Slashdot Mirror
New Dual-Celeron PC's Encourage Overclocking
Steve Nakhla writes "The same people responsible for the iMac-looking PC's are now shipping a dual-Celeron PC, that makes overclocking a breeze. " So the question is, Publicity
Stunt, or a valuable feature that consumers want? A motherboard that
makes 8x overclocking easy is certainly cool, but its definitely gonna void some warranties, and maybe even blow up some chips.
I stopped purchasing and reccammending the bp6 boards simply becuase I could not rely on them to
reboot properly. Installed 98, NT4, DreadHat-6, none of them would reboot without haveing to
powercycle the machine. Needless to say this is
not usefull as a reliable machine.
I'm using an overclocked dual celeron machine right now... This article was VERY VERY misleading about a lot of things related to overclocking these beasts...let's begin:
:)
* You can't change the multiplier for the celerons with any celeron except for the 300a. After the 300a, intel installed a multiplier "lock" on their celerons after that. The 366, 466, etc., all have a locked multiplier that you must use to make the machine boot. I have a celeron 366, and I use the 5.5 multiplier that the chips specifies.
* To overclock the celerons, rather than upping the multiplier, you increase the bus speed of the board. Celerons default to 66MHz bus. If you increase that in small increments, you can get higher speeds. I managed to get my bus speed to 95MHz, so I'm running my dual 366's at 523MHz. It's hot, and requires a BIG case, with a LOT of fans. (8 in this computer)
*Disadvatage: When you change bus speed to anything other than 66MHz or 100MHz on this motherboard (abit bp6), you are overclocking or underclocking your AGP bus, and your PCI bus. This does WEIRD things to hardware, sometimes. If the hardware doesn't have a high tolerance for this sort of thing, you can have problems, like frequently losing data on the hard drive.
I was pretty lucky with my setup. I ordered two celeron 366's "guaranteed" to reach 550MHz. This would be ideal, because I'd be running the 5.5 multiplier at the 100MHz bus speed. The AGP and PCI buses would be running at their standard speeds. Unfortunately, I cannot get the machine to stay stable at 550MHz. To make it reach that speed, I have to change the voltage on the chip from 2.05 volts for 523MHz, to 2.3 volts for 550... If I don't change the voltage, the machine refuses to boot any os before getting an error. If I change the voltage, the computer is only marginally more stable, and quickly overheats and locks up, despite my cooling.
To make a long story short, BE CAREFUL with this offer. The 366->550 is about as far as you can really take the celeron chips. The chance of doing that successfully is 1/4 per chip, I think. The 466 isn't going to overclock to the full 100MHz bus speed, because the celeron fabrication process maxes around 600MHz. Maybe you can squeeze 620 or so out, but you're really pushing it. If you really want to be guaranteed a good overclocking experience, get yourself some 300a celeron chips. They go to 450Mhz (100MHz bus) 80% of the time... But they are rare, and are as expensive as the 366 multiplier-locked chips, sometimes.
Don't buy one of these pre-built machines and expect to run 8x anything... The author was very confused. Go to http://www.arstechnica.com/ and read about overclocking a little, and you'll be a lot safer in the end.
Also, don't ignore the advantage to running these machines without overclocking... Dual processor is nice. Celeron 533's are going to max out the celeron line, before they hit a new chip-fab...so... If you wait a little while, maybe 3 months or so, you can pick up some 500 or 533 MHz celeron chips, pop them in one of these boards, and maybe do some slight overclocking to squeeze another 50MHz out of it... and you've got yourself a great dual processor board that will outperform comparable pentium II's, due to the higher cache speed.
-Larry
sorry about the length of the rant.
...These aren't the droids you're looking for....Move along....
Today's English Lesson: Oxymorons
Sanity.html - Error 404 not found
Yep. My PC will always be a little pipsqueak compared to those billion kajillion dollar super-raging-stonker computers. For my purposes, an overclocked Celeron has a stupendous price/performance ratio. If YOU can afford those machines, bully for you! Leave the rest of us peons alone.
Why yes, I AM a rocket scientist!
>Too bad Quake doesn't utilize SMP.
Really? So why is Quake2 v3.19 faster on NT4 than 95 when software rendering on a dual CPU motherboard?
