hey, any of you guys ever wondered what the DSL before the SOI means on your chip specs? i sure did, so i looked it up.
SOI = Silicon On Insulator to help with die shrink and stopping the possiblilty of electromigration due to the interconnects of the transistors of .013µ or smaller proccess techniques.
electromigration - http://www.csl.mete.metu.edu.tr/Electro ... n/emig.htm
but, you ever wonder what the DSL means? well apparently it's Dual Stress/Strain Liner. a stronger "tighter weaved strand silicon that helps out SOI even more.
there are a lot of people running thier 1.4v cpu's way past the golden rule of no more than .2v overvolt. people are running these 1.4v chips at up to 1.7v.
now i think i know why the cpu's aren't bunring up from electromigration. it's due to DSL SOI. c00l.
i was able to hit 3200mhz at 1.638v and i was thinking that should be my limit on volt. now i think i'll go up to 1.675 depending on the full load temps. i don't want to exceed 55C.
TGM
DSL SOI What? internet on my SOI?
- TheGeekMistress
- Legit Fanatic
- Posts: 227
- Joined: Tue Feb 14, 2006 12:52 am
- Location: SW Michigan
- Contact:
DSL SOI What? internet on my SOI?
+ DFI LanParty-UT NF3 250GB skt754
+ Bios Rev. TMOD's 06/23/2005
+ A64 Newark Mobile 4000+ @ 5000+ (9x341) 3070mhz | Idle 30C Load 39C
+ Asetek Waterchill BIX Dual
+ 2x512 G.Skill 1GBLA TCCD @ DDR512 (CPUz 256mhz 2.0-3-3-6-1T)
+ XFX 7800GS OC'd to 460/1500
+ SATA150 2x WD360Raptors NVRaid RAID-0
+ 21" Hitachi CRT
+ Cooler Master CMStacker case
+ Enermax EG651P-VE(FMA)
+ Bios Rev. TMOD's 06/23/2005
+ A64 Newark Mobile 4000+ @ 5000+ (9x341) 3070mhz | Idle 30C Load 39C
+ Asetek Waterchill BIX Dual
+ 2x512 G.Skill 1GBLA TCCD @ DDR512 (CPUz 256mhz 2.0-3-3-6-1T)
+ XFX 7800GS OC'd to 460/1500
+ SATA150 2x WD360Raptors NVRaid RAID-0
+ 21" Hitachi CRT
+ Cooler Master CMStacker case
+ Enermax EG651P-VE(FMA)
- kenc51
- Legit Extremist
- Posts: 5167
- Joined: Thu Jun 23, 2005 1:56 pm
- Location: Dublin, Republic of Ireland
- Contact:
this thread also explains (in part) electromigration.....
http://forums.anandtech.com/messageview ... id=1773169
I'm currently reading this stuff (dsl-soi) myself.......very interesting but hard to grasp......
great minds think alike ;)
http://forums.anandtech.com/messageview ... id=1773169
I'm currently reading this stuff (dsl-soi) myself.......very interesting but hard to grasp......
great minds think alike ;)
- TheGeekMistress
- Legit Fanatic
- Posts: 227
- Joined: Tue Feb 14, 2006 12:52 am
- Location: SW Michigan
- Contact:
yup, they sure do
i hear the anandtech forums are full of snot nose kids with bad atitudes.
i'm on here, 3dgameman, dfi-strret *rolls eyes* , XS, Guru3D, Mod the Box..... i think that it's.
until i pissed off travis and made him out for a fool, i was most active in dfi-street. 3DGameman, more than a hand full of girls there and my first real forum home. i've got a couple thousand posts there. and a thread i started back in 2001 is STILL going there, lol.
TGM
i hear the anandtech forums are full of snot nose kids with bad atitudes.
i'm on here, 3dgameman, dfi-strret *rolls eyes* , XS, Guru3D, Mod the Box..... i think that it's.
until i pissed off travis and made him out for a fool, i was most active in dfi-street. 3DGameman, more than a hand full of girls there and my first real forum home. i've got a couple thousand posts there. and a thread i started back in 2001 is STILL going there, lol.
