Please create an account to participate in the Slashdot moderation system

 


Forgot your password?
Close
typodupeerror
×
Hardware

SMP Linux on the Cheap 114

Posted by CmdrTaco from the two-scoops-of-CPU dept.
d^2b writes "There is an article using dual Celerons under Linux on cpureview.com. This is even more attractive now that you can buy an Abit BP6 for $130 and plug two socket 370 Celerons into it directly. The good news is that that author gets 183% speedup over a single processor compiling the kernel. "
This discussion has been archived. No new comments can be posted.

SMP Linux on the Cheap More

SMP Linux on the Cheap

Comments Filter:
  • 300a celerons oc'd to 450 on a Tyan Tiger 100, running RH 6.0.

    I had to get all new parts, so it came to around $1600 or so: 2 300a celerons, fans, 2 MSI converters, 128 mg ram, 17 gig IDE, ultra tnt2, floppy, cdrom, 17" monitor, soundcard, ethernet card.

    I did an ftp installation of RH from ftp.cdrom.com; I did it from my school so it only took about forty minutes. The only problem I had installing linux was getting my TNT2 to work properly with X. Had to lynx newer drivers from nVidia. I ran the install script, reran setup and it works perfectly now.

    I had one of my more knowledgable linux buddies help recompile my kernel for SMP. Turned off the computer, switched one of the jumpers on the MSI board (WATCH OUT! they are very easy to drop and lose!!!) to set the bus speed to 100 mhz and I have 450 mhz celerons.

    My computer whoops ass in seti and I can run two copies of bladeenc at once with good performance. One other nice thing about dual cpu's- if Netscape or some other app takes a dumb, it has less of a chance of bogging down or crashing your system, becuase it can only take over one processor.
  • I don't know where you got your numbers, but I have to disagree with some of your assumptions:
    First on prices : (best Canadian prices I found )
    c400 : 149 ( x2 $298)
    K6-3 450 : $245
    Socket 7 board : 119
    Dual Slot 1 board : 400

    And I found that a K6 chip was 40-50% faster for kernel compiles as compared to a similarly clocked Intel chip.

    True, a dual celeron would be faster than a single K6 chip, but I'm saying that a Dual K6 would be a fantastic ( and not too expensive ) machine! The celeron system isn't *that* cheap!
  • I did it! My lone Celeron 300A is now 504MHz. Doesn't seem any faster, but RC5 might be able to crunch a few more keys... I had to change the core voltage to 2.4volts [ars-technica.com] and set the BIOS to ignore the "error" of this higher voltage. The CPU did not like 2.0volts and scrambled the eggs on the hard drive with 2.2volts while compiling the kernel. The CPU does not seem to put out any heat, so 504/113MHz it is!
  • A couple of months ago, the crazies at HardOCP managed to get a Celeron 300a to run "stable" at 600Mhz!!

    Of course, their cooling rig was a work of art...water-cooled TECs...
  • In reality, clocking Celeron 300As *is* that easy. Mine doesn't even need more cooling than the standard heatsink and fan (retail CPU), and it's totally stable with 2.1V core.

    Remember, the Celerons use the same core as the PII-350/400/450, with the addition of on-board cache (cooled by the same HS/fan rather than sitting there in a stuffy cartridge).

    This box has been running at 450 or 464 every day doing RC5. Overclocking isn't something for everyone, I make sure people know the risks assocoated with it before I'll help them build/clock their Celerons. But in reality, the risks are very low. If it dies, just go back to bios and knock it down a bit. Even when I couldn't get into bios (attempt at 600+mhz), it was just a single jumper on the board to reset it back to defaults.

