Slashdot Mirror
Intel Cancels its Timna chip
zensonic noted that Intel has announced that they are
cancelling the Tinma chip. It was an integrated chip that would be used in low end systems... they cited market demand and design problems as the reason.
← Back to Stories (view on slashdot.org)
Software is made to match existing or future hardware. Quake3/UT are a massive improvement, graphics-wise, over (say) Doom. But Doom was smoking fast on my 486/66. Quake3's install CD would physically emit a laughing noise as the drive door closed if I tried installing it on that box. And as fun as Doom was, Quake3 is MUCH easier on the eyes to play and just plain more INTERESTING. It's the rampant increase in hardware quality that gets you that, er, interestingness.
Same goes for Office-type apps. I remember in the days of the aforementioned 486/66, Word would take nearly a minute to load, took forever to spell/grammar check, and don't even THINK about running anything in the background! MIDI player maybe! =) Now on Windows/Linux on my spiffy new-age box, I can have several apps going at the same time, say (under Windows) Photoshop, VC++, Netscape, SQL Server clients... hell, SQL Server... all stuff that I'll regularly tab between during the day.
And there are still apps that bring this new box to its knees. Bryce comes immediately to mind as an app that makes my CPU and RAM cry in pain. Maybe a little 256M upgrade...
My point, and I do have one, is that while it sometimes seems that technology progresses for no reason other than to encourage consumption, the efficiency and kewlness factor of PCs now is far greater than that of PCs 5 years ago, 10, 15... I'm surprised those old XTs got bought by ANYONE for anything besides Lotus 123. They were all but unusable. Sure, products are made to make money, but by labelling the PC consumers as suckers you're really blindering yourself from the big picture.
--
--
The gravitational constant of protein has changed. - Turbine
I posted the original FACE intel comment, I post it on every intel-fscks-up article that slashdot posts. So, for the third time today - here's the original article - and I won't venture any insight as to why the original comment got reset (kinda pisses me off), but it's most certainly a factor as to why Intel has been missing to boat. This is also in the Pentium IV thread.
I post this every time Intel screws up (which is pretty regularly, when you think of it) and every time, no moderators pick up on what's going on. So, for the good of my health, here's the same diatribe yet again:
Intel is losing their edge because they treat their engineers like garbage. Please have a look at the Former and Current Employees of Intel protest web site for the skinny on what's going on there. Intel's abusive human resource policies are coming back to haunt them, because the people with the experience to pull these projects off are getting fed up and leaving for companies like Texas Instruments and AMD.
Always look for what's going on behind the scenes.. Intel has some great spots to work, but the microprocessor division and their manufacturing lines are not one of them.
Don't buy intel. Don't buy their stock. They need to correct their HORRIBLE HR policies - and I wonder why Slashdot hasn't picked up on this earlier.
The problem with Timna was not with Timna--it was with RAMBUS. Timna itself has been ready for over 6 months--the problem was that this "low end" processor was designed only to use Rambus memory. Timna wasn't late, the "bridge" chip was late. Intel made the mistake of assuming that Rambus would be the de facto standard and very plentiful by now...
Bizarre. My first post vanished. That's some serious bad karma!
As I may have mentioned, my first thought when I read "Tinma Cancelled" was "Duron". Intel's roadmap has some serious potholes in it, after recalling PIII's, push back of the P4, supply problems and losing customers to AMD. AMD folks must be heading down to the Tied House (in DT SJ) to get plowed.
I know a number of people within Intel who say expense practices are loose and Intel is hemmoraging large amounts of cash on sloppy management and indecision. I'm expecting after this week there will be some shakeup and possibly some restructuring. This many mis-steps usually doesn't go unrewarded.
--
Chief Frog Inspector
A feeling of having made the same mistake before: Deja Foobar
The new management wanted to change Intel's direction - that is what new management always wants to do. Since Intel was a successful dominant player in a single huge market - they decided to become an unsuccessful player in lots of smaller markets. Of course the consequences of that sort of direction change seems to escape the perceptions of the razor sharp minds who are paid the big money to lead industry.
I am reminded of the huge retail discount chain who spent close to a billion dollars to build branch stores in Mexico - only to discover to their shock that Mexico is a poor country compared to the US and that the people there couldn't afford to buy what they had to sell.
Doh.
That's Tin-Ma, like Rosie the Robot from the Jetsons. We just missed out on robotic maids! Crud. Guess I won't be able to pick up a Tin-Ma at Fry's, and my room is still such a sty, too. I'll have to formulate a plan to deal with this, start with a few beers, then a few more and possibly pass out. Yeah, that works. Guess I don't need a Tin-Ma after all.
