Slashdot Mirror
Via Tech announces buyout of Cyrix
Veck was the first to send us the word that Via Technology (lately of partnership with Intel, as well as being sued by them) has purchased the Cyrix x86 line from National Semiconductor. Despite the huge parity in size, Via Tech seems eager to take on Intel.
Update: 06/30 04:10 by S : To clarify: Via Tech will be purchasing the Texas division
of Cyrix which makes standalone PC processors fitting into standard
sockets (7, 370, and the like). The Colorado division which makes
integrated processors (MediaGX) will be absorbed into National Semiconductor
to address the Information Appliance market. Interestingly
it appears that
Via will also offer integrated products, therefore possibly competing
with National.
I don't think that AMD really wants to get into the chipset business. They just need to provide the initial version until VIA and others come up with their own.
The M2 has the fastest integer performance per MHZ for it's class... if it ran at 400-500mhz it would be competitive. If it had an integrated 128K L2 cache at that speed it would probably eat up Celerys. Alas, the fastest one is only a 250mhz chip, and w/o integrated L2. At least they're cheap.
Dude, you send me the $$$ and I'll buy all the SCSI stuff you can name, but those of us in the real world have to make do with IDE, and on that topic, why should I *have* to add a SCSI controller to make up for the unadvertised shortcomings of the onboard I/O???
As for your experience as a veteran, I commend you, but does that really have a bearing on *all* games, they are not all combat oriented - what about Flight Sims, Strategies or some RPG's, the number of machines I have battled with to try to setup these types of games for customers, despite the best efforts of evil hardware like Cyrix Processors, is quite outrageous?
And why buy a console when your PC can perform more than admirably as a games machine, and output fantastic audio, given a little forethought in it's specification???
I have three letters for you my son:
F - P - U
If your architecture is poor and you have weak Floating Point performance, it doesn't matter how fast you clock the bastard, it will still run like a dog in real world applications...
Odd...I have such a setup here and have never run into the problems you describe. What board are you using? Mine's an FIC VA-503+ with an ATI Xpert 98 AGP and an Ensoniq AudioPCI. I've had two UDMA-compatible hard drives (a Western Digital AC35100 and a Quantum Fireball EX 10.2GB (don't remember the model #)) and both have worked flawlessly with the rest of the system. Maybe if you're running some PC Chips motherboard (or something equally crummy), you might run into problems...hell, PC Chips is almost a guarantee of problems of some sort. Given a decent brand of motherboard, though, you shouldn't have problems.
20 January 2017: the End of an Error.
I've seen benchmarks where (at the same speed) the slower P2/P3 512K cache pulls ahead of the Celery's 128K cache in SMP, whereas in a uniproc environment at the same speed (and same FSB) they're usually even. Heck, mpg123 is a bit _faster_ (slightly) on a Celery :)
A lot of _really_ good stuff is coming from Taiwan... and I wouldn't be suprised to see a couple of the smaller CPU companies get swallowed up by the other bigger companies... (Acer for instance makes or assembles almost everything BUT CPUs, and maybe power supplies. If they had a CPU vendor they'd have more or less the same pieces to play with that Intel does...)
It oughta be interesting to see what happens with Cyrix/VIA now... VIA chipsets have been a bit slower than Intel ones (esp. slot 1), so the infusion of new IP might help. It's also good to see the Cyrix engineers not get laid off either... longer term C/V is probably going to be better off than AMD.
The other thing to look out for is at least one company to start selling a whole 'package' if you will... like Intel does. Some companies can basically buy entire PCs through Intel and slap their name on (or maybe even have Intel put their badge on it in their factories :) AMD does _NOT_ have this, VIA might with the help of some of their subsidiaries like FIC.
Just like Intel fabs there chips in Malasia, and Costa Rica, Via will fab there chips in some other contry as well... and that hapens to be the US :P
so dispite the fact that its a tiwanese company, there products will still be "Made in the U.S.A" heh
_
"Subtle mind control? Why do all these HTML buttons say 'Submit' ?"
ReadThe ReflectionEngine, a cyberpunk style n
what would a thread on slashdot be without a comment about the secretive transmet? not much, that's for sure! :P
_
"Subtle mind control? Why do all these HTML buttons say 'Submit' ?"