Both these figures are on an Abit BP6 with dual Celery 400's (not OC'd)
demo1
'95 400x300: 32.2fps, 800x600: 13.2fps
NT4 400x300: 37.3fps, 800x600: 16.2fps
crusher
'95 640x480: 13.1fps
NT4 640x480: 15.0fps
Note that when using OpenGL mode, NT is about as much slower than 95 as it is faster with software (partly due to the lousy NT drivers most cards have - OK for 2d, suck for 3d).
That is what I was going to say if nobody else came along and already had.
Intel is going to really take notice if there are businesses selling Intel products in a manner that they are not supposed to be sold.
Shame on Future Power for doing this, and for selling an iMac clone. They need to get a new business strategy.
There is a petition for Intel to not take SMP support out of the next Celeron (based on the PIII). http://www.cpureview.com/smp_petition.html
EverCode
On less than 100Mhz FSB speeds, the PCI bus runs at 1/2 the bus speed, eg, 33Mhz for a 66Mhz system bus -- at 100+ speeds, the motherboard makers set the PCI to run at 1/3 the bus speed, so at the officially supported (66, 100) speeds, the PCI stays at 33Mhz, and everything is happy. When you up the bus speed to 75Mhz, for example, the PCI cards are running at 37.5 - over 10% faster. At 83Mhz bus speeds, your PCI bus is clipping along over 40. OTOH, if you clock a 100Mhz FSB chip at, say, 113, you get 13Mhz on the bus, but you're not pushing the PCI bus as hard for it.
because pci runs at 33mhz (1/3 100fsb) and agp runs at 66 (2/3) roughly. anything else and those end up at odd values, which cause problems for some cards, especially high speed agp cards.
i want one! now with duel smp's i'll kill anyone in quake!
So does this mean the store won't try to sell me a warrantee too? Somehow I doubt it.
A dual Celery box is tantamount to OC'ing anyway. (It voids your warranty, and Intel Internet Support won't have anything to do with 'em. :) )
So, OCing and dualing Celeries go hand in hand... _but_ the tolerances are a lot tighter in an SMP setup.
The 8X maximum multiplier in the article refers to the processor speed relative to the bus speed, not relative to the "official speed". Current Celerons use a 66mHz bus, so 8x would mean a maximum speed of 533mHz. These machines will default to 7x for a 466 mHz clock.
don't mind the smell of fresh semiconductor gone wrong. Some of you are recalling the smell as I write this now. This actually harkens back to the olden days of building your own tube radios to get the performance you knew you could have if you were persistant enough. Same thing.
Sure, semis don't spark as much, but that saves you the inconvenience of having to hide behind the couch when you switch things on. Even today if you were to put an electrolytic cap in wrong, that would pop like a firecracker.
Think of it as baptism by fire. Not for everyone, just those who want to push the envelope.
Ahh the smoke theory. For those of you not familiar with it, some electronics techs (I picked it up in the air force) believe in a smoke theory. It states that all electronics run on smoke. If you do something wrong, the smoke escapes, causing an odd smell. Once the smoke escapes, the components don't have anything to run on.
>I trust a board when it's "compiler stable" I use gcc for my tests.
I am running a BP6 at home, dual Celery 400's, not overclocked. Kernel compile time for kernel 2.2.11 (running 2.2.11 SMP) is:
elapsed: 169.379 user: 276.080, system: 26.96
FWIW: I was running X, but no mp3's. I doubt that playing mp3's would have done anything except extend the elapsed time for the compile.
I've been running NT, '95, and linux with no stability problems on any OS (and I admit that's unusual for '95). YMMV. What were Anandtech doing to destabilise their setup? Overclocking perhaps?
If you noticed, there was a quick poll question as to whether or not people should be allowed to overclock their own computers. The results are undoubtedly weighed heavily toward "Yes" because of all the /. traffic.
I looked at the results myself just now, and it was 96% "Yes". Not surprising, although it is surprising that there are 4% who said "No". Apparently these people feel that you should not be able to do things with your own property that are undoubtedly legal and safe (well not for the processor, but has anyone died in a house fire that was started by a P200MMX clocked at 1466 Mhz?) as well as possibly beneficial, just because it voids the warranty and possibly shortens the processor's life about 5 years. Not that anyone with the brains to overclock would be caught dead using 400mhz Celerons in 6 months, overclocked or not. At the time I looked, there were 130+ who voted "No". Those are probably people who were lucky to find the power switch, there's probably 300 more people who are still looking... (scary)
Not only would this be better appeal to hackers in the near term, but the more people that void the warranty, and thus fry their processors/motherboard, the more products they're likely to sell.