    It's simple; know what you're doing, check the CPU temperature and test your chosen speed/voltage thoroughly before settling on it.
  • Posted by SmashPHASE:

    Intel didn't officially go as far as 42Mhz with
    the PCI bus specs, so quite some manufacturers
    have created hardware that dislike PCI busses
    running on 37,5 or 42 Mhz (mobo speed 75/83).
    Common examples are the Voodoo II cards(in particular the ones from Creative), most S3
    cards and (all?) Maxtor HD's.. On the latter, setting the busspeed higher causes busmastering transfers to fail and in the worse case you'll fry your interface.(afaik quite some IBM hd's also dislike it)
    Some mobo's have an extra jumper to
    set the PCI/AGP speed at a fixed rated (33Mhz).
    I'm running OCed Celerons since a year now
    (one 300 at 450 and one 300a at 450 at regular
    voltages) and it runs stable at 100mhz (= 33mhz PCI bus), though the xxxa series do require extra cooling because the cache chips tends to heat up a lot, when ocing.
  • I've been building/selling/tweaking computers for over 15 years. Most of the RETAIL versions of Intel CPUs overclock well and have great track records of longevity in my experience. The RETAIL version is worth spending extra money for if you plan on overclocking. We have many clients running OC'ed Intel PI/II CPUs with absolutely no complaints or failures. We haven't RMA'ed an Intel CPU in over a year.

    The only processor I consistently have trouble with is AMD. Many of our economy line computers based on AMD processors have failure rates like 1 out of 5. We have in a month replaced 4 AMD K6-2 266 CPUs ranging from 3-7 months old. They just die, no boot, a variety system configurations, and locations. Last week a K6-2 350 in a hardly ever used Win95/WinFax machine.
    I don't like SlotKet idea. It reminds of those SIMM savers that broke sockets and got in the way of things. Honestly, I haven't tried them myself but its too much of a kludge.
    I have ordered a ABIT BP6 with retial 366's, PC-133 128MB(Enhanced). Can't wait to play with it.

  • Posted by SmashPHASE:

    I only know about NT4 SMB systems and it won't let you use 2 processors if they aren't of the same
    speed, stepping and manufacturing plant.
    Though I know some people tweaked it to use different steppings. My guess is that using different CPUs under Linux also won't do...
  • Anyone here ever stress-test IDE?

    Using a Diamond Fireport 40 (single channel), two ultra-wide 9.1 gig IBM Ultrastars and one ultra-narrow 4.5 gig of the same family (all 7200 rpm), a reasonably quick Conner tape drive, 32x Plextor CD-ROM, and an 8x Plextor burner (all at 10MHz due to cabling issues), I have done the following, concurrently:

    Run a tar backup of part of an UW drive, while cdparanoia rips an audio disc in the CD-ROM and bonnie is repeatedly stomping out a 700meg file on the narrow drive, and cdrecord burns through a CD-R at 8x from the other UW drive.

    I let it sit for a bit, and noticed that cdrecord was showing no signs of its buffer beginning to empty, so I fired off updatedb and a few fscks on unmounted partitions.

    After a few more minutes of horrid thrashing sounds as the hardware struggled to keep up, cdrecord ejected a perfect CD-R.

    Can IDE do this, let alone on a P-133 such as that which I was using at the time? I suspect not, but I welcome any reports to the contrary.

    It might also be worth mentioning that Netscape (the most cpu-intensive application I regularly use) was very responsive during all of this, probably due to the near-zero CPU time required to tend to a decent SCSI adapter.

    (and for those of you who are thinking of flaming me for using a battery which is highly unlikely to occur in the real world, consider this: he who wants to have a computer which is capable of doing only one difficult task at a time is probably better off with the simplicity of ms-dos. it's faster, too - just imagine wordperfect 4.2 on a PIII/500!)
  • Hey,
    I wholeheartedly agree that a dual Celeron setup is a steal! I've had one going for about a week, sporting 2 Celeron 366s running at 550 Mhz (at default voltage, these are both week 15 SL35S chips, must've been a good batch) and boy it is fast (and rock stable, so far)!

    However, instead of the BP6 I opted for the Epox KP6-BS with MSI 6905 rev. 1.1 slotkets. I just wouldn't feel comfortable being stuck on socket 370. With this setup I can just swap in 1 or 2 PIII-550s by the time they become cheap and get some of that SSE-loving. On the other hand, there are rumours of s370 versions of PIII coming, who knows Intel's real roadmap anyway? Everyone take their chances...

    Cheers,
    Michiel
  • I would I am an ardent believer in the idea that you hold flawed methods up to public scrunity regardless of how unpopular.
  • Just don't spend more than $2,000-$3,000 dollars on a computer. Most people (about 97%) have no need for the processing power of a mainframe at all anyway.

