Slashdot Mirror

Those are already happening; when finally mature, why would you use the cloud?

I'd say, even old MFM and RLL HDDs were mature enough for the needs of the day. I had no need for cloud then, and I have no need for cloud now. We need abstract data exchange between parties, but that's not "cloud." The term covers "paid data exchange between separate instances of you." I guess there are cases when it makes sense, but I am not involved with any of those, and my data is kept away from the Internet except what I choose to publish.

There is one more interesting factor. Data exchange on your LAN is inherently parallel. Alice, Bob and Charlie can send terabytes of data on their personal LANs, and they don't interfere. Alice can even run ten parallel networks at her business if she wants to (some do, for various reasons.) However data exchange with a certain cloud provider has to share the same bottleneck(s) somewhere, especially if done over the wireless link. The capacity of wireless channels is limited by physics; you can jump up and down all you want, but you have to go for wider channels if you want more data... and then you are hit with shorter range of communication. At 60 GHz you are limited by just a few meters. It's amazingly wasteful to do that to a shared, finite (but, fortunately, instantly renewable) resource.

I remember PC Chips doing something similar. They manufactured motherboards with fake L2 cache, from the old 486 into early Pentium 1 era. Do you think I jest? They glued fake plastic chips onto motherboards and then simply programmed the BIOS to report the cache enabled, even though there was none to begin with.

5-12MHz 5-12MHz 16-bit (8086) June 1978 29 thousand

(from http://redhill.net.au/c/c-1.html

So, 29k transistors. Hmm. How many transistors does a wristwatch have? I'm guessing, for instance, that my P.O.S. Timex LCD critter can't have too much semiconductor complexity. OTOH, this prolly has scads and scads more transistor-y goodness.

bad boards - how to recognise and avoid them

http://www.redhill.net.au/b/b-bad.html

This section, however, is not about the normal variation in quality and reliability between typical motherboards. It is about plain old-fashioned greed, and the cheap, shonky boards that sometimes result from it. Here then, is a short gallery of the cheap, the nasty, and the outright fraudulent.

To quote for the Red Hill web page:

PC Chips fake cache 486

Let's begin with the most famous of them all: the fake cache 486 boards that PC Chips produced in the mid-Nineties.

---------------

From the PCCHIPS website we find: http://www.pcchips.com.tw/PCCWeb/AboutCOMPANY/Abou tCOMPANY.aspx?MenuID=8&LanID=2

PCCHIPS has been a leading supplier of motherboards and PC peripherals since 1994. We are committed to provide products of superior value and exemplary customer service to our customers worldwide.

http://www.pcchips.com.tw/PCCWeb/Legal.aspx?MenuID =8&LanID=2

The materials ("Materials") contained in this web site are provided by Elitegroup Computer Systems Co., Ltd. ("ECS") ...

I think these quotes speak for themselves.

Here's a nice piece on CPU history: http://redhill.net.au/iu.html

It's PC centric and doesn't go into anything but x86 though. I found it when i was doing some research on a Pentium chip I had and I liked the guide a lot, it covers most chips from the early Intel's to Cyrix and all the way to the Athlon's. Some homework for the wippersnappers, ey? =)

According to the Chiplist, the original second-sources for the 8088 were AMD, Harris, Siemens, Hitachi and NEC (V20).

For the 286, it was AMD, Harris, Siemens and Fujitsu.

For the 386, Intel cut their second-source contracts, but it seems there were quite a few clones: AMD (SX license, DX clone), Cyrix (clone, manufacturered by TI), Harris (clone), IBM (licensed 386SX), and Chips & Technologies (clone).

There's a timeline, pictures and lots of other information at the Red Hill CPU Guide.

Also, you can look here: http://redhill.net.au/d/d-o.html
For the baracuda 2HP, who used the dual actuator approach....

Maybe those things will be coming again, if the storage density stagnates like the last years.
Bot otoh, if its too expensive, falling ram prices may create alternatives for applications that NEED that much performance...