ReadThe ReflectionEngine, a cyberpunk style n
Close to 80% of the world's hardware is now made by Taiwanese companies. Even Apple is doing it. Apple just signed a new contract here Quanta to manufacture Powerbooks. Check this: http://www.asiacomputerweekly.com/news/99/0628-04/ country_report/showart.cgi?article=cr061 9992803.html
Without verifying this I seem to remember that VIA is using direct mapped L2 (L3 on AMD K6-III:>) as opposed to the set associative caches Intel is using. This goes hand in hand with a study I remember where a direct mapped cache in general needed to be twice as large as a 2 way set associative cache to maintain a similar hit rate (read performance). So basically your 1meg (is equal to the 512k from intel hit wise) Of course there are latency issues here too that other readers have commented on.
I think that with the "collective intake of breath", he is alluding to the fact that so many people use them, that if a bug is at all serious, it could cause headaches for quite a number of people.
(And of course, the fact that a lot of people use them infers that they're fairly good.)
--
- Sean
It's a fine line between trolling and karma-whoring... and I think I just crossed it.
- Sean
Although Via would be able to make chipsets for Slot B - they'd be at a big disadvantage. Currently, if you buy a Super7 board, you're probably getting a Via. If you buy a Slot1 board, you're almost definitely getting an Intel chipset.
If Via has to compete against AMD like they have to compete against Intel for chipsets, they have reason to be scared. And reason to buy the only other real legitimate chip maker on the market - Cyrix.
Well, I've not had any problems with recent Cyrix chips; true, they don't cut it for FP stuff, but I don't need that anyway. Just make sure you have adequate cooling.
As for VIA, no personal experience of their stuff, but I'm aware there's a collective intake of breath on l-k whenever there are bug reports that mention their chipsets.
if so, I think its good that a non-US company is getting in on the act. Maybe they wont be subject to bribery like so many US companies are - i.e. discard their ethics any time a certain large software manufacturer waves a bunch of money at them. I think some of the atractiveness of Linux in the European and Asian communities is from their distrust of US companies (I leave it to you to decide which ones.) I think this is a Good Thing(tm)
Bribery is widespread; are you suggesting that Asian and European companies are immune from it?
In fact, Japan and Korea are home to some of the juiciest bribery cases (and accusations); a Japanese woman of my acquaintance says that bribes were a common occurance when she worked for 5 years at a large Japanese computer company, no names named.
In fact, while Microsoft has been accused of many things, some of which are probably true, can you point out news stories where it is accused of bribery per se? I would be interested in reading them, but I haven't seen any yet.
timothy
jrnl: http://tinyurl.com/c2l8yr / foes: http://tinyurl.com/ckjno5
WARNING: The following is all speculation.
Intel recently kicked VIA out of the Slot 1 boat because of the 133MHz FSB and gave their competitor ALI the rights to produce the Slot 1 boards. So what is a poor chipset manufacturer to do.. oh wait.. Cyrix is for sale? Aha! VIA buys Cyrix so it can use the cross-license to produce Slot 1 boards again, and gain use of Intel's intellectual property. So VIA will now be in the boat whether Intel likes it or not, and as a bonus they get all the cool integrated stuff Cyrix was playing with. That's my take.
JOhn
Campaign for Liberty
Whadda ya bet he and Rob can spell "fully vested" just fine?
(just kidding, I'm delighted for them and us)
I see even classic Slashdot is now pretty much unusable on dial up anymore.
They already make some good stuff -- VIA chipsets have supported proper L2 caches for a long time -- I have a VP1 that must be 3 years old with 512k on the board and a 512k COAS module, a VP2 with 2MB L2 cache, and an MVP3 with 2MB L3 cache (with a K6-3/400; it would be L2 otherwise). They were made by (respectively) Hong Fen (well, it was $45), Abit, and FIC. All do very, very well. The VP1 supported parity and it has 256MB of parity SIMMs on it (recent upgrade when the prices dipped last year), and the other two do ECC. This is so nice. Remember, this was when the low-end Intel chipsets were limited to 64MB and wouldn't use parity RAM at all.
...), they want to make integrated systems, they have been Linux-friendly to a degree for a while (less so than they could be, but hopefully that will change), and they do not try to artificially segment the market into stupid users and servers (i.e., by not making ECC stock with all shipping chipsets, like Intel)(and like Intel crippling the Celeron's SMP to protect their high end).
