Trying a 500 and a 366 together...

[davd_bob]

I was fortunate to have just aquired my 2nd BP6 board.(Thanks Hugoc)
I replaced the caps that were damaged by vibration, or something. I happened to have exactly the right extra caps from a previous BP6 cap job.

I happen to have a pair of spare 500s just lying around. Woo Hoo.

Anyway, I am going to combine a 500 and a 366 on one board and see, um, geez. I just realized I don't know exactly why Im gonna do this, except to see if it works.

CPU1=500 CPU2=366
RAM=2@128MEG PC-133
BIOS=RU with HTP366 not enabled
HD=40GIG,7200rpm,2MEG cache

Does anyone have any tests they want run in this configuration?
purrkur? hyperspace? edited 11/05/2005 anyone else?
Please make any suggestions as someone out there has probably thought of something I didn't.
InactiveX? Dave Rave? edited 11/05/2005 anyone else?

P.S. If I try to OC i will swap the 500 to CPU2 hoping it will OC better if it is carying less of the load.


edited 11/05/2005
ok, my bad. I meant to include everyone.
Im so confused sometimes i forget things.

[Derek]

I guess I'm not allowed to answer this post

[Billl]

I guess I'm not allowed to answer this post

Don't feel bad Derek, I wasn't invited either.


Billl

[Dave Rave]

Does anyone have any tests they want run in this configuration?
purrkur? hyperspace?
Please make any suggestions as someone out there has probably thought of something I didn't.
InactiveX? Dave Rave?

ooooh, I feel special.

I got nothing, though, Derek ? perhaps .....

[purrkur]

Move over guys

I think I am most curious about how the BP6 behaves in a setup like that, how it configures the CPU's and what happens during overclock. I am guessing that it will configure both to run at 366 if not overclocked...

[KliK]

Move over guys

I think I am most curious about how the BP6 behaves in a setup like that, how it configures the CPU's and what happens during overclock. I am guessing that it will configure both to run at 366 if not overclocked...

buuuu, not invited...but do know the answer!!!

one goes on 500, the other goes on 366...although the sandra will report that the L2 cache is either on 366 or on 500MHz...and will tell the cache is overclocked! (that is when you put multiplier 7.5 on FSB clock)

i did run 533 with 433 together...stable system, but i didn't try the games though...maybe you should check the stability in games, 'cause they really demand the 100% of both CPUs...
any other tests (i've been running the UD agent) like calculations would bring yours cels to 80% of the power...and the one that is computing less is the 500 cel!

[davd_bob]

ok, my bad. I meant to include everyone.
Im so confused sometimes i forget things. I do try---kind of like anikin skywalker when he looses his light saber.

All welcome to respond.
All welcome.
All welcome in the light...

[KliK]

ok, my bad. I meant to include everyone.
Im so confused sometimes i forget things. I do try---kind of like anikin skywalker when he looses his light saber.

All welcome to respond.
All welcome.
All welcome in the light...

a good thing would be if you could install the SiSoft Sandra Lite (any version), and on it run some benchmarks...i would relly wanna know what is the difference with the 2x366@550, 366&500 (stndard and overclocked) and 2x500 (standard & ovrclocked)!
that reminds me to do the same...just to see what is mine FPU & other stuff...

Derek, do you have some other program for comparison the procesing capabilities?!

[davd_bob]

Results have already shown dual 366@550 kicks the butt of dual 500s at anyspeed they can OC to due the the faster interaction between the ram and cpus on the FSB100.

I will test Sis Sandra 2003 Standard results on the mismatched pair though. Im sure the results will be lower then dual 500s.
What can I do to give a real test though to compare when the 500 in socket cpu1 vs the 366 in socket cpu1.
I suspect the 366 in cpu1 might allow a higher OC...but it will loose clicks as the slower cpu carring the heaver load like that.

Lets get some guesses on the outcome.
Derek? Tim? Billl? Kilk?
ANYONE ELSE?

[purrkur]

Klick, check this out....

[KliK]

Klick, check this out....

thanks...so that is it! so my cels are slower only in drive speed and large memory tranfers...ok, i can live with that, before i can find some more cels 366!!

'cause, unfortunatelly mine cels 366 are both different ones and they don't boot WinXP!!! so i'm stuck with oveclocked 533s and here is my test of sandra for dual cels 533@544!!! note that those are SL3FZ, but soon i plan to put my SL3PZ onto my board...i'll send you also some photos and benchmarks then!!!

[davd_bob]

purrkur,
thanks for the link.
And for anyone that doesnt know it purrkur did the excelent benchmark tests in the link. purrkur is also the Linux FAQ administrator for BP6.com, and probably forgot more about computers then i ever even learned. Lots of you other guys show that you are on his level of expertise as well.

kilk,
come again on your statement about mismatched 366s not running XP?