Actually, the nastiest drive I can recall were the BigFoots made for Compaq. Some of my personal observations for you:
* was the same form as a cd drive
* had one platter that spun at ridiculously low speed
* access time was abysmal
* was louder than the case fan
* prone to shock damage (you could lightly bump the case with your knee and cause bad sectors on the disk)

It was made to Compaq's request for a really cheap HD to go in their home computers. Anyway, it's kinda fun to take apart. One of the only drives I've seen that makes me ill when I have to leave it in.

Let's just say that I consider the PS2's design (~ 50 Watts) a lot smarter than XBox1's (~ 110 Watts). And just like XBox1, XBox360 is (again) using a brute-force approach.

If a smarter design isn't possible or not economical, then just reduce the clockrate.

As to reducing the polygon count, I don't happen to agree that is a viable option. It is an option, but MS needs to beat PS3 in the marketplace, and surely they don't feel that having a lower-performing console will do that for them.

As far as can be currently seen, the PS3 will have more output connectors (2 screens instead of 1), more controllers (7 instead of 4), more media capacity (40GB instead of 9GB) and full backwards compatibility (instead of "some" games)

The polygon count is the least of Microsoft's problems, also since PS3 is launching later they will probably beat them in polygon-count anyway.

However, there are technical issues to detecting overheating. Basically, you need a temperature sensor in the place that the overheating happens. You cannot just have temperature sensors in every chip, it would mean having custom versions of every chip in the unit, which wouldn't be cost effective, for example you couldn't even use commodity RAM.

That's absolutely no problem if you design it with some reserves in mind. If you know that at around 80 sensor temperature the thing starts crashing, make sure it doesn't reach 70.

Of course Microsoft seemed to have pushed everything to the limit in the XBox360 and that causes lots of (IMO unnecessary) problems.

Well, I have to say I have no way to be sure that PS3 will use as much power. But I would bet significant amounts of money on it. By "as much", I mean within 25%, for a rated of 160W, 120W typical. This would also be too hot to run in a typical unvented stereo enclosure. With an NVidia graphics chip in there, plus the Power PC, plus 8 coprocessors it's not going to be a cool device either. I mean, check the power usage ratings on current NVidia graphics chips, they can easily take 80W on their own, and an optical drive can take 5-10W more easy (depending on whether it is a laptop-type drive or regular one like 360 uses). That leaves 30-35W for the RAM, CPU and cooling system power draw. But we'll just have to finish this discussion when PS3 comes out.

I consider the PS2 (moderate power usage, INTERNAL power supply) a pretty good design, of course that doesn't guarantee that the PS3 will be, too. If the PS3 will be crap, that doesn't make the XBox360 any better.

Anyway: Yes, an NVidia-chip can suck down 80W on its own, however only if run at full speed. Again, if we look at PS2, which AFAIk runs at somewhere around 200 MHz, which was roughly 1/2 to 1/3 of what PC-CPUs were running at that time (see for example here), then it's quite possible that Sony is smart and clocks the PS3 at some speed that doesn't push every part to the limit, hopefully they also keep the internal power supply.

I mean, let's be realistic here: The PS2 and XBox1 are already "fast enough" for most games and there is absolutely no pressing need for more power. Don't get me wrong, of course the new generation has to beat the old one by a longshot, but there is absolutely no need anymore to push it so far that overheating becomes an issue.

But of course we will see. But if PS2 gives any hints about what PS3 will be (and since PS2 was the most successful console ever, they will keep what was successfull if they are smart) then the PS3 will not use a brute-force approach.

Microsoft seems to have not noticed yet, but the console-business lifes off the GAMES and not the h

http://redhill.net.au/b/b-bad.html

ECS has taken advantage of their recent merger with PC Chips

As I recall it was PC chips who produced the fake cache on the 486 motherboards. Look here:
http://www.redhill.net.au/b/b-bad.html "PC Chips fake cache 486"

I do have an ecs board but it was before the merger. It was stable for years.

nevertheless - there are reputable manufacturers out their so why would I care about ECS/PC CHIPS?

AMD provides a better value than Intel, and has since the first generation Athlons. That was what? 6 years ago?

Recent occurrence my ass.

http://redhill.net.au/c/c-f.html

PLEASE, try to know what you're talking about.

A dozen? Don't be so conservative. :) I couldn't find any transistor count number for the MC68030, but this page says the MC68040 had 1.2 million transistors (quite a jump up from the 68,000 of the original MC68000).

