Slashdot Mirror
Intel Tualatin Processors and Motherboard Support?
IntelIntrique asks: "I became intrigued when I learned about the new P3 Tualatin processors with a 512K cache, but was disappointed to learn that they require specific new motherboards to support the chips (Intel claims voltage changes). However, HotHardware features a review where they use one of these new processors in an Intel D815EEA2 motherboard, one that Intel shows as not supporting the new processor. What gives?"
"I have an D815EEA2, and I'd love to grab one of these new processors, but not if it means buying a new motherboard. Is Intel trying to prevent existing motherboard owners from cashing in on this new super chip? Are there any other slashdotters who have tried these new P3's in older motherboards with success? I'm mainly concerned with people using the processors in Intel boards, since it seems as though Intel would be in a unique position to know what types of new chips might be around the corner, and build unofficial support into early board revs."
seems there are slot1 adapters for BX boards in developement here Http://www.powerleap.com/Products/iP3T.htm
I'm running Win2K on one right now. They rock. Upgrade, you'll like it...
accidently hit submit.
.26 to .18 (mobile)), and didn't require a new board. I'm using one right now.
Intel makes the p4 and charge 5x for the same performing Athlon. The P3s were reasonably priced, but intel had competition then. Now that the p4 is out, there is no Mhz competition so they jack the prices up.
AMD boards have previously been able to take a die shrink (The k6s went from
There is no real reason why intel couldn't have made the chip so it works with the old boards.
If the new P3s would have worked... it would have had the same performance of a 1.7 p4, and that would hurt P4 sales... intel doesn't want that. They are doing everything to make the P4 look like a great chip.
They are ripping you off, plain and simple.
Hmm, follows the normal Intel marketing trend. Make new processor, make it use a proprietery motherboard, and clean up on the profits after making everyone else believe that its the fastest thing out there. BUT, won't this backfire? Intel seems to be pushing the "Megahertz Myth" pretty hard. To the average consumer, Pentium 3 1.2 gigahertz is not as good as that pretty new Pentium 4 2 gigahertz processor. We shall see......
http://www.google.com/search?hl=en&lr=&ie=UTF-8&o
Intel has several different versions of the Tualatin, the 256K cache version for the desktop, and the 512K cache versions, one for the mobile segment and one for the lower-end workstation/server market. The Tualatin processors require a different stepping in the 810 and the 815 chipsets to support the new voltages as well as a new revision of 1.25V GTL versus 1.5V GTL+ used in the original Pentium III processors, which aren't compatible. More information about this can be found in the Pentium III-S Datashet here.
As far as motherboard compatibility, I think Intel didn't want the 512K version of the Tualatin (aka the Pentium III-S) to flow through the retail/desktop channels because in a lot of cases, it performed better than their lower end Pentium 4 processor line. That's also the reason why Intel has slowed/stopped production of Pentium 4 processors below 1.6Ghz and will halt production of the desktop version of the Tualatin and shift the current desktop Tualatin into the Tualatin-based Celeron (but without data pre-fetch and only at 100Mhz FSB). More information can be found at the regular sites: Anandtech, Aces Hardware, Tom's Hardware.
There is this site that teaches you to build SMP Linux workstations using Slot 1 BX motherboards at http://www.linuxenvy.com/gened/projects/smplinux.h tml. Pretty neat.
I am the "happy" owner of the Asus TX97e board with Intel's TX chipset. Intel intentionally crippled the chipset so that it caches only 64MB of RAM. If you have more than that, the performance goes down the toilet. After I upgraded my machine to 256MB RAM, Norton System Information benchmark went from 110 to 70 (AMD K6-2/300). It's not a scientific benchmark but still quite telling. Thank you Intel!
And don't even get me started on what they tried (and are still trying) to pull off with Rambust...
___
If you think big enough, you'll never have to do it.
Supposed to be that 10 pins are used differently, and some external timing is different.
Based on this using it in older motherboard should never work. If it does, that is very odd.