TGM
+ DFI LanParty-UT NF3 250GB skt754
+ Bios Rev. TMOD's 06/23/2005
+ A64 Newark Mobile 4000+ @ 5000+ (9x341) 3070mhz | Idle 30C Load 39C
+ Asetek Waterchill BIX Dual
+ 2x512 G.Skill 1GBLA TCCD @ DDR512 (CPUz 256mhz 2.0-3-3-6-1T)
+ XFX 7800GS OC'd to 460/1500
+ SATA150 2x WD360Raptors NVRaid RAID-0
+ 21" Hitachi CRT
+ Cooler Master CMStacker case
+ Enermax EG651P-VE(FMA)
+ Bios Rev. TMOD's 06/23/2005
+ A64 Newark Mobile 4000+ @ 5000+ (9x341) 3070mhz | Idle 30C Load 39C
+ Asetek Waterchill BIX Dual
+ 2x512 G.Skill 1GBLA TCCD @ DDR512 (CPUz 256mhz 2.0-3-3-6-1T)
+ XFX 7800GS OC'd to 460/1500
+ SATA150 2x WD360Raptors NVRaid RAID-0
+ 21" Hitachi CRT
+ Cooler Master CMStacker case
+ Enermax EG651P-VE(FMA)
Good stuff TGM. I'm running almost 1.6v on my Opty and it's hanging in there for now.
This is the forum where I mostly hang out for computer related stuff. DFI-street sometimes but not very often anymore. I dislike Anandtech although there are some very knowledgeable people there. You have to wade through a lot of crap to find good, solid info.
This is the forum where I mostly hang out for computer related stuff. DFI-street sometimes but not very often anymore. I dislike Anandtech although there are some very knowledgeable people there. You have to wade through a lot of crap to find good, solid info.
Joe
- TheGeekMistress
- Legit Fanatic
- Posts: 227
- Joined: Tue Feb 14, 2006 12:52 am
- Location: SW Michigan
- Contact:
it doesn't seem too hard to grasp. just placing electrcally conductive wires closer and closer together would show electomigration. you'd get a static shock.
then coat those wires with plastic and you can get them even closer. but as we know plastic is too thick to be inside a 13 nanometer or 09 nanometer distance apart. and we know that paper would burn out quickly. so what's fluid, can be hardened has billions of billions of particles that are fluid and can be shaped then hardened. sound like anything to you?
Yup, sand. sand makes glass. fluid and then hard. altho standard glass is to fragile at a 90 nanometer seperation. so we use silicon. silicon is sand made with other materials that harden it before it turns to glass. SOI is placing the silicon barrior on the small interconnects that the transistors use to communicate inside the cpu.
now if you were to take all those billions and billions of particles in the Silicon and sort them out to form a line, they are even stronger and more resiliant to electricity. making the electricity harder to get thru the insulating material, therfore allowing "high voltage" to go thru each and every millions of transistors without creating a cascade affect of voltage (electromigration: when an electrical pathway thru the core trips other transistors that start a cascading affect and they all are forced open or closed, that's what fries the core), then you've got a smaller die process, which allows for less voltage, which can run faster since the transistors switches are much smaller and cooler running. DSL SOI.
this has the same effect in metals. if you take a piece of metal and place it in liquid nitrogen, all the particles in that piece of metal will form into lines. you can't see it with the naked eye, but it happens none the less. take that same piece of metal out with the oriented strands and it's much stronger and more efficient at dissapation of heat or cold.
thats the simplified way of explaining it, but it gets people the idea of how it works inside a microprossesor.
TGM
then coat those wires with plastic and you can get them even closer. but as we know plastic is too thick to be inside a 13 nanometer or 09 nanometer distance apart. and we know that paper would burn out quickly. so what's fluid, can be hardened has billions of billions of particles that are fluid and can be shaped then hardened. sound like anything to you?
Yup, sand. sand makes glass. fluid and then hard. altho standard glass is to fragile at a 90 nanometer seperation. so we use silicon. silicon is sand made with other materials that harden it before it turns to glass. SOI is placing the silicon barrior on the small interconnects that the transistors use to communicate inside the cpu.
now if you were to take all those billions and billions of particles in the Silicon and sort them out to form a line, they are even stronger and more resiliant to electricity. making the electricity harder to get thru the insulating material, therfore allowing "high voltage" to go thru each and every millions of transistors without creating a cascade affect of voltage (electromigration: when an electrical pathway thru the core trips other transistors that start a cascading affect and they all are forced open or closed, that's what fries the core), then you've got a smaller die process, which allows for less voltage, which can run faster since the transistors switches are much smaller and cooler running. DSL SOI.
this has the same effect in metals. if you take a piece of metal and place it in liquid nitrogen, all the particles in that piece of metal will form into lines. you can't see it with the naked eye, but it happens none the less. take that same piece of metal out with the oriented strands and it's much stronger and more efficient at dissapation of heat or cold.
thats the simplified way of explaining it, but it gets people the idea of how it works inside a microprossesor.