--
Chief Frog Inspector
A feeling of having made the same mistake before: Deja Foobar
Intel has done, lately:
1. Continue to push through new standards in the industry.
Unlike MS, who introduces new "standards" in software to twist old ones and keep things proprietary, Intel generally pushes through hardware standards that improve hardware in general.
Remember AGP, and everyone bitching at the time that it wasn't needed? I bet everyone's glad that their video cards today aren't running on the PCI bus. Speaking of PCI (which is getting old and crusty itself) Intel is leading the development of its replacement, in the same way that it led the development of PCI to replace ISA. ATA-100, and its future successor, Serial ATA, come to mind as well. Even RDRAM (may it RIP) was just bad execution of a good idea (more serialization, less traces on the memory bus).
2. Merced.
Whoa, before you flame this, think about it for a moment. This oft-delayed processor not only introduces a new computing paradigm (EPIC), it's also a processor that completes 8 instructions per clock cycle. Once the die shrink comes around, and clock speeds start coming up in McKinley, what exactly is going to compete with this 64-bit chip? Not Sledgehammer. Not a whole lot of RISC chips out there.
3. Willamette.
Merced, reversed. Instead of focusing on the number of instructions per clock cycle, it simply focuses on the number of clock cycles. While a lot of people have commented derisively about it, 2 years from now when the Willamette core has been refined and Intel is several GHz ahead of everyone else, I don't think near as many people are going to be laughing.
Light a fire for a man and he'll be warm for a day. Light a man on fire and he'll be warm for the rest of his life.
If I recall correctly, Intel's roadmap included the Celeron and the Timna, along with PIII and P4, along with Itanium. Seems a tad broad, so a thinning was reasonable. I didn't expect the Timna to go, rather the Celeron. That it's almost October and we're still looking at Celerons and PIIIs as the core of their line is telling.
--
Chief Frog Inspector
A feeling of having made the same mistake before: Deja Foobar
It seems to me that Intel has *really* lost their focus in the post-Andy Grove era. They've acquired a million companies, and are just staying alive because their name is Intel. First the Itanium (well, I guess that started in the Andy era), then the P4 being delayed, now this! Too bad Intels competition can't seem to *majorly* punish them (yea, AMD chips are cool, but they don't have any dual/quad processor support & aren't really being placed into high-end servers by companies like HP or Crapaq (oops, I mean Compaq))
Doh!
Who would buy a chip named TIMMAHH!!! =)
God Fucking Damnit
Faceintel.com
The poster's contention was that Intel is working its engineers to the bone, and under a demeaning work environment, and this is why Intel is falling behind - their top talent is leaving for AMD and Texas Instruments (TI) en masse as a result of management. This, the author contended, was the real reason for Intel's recent failures - they've pushed their engineers too far.
I doubt this is a conspiracy, but I'm reposting this anyway, as the comment deserves to be seen, as does the site.
--
Seeing a chip like this dropped is disturbing. Two and a half years ago, I bought a Pentium II 333. At the time, it was the top of the line. The previous machine I owned was sixty some odd megahertz, which I used for software development and running a buziness, so 333 MHz seemed like lightning. I still think it's a smoking system. I use it for development, including coding in what are traditionally seen is heavy-duty processor intensive languages like Lisp. That same machine is also used for 3D modeling. The usual advice is "buy the fastest machine you can get if you are going to 3D work," and that's what I did. Everything I do on that machine is snappy. I played Unreal on it, when it was first released, in software rendering mode--because I didn't have a Glide card--and had great fun with it. I have no speed complaints whatsoever. I'm not like some guy trying to justify that this Commodore 64 is still useful; I honestly think my PC is very fast.
Since I bought my machine, the bus speed jumped from 66 to 100 to 133 MHz. Processor speed went to 350, 400, 450, until the low end machine you can get from mail order catalogs is around 633. High end close to twice that. The Pentium III and Athlon became available, with better throughput and more cache. At the same time, video cards progressed from the then-new Voodoo 2 to the TNT, TNT 2, Rage 128, Matrox G400, GeForce, and GeForce 2. Take a low end machine out of all these specs, say a 600 MHz machine with a 100 MHz bus and a TNT 2. That's at least twice the performance of my machine. And in all honesty, I don't know what to do with all the power of my current set-up.
These low-end chips that get kicked around, like those from Transmeta, are still more powerful than what I currently have. And yet the constant wisdom that is spouted is "there's no market at the low end."
