Let me break things down via your quoted post.
Jeez man this is insane... Seems you got there diamond level chips, that can reach insane speed, for not much voltage too....
TBH they really aren't any different that yours by steppings or dates.
As for pics, I'd have to do some real hunting to track them all down since the move here. I haven't unpacked anything related to Socket A do that's the deal with it.... Plus it's been awhile since those runs were done anyway.
The best i got could not boot at 2.8GHz i think, as i OC with the intent to use 24/7 and also to be on the safe side, as sure AthlonXP running at 2.9GHz would be great, as it will wipe the floor with about every 478 Pentium 4, aside from the ones that OC up to 4GHz lol... But the chip will prob fry itself or at least degrade, as much GHz and voltage too...
Running an A this high in MHz 24/7 can spell a short life for them.
Also, I didn't boot the system at that speed, instead I used ClockGen to get them up there once I was in the OS.
Can u post pictures of these diamond grade CPUs that reach 2.84GHz and 2.9GHz to see how they look, steppings and shht.... Is the 2.87GHz 3000+ Mobile Barton, cuz i have one of these and was total sucker for OC, did read the same in CPU world in the comments about the model..
The 3000+ XP-M is a good one itself and believe me I did pay to get my hands on it when it appeared.
3000+ XP-M's are rare anyway and you just don't see them that often, that's why I didn't mind the price to get it.
Also saw that one of them is on 177 FSB, i look only at 200 FSB, since i want usable system. The other one is better at 230 FSB, tho i always use 2GB of memory, so nothing above 200 FSB works for me, even damn DDR500 Mushkin memory that was designed for s939, would run at 240 FSB, just so it can start spitting ton of errors and critical errors 30 minutes later lmao... Also i am mad, or i saw somewhere that u did OC on DFI LanParty NF2, as i cant see it now for some reason... Tho the NF2 LanParty was never a stable mobo, not like NF7 V2 with modded bios...
Related to the 2800+ XP-M validation:
That's because I was going for max speed to validate it but
didn't have it frozen, instead it was on chilled water to get there.
Sometimes running the highest multiplier it can handle is the way, sometimes not because every chip has a limit to how much FSB it can handle based on things like load temps and that must be determined on an individual chip basis.
The board I was using to get it done (DFI Ultra II Infinity) does have an Ultra B BIOS flashed in it so that helped some.
However if I should freeze it, no doubt it could go a little higher.
U did use ice cooling, on the 2v m barton seems... I think big a$$ tower cooler will keep them in check, as i use TT Sonic Tower with x3 heat pipes, and x2 fans on higher speed, and the damn thing hardly gets above 35-36c no matter the load... So i dont think extreme cooling is needed for these even at 2v, just big chunk of metal with copper pipes... Tho the die is so small, and it produces a lot of heat on its own... I dont bother OC other parts like 3Dfx Voodoo V 5.5K or other stuff, cuz most of them are rare as hell these days, expensive too, and also old enough to die even on stock speeds, but for some reason i like OC 462 mobile bartons for fun, and some 939 CPUs...
No, that kind of speed and voltage isn't safe if just on air or regular water.
It takes COLD to make them go and not turn into an extra-crunchy chip (Cheeto?) along the way.
It's not a matter of how big your cooler or tower is, that can only get it so far before thermal saturation sets in
based on the size of the area in contact with the cooler/block's die contact surface which actually makes contact with it.
Die temps UNDER LOAD, not at idle are the thing about it.
What's seen in the BIOS or with a monitoring program like HWmonitor while it's sitting idle at desktop aren't what matters here.
Only true way to know is to run it
under load while running HWmonitor or some other monitoring proggy in the background while it's benching - That will tell you exactly what's going on.
The colder it is, the more efficient your cooling is too as to how much heat energy it can absorb before thermal saturation of this area happens so it takes cold, not just bigger cooling to attract heat that way and we are talking about a little rectagular die on the chip so you really want to help it as much as possible if you can.
For desktop Barton's it means this really small area is already radiating 62W's of heat at STOCK voltage and speeds, not while OC'ed!
NOTE:
XP-M's can draw just as much power as any desktop chip will, once either one is OC'ed.
In my findings with few mobile bartons, the IQ stepping seems to be the best (interesting stepping name huh)... I do wonder if the color of the die indicates higher bin, or the date code... I believe that the darker purple dies are the better ones, compared to the pink ones...
There are only two PCB colors for Barton and T-Bred chips.
Brown and Green and that makes no difference in how they perform - It's just the color of the chip's PCB and that's it.
Purple PCB's are based on older Spitfire, Thunderbird and Morgan cores which are all older architecture chips anyway, so it's natural they won't clock as high as a T-Bred or Barton core.
As for the die/core looking purple or pink - That's simply a light-refraction "Thing".
Doesn't mean anything I'm aware of BUT that also doesn't mean it's "Nothing" either.
I'll need to check that one day and see if my good ones are like that or not, one way or the other.
Yes, the "I" stepping chips are best as said earlier and your chip's date plays a role too.
The preferred chip date is one newer than 0339, even though that's the cutoff date for Superlocked chips. They tend to do better (For non-mobile chips) and that carries over to XP-M's as well.
Years ago I got LUCKY and snagged a pre-superlock 3200+ Barton core that's an E chip (0336 date) and it can hit over 2700 itself when it's cold (Not frozen though - Chilled water).
Referring to Barton cored chips:
C,D and E chips (Very last digit on the chip's model name/string info) that C's and E's tend to do best, D's alot of the time fall short because they don't seem to like alot of FSB, instead leaning on CPU multiplier to get there if they can.
There are some D's that can go too but the tendency (Not an absolute of course) is either C or E chips to be good vs a D chip overall.
You may know this but if not:
C means it's a 133MHz (FSB) chip.
D means it's a 166MHz (FSB) chip.
E means it's a 200MHz (FSB) chip.