A Pentium 4 seems to have either 55 million or 125 million, depending on the core generation (those are "Northwood" and "Prescott" cores, respectively), all according to this page. There might be newer generations still, I'm not 100% up to speed on Intel CPUs.

Thus, you can fit either 55/1.2 = ~45 or 125/1.2 = ~104 MC68040s in the transistor budget of a single Pentium 4. Of course, this is just back-of-the-envelope numbers, and I'm not enough of a silicon geek to know how much "glue" would be needed. Plus, trading cache transistors for gate transistors as this calculation does is probably not OK. Not to think about memory bandwidth requirements to feed 55 cores ... But still--one can dream!

Isn't ECS the parent company of PCChips, the company that not only used to sell crap boards with fake cache chips (which even had modified BIOSes that claimed that the cache was real), but even had relabled chipsets and sometimes exploding capacitors? Newer boards by them are also usually flakey, mostly due to the fact that they use the cheapest possible components.

http://www.redhill.net.au/b-bad.html
http://www.rainbow-software.org/hardware.html

One report says that PCChips/ECS is all of these brand names:
PCChips, Amptron, Protac, Aristo, Minstaple, Eurone, Matsonic, ECS, and possibly more.

(I should get a Protac motherboard!!! oh yeah!! lol)

Here's something interesting I just found:
http://www.redhill.net.au/b-02.html
quoted:
------------------
"ECS K7S5A

The SiS 735 chipset is a particularly interesting one. It was first previewed in mid-2001 when the DDR main board market was in the doldrums. The ALI entrant was considered a non-starter, the VIA KT-266 buggy and very late, the well-performed and stable AMD 760 was dear and in short supply, and VIA's SDRAM-only KT-133A was taking all before it. Nvidia's much-touted Nforce was still vapourware. Unless you just didn't care about the cost, a KT-133A was the only rational choice.

Then, from out of nowhere, came the SiS 735: an entrant from a firm that had all but foundered in its sudden rush to build its own fab facilities and cut ties with its former manufacturing partners, a firm that had little left but a reputation, and that a poor one. To everyone's astonishment, the SiS 735 was the clear benchmark leader, and in most respects it still was right up until the transition to 166MHz FSB chipsets began: if we are to disregard the weird all-in-one Nforce, only VIA's second-effort KT266A could beat it. SiS had three hurdles to overcome: the first was demonstrating competitive performance. This they had already done. The second was demonstrating stability and compatibility: this too was within their measure. And the third was getting mainboard makers to adopt it. This was perhaps the hardest task, as mainboard makers are reluctant enough to offend Intel by making VIA and Athlon products, offending both Intel and VIA at the same time requires more than the usual bravery. SiS chose to overcome that reluctance by making the 735 an offer just too good to refuse. It was very cheap. For a high-tech state-of-the-art DDR chipset, it was amazingly cheap.

Elite are surprisingly little-known for a company that is one of the largest mainboard manufacturers of all. They are bigger than ever since their merger with the infamous PC Chips group (the fake cache people) in the late 1990s. They made quite a splash on the overseas markets with this board, one of the very few to use the SiS 735 chipset, and once they overcame a well-publicised BIOS problem, were very successful with it. Here at Red Hill we had been very happy with our KT266A mainboards and had no need to switch, but with the Elite coming in anything up to fifty dollars cheaper than a KT266A, it would be foolish not to try them.

In the flesh, the boards had that familiar PC Chips look about then: they were alarmingly thin and very cheaply made. Our first impression was that there was no way these could be as reliable as our Epox and Soltek KT266As (or our Soltek KT133As, for that matter, for these were a dual mode board that can take SDRAM or DDR), and our past experience with PC Chips associated companies did little to encourage us. Still, we gave a pair of them every chance to show their stuff. We soon found that they were fussy about RAM and incompatible with Athlon Thunderbirds. Not a great start. From there it got a good deal worse: the more we tinkered with them in the workshop the more apparant it became that they were unstable. Quite often they wouldn't even run error-free

Isn't ECS the parent company of PCChips, the company that not only used to sell crap boards with fake cache chips (which even had modified BIOSes that claimed that the cache was real), but even had relabled chipsets and sometimes exploding capacitors? Newer boards by them are also usually flakey, mostly due to the fact that they use the cheapest possible components.

http://www.redhill.net.au/b-bad.html
http://www.rainbow-software.org/hardware.html

One report says that PCChips/ECS is all of these brand names:
PCChips, Amptron, Protac, Aristo, Minstaple, Eurone, Matsonic, ECS, and possibly more.