I cracked open my machine yesterday to add a new card, and it really struck me how much junk there is in the average PC. Mine must weight 30 pounds. There are a couple of fans, and two absolutely enormous heat sinks. It bothers me to see people tossing these out and buying new machines, just so they can surf the web, listen to MP3s, run Office, and play horribly broken game demos (that is, game demos that don't look like anyone gave a moment's thought to making them run fast on more than reasonable machines). Nobody cares about power consumption or form factor either, just so-called "performance." Even if you need a car battery to power a video card, some people don't care. "My bubble sort is too slow! I need an Athlon!" At some point, this has to stop. People don't realize how much they're being suckered here, which is surprising for the typical anti-corporate college student geek.
problem 1 for Intel
x86 is a screwed up architecture and you have to have ever decreasing die sizes and people to rejig everything (memory access/SMP crossbar) to get real improvements (AMD know this and are dumping/saying it is not surported parts for their upcomeing 64bit chip )
2
the DIE SIZE is BIG putting it mildly
this means less profits something intel does not like
good old digital has a bit of problems with patents
enter Intel's lawyers (these are people who I give respect to ! )
Intel gets strongARM + a whole bunch of people who only have 3-5 years left to get shares/pension now sorry but Digital's engineers (who have been doing this for a long time) + Intel's budget and we get
(drum roll please)
Xscale
die size/performance that beats the pants off anything MOT has produced and part of the ARM arch so yes all those routers/palms/watchs (IBM linux watch)/4G phones can use it and have code that runs on it
er and you want to prat around with x86 compatible stuff which is huge die/power requirements
I don't think so
kiss it all goodbye because x86 is going bye bye dorthy
john jones
(a deltic so please dont moan about spelling but the content)
This is one of the many mistakes they have made recently. Not only has Intel planned on using RDRAM for the Timna platform, but also there is a extremely high transistor count. That would make it a lot more expensive to manufacture, inducing much higher failure rates, higher costs, and much slower clock speeds. These in combination with the already very inexpensive i810 platform which would offer similar performance at a much lower cost is most definantly what caused Intel to can Timna.
Now, had Intel released Timna, with RDRAM and all (Assuming RDRAM was, say, half as expensive as it is now)it would bring the low end PC market to even smaller profit margins, with higher prcies too.
In mass quantites, a Celeron 600 would probably cost around $50-60, and a cheap i810 motherboard with built in sound, video, and winmodem for another $50 or so. In the case of the Timna, the CPU would probably be in the $160 range, going by past "low end" CPUs from Intel, but the motherboard will naturally be cheaper, so maybe $30-40 for a motherboard. Add another $100 for the same amount of memory, and you have something that costs a bit more then $200 over a similarly equipped Celeron system.
Had Intel chosen to go this route, they would have shoved Timna down the throats of companies, stopped making Celerons, and we would have more expensive "cheap" computers that aren't any better then the old ones. Not only this, but also consider that because the computers would be in the "cheap" range, they would need to keep profit margins low. Very low. In order to offset the cost difference, they would have probably had to sell Timna systems at a gross loss (Not a Net loss, a Gross loss), which would cause a mass exodus amongst a lot of PC building companies, who would revert over to AMD where they could see more profits, better sales, etc.
This situation would see Intel almost completely drop out of the low end market, and their midrange stuff would be sorely lacking. While this may have been a good thing for AMD, it is not for the consumer, as less competition on any level (As in AMD owning the low end market) causes less price competition, and we would see prices drop slower, slower releases of new processors, and higher pricing.
It truly is a good thing Timna was cancled.
Only those who dream can grasp reality.
Ace's Hardware has been covering AMD's advances fairly well - they have an SMP Capable chipset (the 760MP) in the works, which should be out by december.
p ost_id=10000240
4 .html
A little bit of info here: http://www.aceshardware.com/Spades/read_news.php?
As for compaq: well, make what you wish of this: http://www.theregister.co.uk/content/archive/1222
I'd say that AMD is setting it's self up to replace Intel rather quickly. Already, many OEMs are droping their Intel only policy...
I think that this is a move in the right direction. We already have enough stinking bargain chips (now that VIA is starting up the damned Cyrix series again with the CyrixIII, that puts the total of cheap-ass CPUs from 2 [Intel Celeron and K-6 x] to 3). Maybe now the Intel hardware designers can work on a chipset that doesn't suck, like a DDR chipset for the P4, or maybe a new 440BX-esque chip for the Coppermine series. I'm sick of VIA dominating the chipset market with their "0.25 micron crap on PCB" strategy.
"Ancillary does not mean you get to rule the world." --U.S. Circuit Judge Harry Edwards, speaking to the FCC's lawyer