Whast the difference between 875P and 865PE now that PAT can be enabled on 865PE?

[AgaBoogaBoo]

The only difference I can note between the two is ECC memory support. Abit, Asus, Chaintech, and Albatron have all enabled PAT with new BIOS revisions.

Loads of mainboard makers try to follow ASUS with enabling the Performance Acceleration Technology on i865PE-based mainboards in order to compete with the largest mainboard manufacturer. As we understand now, it is hardly difficult to enable the technology on the chipsets that do not officially support it.

Albatron, ABIT and Chaintech have reportedly released new BIOS versions for their i865PE mainboards that enable certain memory performance optimizations that increase the speed dramatically. I now think that these optimizations have the same nature as Intel?s own PAT, however, since Intel denies using this brand-name when referring to its mainstream products, mainboard makers do not state the they actually enable this feature. As you probably remember, ASUS announced a week or so ago that it had enabled PAT on i865PE (see this news-story), but now the company refers to this ?PAT? as the ?Hyper Path? mode.

If the substance of [i865PE] optimizations implemented by mainboard makers includes the same techniques as in Intel?s PAT, it makes buying i875P instead of i865PE practically useless for end-users and especially for hardware enthusiasts.

We still have to reveal if i875P mainboards have some real advantages over the i865PE-based devices in addition to ECC support, but at this point we can conclude that mainboard makers try to do everything in order to boost performance of i865PE towards i875P level. They certainly do understand that they lower the attractiveness of i875P solutions for a lot of end-users, so, there should be a logical reason for the action. Maybe after all mainboard makers will begin to boost i875P performance substantially as well?

http://www.xbitlabs.com/news/chipsets/display/20030606181831.html

 

[Darkhawk28]

Very nice question.

Personally, I'm waffling between upgrading to a new Athlon platform or switching toa 865PE platform. This kinda tips things to the Intel side again.

{note: This is not an AMD vs. Intel statement. Just a personal thought I've been having. I love both platforms, just thinking about future upgradability}

😀

 

[First]

Now, the only difference is ECC support.

 

[AgaBoogaBoo]

Originally posted by: Evan Lieb
Now, the only difference is ECC support.

Then I guess there isn't any realistic reason to purchase an 875P for a home/enthusiast user, is there? I think I'll just get the P4P800 then. Thanks.

 

[aggie113]

It's not true PAT technology. It's a little work around that basically is like the old radeon 9500 128MB hack to turn it into a 9700. They are tricking the motherboard into thinking it is the 875 chipset. Both chipsets use the same northbridge but only a certain % make the grade to use PAT technology. So my problem with this is the company is taking chips that intel says aren't good enough to run PAT on and are doing it.

 

[ViRGE]

Is it me, or are they shooting themselves in the foot? If they get the performance of their i865 chips up to i875 levels, they're going to be making less money, since the i875 most likely has a higher profit margin than the i865. They of course need to stay competitive, and I'm not going to complain about things being a good deal, but if this lowers the profit margins too much, it can't be a good thing for the industry.

 

[charlie21]

Originally posted by: aggie113
It's not true PAT technology. It's a little work around that basically is like the old radeon 9500 128MB hack to turn it into a 9700. They are tricking the motherboard into thinking it is the 875 chipset. Both chipsets use the same northbridge but only a certain % make the grade to use PAT technology. So my problem with this is the company is taking chips that intel says aren't good enough to run PAT on and are doing it.

I don't think that PAT even qualifies as a "technology". When I hear the term technology used, I think of a property that fundamentally seperates one product from another. Speed binning chips that can shave a couple of clocks off of memory accesses? That's like saying an XP 2000+ uses a different technology than my XP 1800+, even though I can OC it to that level. Marketing BS.

I'm sure that Intel's yields are good enough that the majority of the chips that aren't "PAT-certified" are capable of being so. Look at it from their perspective. Design and produce one chipset, but artificially cripple some of your product according to market demand. Genius.

 

[aggie113]

I think the people who designed the pathway implementation of the northbridge of that chipset would disagree that it's not technology. It's very real technology. And did you even know what I'm talking about when I refrence the ATI 9500 hack? About half the cards that hack was applied to got nasty visual artifacts because the plain fact is that no FAB process will output 100% on a silicon wafer. You WILL get defects and fuckups on it all the time.

 

[AgaBoogaBoo]

Originally posted by: aggie113
I think the people who designed the pathway implementation of the northbridge of that chipset would disagree that it's not technology. It's very real technology. And did you even know what I'm talking about when I refrence the ATI 9500 hack? About half the cards that hack was applied to got nasty visual artifacts because the plain fact is that no FAB process will output 100% on a silicon wafer. You WILL get defects and fuckups on it all the time.

Look, I don't know for sure because I haven't seen Intel's production factories or the design of the wafers, but all I know is that if Asus, along with Abit, Albatron, AND Chaintech, have all released a way to enable it, then it must be fine, especially considering that other motherboard makers will soon follow.

 

[charlie21]

Originally posted by: aggie113
I think the people who designed the pathway implementation of the northbridge of that chipset would disagree that it's not technology. It's very real technology. And did you even know what I'm talking about when I refrence the ATI 9500 hack? About half the cards that hack was applied to got nasty visual artifacts because the plain fact is that no FAB process will output 100% on a silicon wafer. You WILL get defects and fuckups on it all the time.

Yes, I do know about the 9500 -> 9700 mod trick. In fact, that's analogous to the 865PE -> 875 situtation. A 9500 is just a 9700 with a 128-bit memory interface and a slightly lower mem/core speed, IIRC. By doing the hardware or software mod, you enabled the full 256-bit memory path. Then you OC your memory and GPU, and *poof* you've got yourself a 9700. The artifacts are present in chips that are unable to handle this. Some cards worked perfectly, though. The core technology is the same, but due to "the plain fact is that no FAB process will output 100% on a silicon wafer" the 9700's just passed stricter validation tests.

Anyway, I was just trying to complain about stupid marketing BS.