Very strange business model, if they actually cooked it up.
These types of announcements are a sure fire way to push Intel to add restrictions to the Celeron processors. I fully expect we'll see this when Coppermine comes out, the packaging will change and the chip will no longer support any overclocking or Dual mode.
And yet you're wrong. The Pentiums and the Celerons come off of the same assembly line. Therefore, the differences are minor by definition. The Celerons are just the lower end of the batch. But Intel's got their manufacturing process down, so even the low end of the batch is pretty damn good. But Intel needs lower end/lower price chips to compete with AMD, so they ship them out locked down and at a lower price.
Intel wants the Pentiums to be capable of operating in a dual-processor system. However, they don't want the Celerons to work that way because they want people to spend more on the Pentiums if they want higher performance, thereby increasing their profit margin. So they lock the Celeron so you can't use it in a dual-processor system. But the lock has to be simple, or else Intel has to create a new assembly line for Celerons, which defeats the purpose of having them.
As far as new chips from Intel goes, I suspect they will continue to want a lower priced processor on the market to keep competing with AMD. How they go about that is anyone's guess. I don't know that it will ever be cost effective for them to run a separate assembly line for the lower end chip, however, since those are by definition lower margin sales. It may not justify the overhead of having a separate line. Only time will tell.
-Todd
---
"The details of my life are quite inconsequential..."
The PII and PIII place the 512k L2 cache on two separate 256k modules and a tag RAM chip, outside of the CPU die but on the same PCB board. The Celerons have the 128k L2 cache on the same die as the CPU. Therefore, the Celeron die is LARGER than the PII or PIII CPU die. Similarly, the PII and PIII CPUs (just the CPU modules, not including the cache) have a higher yield than the Celerons.
The Celerons are easier to overclock than the PII and PIII because they don't have the external cache modules holding them back, which are the limiting factor in overclocking the PII and PIII. If you disable the L2 cache, you can usually get a PII or PIII to clock faster than a comparable Celeron.
Finally, nobody outside of Intel really knows whether the Celeron is much cheaper to produce than the PII or PIII. The larger die size of the Celeron may make it more expensive, but the lack of external cache modules (which are cheap) and lack of a plastic case may offset the difference. My belief is that the production cost of the entire Intel CPU line is relatively constant, and also very small compared to their wholesale prices. Their pricing structure is artificial and largely market driven, not based on their production cost.
The performance decrease suffered from going from a 512k to 128k L2 cache is more or less offset by making the Celeron cache full speed vs. 1/2 speed for the PII & PIII. At equal clock speeds, the Celerons actually perform slightly better in games and other FP intensive applications while the PII and PIII perform slightly better in multi-tasking loads. Overall, the performance differences are very small (less than 1% in either direction) in most cases.
Yeah but look it bring a system to the table. And another thing, the system is really cheap and comes well packaged. It also runs linux out of the box, if it is RH though I will look the other way :-), what we need are some slackware running PCs ;-).
So far I've overclocked 466s and 366s. There is a correlation between batch numbers and overclockability. Unfortunately when you overclock, you reduce the environmental tolerance of the chip. You can't run it outside in 90o weather anymore. You can't use it on the road when you're unsure of the room temperature.
Also when you get pretested chips, the tests they use aren't very comprehensive. The same chips which run Prime95 forever crash easily with Broadcast 2000. Also one reason they're selling CPUs matched to the dual motherboard is that the two sockets are electrically different. Some CPUs are only stable when inserted in one order. Instead of getting those CPUs back as defective they're matching them to the socket they work in, but if you accidentally change sockets without knowing this, you're out of luck.
not true, some of that Lucent stuff has drivers. Still utterly cheap shite.
matt
www.cpureview.com did some dual celeron reviews and linux kernal compile stuff. I think he came up with -j (# of cpu +1 ) for best compile performances over one or two cpus.
matt
I have been running a BP6 w/dual 400mhz Celerons for about 4 weeks now. No overclocking here (I have tested the CPU's running at a high FSB) but not for any length.
