could get this info so easily in such a short period of time, then perhaps the angry blow shit up type terrorists already took that same information (and more) several times over.

Re: FLOSS nutter

"If a Windows user (more likely to be lay computer user) has a good experience with things like The GIMP or Libre Office, they are more likely to try a Linux distro."

It's possible that might happen, but I doubt many people would choose Linux because their favourite Windows app happens to be open source - because many of them will think it is easier to carry on running OpenSource apps on their current OS than to migrate to a new OS.

I agree with your argument to a certain extent though... I think it's fair to say they may be more open to moving to a Linux distro having had good experiences with stuff like GIMP & Libre Office.

Re: Molehill->mountain @roo

"You missed the bit about an iPhone being usable as a Wifi hotspot?"

No, but you clearly missed the point about mobile data access not being available 100% of the time.

"Most road warriors that I know connect back to home base. Do you have a particularly ineffective IT department where you work?"

The efficacy of an IT dept is irrelevant if you don't internet access in the first place. It is also irrelevant if you are simply on holiday and don't want to share your holiday experience with the office.

Why do you find it so hard to accept that people might legitimately have a use for a USB cable now and then ?

Re: Who is Mike Shema ?

"Then again, I would not let a "current or former" Windows admin into my office with or without broom, never !"

I'd give them a chance to renounce their ways/learn to admin a UNIX box before I barred them. With broom would be preferable in the instances where they need ejecting from the premises.

Re: Molehill->mountain @roo

"Secondly your use cases are, to say the least, extreme."

I really don't see the lack of unfettered WiFi and/or roaming data as being extreme. It's a fairly routine occurrence when you are travelling in my experience, I guess your experiences are different to mine, lucky you.

From my personal point of view I would rather use a USB cable than sync by hand because it's faster, easier and more accurate. Also given the choice of using a USB cable or syncing via an untrusted third party I would rather use the USB cable too. I don't see any benefit in making the lives of marketing wonks and script kiddies any easier than it has to be.

I am quite happy for you to mess about entering stuff by hand, set up servers or simply punt you and your friends personal data to a third party that you have absolutely no control or governance over.

Re: Molehill->mountain

I have some sympathy with your assertion that a mountain is being made of a molehill, but I think the underlying complaint is a fair one.

"So someone sufficiently paranoid as to air gap a PC (even to the point of not allowing a bluetooth PAN) is going to risk their data on a mobile, Internet connected phone?"

Firstly it is not down to you to decide whether giving away all your contacts to a third party is a good thing for everyone or not. There are examples of organisations disciplining people for using that data to stalk people (British Police Forces and NSA included), so for some people I think there are very good reasons for NOT spreading that kind of information any further than necessary.

Secondly, there are other uses cases that don't have paranoia as a source of the air gap. Here's a couple of examples...

Sometimes people do not have roaming data access (or WiFi) but they would still like to transfer some recently acquired contacts to their phone from their laptop.

In some cases even if they do have internet access on both their phone and their laptop they still find it easier to hook them up with a USB cable instead of farting about with entering WiFi Keys or working around web content filtering.

I think most people who travel have experienced those two use cases at least once in their lives. While the world won't end because folks with Mavericks can't sync their contacts via a USB cable, I think it's fair to say that given the choice most of them would prefer to have the option, right ?

Re: Molehill->mountain

"They've provide users with a perfectly good alternative to cabled synch but all we get are haters ascribing negative reasons for it."

As I understand it is NOT "a perfectly good alternative" because it takes away the ability for you to sync using a cable hooked up to an air-gapped host...

Re: Let's not even go there shall we. @Muppet 09:00

"Yes - that's the main use for Linux - isolated command and control systems where the vulnerability risk from Linux is not too great. For networked systems they primarily use Windows..."

Right, so they can get hacked by amateurs in their bedroom like Gary McKinnon.