BTW, and update on the 'new board. I still have it in "uni" configuration running a 500 on 66FSB. It is preforming correctly on Win98.
I changed 6 caps and noticed some kind of goo/dust stuff a couple places on the board. I used scalding hot water to rinse it off then dried the board. It had trouble booting the first time then gave me a memory error. It was rebooted with no errors and formated a HDD ok then powered off. Couple hours later booted perfectly several times with various HDDs/OS. Not a hicup or problem with the board since.

Also I have a Maxtore 40GIG that has a bad spot starting at about 5% through at least 9% and I can't get it to work. Other then partitioning around that spot can anyone suggest a solution?

[KliK]

purrkur,
thanks for the link.
And for anyone that doesnt know it purrkur did the excelent benchmark tests in the link. purrkur is also the Linux FAQ administrator for BP6.com, and probably forgot more about computers then i ever even learned. Lots of you other guys show that you are on his level of expertise as well.

kilk,
come again on your statement about mismatched 366s not running XP?

BTW, and update on the 'new board. I still have it in "uni" configuration running a 500 on 66FSB. It is preforming correctly on Win98.
I changed 6 caps and noticed some kind of goo/dust stuff a couple places on the board. I used scalding hot water to rinse it off then dried the board. It had trouble booting the first time then gave me a memory error. It was rebooted with no errors and formated a HDD ok then powered off. Couple hours later booted perfectly several times with various HDDs/OS. Not a hicup or problem with the board since.

Also I have a Maxtore 40GIG that has a bad spot starting at about 5% through at least 9% and I can't get it to work. Other then partitioning around that spot can anyone suggest a solution?

i have SL35S & SL36C, so that are the two mismatched cels 366...everybody here has said that they are identical, but if i put them with no overclocked in mbo, and tweek only a voltage to 2,1 or 1,9 the system wouldn't boot!!!
so now i'm searching to buy another 366...maybe i will got them one day, so i can make it running!!!

also, about that problem you have...make two partitions, first to 12GB, and the other on the rest...then select second partition for your Win system, and it will only format your HDD from the 12GB up to end of it!!! that should work, so try it!!
also, the first partition is unusable, unless you have some tool like Prtition Magic, and format it without those bad sectors!!!

[kuun]

im curious as to the actual difference you get when you run a bench mark with the mismatched pairs when both are in..

the 366 in socket 1 and then again with the 500 in socket 1...

the difference in theory shouldn't be different.. but we all know better..

davd, what'd your tests reveal?

[jaybird]

Don't know if this helps but according to ABIT, if you run different speed Celerons' "the BP6 will only run at the speed of the slowest Celeron mounted on the board".

regards,

jaybird

[KliK]

Don't know if this helps but according to ABIT, if you run different speed Celerons' "the BP6 will only run at the speed of the slowest Celeron mounted on the board".

regards,

jaybird

well that is what was supposed to be...but it is not, after the last BIOS patches (before they would not work if they were different, or would not be stable)...Cels just have their hardware multiplier, and they will run at their speed internal speed which is set by FSB*mult...
and all the components of the processor will run on that speed, except maybe L2 cache...

[kuun]

you can still OC them to the higher CPU's speed

[jaybird]

Ok, I just finished digging back through my "archives"(hand scribbled notes from over 5 years ago)

I ran a 366 and a 466 together on my bp6 and then ran dual 366's (not over-clocked) through both SISandra and several units of SETI. The results showed no difference!

However, when I tried to over-clock, the dual 366's went to 550 with no problems but the 366-466 combo would only over-clock as high as the 466 would allow, about 75 mhz and was VERY unstable!

Since the multipliers are locked and you cannot overclock each CPU differently, the above seems reasonable and, based on SISandra and SETI, it also seems reasonable that through the board and/or the OS (in my case its' been NT4.0, SP6 and WIN2K Pro, SP4) "sees" or processes the duals, as the same type of CPU (366 or 466, not 366 and a 466).

Now I was not able to compare dual 366's Vs dual 466's but I think its' pretty obvious whats' happening here unless I am miss reading the results.

OBTW, as a side note, although not a degreed engineer I was an advanced software/hardware applications engineer for companies like Johnson Controls and Honeywell before I "retired" so, in my mind, I think I have a fair understanding as to what I think was happening .

Comments please!

Regards,

jaybird

[davd_bob]

Move over guys

I think I am most curious about how the BP6 behaves in a setup like that, how it configures the CPU's and what happens during overclock. I am guessing that it will configure both to run at 366 if not overclocked...

buuuu, not invited...but do know the answer!!!

one goes on 500, the other goes on 366...although the sandra will report that the L2 cache is either on 366 or on 500MHz...and will tell the cache is overclocked! (that is when you put multiplier 7.5 on FSB clock)

i did run 533 with 433 together...stable system, but i didn't try the games though...maybe you should check the stability in games, 'cause they really demand the 100% of both CPUs...
any other tests (i've been running the UD agent) like calculations would bring yours cels to 80% of the power...and the one that is computing less is the 500 cel!

I ran a 366 and a 466 together on my bp6 and then ran dual 366's (not over-clocked) through both SISandra and several units of SETI. The results showed no difference...

