Madshrimps Forum Madness

Madshrimps Forum Madness (https://www.madshrimps.be/vbulletin/)
-   Hardware Overclocking and Case Modding (https://www.madshrimps.be/vbulletin/f10/)
-   -   Optimal For Overclocking: P4/2500 With 100 MHz FSB (https://www.madshrimps.be/vbulletin/f10/optimal-overclocking-p4-2500-100-mhz-fsb-673/)

Syco 27th August 2002 22:12

Optimal For Overclocking: P4/2500 With 100 MHz FSB
 
quote:

"An important rule for overclocking Intel CPUs is the following: you need a lower FSB clock, plus a high multiplier value, because it's not possible to adjust this factor as you can with all AMD CPUs. The P4/2500 variant, for example, is factory-set to run with 25 x 100 MHz (2500 MHz), and this alone makes it optimal for extreme overclocking.

By contrast, the P4/2666 and P4/2800 don't look too good when it comes to overclocking. Increasing the FSB clock stops at 150 MHz at the latest (based on a motherboard equipped with the Intel 850 chipset and RDRAM). For both CPUs, this means that only 17 MHz is left available for a clock increase. The clock limit of the P4/2800 thus lies at 3150 MHz (150 MHz x 21), or an increase of 12.5%. It's different with the P4/2500 with 100 MHz FSB: by increasing the FSB clock from 100 MHz to 133 MHz, the CPU runs at 3325 MHz (133 MHz x 25) - that's a 33% increase. "

So, will all P4/2500 be able to run @ 3325 Mhz air cooled? Using 16-bit PC-1066 RIMM's, should a P4/2500 be an better option than the P4/2533, of course running the P4/2500 at 133 FSB. (3325 MHz). Using 16-bit PC-1066, u won't get the P4/2533 much higher than 2800 Mhz I guess.

Any advice or comments are welcome! :)

jmke 27th August 2002 22:23

Quote:

When the Intel Pentium 4 2,8GHz CPU arrived to our testlab we ordered 10 liters of Liquid Nitrogen (LN2 -196°C) and decided to run some tests in very low temperatures.

After some adjusting and testing we were able to run SiSoft Sandra CPU and Memory benchmarks and Pifast benchmark smoothly when the CPU was running at 3917MHz. We raised the FSB one more step and managed to run succesfully SuperPi benchmark while CPU was running at 3998MHz. The result was 39 seconds.

Test setup: P4 2,8GHz, Modified Asus P4T533-C, Samsung PC800 RDRAM, PNY GeForce 4 MX440 and Windows XP OS.




ps: I have reached 180+ FSB with my "older" RDRAM setup (Abit TH7II + PC800 + P4 1.6A)

The Senile Doctor 28th August 2002 09:42

I would advise against it...
why?

take this...

you have a 2.4. It could max out at 3 ghz, it could max out at 3.3.
If it maxes out at 3 ghz... you won't get over 125 mhz fsb.... even quad pumped, that's not special, you'll have extreme cpu power bottlenecked with a slower bus...

if you get to 133 mhz fsb, you'll get 3200 mhz...
quad pumped : that's simple plain pc1066.
cool, however, your cpu would happily use more bandwith, but the cpu is maxed out...

if you have a 150 mhz block on the motherboard : that's a problem with the 2.66, however, people tend to forget the extreme power the abit th7-II (the oldie) still has... that goes all the way up...the abit th7-II I have does 191-193*3...
so there you have it... that machinery will get your cpu to its max, and it will have due to higher fsb, about the same rampower as the quadpumped 133...

you think you'll get a 2.4 100 mhz to 156 mhz fsb?
have phun, but it won't work.

get a gigabyte if you want to overclock softly to 3 ghz
if you wanna hit the rooftops on your cpu, get an abit th7-II oldie, and run it @ 3*. or an asus p4t533 if you can stand possible failure


Da`Hitch 28th August 2002 14:04

sorry Calantak, but you lost me here.
why exactly isn't it a good idea to team up a 2.53 w/ a Gigabyte motherboard?