  • The memory bottleneck (Score:3)

    by Admiral Burrito ( 11807 ) on Sunday July 11, 1999 @07:33AM (#1808706)

    The idea of using Celerons only works if you can overclock them. It may go well for hacks and tinkers, but for the heavy video and I/O, a 66Mhz bus frequency is unacceptable.

    In the article, the guy ran one of the chips at 2.1 volts... I wonder how long it stayed stable at that speed/voltage. I have a C300A that runs perfectly stable at 450 MHz 1.9v, but crashes regularly if I use default 2.0 voltage. These PPGA Celerons don't dissipate heat as well as their Slot-1 brothers can.

    I would expect memory bandwidth to be an issue with SMP Celeron systems, especially if you're not overclocking. The last part of the article touched on this, but I think a lot of people still don't get it...

    The Celerons only have 128k cache, compared to 512k on the P2. The Celeron cache runs at twice the speed of the P2 cache, often resulting in better single-processor performance, but because the Celeron cache is smaller it has to go to RAM more often. This means that the Celerons consume more memory bandwith than the P2s.

    Combine this with the fact that an un-OCed Celeron runs with only a 66 MHz bus, compared with 100 MHz for a P2. Even though the Celeron needs more memory bandwidth than a P2, it gets less.

    Now stick two of these bandwidth-hungry chips on the same board accessing the same 66 MHz RAM...

    I suspect it would not be worth it unless you used 300As at 450+ (100+ MHz FSB).

    Does anyone have performance numbers for un-overclocked Celeron SMP?

  • Its not a SMP with my P2B motherboard, but I have tried my Celeron at 503MHz with the bus speed at 113MHz and it will run for about ten minutes before the kernel oops with a memory page error. Its been running at 463/103MHz since October without a glitch. Has anyone had success with a 113MHz bus speed?
  • After reading this article, I'm now very interested in building a cheap SMP system. However, I want to get the most bang for the buck.

    In your opinion, what is the best celeron to buy for the money (esp. considering the overclockability of the chips). I saw pricewatch advertising 300a PPGA's for $51 and 400PPGA for $76.

    Also, what is the best mb to buy for the money? Is there a good one that has sound and video integrated and works well with linux?

    Anyway, I'm trying to keep the price close $400 (not including shipping) so I can do it before summer is over.

    Here is my tenative(sp?) plan:

    2 - 128meg PC100 Dimms - $140 ($70 a piece)
    2 - 300a Cels - $102 ($51 a piece)???
    1 - m/b ?????
    1 - 8 gig drive free (already have)
    1 - 10/100 d-link nic $30.00
    1 - cdrom ~$40.00

    Anyway, any advice on how to do it/what to buy would be appreciated.

    Thanks,

    Waxman

  • If you'd bother to read the cpu review which this thread is after all about, you would know that ATA-66 drives are comparable in speed to SCSI 2UW. Note: I say "comparable" not as fast as. But for the price, you can't beat it.

  • But there are places that will sell you a tested overclocked celeron system.

    I cam across this one the other day:

    http://www.becomputing.com

    Ben

  • While I have not reviewed it (YET) the Abit BP6 seems like an excellent board; and it has an AGP slot so you can add a better video card.


    Two 300A's @ 450A (if it is stable, no guarantees) are an extremely good bang for the buck. I've been hearing that the 366's often will run at 550; but that is even riskier than the 300A's at 450.

    hope this helps,

    Bill
  • The celerons are underclocked to begin with. The core is the same as that of the PIIs running at 100 FSB.. The only issue should be the cache. I run dual 300As at 450 with no issues. The CPUs don't even really heat up that much.

    From what i've seen, the biggest issue when doing SMP OC'ed celerons is the MB and the slockets.

    Intel CPUs are tanks.. I've worked at a computer store for over two years.. Since the day I've started we've had a total of 3 intel CPUs ever fail (including DOA).. I've seen some techs/customers do some pretty evil things and the CPU just gets up and runs again.. In one case a computer was caught in a lightning storm, and the line surged. Took everything out but the CPU.

    Besides, even if you did manage to break a celeron, they're cheap as hell. My MB cost more than both CPUs and the adapters.

