845 vs 865 vs 875

/dev/null

[H]F Junkie
Joined
Mar 31, 2001
Messages
15,182
How bad, speedwise, is an 845 vs an 865 or 875? I know the 845 doesn't support 800fsb, but outside of that, is it resonable? I'd like to drop a celeronD that I have here into a SFF box for lanparty gaming. I can currently smack the celeronD up to 3ghz w/o problems in it's current case. I imagine this is at least as fast as a p4-2.2 or 2.4 without HT.

Rob
 
The biggest problem is the lack of dual channel memory. That hurts. But with a Celeron, you probably won't lose as much performance compared to using a northwood C CPU on an 848 chipset (oh, btw, the 848 chipset supports 200MHz FSB).
 
Is the CeleronD/P4 Prescott w/533fsb compatible with 845?
 
That's a good question. If the mobo manufacturers have a good power delivery circut and BIOS support, it should work.
 
The single channel memory vs. dual channel memory isn't as big of a hit as some people make it out to be. I had to limp around on an 845 motherboard while getting my 875 chipset motherboard voltmodded and I was still able to play all my games smoothly. I only noticed the performance hit when I was playing quake 3, but that is because I have it push 125fps or 333fps depending upon the map and needs.
 
I tried an 800 FSB 2.8 northwood on a motherboard that didnt support 800 FSB and it it didnt work. It only ran at half speed even with the FSB set to 200. I had to buy anothermotherboard to use the processor.
 
From what I saw when I looked around Abit (I was feeling nostalgic about my old IT7-Max2 Ver. 2) I couldn't find any new Bios with Cel D/ Prescott 533mhz support. Maybe other manufacturers do though. Maybe at that point I'd look into the PT880 or something though :eek: ...Via.
 
I'd opt for an 865 board. By the way at 533FSB the chipset only allows you to run up to DDR333 (well at least on all the Asus 865/875 boards I've seen). Also, I think the dual-channel helps the Celeron D's more due to the smaller L2 cache and greater reliance on the FSB. Well in my experience, I got a noticable performance boast going dual-channel on a Celeron D 320 (2.4Ghz) (mostly just gaming).

BTW what model Celeron D is it?
 
while we're at it, what exactly is the difference between i865 and i875, I used to think it was PAT, but my i865 had PAT settings =-O do tell!
 
XeroHouR said:
while we're at it, what exactly is the difference between i865 and i875, I used to think it was PAT, but my i865 had PAT settings =-O do tell!

Pretty much it's the PAT. The 875 and 865 are the same silicon just Intel disabled PAT on the 865. But some motherboard manufacturers re-enabled the option. However I hear the 875's overclock better with PAT on than the 865's do (with PAT on). Wouldn't know though as I don't have an 875 board to compare.
 
Yiffy said:
Pretty much it's the PAT. The 875 and 865 are the same silicon just Intel disabled PAT on the 865. But some motherboard manufacturers re-enabled the option. However I hear the 875's overclock better with PAT on than the 865's do (with PAT on). Wouldn't know though as I don't have an 875 board to compare.

Your correct. The i875P chipset does generally OC better. As for the rest of it. The i845 chipset based boards don't have dual channel ram. That hurts more than you might think. They can't use DDR400 ram, that hurts, although not with a Celery. Finally the Celeron-D is a Prescott core, and I know for a fact that the Prescott power design is VERY different than that of the Northwood. Any non-800MHz FSB capable board will NOT work with a Prescott based CPU. End of story. Not to say you couldn't mod it to by changing the voltage components, but officially the answer is no.

The Prescott power requirements were sent to the motherboard manufacturers after the 800MHz Northwoods came out. There are only a couple of early i875P and i865PE board makers that didn't have to revise the boards at some point for Prescott support.

Intel boards made before a certain date with the above chipsets wouldn't do it. The latter revision had different voltage components and a newer BIOS.

Cliff notes version:

Single channel ram=sucks
no 800MHz FSB support=No Prescott support which means that 533MHz P4's and Celeron D's will NOT work.
BIOS update needed also.
 
Sir-Fragalot said:
Your correct. The i875P chipset does generally OC better. As for the rest of it. The i845 chipset based boards don't have dual channel ram. That hurts more than you might think. They can't use DDR400 ram, that hurts, although not with a Celery. Finally the Celeron-D is a Prescott core, and I know for a fact that the Prescott power design is VERY different than that of the Northwood. Any non-800MHz FSB capable board will NOT work with a Prescott based CPU. End of story. Not to say you couldn't mod it to by changing the voltage components, but officially the answer is no.

The Prescott power requirements were sent to the motherboard manufacturers after the 800MHz Northwoods came out. There are only a couple of early i875P and i865PE board makers that didn't have to revise the boards at some point for Prescott support.

Intel boards made before a certain date with the above chipsets wouldn't do it. The latter revision had different voltage components and a newer BIOS.

Cliff notes version:

Single channel ram=sucks
no 800MHz FSB support=No Prescott support which means that 533MHz P4's and Celeron D's will NOT work.
BIOS update needed also.

:confused: Celeron D most definately works on i845... Prescott 533's will work with a bios flash if they don't already
 
pigpen said:
excuse my ignorance. what is PAT?

Performance Acceleration Technology. Basically it increased memory throughput by 3-5%. The real differences between the i865PE and the i875P are ECC memory support, and PAT. Some mobo manufacturers like Abit and Asus found a work around for the i865PE to use PAT, however it sometimes wasn't as stable and the i865PE chipsets weren't near the overclockers that the i875P's were.
 
ScHpAnKy said:
:confused: Celeron D most definately works on i845... Prescott 533's will work with a bios flash if they don't already

On the 848 chipset I would agree. Which is single channel and supports both the 533MHz and 800MHz FSB's.