Probably much more interesting is that the tualatin core has shown a lot more promise than current P4s. This review (http://www4.tomshardware.com/cpu/01q3/010919/inde x.html) over at Toms shows how a measly 1.2 tualatin holds it's own with the P4.. and overclocked to 1.5 it can be see that it has much more potential than the P4, even with the P4 running on rambus and the P3 on SDRAM! At the end of that review Tom mentions how the tualatin core is capable of 'much higher clock speeds', but it seems Intel is keeping them down because they don't want it to compete with it's 'Big Brother', however underachieving he is.
Personally I have ordered myself a Tualatin 1.2, I choose it over the P4 offerings.
From the refferenced page:
Date last revised: 31 March 1999
And they only talk about the 2.2.x kernels. Kinda old, man. There's been a lot of work done with the 2.4.x breeds (most back-ported to 2.2.x) that make SMP a God Send (TM).
Also:
AMD and Cyrix do not support SMP.
Hmmm...
I'm not a prophet or a stone-age man,
I'm just a mortal with potential of a super man.
I have a Tyan Tiger 100 (rev. F) with 2 P2-450's. I'm planning to upgrade, but the fastest officially support chip is the P3-850 (100 MHz FSB). Does anybody know if I can us Tualatin processors with this board, perhaps with some kind of slotket? (Or, does anybody know if it's stable with the 100 MHz FSB P3-1GHz?) Any advice is really welcome... the processor upgrade will be the last uprade I do before getting a new motherboard, so obviously, I want to do the best I can.
It says that the chip they reviewed had only a 256K cache, and that 512K would come in future versions. Without 512K, the chip is compatible with current motherboards.
actually, only people who work in design at that level have the skills needed to comment at that level completely competently.
Now you can piss on everyone else, or get a life.
Intel needs to sell more chipsets because via DARED to touch into their lucrative chipsets market.
No but seriously, tell me you can design a 40 million transistor chip that runs in gigaherts, and not make a design for it to be compatible with current motherboards out there? Talk about bad will. I was hoping to upgrade my serverworks-based workstation with 2 of these monsters, well seems like I'll keep the 800mhz... See intel? your plan on selling chipsets for what, 25$ each, costed you the sell of 2 processors, of what, few hundred bucks each?... I'm sure I am not the only one who'll react like this or turned his head towards the TigerMP platform.
--- Metamoderating abusive downgraders since my 300th post.
or the latest version of the comparable VIA Apollo Pro chipset. Search pricewatch for the terms "FCPGA2" or "tualatin". Such board include:
Abit ST6/ST6-RA, Abit VH6-T Asus TUSL2, Soyo SY-TISU
Some board makers have been using the i815-B but don't mention it in the board specs ; marketing hasn't caught up with production yet.
BTW, the newest Celeron 1.2GHz uses the Tualatin core; no hardware prefetch and it uses a 100MHz FSB, but it can overclock to 1.5GHz with at most a minor voltage jump and standard cooling.
---
since 1995 I've been doing cryptography and also digital graphics (3D) and for a long time (well, in terms of technology it is long), Intel was king b/c they were better at floating point calc, and both of those were almost all reliant on that. But then the amd athlon came along and now it kicks ass in that area too.
I haven't upgraded my current systems yet, and even the newest one is a dual PIII b/c at that time the athlons weren't able to do dual - but now they have the MP and great boards (see the articles about the tyan tigermp).
So I don't understand why you would seek out the Intel...
There are some odd things afoot now, in the Villa Straylight.
There are slot 1 100mhz bus 1ghz coppermines. I am using one now in an Abit BE6. This board only "offically" supports up to an 850mhz pIII but I am using it to type this now. Is is boxed for server use only but works great form me. Search for "1ghz 100fsb secc" on www.pricewatch.com.
Be warned, these pups are a bit pricey, I paid $212 from www.googlegear.com (Item # 80607) I think these are being phased out so act soon. Google is a good place to deal with so check them out.
Email Tyan support to verify for sure, I suspect it will work fine. Good luck!
The part might work in the non-specified board, but if you take another of the same part and board, it might not.
It's not "supported" because Intel didn't spend on the testing to ensure that it's supposed to work at levels that don't cost Intel a lot of returns.
They changed the voltages on the part to get some other benefit (lower power, more speed, better yield, whatever), and couldn't guarantee the old boards would work, so they designed a new board.
It's not that big a deal (unless you think it is) because people who pull their CPU and replace it are relatively few and far between.