    I think everyone should at least look into Dual Celerons for their next upgrade. Highest bang-for-your-buck ratio i've ever seen.
  • A few months ago (I really should have waited) I bought a tyan tiger MB, the cheapest dual bx I could find. I got some modified msi converter boards, and a couple 300a's. The total price was about $80 more than it would be now with that dual 370 board. They both overclock to 450 just fine. I keep a big fan next to it, so they only get up to about 34 degrees on a really hot day. It's crashed a few times in a virtual terminal, but it even does that in cold weather and not overclocked. never crashes in X though... odd.

  • Re:Very risky (Score:1)

    by Anonymous Coward
    My Utra 10 does 8.7mkeys/ec, so neener :)
  • Thats the thing then, IMO. I have seen many MB at work, but Asus are the best so far. Can't wait to get one. Hell, can't wait to try this setup, and give my AMD K6-2 366 to a friend of mine still running a P166mmx. But remeber, Ram is key. Get at least 128megs if you want a realy kickass machien, and 256 later on, when you've saved up another $100. Also, Word to everyone, for real performace, get CAS2 rather than standard (CAS3) 100MHZ sdram. It has a lower latency, and is faster. It is also about $15 more than CAS3 stuff, but worth every penny.
  • Takes me 2-3 hours he he he
  • A brief log of my experience can be found at my website [209.233.19.231]. Boring ... it just worked.

    Stan
  • First, ram is truely the key to a fast system. Yes, having a badass, dual CPU Celeron system OC'd to 464 is awsome, but if you are going to spend the $240 or so, bare minimum, on this, spend another $100, and get 128M of ram. And not just any ram. Be sure that when you are buying it, it is CAS2, rather than the standard CAS3. Of corse make sure it's PC100 as well. Don't go for PC133 just yet, it is far to expensive, and no MB or CPUs support it yet.

    Second thing, send me lots of money for my advice! I want to get one of these too. %^> Oh well, guess thats not going to happen. Looks like time to save every dime I get again.
  • I shure would like to see a poll with something like "what is your processor"?

    1. 286
    2. 386
    3. 486
    4. Pentium
    5. PPro
    6. PII/PIII
    7.AMD
    8.Cyrix

    mine is in class 3 and it apparently sucks with the above benchmarks in comparison
  • I think this is an excellent idea and think more
    people should take advantage of this feature intel
    poorly removed. I have a dual celeron 300 mhz
    system using a tekram dual slot 1 motherboard
    and i am using two msi slocket adapters. I
    overclocked both 300 mhz chips to 503 mhz using
    fans (over a GHZ!) I did use a whole lot of fans,
    2 x 3 fan units for the cpus and 3 case fans. The
    case runs at 90 deg avg. This system has been
    running this for a month and a 1/2 totally stable
    running 2.2.6 It is the fastest thing I have ever
    seen. I can compile a kernel in 3 minutes flat.
  • I just got my celeron 366s in today and I have the bp6. The only thing I've gotten to work with the hpt66 controller is win98 :|

  • I saw a reference to doing exactly this just yesterday. It is possible to run SMP with Celerons of different speeds on the new Abit board. It seems that it doesn't actually unlock the multiplier though.

    Check out Hot Hardware [hothardware.com] and HardOCP [hardocp.com] for the scoop.
    --

  • Has anyone put together a Linux SMP system based on the dual-socket 370 motherboard from Abit? I saw an article about it a few weeks ago but nothing about pricing, availability, or technical details of the dual-S370's SMP capabilities/limitations...

  • I put together a Dual Celeron machine (Asus P2B mobo and 2 PPGA Celerons with Slot-1 converter boards) and it does indeed fly. You can't beat the bang for the buck on this one.

    and oh yeah, first p0st!@!#@@! :)

  • Very risky (Score:3)

    by heroine ( 1220 ) on Sunday July 11, 1999 @05:33AM (#1808733) Homepage
    The idea of using Celerons only works if you can overclock them. It may go well for hacks and tinkers, but for the heavy video and I/O, a 66Mhz bus frequency is unacceptable. The chances of overclocking single celerons to a 100Mhz bus is 75%. For both processors in a dual processor system, the success drops to 56%. Also since people are much less willing to report failure publicly, the percentages may even be lower in real life. You won't see anyone post a failed overclock on slashdot.