...although not a degreed engineer I was an advanced software/hardware applications engineer for companies like Johnson Controls and Honeywell before I "retired" so, in my mind, I think I have a fair understanding as to what I think was happening .

Comments please!

Regards,

jaybird

Never enough time...
I kinda wish I wouldn't have started this thread cause I need to figure out some how to actually post some kind of results.

Jaybird, I got an AS in computer service techonology, and quit one elective exam short of MCSE(only NT4) but that paper stuff doesn't count for much. Your credentials get a high rating for coolness that doesn't come across as geekyness. Mine sound geeky but I'm not very cool.

Anyway, Klik, you said "the one that is computing less is the 500" and Im guessing you mean that the work is being supplied equally to the processors so the 500 at a higher multiplier is doing less work then it is capable of.
Jaybird, you said the dual 366s at FSB66 and the mismatched 366/466 produced the same WUs on SETI? That seems like something in the system is dividing the work equally and sending it to the processors...and the processors are all viewed as 366s. I hadn't thought about it but that sounds like ONE pretty fair bench result. Sis-Sandra will show information it reads from the board but I don't think Sandra has the capability to properly handle bench tests of miss-matched processors.

By the sound of it from both Jaybird and Klik, the work is supplied by the software equally to the processors(maybe from the OS) as though both were the slower speed.

A question for anyone with specialised experience:
What part of the system would divide the work being sent to the processors. Is that a function of the OS or something else? Can it be over ridden insome simple way.

Guess I will play Dungen Seige on each configuration and see how it "FEELS" during play. Im sure the end result will be missmatched will get the lowest score, dual 366@550 the highest and 500s somewhere in-between.


BTW, everybody remember to click on the Google adds on BP6.COM.

[jaybird]

If I read the last post correctly, your looking for a way to direct the work to certain CPU's?

the answer is "yes" there is a way it is called the MS Affinity Filter.

I stumbled across this when struggling with my old CBLive 5.1 Surround Sound system and the fact that Creative did not support WIN2K Pro with surround drivers. To make a long story short, my Platinum drivers trashed WIN2K!

I also found a software package call the "XProject" which corrected the driver problem (replaced Creative drivers with there own) and corrected the "lock-up" problem buy using the Affinity Filter to direct all multi-media tasks through one CPU!

In short (and as you know, I NEVER leave shot answers ) ) the affinty filter lists all currently running tasks and allows you to decide which CPU you want to run which tasks!

Neat little tool for free!

Regards,

jaybird

[davd_bob]

Thanks Jay.
I will have to scope that out(MS affinity filters) because it sounds like the solution for an issue here at work too.

BTW, when you were running your 'mixed dually' do you think you could have gotten the 466 to do more work by using this tool?

[KliK]

Thanks Jay.
I will have to scope that out(MS affinity filters) because it sounds like the solution for an issue here at work too.

BTW, when you were running your 'mixed dually' do you think you could have gotten the 466 to do more work by using this tool?

i would agree with you...when i had 466&533 on my board, the UD agent got the same result...but one time it say it was 466, and another there was 533 on the board...
also, i did notice that the one processor was not stress, sthg like about 50%, but another did the job on 80%...so that must be it, the difference in the multiplier...

[jaybird]

No, sad to say, I didn't even know the affinity filter existed at the time.

Once I finish my "new" BP6, I will be using it to try to resolve some video editing and capture problems that I am having by using the affinity filter.

"NEW" BP6

dual 1100/100/256/1.75 @ 110 FSB (I hope )
768 meg PC133 RAM
PNY FX5500 128 meg dual hd vid card
Chaintek VR710 7.1 surround sound card
SIIG multi-purpose card (USB, Firewire,ATA100/RAID)
10/100 NIC
Adaptec SCSII for X6EL Microtek scanner
analog capture card (can't remember the name at the moment, I'm at work)
100 gig hdd for dual os's & other software(98SE/WIN2K Pro)
200 gig for video work
450W P4, Athalon PS
Peltier cooling on Northbridge
Pioneer slot type DVD player
Sony DVD/DVD-RAM 'burner
Liteon 52x 24x 52x 'burner
ZEN Multi-Beam True X, 52X CD-ROM player
SuperMicro 750 server case
250W Cambridge Desk Top Surround Home Theater system
dual Viewsonic 20" high end grapics monitors

This thing should "blow the doors" off my Gigabyte duallie Apollo Pro MOBO!

I could go on as to the rest of the "stuff" connected to it but you get the idea (you can never have enough "toys")

I'll post pic's when completed.

Regards,

jaybird

[davd_bob]

those 1100/100/256/1.75 C'mines are the best option for a mod'ed BP6 upgrade in my opinion.

Good luck going over 100fsb. On 3 different boards so far I have never got 110fsb. I can usually get Turbo-100 or maybe 104 but nothing stable beyone that. Active cooling on the BX doesn't help.

[jaybird]

Yeah, I know, but I did make it on my first BP6 to PIII conversion, maybe I'll get lucky again. If not, I'll be satisfied with dual PIII's stable at 100 fsb.

Regards,

jaybird