--Blair
I just bought a Supermicro P3TDLE, a dual Tualatin board, with Serverworks chipset, so i have been researching the chips myself. It seems as if the clowns at intel have released two versions of Taulatin at 1200, one with 256K cache at 1200 (Cost $200) and a 1266 part with 512K cache(Cost $390+), which is also a dual processor chip, whereas the 256K cache Taulatins will not be dual processor capable. I also note they are using a Pentium III-S designator on some chips. What a bunch of idiots, just try and read the processor PDF files from thier own site!! It will give you a headache. I believe i will stick with my dual 850-FCPGA for a while, until the price settles on the 512K models.
I'm looking for a tiny case to go with a micro-atx motherboard for a Linux router project (Tualatin based ;). I'm looking for something on the order of 12H x 16D x 6W or smaller. I don't want a 1U case. Any recommendations?
The C in INTC is for "crooks".
Depending on the context, it can also
stand for "crap".
Greetings!
There are two versions of the D815EEA2. The new version, which supports Tualatin processors, is the 'Universal' version. It's -supposed- to have a 'U' appended to the model number (D815EEA2U). Looking at the picture of the motherboard on the site doing the review, I notice the white square with a 'U' in it toward the end of model number printed on the board.
I've had a -heck- of a time trying to find a supplier who can verify that they're selling this 'Univeral' version. I finally gave up and got an ASUS TUSL2-C. At least they had the decency to use a different model number for their Tualatin-capable product.
It's kinda neat to see Intel using familiar places in the Northwest as their product names. Willamette (which nobody knows how to pronounce - it's "wil-LAM-et", not "wil-la-MET"), Tualatin... what can we expect next? Multnomah? Clackamas? Columbia perhaps.
:-)
For those of you not familiar with local geography, the Clackamas river flows into the Willamette river (a mile from my house), which flows north through downtown Portland into the Columbia, which of course flows into the Pacific. The Portland metropolitan area spans Multnomah, Clackamas and Washington counties. Multnomah Falls is the second highest year-round waterfall in the country. Most things in the Northwest have either Native American names (mmm, Tillamook cheese), or names that were brought from the east coast by Lewis & Clark, John McLaughlin and other settlers.
Sorry for babbling off-topic.
$x='S24;r)>63/* h@<5+oZ)32"5cz';$me='phroggy'x$];
$x=~y+ -xz+\0-Tx+;print$_^chop$me for split'',$x;
It's because the two have different FSB voltages - 1.5 vs. 1.25. If you wanted to try a tualatin in a p3 motherboard, it may work (depending on the motherboard) but you'd be overshooting on your FSB buffers. This may not be that big of an issue because we typically overbuild our processors way more than is necessary.
So, depending on the motherboard, I think you could probably do it.
-anonymous intel employee
Yes, there is a couple of reasons Intel is pushing the 830 chipset with the Pentium III-M chips. First, the 815 needs a stepping to be compatible with the voltage requirements of the PIII-M 512K version. Second, the PIII-M is marketed with the 830 for mobile platforms and 1U servers... Low power consumption = Low heat dissipation. Also, the 830 comes with a GeForce 2 embedded graphics adapter which is another marketing point for low end machines and notebook computers.
HardOCP.com sez:
There will be a few of you excited about this. Powerleap has put together an adapter that will allow you to stick a S370 Tualatin chip on your current Slot 1 mainboard. Looks as if some of you might have some server upgrade options that you did not before. w00t!!1
The PL-iP3/T(TM) employs patented technologies to adapt Slot 1 systems to the voltage and signal requirements of the new generation of Intel's Pentium III (FC-PGA2) and Celeron-II (FC-PGA2) processors. With the PL-iP3/T(TM), a typical* P-III system can reach speeds up to 1.26 GHz with the latest Pentium III-S CPUs (133 MHz FSB required), and up to 1.2 GHz when used with the latest Celeron-II CPUs (100 MHz FSB required).
The ultimate network admin tool needs HELP!
Who cares about Intel. I'd rather eat hot shit than buy an intel product.
Who cares about Intel. I'd rather eat hot crap then buy another Intel product.
I wish corporations realized that they can save money buy buying Athlons. Then they would have faster and cheaper computers!!! my god... so many people deserve to be shot it's not funny.
