• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

color of the core

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

oc_byagi

Member
Joined
Apr 28, 2004
Location
Sunnyvale, CA
do the color of the core has anything to do with overclocking capabilities? I have two 2600+ mobiles in front of me with same steppings but one core has greyish glow and the other has reddish/purplish glow. (no i'm not stupid enough to not know the angles of light reflection)
 
It's a myth that colors play a role. They do not.

Codes (to a certain extent) and core type does.
 
Its strange how they would use different martial, maybe the different types disapate hate better or something. Or maybe its just a dye hmm. weird anyway.
 
I think the color comes from the oxide that's on the back of the die and not from polishing, since it's silicon polishing by any means wouldn't change the color.

You can tell the thickness of the oxide by the color

so if you wanted the best heat transfer you would be looking for the thinnest oxide layer (but I guess it wouldn't affect the overall heat transfer)



Film Thickness (µm) vs Color
0.05 Tan
0.07 Brown
0.10 Dark violet to red violet
0.12 Royal Blue
0.15 Light blue to metallic blue
0.17 Metallic to very light yellow green
0.20 Light gold to yellow; slightly metallic
0.22 Gold with slight yellow orange
0.25 Orange to melon
0.27 Red violet
0.30 Blue to violet blue
0.31 Blue
0.32 Blue to blue green
0.34 Light green
0.35 Green to yellow green
0.36 Yellow green
0.37 Green yellow
0.39 Yellow
0.41 Light orange
0.42 Carnation pink
0.44 Violet red
0.46 Red violet
0.47 Violet
0.48 Blue violet
0.49 Blue
0.50 Blue green
0.52 Green (broad)
0.54 Yellow green
0.56 Green yellow
...
 

Attachments

  • reference_si.gif
    reference_si.gif
    11 KB · Views: 273
c627627 said:
It's a myth that colors play a role. They do not.

Codes (to a certain extent) and core type does.


Hmm.. I may just have a crappy barton, but I have a week37 2500+, and a week 50. the week 50 is brown, and has a blueish core, and I cannot get it to post past 210fsb, even at 20C >_< My week 37 is a green PCB, and has a reddish core, and it does 223fsb easily at 1.7V...

Bad chip?
 
If this is happeing on exact same system with exact same components, yeah the chip is the one to blame - the chip itself, not the color of it.
 
c627627 said:
If this is happeing on exact same system with exact same components, yeah the chip is the one to blame - the chip itself, not the color of it.


Aww... but I liked blaming the brown chip :D

The one problem that I see, is that the **ONLY*** difference in CPU is **COLOR** and **WEEK**.. They have exact same stepping (AQXEA) as well.

Either way, its the brown chip :p
 
Early on, rumors were the best overclocking cores had a silver-green tint. Worse overclockers were the red-purple ones. Yet all the highest frequency CPUs at the time XP3200, etc. had purplish cores. ???

Supposed the silver-green ones came from the center of the 'sheet' or I forget what you call it when they make CPUs.

That being said, I tried to get a whole bunch of the silver green ones. However, after testing 2-XP1700s and 2-XP2100s (all of which did 2350-2430), the best one which did the lowest VCORE had a purple-red core - so much for that. The others were bluish to greenish silver. Note they all did about the same, not too shabby on air.
 
Clevor said:

Supposed the silver-green ones came from the center of the 'sheet' or I forget what you call it when they make CPUs.

Its called a waffer
 
i only ever get purple chips....

the colours or the rainbow... lalalla...

kind or intresting why they have different colours though. note the colours not the colors.

:)
 
I'm in Japan where they sell a lot of OEM stuff in trays, so you can check out all the colors and pick out the one you want. The wafers on the lower frequency CPUS like XP1700-2000 are still various colors.

I believe another trick was check the s/n of the chip. Usually you can make out a 2700 or 2500 in there, which means the XP1700 is a downgraded XP2500 core, etc. Not sure this correlated with better overclocks. Or the 'XPMV' at the end. Supposedly the lower the first alphabet, the better the core, though a lot of Xxxx chips were real good overclockers.
 
The two digits supposedly hidden within the serial number was proven to be false as when the Tbreds first came out at 2200 there were cpus with these numbers as low as "20".
Again can I just say that when the cpu's are tested for frequency they are merely tested at the highest possible rating (eg 3200+ for Bartons) and at their given voltage. They dont have time to check their overclocking potential! Also at present I would suspect due to the maturity of especially the Bartons - virtually all cpus will pass first time and wont require testing at lower ratings (3000+ etc). This maturity is especially noticeable in AMD's ability to release mobiles at 2800+ etc rating with a lower vcore.
One other thing I should point out at present the TMP (testing, marking and packaging) is done in Malaysia for all cpus but additional testing is done in Singapore on the A64's and FX's. Some or all of the TMP will shift to China once the new facility is built there.
 
I heard a rumor that the color of the core does matter. From what I heard, in the light pure silverish does the best, then blueish, then purple/red, then mixed colors, but, I have a silverish 2100 that did horrible and a purple/red 1700 that does really well. Truthfully, I think it is just better to get a good stepping and week then to judge by the color.

As for the color of the chips, I think that is random. I asked AMD once, and they said it was random. Sorry, but that e-mail is long gone. Anyways, I have seen green and brown of the same week and stepping before. I think the only difference is that the green ones are harder to to bridge mod.

-0cer
 
The rumors that the color of the core does matter have been conclusively proven as false. It was just a coincidence that people had a couple of CPUs that did well and they thought the color had something to do with it. It does not.

It's not the color, it isn't even the the part of the CPU markings that says something like 'MPMW.' That has also been proven as a false indicator of how well the CPU overclocks.
 
c627627 said:
The rumors that the color of the core does matter have been conclusively proven as false. It was just a coincidence that people had a couple of CPUs that did well and they thought the color had something to do with it. It does not.

It's not the color, it isn't even the the part of the CPU markings that says something like 'MPMW.' That has also been proven as a false indicator of how well the CPU overclocks.


lol! This could be a never ending circle...

According to you, my CPU is just a bad chip... but they just happened to have a good chip...... Then again, what isn't a pattern, if you exclude the oddballs? LOL!
 
The core: mobile Barton > desktop Barton > Thoroughbred B > Thoroughbred A > Palomino.

dates of production to certain extent: New multiplier locked desktop Bartons overclock further than old unlocked desktop Bartons. (But you need quality RAM to overclock the FSB part of the MHz speed = FSB x multiplier equation with them.)

and the opposite may be true for the mobiles: old 2400+ mobiles apparently overclock further than 2400+ with most recent dates on them.

+ maybe the stepping codes to a certain extent...


not the color of cores and not the CPU marking where it says something like MPMW .
 
Back