VIA has made some good stuff, the MVP4 (look it up) integrates basic video and sound if you want that. VIA stuff has always worked with Cyrix chips (I know that this is more a mobo issue, but
VIA has been a cool company for a while. I will not by a board without a VIA chipset. Cyrix has been pretty cool for quite a while too.
This is good.
The sentence doesn't even finish either. Paying what?
bp
And this news was in the local papers here 4 days ago. Now the Taiwanese will be making computers completely independent of the US. Should be an interesting ride.
VIA has a killer chipset to Intel's BX.
Basically, it has the same features of the BX, but supports the faster PC133 Memory and has UDMA/66 support.
Intel doesn't have anything like this until September, so Intel sues VIA to scare 'em.
Now VIA can give 'em the middle finger and continue to give Intel competition. Only if VIA could take a Cyrix chip, not make it run as hot and make it the controller for a K7.. =)
A Pentium equivalent processor that handles all IO for the system -- that would be plenty fast.
If VIA succeeds, then we won't have to upgrade to DDRAM that Intel has invested significant amount of money into. I enjoy paying $70 for 128 megs of RAM and I'll enjoy paying less later. With DDRAM, I'd have to pay about 4x times as much.
Good luck VIA!
- A.P.
--
"One World, One Web, One Program" - Microsoft Promotional Ad
"Remember when the U.S. had a drug problem, and then we declared a War On Drugs, and now you can't buy drugs anymore?"
Cyrix chips (M1 and M2 for instance) are very well designed.
M2 is smaller than Celeron
M2 is smaller than K6
Mhz for Mhz, M2 is faster than P3.
The only problem is they can't run at high mhz.
But it doesn't, so it isn't.
You might want to ask why it doesn't. Perhaps the design isn't capable of going to high clock rates. Perhaps their process technology isn't good enough. Perhaps both.
I hate to rain on your parade pal, but the VIA MVP3 chipset is a piece of crap, I ought to know, I got lumbered with one!
It's implementation of both the PCI/AGP and IDE bus is woeful...just try using any PCI sound card with one setup for UDMA...Snap, Crackle and Pop!!!
Cyrix - cool, yeah right! You ever tried to install or play any games with one???
And you're basing this on what? Personal, inside knowledge? I don't think so!!!
:-)
Cyrix has some excellent engineers and pound for pound I'd stack their engineers against Intel any day. Intel doesn't have any advantaqe in intelligence, just in brute-force manufacturing prowess. Intel can afford to make dumb engineering decisions because they have great control over their process.
And.... *yes*, I am basing these opinions on personal experience with *both* companies
Cheers,
- Brendan
"...The smart and lazy ones I make my commanders." - Erwin Rommel
I would say Cyrix more than likely has rights to Intel's Slot 1, but how that transfers to VIA will be up to the lawyers. My guess is VIA is in the game and they will begin legally producing slot 1 chipsets. I can see no other compelling reason for VIA purchasing Cyrix. Sorry for not doing the research for the first post.
February 3, 1998
Intel Corporation (Nasdaq: INTC) and National Semiconductor Corporation (NYSE: NSM) announced they have settled a patent infringement lawsuit and extended the term of the existing patent cross-license agreement between the two companies. The settlement dismisses a patent infringement lawsuit filed by Cyrix Corporation against Intel in May 1997. Cyrix became a wholly owned subsidiary of National Semiconductor on Nov. 17, 1997. Under the settlement, Intel and National agreed to extend the term of their long-standing broad patent cross-license. Other terms of the settlement were not disclosed.
Links
National and Cyrix Press Release
Intel's Press Release
This info found easily thanks to Google. Google rules!
Campaign for Liberty
You can find some good stuff on this new K7/Alpha archicture here.