:)
The machine has been running fine, no problems that I can see.
I have compiled the kernel several times, along w/Wine, etc etc... No problems, and it is fast
r_smp 1 enables SMP in Quake3
Overlocking a pair of 466s? You have to be joking...
The only FSB speed I would consider running anything at is either 66 or 100mhz, and to overclock a 466 celeron up to a 100mhz FSB, it would run at 700mhz. Sounds nice, but is, as far as I know, impossible to attain. Even with 2.3v through the cpu.
The easiest to overclock have always been the 300As, (which I myself have running at 4.5x103), but the 366s became popular for running at 550 with a bit of tweakage.
I have a mate with a Abit BP6, and two 366s running at 550 mhz each, which totals to 1.1GHz, if you use a SMP-enabled OS, such as linux or NT.
However, his case temp is up to 40 degrees celcius, which I regard as ridiculous (as mine runs at 20).
My point is, the higher the clock multipler, the harder they get to overlock. Anything above 366 is very hard to reach 100mhz (FSB), and running the FSB at 75 or 83Mhz is just plain stupid, due to the AGP/PCI overclock, which can (and ofter does) result in destroying your cards.
So buying a dual 466mhz system, with hopes to overclock, will leave you disappointed.
---Transmission Ends.---
Overlocking a pair of 466s? You have to be joking...
The only FSB speed I would consider running anything at is either 66 or 100mhz, and to overclock a 466 celeron up to a 100mhz FSB, it would run at 700mhz. Sounds nice, but is, as far as I know, impossible to attain. Even with 2.3v through the cpu.
The easiest to overclock have always been the 300As, (which I myself have running at 4.5x103), but the 366s became popular for running at 550 with a bit of tweakage.
I have a mate with a Abit BP6, and two 366s running at 550 mhz each, which totals to 1.1GHz, if you use a SMP-enabled OS, such as linux or NT.
However, his case temp is up to 40 degrees celcius, which I regard as ridiculous (as mine runs at 20).
My point is, the higher the clock multipler, the harder they get to overlock. Anything above 366 is very hard to reach 100mhz (FSB), and running the FSB at 75 or 83Mhz is just plain stupid, due to the AGP/PCI overclock, which can (and ofter does) result in destroying your cards.
So buying a dual 466mhz system, with hopes to overclock, will leave you disappointed.
---Transmission Ends.---
Sorry about the repeat post!
isn't the overclocking itself, but the chips that you would be overclocking are the Celeron 500s. They have an 8x multiplier locked into them. If Joe Shmoe sets the bus speed to 100mhz he now has an 800mhz system which melts without one hell of a cooling system. Of course he could overclock it to a 75 or 83 mhz bus speed which would give him either a 600 or 664mhz system which would melt without some nice sized fans, but then the PCI bus acts like an abused monkey. Intel doesn't want the bad publicity of their chips melting. They also don't want hardware and software companies breathing down their back because people are complaining about their products not working with these overclocked systems. Joe Overclocker wouldnt complain, he knows what he's doing but Joe Shmoe who thinks it's cool to change numbers in the BIOS or play with jumper settings will complain.
I'm a loner Dottie, a Rebel.
The two things that matter when you overclock is the speed of the system bus and the clock multiplier. In the Celeron the multiplier is locked but you can increase the system bus for example 66->100 MHz so what you actualy do is overbusing. The thing with the Abit BP6 is that is has alot of bus settings, it has 1 MHz steps between 75 and 100 Mhz and that is why it is so good for overclockers.
Quick primer:
Now, for the explanation. For heavy processing of repetitive data (searching a database, for example) a bigger L2 cache is better, so from worst to best is: Celeron, CuMine, P2/P3, Xeon. Yes, that means that for repetitive data, a Coppermine P3 is worse than a Katmai (current) P3. For non-repetitive data (games, office apps, distributed.net...) a faster cache is better, regardless of size. So, for THAT, worst to best is: P2/P3, Celeron, Xeon, CuMine (Xeon and Celeron's positions are debatable. Because the Celeron's cache is on-die, it is marginally faster than the Xeon's. If all data is truly non-repeating, a Celeron will beat a Xeon. Proof below.)
Now for the benchmarks. Distributed.net is a great example of non-repetitive data. Because it is very CPU-intensive, and key searching never sees the same data twice, it is a good example for non-repetitive benchmarking.