  • Re:vr000m (Score:1)

    by Anonymous Coward
    0h y34h, 1'm sUp3r l33t..1 g0t th3 f1rst p0st!

    ph33r m3, 1 c4n r34d pr0n tw1c3 4s f4st w1th tw0 c3l3r0ns!#@
  • I will be trying it out tommorow hopefully. I have already received my Abit bp6 and I'm just waiting for my celerons to be delivered. Their sitting in a ups station 20 mins from my house, but yet I can't get them till monday *sigh*
  • I read somewhere that Intel was going to start producing the pentium !!! in the ppga style.
  • Actually, the performance difference between a celeron 400 and a p2-400 is pretty slim. AFAIK, it's only about 3%. But I can't wait for my 2nd 300a and my new BP6 to arrive so I can run dual 464 :)
  • I have 2 366 MHz Celerons on my BP6. Mandrake 6.0 and BeOS 4.5 run very nicely, thank you.
  • Abit is a fine motherboard company, and benchmarks have shown that dual celerons run just as fast, if not faster than pentium IIs. This means Intel will try to take out the celeron's ability to do smp. (forcing us to go with the more expensive pentium IIs/IIIs) Wouldn't that render the BP6 useless in the (near) future?
  • The people I'm buying my CPUs from test them first, and guarantee a given clock speed at a given voltage.

    I've ordered two Celeron 300A PPGAs, each guaranteed to clock to 464MHz at 2.0V, and the Abit BP6 that (I discover a week later) everyone's raving about.

    Woohoo!
  • It's so risky that companies are marketing
    products based on the idea? (http://www.computernerd.com)

    and setting prices on the guaranteed overclock
    MHz?


  • Mine is on order and due to arrive next week. Price tag on my custom built system from KC Computers [kc-computers.com]? $2500~. Here's what I got:

    • Abit BP6 MB
    • (2) Celeron 466MHz PPGA cpu's
    • (2) PC Power heatsink/fans
    • 18GB IBM ATA-66 HD
    • CD-RW Ricoh
    • DVD 5x player Toshiba
    • LS120
    • 19" TE995 Relisys Monitor
    • ASUS V3800TV nVidea TNT2 Ultra Video Card w/ 3D Goggles
    • 256 MB CAS2 RAM
    • kb,mouse,floppy,case w/ (extra fan)
    • 10/100 Mb NIC 3Com 3c905b

    As you can imagine, I'm chomping at the bit in anticipation. And with ADSL services at $49.95...

  • What happens when a CPU gets too hot is that the system locks up. You wont get any burning plastic from an overclocked CPU unless you try something really stupid. I've got a standard Celeron 300A, no extra fans, no extra voltage, just bang it in and run it at 450MHz. I've had it for nearly one year now ane not one problem.
    You're just spreading the Intel FUD.
    (Besides, the Celeron beats the PII in some tests when you run them on the same clock frequency, due to the fact that it's cache runs at processor speed)

    Anders.



    ____
  • yes, there isn't much of a performance difference, but when it comes to overclocking, there's always the risk of killing a cpu, the risk of frying your whole system, and the risk of getting an unstable system. Maybe it wouldn't be so risky if you are just going for an extra 10 to 50 mhz, but this is going from 300 to 450/464/500+ !@#$? Of course we all heard the success stories of overclocking, but heroine brings up a good point -> no one ever reports their failures publicly. While many people make it sound so easy to overclock (set it to 100 mhz bus, up the voltage a little, be sure to have good cooling) in reality, nasty stuff can happen even if you just make one small mistake. (i remember smelling burnt plastic) So think about the risks first before you try to overclock your ~$100 ± 50 celery
  • I've heard positive reports from people who have, and my dual-370 stuff is on the way.
  • I've been using a dual-celeron system for ~ 1 month now ( 366 on 83MHz bus - 458MHz ) and have had NO trouble at all ... standard fans / heat sinks, and they're cooler than standard PII's ... and as long as you've got enough ram this is the fastest machine I've ever had the pleasure to use. Now if only I could find an OpenPIC mobo and run dual K6-3's !
  • I don't doubt the sustained transfer rate to a single device is very similar.