VP2 chipset: 2MB
MVP3 chipset: 2MB
MVP4 chipset: 2MB
Cache makes a huge difference. When the PPro came out, I was unamused to see that Intel was shipping with 256k L2 cache. With 256k, the PPro at 200Mhz was as effective as a 166MHz Pentium with 512k on the board, and it cost about 3-4x as much. The usual rule of thumb has always been (IIRC) that doubling the cache is the same as a 30% speed improvement. In my experience, VP1 chipsets with 133MHz Pentiums and 1MB cache would equal 200MHz Pentiums with 512k of cache (which I tested on the same board, pulling the second COAS module out and changing the jumper for the test and using the same chip with 2x as opposed to 3x multiplier). I recall using a Byte benchmark and (I think) Linpack. This was a few years ago, but with a baseline of a 133MHz Pentium with 1MB of cache, the results were basically:
133MHz/512k 35% worse
133MHz/1MB baseline
166MHz/512kb 15% worse
166Mhz/1MB 20% better
200MHz/512k within a few percent of baseline
200MHz/1MB 35-40% better
I haven't tried anything like this lately (the cache has generally been on-board since then), but it really changed my mind, and made me really dislike Intel and Sun for not maxing out the cache. Cache is a good deal and worth the cash, not MHz, generally. Look at the best perfroming Ross boxes -- 1MB cache on 140MHz chips. Look at one of the reasons the PA-RISC kicks ass (HINT: it ain't HPhUX). Look at the benefit that cache gave to the Celeron.
VIA has given us cache and continues to do so. For this reason alone, I will keep buying VIA only.
I do wish that they would release more info and support Linux better, though.
Not to be obnoxious, but Slot B is for the Intel Xeon. AMD/Alpha is Slot A.
Anyway, I don't think Via (or the other Slot A mobo chipset makers) will have to compete against AMD as they currently do vs. Intel. AMD is creating a chipset just to get things rolling, not to dominate the market as Intel does. I doubt AMD will push anyone to license their chipset, or make it difficult if that's what a company wants to do. The proliferation of Super7 boards has helped AMD and they will continue to groom these partners.
Queens of the Stone Age - they rule
Oh, give me a break. The market a 233MHZ processor as M2-300 and and with their PR ratings clearly try to imply that it is just as good as PII-300mhz. But their integer performace is only just as good as the P300, and the floating math is worse than even the old good Pentium 233 mmx. I wish their stoped this PR rating crap. I myself try to evaluate the CPU speed by looking at SPEC results not the MHZ it can clock ..
As far as I know, Cyrix never had a license...
Cyrix got use of the IBM x86 cross-license (back from the IBM/PC days) when Cyrix had their chips
fabbed there, but no Slot 1 coverage.
Cyrix got use of the SGS-Thomson (originally from Mostek) cross-license when they (almost) had their
chips fabbed there, also no Slot 1 coverage.
After they were purchased by National, they got the use of National's broad cross-license (both
x86 and Slot 1 are covered).
Via is in cross-licensing hell and they just bought Cyrix who never even had a license.
Maybe, but IDT just announced more details on Winchip4. Why would they let all those investment dollars go to waste? Supposedly the W4 will hit 400Mhz+ this year with an improved FPU. The K7 will likely hit 800Mhz by the end of the year, but that's the extreme high-end where IDT has no intention of competing.
Queens of the Stone Age - they rule
You're right, we're NOT talking about sneakers. I'm refering to their actions taken in Silicon Valley.
Think about it. (They've probably already got a factory site located in either Fujian or Shenzhen.) They send their top people to the States to work with the Cyrix design team (remember, the Taiwanese aren't a bunch of rice farmers). Meanwhile, another group is working in the manufacturing wing. In the not too distant future, manufacturing will be shifted out of the US and shortly afterwards, the design work also. Selected American employees will be kept on.
Case in point, know the motherboard maker Tyan? They *were* the largest American motherboard manufacturer....they were bought by Mitac of Taiwan on 3 Feb 1999. Even before they were bought, Tyan boards were actually made here in Taiwan.
FWIW, here is a quote from the Tyan-Mitac press release:
"The MITAC-TYAN alliance, therefore, offers a unique combination that's bound to play a big role and leave its mark in the PC industry, a winning synergy that will add more muscle to MITAC's manufacturing and logistics capabilities. For starters, it will boost MITAC's motherboard capacity, already at 300,000 per month to 500,000 per month. It will also enhance the company's R&D capabilities, considering the fact that TYAN is undoubtedly an R&D powerhouse."
According to the people I talked to at Rise (in Taiwan) their new CPU 400mhz) should be out in August. It will be SMP capable but NOT over-clockable.
A note, their product specs state that their CPUs have all been tested and run Linux, Unix and BeOS.
If you can read Chinese, their site is at:
http://www.risetec.com.tw
For English:
http://www.rise.com