All tests run in Windows NT Server 4.0, Service Pack 5, 128MB RAM, U2SCSI drive, with distributed.net as the only user process running, with all services killed, and explorer.exe killed. The Pentium II and Pentium ||| tests were run on an Intel L440GX+ board, The CuMine test was run on a beta L440GX+; the Xeon tests were run on an Intel C440GX+ board (essentially the same board), and the Celeron tests were run on an Intel CA810 with an Adaptec 2930U2. The tests were run by using the same buff-in.rc5 file, and running the "Long RC5 benchmark".
So, what do you think?
As for the CuMine in socket form? Well, it will be PGA370 starting in December.
Anyway, Anandtech's review of the BP6 (yes, I actually read it before I bought my BP6) is the only negative review I've read about the board. They are somewhat vague concerning the instability of the BP6, and don't really provide any details (or perhaps I didn't read it correctly).
From my personal experience, I've had zero ramdumps in NT with the BP6 (except with Winamp which is due to the SMP code in the CLabs SBLive! drivers), while ramdumps were common when doing a lot of multitasking on my Asus P5A w/ K6-2 400 and 128M of RAM (under NT 4.0 w/ SP5). http://www.ocabj.net/pcs/aurora.html. It's got a few tidbits to keep it half interesting.
The dual Celeron 500's in my BP6 work just fine. It's nice to run a cumulative total of 1GHz. :-)
These socket-370 Celerons are multiplier-locked. I never overclock anything though --- I just don't have time for worrying about instability.
"The question of whether machines can think is no more interesting than [] whether submarines can swim" - Dijkstra
If those commercial boxes are up-market, what does that make my latest machine: Abit BP6 m/b, dual Celeron 500, 256M memory, dual 80MHz Ultra2 SCSI controllers, dual 100Tx Ethernet, 6 x 9.1 Gb LVD/Ultra2 drives split across the two SCSI buses, 19" rack chassis with dual redundant power supply.
:-)
I'm in heaven.
"The question of whether machines can think is no more interesting than [] whether submarines can swim" - Dijkstra
No, you don't understand! The Pentium III (r) makes the Internet come alive! It gives you dancing Intel bunnymen and strange viking-type people to enrich your Web experience! If you're not using a Pentium III (r) processor to browse the Web, you're missing out!
http://www.intel. com/home/pentiumiii/surf.htm?iid={showroom=body}-- Ed Avis ed@membled.com
Nah, make -j2 is fastest for me :
make --version
GNU Make version 3.76.1
http://rareformnewmedia.com/
Here's a cool review/comparison I found on Ars-Technica.
There's a lot of benchmark comparison stuff, so if you want to skip to the final analysis....
I just got my BP6 and 2 Celeron 466's in the mail yesterday...maybe I'll call in sick today.
Uh...what?
Yeah, like my Celeron 300A running at 450 MHz that's been running 24/7 running seti@home for almost a year now. I had an uptime of 50 days before I last changed kernel.
%japh = (
'name' => 'Niklas Nordebo', 'mail' => 'niklas@nordebo.com',
'work' => 'www.pipe-dd.com', 'phone' => '+46-708-444705'
The BP6 is Socket 370 only, so it depends on Intel's move in random socketing at that time.
%japh = (
'name' => 'Niklas Nordebo', 'mail' => 'niklas@nordebo.com',
'work' => 'www.pipe-dd.com', 'phone' => '+46-708-444705'
All you need is a good heat sink/fan combo. I have a dual celeron 366 running at 550 MHz 24/7 and I have yet to have any problems. My motherboard (ABIT) has TCs to monitor the temp of the CPUs and they claims my CPUs are under 100 degrees F each. I'm not too worried about smoking these puppies.
If you are worried about smoking your chips, you can always buy your Celerons from a place like computernerd.com who will give you a year guarentee they won't smoke.
BTW. The heatsink/fans I use are awsome. They are the cool-it-dudes! from computernerd.com.
later...
Quack
ps. Did I mention deal celerons rock?
8x, I would suspect, is the maximum multiplier "officially" supported by the motherboard. Since Celerons all are multiplier locked, you can't increase the multiplier - the system will refuse to boot. HOWEVER...
You CAN increase the FSB speed. Heh.