(I should get a Protac motherboard!!! oh yeah!! lol)

Here's something interesting I just found:
http://www.redhill.net.au/b-02.html
quoted:
------------------
"ECS K7S5A

The SiS 735 chipset is a particularly interesting one. It was first previewed in mid-2001 when the DDR main board market was in the doldrums. The ALI entrant was considered a non-starter, the VIA KT-266 buggy and very late, the well-performed and stable AMD 760 was dear and in short supply, and VIA's SDRAM-only KT-133A was taking all before it. Nvidia's much-touted Nforce was still vapourware. Unless you just didn't care about the cost, a KT-133A was the only rational choice.

Then, from out of nowhere, came the SiS 735: an entrant from a firm that had all but foundered in its sudden rush to build its own fab facilities and cut ties with its former manufacturing partners, a firm that had little left but a reputation, and that a poor one. To everyone's astonishment, the SiS 735 was the clear benchmark leader, and in most respects it still was right up until the transition to 166MHz FSB chipsets began: if we are to disregard the weird all-in-one Nforce, only VIA's second-effort KT266A could beat it. SiS had three hurdles to overcome: the first was demonstrating competitive performance. This they had already done. The second was demonstrating stability and compatibility: this too was within their measure. And the third was getting mainboard makers to adopt it. This was perhaps the hardest task, as mainboard makers are reluctant enough to offend Intel by making VIA and Athlon products, offending both Intel and VIA at the same time requires more than the usual bravery. SiS chose to overcome that reluctance by making the 735 an offer just too good to refuse. It was very cheap. For a high-tech state-of-the-art DDR chipset, it was amazingly cheap.

Elite are surprisingly little-known for a company that is one of the largest mainboard manufacturers of all. They are bigger than ever since their merger with the infamous PC Chips group (the fake cache people) in the late 1990s. They made quite a splash on the overseas markets with this board, one of the very few to use the SiS 735 chipset, and once they overcame a well-publicised BIOS problem, were very successful with it. Here at Red Hill we had been very happy with our KT266A mainboards and had no need to switch, but with the Elite coming in anything up to fifty dollars cheaper than a KT266A, it would be foolish not to try them.

In the flesh, the boards had that familiar PC Chips look about then: they were alarmingly thin and very cheaply made. Our first impression was that there was no way these could be as reliable as our Epox and Soltek KT266As (or our Soltek KT133As, for that matter, for these were a dual mode board that can take SDRAM or DDR), and our past experience with PC Chips associated companies did little to encourage us. Still, we gave a pair of them every chance to show their stuff. We soon found that they were fussy about RAM and incompatible with Athlon Thunderbirds. Not a great start. From there it got a good deal worse: the more we tinkered with them in the workshop the more apparant it became that they were unstable. Quite often they wouldn't even run error-free

The 486DX/50 (note, not the 486DX2/50) ran on a 50Mhz system bus (or "Front Side Bus" for the young'uns). At the time, this caused hardware developers massive headaches because it was a nightmare trying to build a motherboard that could keep up. VLB (VESA Local Bus) machines, in particular, were the worst as only one (of usually 2 or sometimes 3) slots could run at 50Mhz without adding an extra wait state. Indeed, most VLB 486 boards at the time had a specific jumper to use if you had a 486DX/50 with more than one VLB slot in use that added an extra wait state to keep the system stable.

Here are some websites that confirm this. Alternatively, if you want to trawl through google groups there should be literally hundreds of "DX/50 vs DX2/66" threads arguing about how the DX/50's higher bus speed makes up for the DX2/66's higher clock speed (like this one). Also, if you can get hold of a 486 board, pretty much anything with VLB on it should have a bus speed setting for 50Mhz and another jumper to add a wait state to the VLB. Or, if you're feeling particularly adventurous, Intel's design docs for the 486 should provide the information - although you may have trouble finding info on something so old, the DX/50s were never common because they were quite expensive.