The operating voltage of these chips is lower because they are built on a 0.13 micron process, however the power requirements are still significant (though only about 1/2 to 2/3 of a top line Athlon or P4). One problem is that the CPU's current requirements are higher than for regular P3s. Even though the power requirement is lower, some motherboards may not be able to supply enough current for stable operation, or the motherboards power supply circuits may be over taxed and fail prematurely because of the increased current being drawn.
The lower electricity useage, heat production of these chips makes them ideal for 90% of the desktop machines out there. To bad that CPU marketing is so driven by "performance". Via C3 anybody?
Here's a hint: The people who buy the most Tualitins _will_ be buying them with motherboards. Those people have names like "Dell" and "HP" and "Gateway". Intel, fully aware of this -- though they'd rather be selling Pentium 4s -- determined that they could get the Tualitin to market faster and with better features (for mobile users, mostly) if they altered the motherboard slightly.
Well what does that mean? It means that you, as an individual, walking into an OEM shop, really don't matter very much. It means that for what you do count for, Intel would much rather sell you a Pentium 4 system [so they can take market share to software companies and say "optimize for us!" and thus get their benchmarks up -- that's why P4 CPUs cost *less* than P3s, or hadn't you noticed?]. But mostly it means that if you're freakish enough to want to upgrade to the next generation of Celeron, you should probably just plan to waterfall your current motherboard with your current CPU.
Heck, if you're shelling out the premium "why not buy a P4 instead?" price Intel's put on Tualitins, why not go whole-hog and get the motherboard as well?
(And some people are saying that this consitutes some kind of conspiracy. Sheesh. The conspiracy is -in your head- people; it's those little voices you hear whispering when you go to bed at 3am...)
From the story:
"... HotHardware features a review where they use one of these new processors in an Intel D815EEA2 motherboard, one that Intel shows as not supporting the new processor. What gives?"
If I understand the table correctly, it is probably just a case of not having updated their web site. Intel is notoriously sloppy about things like that.
About a year ago I called and talked to an Intel employee about a huge mistake on their web site. He said it would be fixed immediately. Eight months later the error was still there. I called and talked to the same man again about the same error. He didn't realize I had called before. He told me again it would be fixed immediately. Again it was not fixed. This is just one example.
Be careful with the D815EEA2 motherboard. If you remove a removeable drive, it may re-configure the BIOS, without any warning, and boot from the wrong drive.
Be careful with the network adapter if it is built into the D815EEA2. It assumes that it is attached to a huge network. If it is attached to a peer-to-peer network, you may not be able to make it function. An Intel technical support person and I worked on this problem for more than an hour. The final answer was to buy a CNET network adapter for $12.00 and disable the network adapter on the motherboard.
Also, if you are running a Raid 0 controller like the Promise Technology FastTrak 100, the D815EEA2 BIOS has a very weird configuration. It is not obvious how you get the motherboard to boot from the Raid controller, because the way you select it is hidden.
What should be the Response to Violence?
Bush's education improvements were
BX is the key, unfortunatly it does not look like there will be any Via chipset support since Powerleap says it add a lot to the cost. Also note that they will only realease and make this when the market has a large supply of these chips.
Sorry, I meant new pinouts, not 10 additional pins. Here's the skinny:
New Pinouts
Who cares when 1.4GHz Palominos are like $90, 2100DDR is like $35/256MB, and - if you haven't already gotten an Athlon mobo in last 1.5 years (compatible) - rock-solid DDR mobos are less than $100?
Seeing as no one has mentioned this yet...
The reason for the new motherboards is because the chipsets have to recognize a lack of voltage on pin AF36, so that they run the chip at 1.25v. Otherwise, they will run at 1.5v. That should answer all the questions.. yes the tualatins should run in older boards, however it will be at a higher voltage than specified. There are a bunch of other new pinouts (and not NEW pins), for full details, see Toms Hardware
So is it stable under linux, unlike AMD's crap?
Their Irongate / AGP issues, which have been known for NINE FUCKING MONTHS now, are driving me back to Intel. I can't type 'startx' without first doing an emergency disc-sync in case the whole system locks hard. And it usually does. Well done AMD. I know support the more expensive and most likely technologically inferior Intel chips over your unstable crap.