Normally, your Celeron 433 will run at 66.6_x6.5, or 433.3_ MHz. Now, just up that FSB to 100mhz, and you've got a machine running at 650mhz. PROBLEM! You probably also just melted your motherboard!
Don't worry though - just jump down to 75MHz (487MHz) or 83.5 MHz (542.75MHz) front side bus speeds. Much more friendly on your CPU, but a bitch on your PCI cards, and let's not even talk about AGP. ^_^"
InThane
I mean, I know the Xeons have more cache, but with Celerons approaching 600 MHz, why doesn't Intel concentrate on the Celeron for consumers and leave the overpriced PIII to the highend / server market? What kind of advantage is Joe User going to see with a 600 MHz PIII that he won't with a 500 MHz Celeron?
What makes the PIII "better" than the celeron, anyway? All those useful extra instructions?
I have so many questions today!
rooooar
And here I was, all excited to crank my celeron 400 up to 3200 Mhz.... heh
is overclocking. I got a celeron 466 and I'm running it at 525 by raising the front side bus speed up to 75 in the softmenu bios of my abit motherboard. I love celerons and overclocking :)
-Dyl6
Pentium 3's have 512k lb cahce and the celerons only have 128k. So naturally a pentium 3 600 will perform better than a celeron 600. They had to take som of the cache away from the celerons so that it could be in the "low end" market.
-Dyl6
I'm using the BE6 motherboard and I'm using windows (*gAsP*) and so far I've had no stability problems as long as I have the proper cooling.
-Dyl6
Yawwwnn...
Actually the Celeron and P3's come off of the same (or similar) lines, but are different chips... my guess is that if it weren't for AMD the Celeron 300A would have been labeled (and priced as) the Pentium II/450. :)
I have an ABit BX6 version 2 running a Celeron 400 at 450 Mhz, which is a pretty conservative overclock.
It is very stable - at least, I have had no problems running Quake 2 and Unreal time demos for over 8 hours at a time. Those are fairly stressful system tests.
And, the extra 50 Mhz (plus bus speed at 75 instead of 66) gives a small but noticable speed improvement.
Azog
Torrey Hoffman (Azog)
"HTML needs a rant tag" - Alan Cox
wierd, I thought celerons we're only good for coasters, but maybe not... pretty cool.
either we are networking or we areNT networking
Why would you use -j2 unless you have a very small amount of memory... if you have say, 128 megs, try -j 9, if you have 64, try -j5, if you have more than 128 (yes, including swap) try just
plain ole -j and watch it fly =)
(no, this cannot damage your system, dont worry)
Just because you disagree doesn't make it offtopic or flamebait.
You have no clue. I have a 450 with 128 megs of RAM for 300 dollars. All I did was buy Celeron 300 A and some special cas2/2 ram and a cheap motherboard. and I probably payed a thousand dollars less than you did.
Forget their marketing babble. I wouldn't buy their overclocked 466 Celeron boards because they have chosen the wrong CPU. 466 Celerons don't perform as good as 366 when it comes to oc. The only 466++ bonus: these chips run cooler than their slower brothers, 366: about 33W, 466: about 30 W booth at 550 mhz (calculated from INTEL specs. Search for the INTEL 243658-009 datasheet). But what about this 8 fold speed increase they are talking about? Do they mean the 8x multiplier setting of the ABIT BP6? Sorry to say that this doesn't matter, because Celerons have locked multipliers: The 300A is locked at 4,54, the 366 at 5.54 and the 466 at 7.0. So using a 2x multiplier, your 466 will run as fast as using a 8x multiplier. Your Celereon simply ignores your multiplier settings. Your only way to oc the cpu is by increasing the speed of the fsb. INTEL thinks that Celerons should run only using a 66 MHz fsb (perhaps Celerons would be too fast otherwise ? :) Now if you change the speed of the fsb to a healthy 100 MHz, your 466 Celeron would try to run at 706 MHz. To get it to boot (if it posts at all), you'll have to increase its core voltage to about 2.2 - 2.3 volts. But be prepared to have a fire extinguisher at hand after 10 minutes :) So you have to go down with your fsb speed back to about 80 MHz. Now your CPU runs stable without signs of smoke at about 550 MHz. But which CPU will be faster: A 466 Celeron running at 550 MHz with a 80 MHz fsb or a 366 Celeron running at 550 MHz with a 100 MHz fsb? BTW: The main problem when overclocking is heat dissipation. A 366 running at 550 will release about 33 W. Air coolers big enough to remove that amount of heat will be pretty bulky (and don't forget your BX chip with that nice tiny green cooling cap). My BP6 runs perfectly stable at 366@550 (Vcore 2.2 Volts) as long as you let it run cool. At the moment I'm using an additional big room ventilator which blows cool air into the open case. (CPUs about 50 grd C, BX chip 48 grd C). Upgrading to a water cooling system and 2x600 mhz comes into mind. Just yesterday I happend to buy a nice litte silent water pump and some copper:)
--
not sure if anyone else has posted this yet, but arstechnica did some benchmarking, and you can read the article here.