The Senile Doctor 28th August 2002 14:23

the problem is that the gigabyte won't go over 156 mhz fsb (it stops there, it's blocked), so if you've got a 133 mhz cpu, you can get a max increase of fsb of 23 mhz... meaning if you have a 2.53, that is 23*19... that is 430 mhz.
you'll never get past 3 ghz that way.

however the board is a true beauty, and if you're happy with near 3 ghz speed, you should get this one.

if you want more than 3 ghz you'll have to get another board or a cpu with lower starting fsb (100 mhz type) however, you're not sure you'll get to 3 ghz that time either.
and what's more... to get to the same memory bandwith (156) you'll have to get the cpu over 3.6 ghz if it's a 2.4.

so... get the gigabyte and the 2.53


damn, it's not easy explaining this ;)

Syco 28th August 2002 17:54

and now we are waiting for 16-bit RIMM's that can handle 156 Mhz quad pumped :grr:

Any idea when these baby's will be available? (and at what price)

Da`Hitch 29th August 2002 11:16

after carefully examining the first post, I just started catching on to what your saying here, I think, I could also just be plain wrong ;)
correct me when & where necessary. :)

so, the 2.53GHz has a FSB of 133 MHz, in combination with a gigabyte, that only leaves room for a 23MHz increase, because the mobo is limited to 156MHz.
so let me do some calculations.
so, 2.53GHz = 2590,72MHz (let's take it at 2590MHz, assuming this value is correct)
this CPU runs at a FSB of 133MHz, so the multiplier should be 2590/133 = 19.47 (19)
so, the max FSB of the gigabyte is 156MHz, so the theoretical max OC of the 2.53 CPU is (19*156) 2964 (just a little over 3GHz, if calculated with the actual values)
that's like a 14% increase.
if the 2.50GHz CPU is used (wich runs at 100MHz with a multiplier of 25) incombination with the limited gigabyte, then the theoretical max is 3.9GHz wich is a pretty impressive 56% increase (but can probably only be achieved with a phase changed cooling, if even possible at the moment)
but still, even if the CPU runs at 133MHz, it can still run @ 3.32GHz wich is still quite impressive. (but also here probably not attainable with a HSF-cooling)

the only thing I did not take into account is the memory, what kind of memory would one need for this, off course RIMM, but 16bit PC800 RIMM, or is use of 32bit PC1066 RIMM required? (haven't quite figured out exactly what the main difference is, except for the fact that the latter seems to be faster.)

so the question now becomes when will this 2.5 be available, and what will it cost :D
because if my calculations are correct (but they also could(edit) be so very wrong) this CPU is a lot more impressive then the 2.53GHz.

could someone plz shed some light on this (and while your at it perhaps check if my theory is correct, cuz I'm PIV-n00b, but learning ;))

The Senile Doctor 29th August 2002 13:05

your theory is acceptable :)
32 bit only works with the asus p4t533 : no go
16 bit = okay with the other pc1066 mobo's

16 bit pc800 can be put at 3*fsb for higher fsb.

what you're not getting is that a 3.32 ghz cpu @133 fsb, quad pumped (pc1066), will not be as fast as a 2.950 mhz cpu running 156 mhz fsb (quad pumped with good pc1066)...

Da`Hitch 29th August 2002 13:11

OK, since my last post, I've been reading up on an old review of the P4Northwood, read quite some interesting stuff. ;)

but I have also been looking for some prices of the 2.5GHz proc.
Comtech doesn't have it.
wallabie is waaayy to expensive (I'm talking 900€'s here :wtf: )
I haven't been able to check other belgian shops, because this old crappy browser at work (netscape 4) can't acces those pages. *sigh*
but I did manage to get onto the site of an american shop, namely New Egg and I checked the prices there. (I know prices are lower there, but still, they can be used as a reference)

http://www.newegg.com/app/ViewProduc...tby=14&order=1
Quote:

Intel Pentium 4 / 2.5GHz Northwood 512K Socket 478
Processor 400MHz Retail box
Specifications:
CPU: 2.5 GHz
Type: Pentium 4 Northwood
Cache: 512K
BUS: 400 MHz
Socket: 478 Retail (Box with Heatsink and fan)
price: $276.00 (which should be about the same in €'s)