Stop complaining about Intel, and do something to help... BUY AMD! Why would you even consider a Taulatin for $400, when you can get Athlon MP for $150? Get 2 of them, and a TigerMP Mainboard for $500, just $100 more than on Taulatin. Btw, the Athlon MP's overclock quite well to 1.4ghz, a speed that beats 2.2ghz p4 easily, get 2 of them... big bragging power! :-)
"And we have seen and do testify that the Father sent the Son to be the Savior of the World"
1 John 4:14
Intel = Microsoft
That simple... someone from intel broke into MS and stole Bill's marketing techniques.
Get your Unix fortune now!
AUGH, yes, this has been driving me bananas.
I was seriously thinking about building myself a nice box with dual 1.26GHz Tualatins, but have been unable to find SMP mainboards that support Tualatins in an SMP setting.
Well, actually, that's not completely true...I did find a couple, but none of them use Intel chipsets. There are a couple with VIA chipsets, and one using ServerWorks. I would prefer a board built around an Intel chipset, but the problem is that the ONLY Intel chipset that supports Tualatin so far is the i815 series which, although reportedly a good performer, is severely crippled in a number of respects: 1) It officially does not support SMP, and 2) RAM is restricted to 4 banks of SDR SDRAM which cannot exceed 512MB total.
Despite the first restriction, I did find one SMP board built around the i815 chipset (quite an engineering feat, it would seem, considering they probably didn't get any help from Intel): the ACorp [acorp.com.tw] 6A815EPD. It looks like a good board, seems to be getting favorible reviews, and I could live with knowing that I was limited to 512MB. ;-) The only problem is that ACorp has yet to come out with a revision of this board based off of the i815EP "B stepping," which is the revision of the i815EP chipset that supports the Tualatin (I even e-mailed them, and they confirmed that they do not yet have a dualie board based off of i815EP B step).
And, naturally, most "respectable" mainboard manufacturers are not about to release an SMP board based off of a chipset that does not officially support it. And Intel has yet to release a chipset that supports both SMP and Tualatin (and this is not because Tualatin cannot SMP, because it can; see VIA, ServerWorks).
So, basically Intel is driving away my business.
I am now considering VIA and ServerWorks-based boards as options. Can anyone comment on the performance and stability of these chipsets (VIA Apollo 133 and Apollo Pro 266T, ServerWorks HE) as well as their compatibility with Linux? I've heard some negative things about each with respect to Linux compatibility, and am curious to know if these rumors have any basis in truth. Thanks.
-- Nathan
IT will rock
http://www.windrivers.com/
1.6ghz P4 = 163$
1.0ghz P3 = 164$
Man intel either wants to push those P4 out the door fast!!! or they want those P3s to die faster.
Why do they bother with 1.9ghz and 1.8ghz, thats like selling a 300mhz vs a 315mhz cpu in 1997
What a joke intel, you guys a overun buy marketing no hoppers. I expect a min of 25% diff in each model, not 5% shit.
Why not just sell ONE CPU that can work up to 2.2ghz but boots up at 500mhz for FREE, and makeit register online and you can pay per mhz (25cents/mhz), then when you hit submit, your CPU springs up to the mhz value.
Remember, marketing requires less brains and less months study, compared to engineering requiring 100x more brains and 10 more years learning.
using the "plus" processors (k6-2+, k6-3+) gets around this foolishness as those cpus have their own onboard cache.
proud owner of a tx97-xe with 2x 128mb dimm running at 83mhz fsb, k6-3+ at 500mhz.
Because we are buying them for where we w*rk adn we aren't about to have the PHB see an invoice for a mb/cpu/mem upgrde under $200.
Heh. I like AMD's, I've been getting them for the desktops (my boss had been buying Celerons! Ack!). I'm completely happy with them. They take abuse well too (which is a requirement for our location, dust, heat, humidity). But for the server? I'm looking for Tueys. Twin Tualatins, please.
And what's next for this happy company? StongARMs...
...Time is the best teacher, unfortunately it kills all of its students.
A pair of Athlon MPs along with a Tyan MP board would be something like $700 - that wouldn't be expensive enough?
there are new versions of the 815 mb that support the p III-s.
get d815eea2u or d815eea2lu