2 celeron 300A's OCed to 504MHz each
vs
2 PIII 500's OCed to 560MHz each
there were a couple of differences in the systems, but the celerons beat the pIII's in almost every test... they're the only numbers i've seen on this issue - interesting at the very least
'Thats they exact same thing a banana wrench monkey.'
It's been my understanding that all celerons since the 300A have had 128K of on-chip L2 cache running the same clock speed as the processor. If the front side bus is running at 66MHz and your HD controller doesn't have a multi-threaded driver, you probably won't see an improvement over your k6-3/400 which has a FSB of 100MHz. If that's the case, not including the "-j 2" and making sure that -pipe is included in CFLAGS might be faster than if you had the "-j 2". I have a dual slot machine, but I need to get a pair of PPGA370 celerons in slockets to test it out.
the good ground has been paved over by suicidal maniacs
I've never played with tubes, but I know that smell you're talking about. There's only one thing in the world that smells like melting semiconductors. Psychologists say that smells are the sense most linked to memories and emotions. The dentist's office has that one smell, Grandma's house has one smell, and melting semiconductors have that smell that always makes you think "Ooooh Shiiiiiiiiit!"
-Barry
Not the old... The Celeron is the exact same CPU as the Pentium argument again. Hello? They're completely different internally... Same technology yes, but it's not like the Celeron is just the PIII's which failed some tests. Sheesh
Is anyone home in there? Did I say Pentium III? No, I don't think so. I said Pentium. The Celeron is the exact same die as the Pentium II. Just like a Pentium II is a different die from the Pentium III, the Celeron is a different die from the Pentium III.
And that's exactly what a Celeron is. A Pentium II that failed some tests.
-Todd
---
"The details of my life are quite inconsequential..."
Yep, the Celeron 300A and above all have 128 K of L2 cache running at full core speed.
...But I'm still gonna make one! ;)
However, for SMP purposes, the cache size matters a lot. Since the two CPU's are sharing the memory bus, every cache miss means potential contention. Running large memory intensive programs on both CPU's simultaneously would almost certainly show significantly lower performance than same-clocked PII's with their 512K caches and much higher cache hit rate.
That is, in single CPU comparisons, the doubled speed of the Celeron almost makes up for the reduced cache hit rate relative to the PII. However, in dual-CPU configurations, the decreased cache hit rate of the Celerons carries the additional penalty of increased memory bus contention.
Geeky modern art T-shirts
Anybody who tries to compare any PC to a Starfire is obviously going to look silly -- if you're running a workload that the Starfire can run effectively (nicely parallel). But then again, throw together a Beowulf cluster with enough cheap PC's and run something that's reasonably close to embarrassingly parallel, and you're right up there in the list.
Somehow, it seems awfully cynical to call anything that's in the 3-digit mips and Mflops range a "puny toy". Any of these things has the raw CPU to blow a Cray-1 out of the water (of course, it doesn't have the memory bandwidth, but that's another matter). It's also a very large fraction of what we could do on a Connection Machine (CM-2) 10 years ago. Sure, I'm well aware that I can spend less on a processor than on the latest Star Wars toy for my nephew, and likewise that processor technology's advancing at an incredible rate, but these chips would have been called supercomputers not too many years ago.
And then I shudder to think how these things actually slow down as much as they do running Windows, and I can't imagine what must be going on that mere UI display (without any speech recognition) can bog one of these things down. Oh well, I guess I can just be smug about the fact that I can't even *run* Windows any more, since my last mobo upgrade...