    But ATA-66 is still IDE, which is still a positively braindead way of interfacing a hard drive.

    The point of scsi is not that scsi is fast

    The point of scsi is that scsi is a fast interface that doesn't have to bog down your motherboard and cpu resources.

    What's more, SCSI can multitask i/o between devices. ATA-66 can't.
  • Posted by Open Matrix:

    Well you've never tried to overclock a Cyrix (I know!!!! That was back when I didn't know anything)P150+ to 200MHz. Mine overclocked fine and ran fine and semi-stable but after about ten seconds you start smelling ugly smells but it still ran. One time I actually put a small box fan on it blowing straight from my AC vent and it ran for quite a while without crashing and all this with the default voltage. That was the most overclockable non Intel chip I've ever had. Too bad about the FPU on it though.
  • the memory price was from pricewatch. PC100 - 128 megs for $70 a piece
  • I built this system with an MTech M668DS LX chipset dual slot 1 motherboard, matrox G200, 128 megs of 6ns ECC SDRAM.

    I'm using two celeron 366's on pre-modified MSI MS-6905's. I am not overclocking.

    Doesn't work worth a damn. My uptimes range from 3 hours to 2 days. It's like using Windows or something.

    If it were a memory problem, I'd see ECC complaints all over my logs. My cpu's have huge heatsinks and lmsensors report that they rarely get over 44c, so it can't be heat.

    It usually crashes when the system is doing nothing more complex than animating the "Ifs" xscreensaver module. It never has crashed during a kernel compile, though i recompile every week or so, and use make -j4.

    At this point, all I can figure is that the slotkets are defective. I don't expect to get a refund from the place i bought them from. I've ordered (far more expensive) PowerLeap slotkets on the off chance they might work better.

    If that doesn't work out, I'm probably tracking down a pair of PII-333's. Which would suck, because this system performs very well when it's not locked up.
  • What speed does the PCI bus run at? I've had problems with devices where the PCI bus is run at over 36Mhz (i.e. bus over 75Mhz) since there is no clock divide suitable (at 100Mhz, you divide by three, and at 66Mhz you divide by two)
    John
  • Here's [arstechnica.com] a story at Ars Technica comparing dual Celerons to dual PIII's, both oc'ed. Results were suprising, to say the least.

    (I don't believe this is a fluke, either. My machine is basically the nerd box, but at 464, and it is thus far is proving to faster than I expected, close to PIII 500 speed (without SSE, of course).)
  • Really? I have a very similar system ( except ASUS P2B-D ) and have had no trouble ... Have you tried running uni-processor? Help in diagnosing ...

    If it boots and is unstable, that says to me that it is a heat-related problem ... once the machine gets up to running temp it crashes. Since the cpus and adapter cards would reach their running temp MUCH faster, I am tempted to point to your motherboard...

    Something to try, at least!
  • Most people I hear hitting 503 have to bump up the CPU voltage to 2.3 or 2.4V. I know of one that has done it (a co-worker). I, however, am afraid to run it any more than 464/103. Both are with 370 Celerons. From the word on the street, the 370 Celerons are about 90-95% effective overclocking to 450/100. About 80% are effective at 464. I've even heard of one company that guarantees 504/113 at 2.4V. I'm just afraid to run my CPU that hot.

    Now, I just want to sell my existing mobo and get one of those Abits for dual Celerons. I'd love to see that...
  • If enough (who knows how many that would be?) people state they want the feature left in, Intel might cave to public pressure. Remember the DIV bug? F00F bug? In both cases they gave in to public pressure.
  • We just received a couple new systems, similarly spec'ed out, where I work. Two of the systems are nearly identical, except for the fact that one is running an IDE interface (ATA-66 hard drive, IDE DVD), and the other is running UW SCSI and narrow SCSI (UW hard drive, Narrow CD-ROM).

    The SCSI system is so much faster that the IDE system, I really find it hard to believe that ATA-66 is all that great. IDE is IDE, and still has a long way to go before it will come close to the efficiency of SCSI.

    Of course, from a performance/price standpoint, today's IDE is pretty impressive. I wouldn't mind having IDE in my home computer, but my servers will NEVER have an IDE drive system, period.