TGM
+ DFI LanParty-UT NF3 250GB skt754
+ Bios Rev. TMOD's 06/23/2005
+ A64 Newark Mobile 4000+ @ 5000+ (9x341) 3070mhz | Idle 30C Load 39C
+ Asetek Waterchill BIX Dual
+ 2x512 G.Skill 1GBLA TCCD @ DDR512 (CPUz 256mhz 2.0-3-3-6-1T)
+ XFX 7800GS OC'd to 460/1500
+ SATA150 2x WD360Raptors NVRaid RAID-0
+ 21" Hitachi CRT
+ Cooler Master CMStacker case
+ Enermax EG651P-VE(FMA)
+ Bios Rev. TMOD's 06/23/2005
+ A64 Newark Mobile 4000+ @ 5000+ (9x341) 3070mhz | Idle 30C Load 39C
+ Asetek Waterchill BIX Dual
+ 2x512 G.Skill 1GBLA TCCD @ DDR512 (CPUz 256mhz 2.0-3-3-6-1T)
+ XFX 7800GS OC'd to 460/1500
+ SATA150 2x WD360Raptors NVRaid RAID-0
+ 21" Hitachi CRT
+ Cooler Master CMStacker case
+ Enermax EG651P-VE(FMA)
Actually there is a beter way of doing it.. and thats using carbon.
The reason that sillicon is stronger in a line is because it can "cantenate" (a chemistry term). This literally means it can form bonds between its molecules to form a chain. However, Si can only form 6 or 7 bonds before the chain becomes unstable. Carbon on the other hand can cantonate until forever and also is not a semi-conductor either.
Now what they could try and do is use carbon nanotubes for the links inside the processors. Carbon nanotubes are cylindrical tubes of carbon with a hollow inside, kind of like a sheath, a fully non-conductive and only 1 atom thick sheath (anone heard of buckminster fullrerenes or bucky balls).
Im not sure of the exact purpose they are trying to get out of the nanotubes, but IBM and intel are looking to see how they can use them!
The reason that sillicon is stronger in a line is because it can "cantenate" (a chemistry term). This literally means it can form bonds between its molecules to form a chain. However, Si can only form 6 or 7 bonds before the chain becomes unstable. Carbon on the other hand can cantonate until forever and also is not a semi-conductor either.
Now what they could try and do is use carbon nanotubes for the links inside the processors. Carbon nanotubes are cylindrical tubes of carbon with a hollow inside, kind of like a sheath, a fully non-conductive and only 1 atom thick sheath (anone heard of buckminster fullrerenes or bucky balls).
Im not sure of the exact purpose they are trying to get out of the nanotubes, but IBM and intel are looking to see how they can use them!
- TheGeekMistress
- Legit Fanatic
- Posts: 227
- Joined: Tue Feb 14, 2006 12:52 am
- Location: SW Michigan
- Contact:
i would suspect that nanotubes with their infinite structure length and hollow properties would allow not only insulation, but the hollow struture would aide in creating a "cooling space" inbetween the transistors.
TGM
TGM
+ DFI LanParty-UT NF3 250GB skt754
+ Bios Rev. TMOD's 06/23/2005
+ A64 Newark Mobile 4000+ @ 5000+ (9x341) 3070mhz | Idle 30C Load 39C
+ Asetek Waterchill BIX Dual
+ 2x512 G.Skill 1GBLA TCCD @ DDR512 (CPUz 256mhz 2.0-3-3-6-1T)
+ XFX 7800GS OC'd to 460/1500
+ SATA150 2x WD360Raptors NVRaid RAID-0
+ 21" Hitachi CRT
+ Cooler Master CMStacker case
+ Enermax EG651P-VE(FMA)
+ Bios Rev. TMOD's 06/23/2005
+ A64 Newark Mobile 4000+ @ 5000+ (9x341) 3070mhz | Idle 30C Load 39C
+ Asetek Waterchill BIX Dual
+ 2x512 G.Skill 1GBLA TCCD @ DDR512 (CPUz 256mhz 2.0-3-3-6-1T)
+ XFX 7800GS OC'd to 460/1500
+ SATA150 2x WD360Raptors NVRaid RAID-0
+ 21" Hitachi CRT
+ Cooler Master CMStacker case
+ Enermax EG651P-VE(FMA)