Quote:

Intel Pentium 4 / 2.53GHz Northwood 512K Socket 478
Processor 533MHz Processor Bus
Specifications:
CPU: 2.53 GHz
Type: Pentium 4 Northwood
Cache: 512K
BUS: 533 Mhz
Socket: 478 Retail (Box with Heatsink and fan), 3-year MFG.
which is priced at $285.00

so we can conclude that the 2.5GHz should be cheaper than the 2.53GHz. (whenever we will be able to buy it here in good 'ol Belgium, unless it's allready available in a shop that i haven't been able to check yet)
anyway, that's good news :)


now, back to the RAM, still a bit confused here.
I know that 32-bit RAM has a greater bandwith than 16-bit RAM, but what exactly does this mean when it comes to performance.
ok, then there is the PC1066 RIMM & PC800 RIMM.
so the PC1066 RIMM is supposed to be used in a 533MHZ FSB-system. in that instance the RAM is "quad-pumped", meaning that it is running at 4*133MHz. correct?
then there is the PC800 RIMM, which is supposed to run in a 400MHz FSB-system (4*100MHz), but it can also be used in a 533MHz FSB-system, and then it runs at 3*133 (?? not sure) which is then 1MHz slower :p

but now here's the catch, what the hell does 3*133 & 4*133 mean??
does it mean that you have to use 4 sticks of RAM, or is it merely a BIOS-setting?

I know it, this is probably the wrong thread for it, and I'm asking a lot of *stupid* questions, but I would really like to know more about the PIV, and how it works, before I actually purchase a PIV system.

any links to webpages that could help me would be most welcome :^)

The Senile Doctor 29th August 2002 14:41

Quote:

Originally posted by D@ Hitch

now, back to the RAM, still a bit confused here.
I know that 32-bit RAM has a greater bandwith than 16-bit RAM, but what exactly does this mean when it comes to performance.
ok, then there is the PC1066 RIMM & PC800 RIMM.
so the PC1066 RIMM is supposed to be used in a 533MHZ FSB-system. in that instance the RAM is "quad-pumped", meaning that it is running at 4*133MHz. correct?
then there is the PC800 RIMM, which is supposed to run in a 400MHz FSB-system (4*100MHz), but it can also be used in a 533MHz FSB-system, and then it runs at 3*133 (?? not sure) which is then 1MHz slower :p
but now here's the catch, what the hell does 3*133 & 4*133 mean??
does it mean that you have to use 4 sticks of RAM, or is it merely a BIOS-setting?

MAN MAN MAN
you need to read a little more :)
four stix of ram? where do you keep getting those funny things :P

(it's a bios setting)
32bit ram is actually not faster (read the thg review concering i850e...)
there is pc800 16 bit : designed to work at 100 fsb, quad pumped (*4)
there is pc1066 16 bit and 32 bit (only for asus p4t533) : designed to work at 133 quad pumped (*4)

some pc800 will be able to work overclocked @ 133*4... however not all of the ram will be able to do this...and when the ram cannot run at a certain fsb, some mobo's have the option to run it at 3* instead of 4* (quadpumped), so then you have a backdoor when your cpu can go higher and your ram can't keep running at 4*. (example : your ram does 138*4, but your cpu is a 2.8-133 and it can go higher coz you're watercooling it... so in order to be able to go higher, you need to unstress your ram (coz the ram is at it's max), so you put the ram @ 3*, then your ram is working slower (4*138 compared to 3*138, and you can go higher, for example, your cpu goes to 152 fsb, that would mean getting to 3*152), you will lose bandwith (4*138 = 552mhz vs. 3*152= 456mhz) but you'll gain cpu power.
Most pc800 will do 3*160.

pc1066 will sometimes go to 4*170 but it could also stop at 4*150. luck of the draw...


All times are GMT +1. The time now is 11:41.

Powered by vBulletin® - Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
Content Relevant URLs by vBSEO