    SaDan
  • > I can compile a kernel in 3 minutes flat.

    .. and that does NOT sound impressive.
    My K6-233 (Not K6-2 even) does compile in under 5 minutes
    with somewhat old HDD
  • This sounds like a bad idea. Intel obviously dosen't want people to do smp with celerons, they want people to have to shell out cash for the p2/p3's. If alot of people sign the petition, intel will know how many people are using celerons for smp. Then they will figure out how much money they are losing and this will justify them taking smp out of the celerons.
  • i guess you could count me. just powered them on yesterday. i'm using sl36c maylasia parts and the globalwin fans. they seem to go 550 at 2.05V.

    "guess" and "seem" are because i see occasional weird behavior, not outright crashes. kernels compiled -j 4 do work.

    this is all on mandrake 6.0, which does seem to be an improvement over redhat 6.0.

    i was fully prepared to go as low as 506 (92 FSB, still with 1/3 PCI divisor) but the board absolutely doesn't want to hear about that FSB frequency.
  • lmsensors reports core1 voltage at 2.01vdc and core2 voltage at 1.47vdc -- given the likely margin of error with an on-board sensor that hasn't been calibrated by a human, those seem fine. Or maybe they're lower than reported.

    The PowerLeap boards have their own regulators onboard, that might help the situation. Or not.

    Interesting you should point out AGP as a possible culprit. I tried like heck to use a PCI video card in this system, but for some reason it refused to work. I really, really didn't want to blow $100 on a good AGP video card when i had a perfectly good PCI card already.

    With a PCI video card, I couldn't boot multiprocessor, and couldn't start X in uniprocessor mode. Probably a bios or MTRR problem. MTI hasn't been helpful. I don't recommend their products.

    Other cards in the system include a pair of RTL-8029 pci nics, an AWE32, an Aztech FM radio card, and an ADS ChannelSurfer TV (similar to Hauppaugge WinTV)

    Oh, and there's integrated AIC-7880U on the motherboard. AHA-2740UW for the uninitiated. This, for some reason, always shares it's irq with pci slot 1, which is the nic driving my DSL connection.

    Nothing in the box is at all out of the ordinary, and like i said, it crashes at the dumbest times.

    If it crashed during hard disk activity, I could point a finger at the SCSI. But the disk is usually idle when it crashes.

    It sometimes crashes during MP3 encoding, but I can't point the finger at cpu load because it has never crashed during a heavy compile, which puts far more load on the system in terms of memory i/o and processor use.

    The only commonality i can find between the common crashes is floating point operations. Ifs is a fractal animation.

    The motherboard is indeed properly mounted. I trust the power supply, but have considered getting a higher quality power supply with a higher volume fan.

    Frankly, I'm baffled. The only parts of the system that isn't 100% on the level are the wire jumpers on the slotkets. I suspect cold solder joints. This is why i have ordered new slotkets.
  • aww geeze, isn't where i ment to say aren't. Man, I feel smart.


  • Maybe this petition should be forwared to microsoft (yes i know) because AFAIK win2000 will not be supporting dual celerons (and believe me build 2031 does not support dual celerons)

    Surely if intel was to remove the smp capability of the celeron they would lose market share to another cpu company.
  • My understanding is that the K6 does not have the APIC (interrupt controller) needed for SMP. The K5 had a SMP capable PIC but nobody supported it. I've read that the APIC is patented by Intel.
  • AFAIK the only dual-configuration socket 7 boards use APIC, the Intel patented processor communication protocol ... Cyrix and AMD consequently use OpenPIC, which is used by manufacturers like Motorolla. Hence, you can use Intel chips in smp configurations, but not AMD or Cyrix ... (but who'd want to, Pentiums have no on-die cache, would have to share L2 between cpus and that would SUCK! )
  • I'll try a BP6 (hopefully in the near future); when I picked up the M750i the BP6 was not announced yet.
  • Maybe it doesn't matter. The people at Rise (http://www.risetec.com.tw) told me that their new cpu will be a Socket 370 400 MHz Celeron clone. It won't be overclockable but it will be able to run SMP. Should be out 3Q 99.
    BTW, they test their cpus with Linux, Be and QNX.
  • I'm running a petition to ask Intel NOT to remove the SMP capability from Celerons... I can use more signatures :-)


    4957 signatures to date.
  • He he he...