As someone who owned (still do, it's just packed away in a cupboard) a very expensive (at the time) 486DX/50 system, I take offence at you implying my old workhorse doesn't exist ! :). Heck, in the closing days of the 486, there were even a few DX4 chips that could be coaxed into running at 3x50=150Mhz, if you could keep them cool enough.

Sure. But what exactly is waiting on the memory? Your joystick?

The point is if there's wait states there it's the fault of slow memory, not the CPU. Your DX/25 might have been faster than a slow-memory crippled DX/33, but that was because of the memory, not the CPU - and I'd be highly sceptical of even such a crippled DX/50 or DX2/66 being slower at anything except a few corner-cases. The other thing to consider, of course, was those were back in the days where the market was rife with people selling motherboards that didn't have any - or fake, nonfunctional - L2 cache. A decent 386 would probably be faster than a 486/33 without any L2 cache., so if the machine you were comparing to was hamstrung like that, it would also have been (*much*) slower.

When the 386 was introduced it was, clock-for-clock, twice as fast as the 286, no recompiling necessary.

No, quite the opposite, actually.

Thanks for confirming my fears that ECS WAS PC Chips... Looks like ECS is another brand I won't buy...

Umm... I'll replace the chips on the RAM I was going to send you with fake plastic chips if you forgot about coal...

seagate barricuda 2hp

explanation. Still don't like it. We needed something like this. I did see a Bestbuy computer that had raid by vector i think.

seagate barricuda 2hp

Should have kept making them. Raid is not great for the desktop.

It even has a picture of the Media-GX in there.

This site actually has a lot of good info on the early microprocessors, complete with pictures. They've even got info and pictures of the 4004! :)

60 GB capaciy and cool and all, but what really makes it sweet is that it's got the classic styling of a 1997 Western Digital hard drive. Beauty.

Nope, it's not of the future. The HDD future belongs to something else: platter-level RAID systems.

I'm sorry to be blunt, but that is incorrect. If you are interested in the future of storage, follow StorageReview.com's forums. Every other week or so when someone comes into the forums complaining that nobody makes such a drive, or comes up with this "great new idea!!", they are usually referred to the website of Tannin, one of the more knowledgeable members, who keeps track of this sort of thing. Here is that website: Odd drives

I don't mean to be rude, but you made your statement about interdrive RAID authoritatively, as if you knew this for a fact, which you do not, because it is wrong. Further, you are incorrect about the reasons for not implementing such a system. The real reasons are:

1) Too expensive. It simply costs more to make. It requires a small processor to split and recombine data, it requires extra testing, it requires extra validation, more complex firmware, and is less reliable.

2) Time to market. See 1

3) There is absolutely no point. As the website linked states, you can achieve the same results with two ordinary hard drives in RAID. If you are going to spend extra on such a drive, you mas as well get better reliability/performance/capacity by just using the tried and true method of more than one drive.

To further compound the incorrectness of the statement, the primary advantage of RAM and Flash-ROM based drives--the reason they are so good, is their access time. The world's fastest hard drive, the Seagate Cheetah 15k.3, access data in milliseconds. Even poor quality RAM can access in microseconds or nanoseconds.
One company used as a "case study" by Platypus, a company making solid state disk drives, was able to replace their mail server drives with platypus drives and REDUCE the number of servers (which had RAID) rather than increase the number. Mail servers do not give a damn about sustained transfer rate (STR), but they certainly do care about access times, especially when dealing with tons of small files like Maildir-bases systems use.

Again, I don't mean to be an asshole. I have certainly been dead wrong my fair share of times, but I am usually rather bluntly corrected. Make a mental note and drive on.

>maybe one day they'll figure out how to make dual read heads with independent actuator arms (i.e at 0 degrees, and at 180 degrees) on the same platter.

The Seagate Barracuda 2HP used dual read/write heads to achieve blazing (for that time) performance - 2HP is short for "Two Heads Parallel".

See also this page, which has a write-up on the 2HP and why the concept ultimately did not catch on, as well as some other examples of unusual disk technologies.

You are so right, alas. See Red Hill's comments on PC Chips motherboards.