Have you ever tried to install a Prescott based processor on a non-800MHz FSB board? What motherboards would do it? The power situation with Prescott was very different than with the Northwood. Many 800MHz FSB boards could NOT run the Prescott. Even some early i875P and i865PE boards couldn't. Of Intel's OWN design no less. Just because the board and processor have a 533MHz bus and the same socket type, doesn't mean it would actually work.

I've never seen it work. From what I know of the power design of Intel chipset based boards which were changed somewhat when HT Pentium 4's with the 800MHz FSB were introduced, I would find if VERY unlikely that a Prescott CPU could work at all on a i845 chipset based board. Unless for some reason someone made an i845 board AFTER the introduction of the i848, i865PE, and i875P chipsets.

There might be one or two freak boards that could do it, but the bulk of them will not. They haven't even updated the BIOS's for Prescott support BECAUSE the boards can't do it. Generally speaking if the board doesn't at least support 800MHz FSB Northwoods, it damn sure won't support Prescott.
 
I used to run a 2.80A Prescott on a board that had a 865 Chipset, thats right, 865, not P, note PE, just 865, it only supports 533mhz FSB and Dual-Channel ram, worked fine with a bios flash.
 
Peach said:
I used to run a 2.80A Prescott on a board that had a 865 Chipset, thats right, 865, not P, note PE, just 865, it only supports 533mhz FSB and Dual-Channel ram, worked fine with a bios flash.

Not quite true. There is no "plain i865" chipset.

http://www.intel.com/design/chipsets/linecard.htm

This is Intel's chipset comparison chart. There is no plain i865 chipset. All i865's that didn't have video integrated are PE's or the P. You are right that a dual channel i865P could run a 533MHz Prescott or Celeron D. That chipset was a value chipset and was NEW enough to be used on motherboards that would support Prescott core CPU's. That does not mean that every board with that chipset would though.

Support for the processor isn't so much about the chipset as it is the voltage components and BIOS of the processor. Now there are certain limitations. On the above chipset 800MHz FSB processors would NEVER work.

I am saying that the i845 chipset based board would NOT work with the Prescott CPU's. Not because of the chipset, but because of the old ass boards it was on.

Most of the boards I've seen won't do it for sure. Check with your board manufacturer for a BIOS update and see what they say. You'll find that most of the boards that don't have at least an i865 or i875P based chipset will NOT support a Celeron D or Prescott "A" series CPU.

If a new enough board with the proper voltage and power design, were equipped with a i845 chipset and a Celeron D or Pentium 4 Northwood 533MHz or Prescott "A" series processor it could work.

Most of the boards I've seen won't do it for sure. Check with your board manufacturer for a BIOS update and see what they say. You'll find that most of the boards that don't have at least an i865 or i875P based chipset will NOT support a Celeron D or Prescott "A" series CPU.
 
It's not like an 865 based board is that expensive anymore anyway if you needed to replace that 845 board :).
 
Robstar, btw I did a bit of testing of Single-Channel vs Dual-Channel:

CPU: Celeron D 320 2.4Ghz
Board: Asus P4P800-SE (i865PE)
Memory: PQI DDR400. 2x 512MB @ DDR333 2-2-2-5
Video: Radeon 9700Pro

Serious Sam: Second Encounter: Grand Cathedral Demo
1024x768 0AF 0AA, Audio ON

Single Channel:
Avg: 64FPS
Low: 40FPS

Dual Channel:
Avg: 75FPS
Low: 47FPS

As you can see, the low peak gained about 17% ! I can't say all your games will benefit this much but it does show what dual-channel can do for a Celeron D. Worst case, maybe a 5% gain but when you're talkin' low peak framerates every bit definately counts.
 
For the Celeron I'd get the 865 board, but if I was going to upgrade to a 800 MHz P4 later down the road I go with the 875 as they seem to have a higher stable overclock. With a 2.4C I've tried 2 Abit IS7's and I could go to 3 GHz with the IC7-G I got to 3.2 GHz. I guess it could be luck but I know others with simular experiences.
 
Sir-Fragalot said:
I am saying that the i845 chipset based board would NOT work with the Prescott CPU's. Not because of the chipset, but because of the old ass boards it was on.

Most of the boards I've seen won't do it for sure. Check with your board manufacturer for a BIOS update and see what they say. You'll find that most of the boards that don't have at least an i865 or i875P based chipset will NOT support a Celeron D or Prescott "A" series CPU.

If a new enough board with the proper voltage and power design, were equipped with a i845 chipset and a Celeron D or Pentium 4 Northwood 533MHz or Prescott "A" series processor it could work.

Most of the boards I've seen won't do it for sure. Check with your board manufacturer for a BIOS update and see what they say. You'll find that most of the boards that don't have at least an i865 or i875P based chipset will NOT support a Celeron D or Prescott "A" series CPU.

I know for a fact that a prescott core will run on a i845 board, unless you mean specifically the 800FSB/1066FSB ones, but every other will run. Photographic proof:

celeron6ja.jpg


Celeron D = prescott core (90nm, same core with less cache, to the T).

Definately a i845 board (as shown), also it had a bios from before last year, so I know it was an old board
 
ScHpAnKy said:
I know for a fact that a prescott core will run on a i845 board, unless you mean specifically the 800FSB/1066FSB ones, but every other will run. Photographic proof:

celeron6ja.jpg


Celeron D = prescott core (90nm, same core with less cache, to the T).

Definately a i845 board (as shown), also it had a bios from before last year, so I know it was an old board

Well that surprises me. Typically the VRM's on the older i845E boards wouldn't do it. Hell alot of early i865PE and i875P chipset based boards. Well I stand corrected.
 
Back
Top