    Let me quote myself:
    ...unless you do something really stupid...
    ;)

    I still don't think this would be an issue with the Celeron, though.



    ____
  • I've heard that when you overclock the 466's the bus speed is so out of whack that alot of peripherals don't function anymore. The guys at computernerd.com say the best setup is a pair of 366s overclocked to 550.
  • CAS2 RAM is the only way to get a stable OC'd system. It can operate on elevated BUS speeds, and has a tendancy to dissipate heat more readily than your normal RAM. CAS2 RAM is also more expensive. Check out pricewatch.com. --Bryan

  • Would that be considered monopolistic behaviour? After all, people are buying more and more celerons because of it's ability to use SMP, yet being cheaper than P2s and XEONs. Yet Intel forcibly wants them to buy the more expensive chips if we wanto to do SMP.

    Inquiring mind want to know.

  • Last weekend I took the plunge and
    bought an Abit BP6 and two - week 14
    Malay Celeron 366's. I have them
    running rock steady at 450MHz under
    Mandrake6 [1009.26 bogomips SMP] (and NT-4 and
    win98se). I've had it compiling kernels and
    plowing through setiathome for a few days
    now and so far not one crash.
    Now if there was just a way to watch
    DVDs under Linux life would be perfect!

    Todd
    system desc:

    Abit BP6= $145
    2*366 Celerons @ $78 = $156
    128MB PC-100 ECC Memory = $128
    Quantum - FireBall CR (13GB 66MB/s) = $173
    Toshiba DVD SD-M1212(OEM) = $98
    InWin A500 Case = $58

    Rest I already had laying around (Matrox G200
    etc.)


  • I was just wondering how many people out there have had sucess o/c their celerons 366s to 550 ppga only. I ordered them with the abit bp6 and would love to have them run at 550.
  • It would be hard to call Intel a monopoly. They are definitely the largest CPU maker, but they do have serious competition from AMD, Cyrix, Rise, IDT (uh... forget IDT), and in the Linux community the PPC and Alpha.
    2/3 of my computers use non-Intel CPU's. And the only reason one does is because the mb wouldn't do the 83MHz the Cyrix wanted, so my friend swapped his P200 for a 6x86-PR200.
    If anyone knows if Cyrix's can be used for SMP, please e-mail me.
    Digital Wokan, Tribal mage of the electronics age
  • Just check this out:

    Make sure that you're mobo is mounted correctly and specifically that there is no extra metal spacers underneath it (you know, those that you use to secure the mobo to the chassi). I have had random lockups because of those short-circuiting everything, and I could have sworn that it was a defective PCI bus that was causing it.

    Anders.



    ____
  • What's REALLY fun is when 12 volts gets put into one and you see a jet of flame coming out of your 200MHz Pentium Pro:)

    (several years ago - Bad motherboard/power connection)
  • how much are you paying for the chips..?
  • I'm not spreading FUD, I'm speaking from personal experiences. I have seen cpus get fried and it ain't pretty. Maybe your celeron can run at 450 with no tweaks, but that doesn't apply to every celeron out there.
  • I aint goona spending $600-$1000 for dual pentium IIIs, the whole point is to use two cheap celerons for smp, right?

  • Thanks for the info.

    I don't understand how the FSB speed would affect a C466 running at 75 MHz FSB any differently than a 300A. What I've heard, is that the 466MHz and higher speed Celerons just don't overclock as well.

    FYI: I don't intend on overclocking this system until it is pretty out of date. I.e., a year or two.

  • Just curious, now that Celerons have the locked multiplier, could I run a 4.5x and a 5.0x (300a and a 333a) in SMP? I currently have a dual 300a/504 system, but it has occurred to me that:
    a) On old systems, the multipliers were set at one spot on the motherboard
    b) The processors don't know what the multiplier settings of other processors are
    c) They communicate to each other at the bus speed

    This to me would mean it is entirely possible.
    I know people who have gotten a PII and a Celeron to work. Anyone want to take a stab if you can have to different multipliers?

Slashdot Top Deals

Don't tell me how hard you work. Tell me how much you get done. -- James J. Ling

Close