Essay time!
The era of 45nm is upon us, and with it comes a hurricane of uncertainty. What's the official TJMax? Why so many stuck / high-reading DTSs? What does high voltage really do to one of these things?
I recently got hold of GRUnion's old q9450, and the first thing I noticed about it is that it takes a LOT of juice to kick up the clock on it. With my old e6750, if it didn't pass Orthos I'd kick the vCore up two notches and usually that lead to success. With this guy... it takes 4 notches or more. To jump from 3.44 to 3.472 required an increase of .025v, twice as much as I'm used to changing. It looks like it'll take a lot of vCore to get it to 3.6+.
Here lies my dilemma. I've read lots and lots of posts on this topic - Intel says that you shouldn't apply more than 1.3625v and a lot of us seem to agree with that, but here's what a quick google search reveals:
According to this, this, and this the safe voltage is actually 1.45v - it's not the vCore that will kill a 45nm, it's the VTT (FSB voltage). I looked into this (here's Intel's documentation for 45nm) and here's what I found. Quote:
(I attached a picture of it in case you don't have Adobe Reader... or you're just lazy 
)
According to that, this MAX voltage that is "outside of functional range" is 1.45 volts. Under that, there are no guarantees; over it and you're gonna break something. When I looked for this alleged "functional" voltage range, I got a chart that documents vDroop, and nothing really helpful. Anyway, what that says in a nutshell is that stay under 1.45v and essentially you're just stealing years away from its life, but it will probably work. I personally upgrade my CPU every year or two, so this doesn't bother me too much - especially since the 65nm chips were so notorious for their long life.
I decided, since I know much more about 65nm chips than these, I'd look to them for some advice. Note: I know that the two don't translate between each other very well. Intel's specs shows that the VID range goes up to 1.5v for these, yet the spec sheet for 65nm (second attachment for you lazy guys) shows that this "max voltage" outside of functional range is 1.55v! Now I had my old e6750 running at 1.55v for almost a YEAR before I replaced it with this new chip. It makes me think, maybe there's a little leeway between this MAX voltage they give, and the highest VID voltage.
So, like 1.5v was the safe limit for our previous processors, 1.3625 is assumed to be the safe limit for the new guys. But the difference between 1.3625 and the 1.45 MAX is considerable - almost twice the difference for the 65nm chips!
There are people running them above 1.3625v, and hell if you look at the records for my chip alone, people have run them as high as 1.8 volts! I'm not suggesting you try that, just pointing out that they were not instantly reduced to a pile of ash, even at a voltage I wouldn't run through my OLD processor. I know it's way too early in the game to tell what the effects of higher voltages will do to these things, but I feel like we might be able to push them harder than we tend to. While these chips have been lauded as super-sensitive to voltage, they could possibly take some abuse like their larger-die cousins.
---Stuff that people have come up with---
ericeod: for quad cores, raising NB voltage to 1.4v has a noticeable advantage, and may help attain higher overclocks without taking the vCore too high - a nice workaround if your motherboard can take some abuse!
Blameless: stay under 1.3625v for dual cores, but quad cores might be a little more resilient. Either way, you want to park your FSB voltage at 1.2-1.3. Taking it higher than that is bad news. A P35 Northbridge can easily take voltages of up to 1.65 (+0.4), so don't be afraid to take it up there for stability - as long as you keep it cool, it'll be fine.
JadeMiner: 1.3625v is the functional safe limit in the BIOS, not under load. You don't want 1.36v running under load, you want much less.
Edited by The Vector Kid - 8/18/08 at 10:39pm
The era of 45nm is upon us, and with it comes a hurricane of uncertainty. What's the official TJMax? Why so many stuck / high-reading DTSs? What does high voltage really do to one of these things?
I recently got hold of GRUnion's old q9450, and the first thing I noticed about it is that it takes a LOT of juice to kick up the clock on it. With my old e6750, if it didn't pass Orthos I'd kick the vCore up two notches and usually that lead to success. With this guy... it takes 4 notches or more. To jump from 3.44 to 3.472 required an increase of .025v, twice as much as I'm used to changing. It looks like it'll take a lot of vCore to get it to 3.6+.
Here lies my dilemma. I've read lots and lots of posts on this topic - Intel says that you shouldn't apply more than 1.3625v and a lot of us seem to agree with that, but here's what a quick google search reveals:
According to this, this, and this the safe voltage is actually 1.45v - it's not the vCore that will kill a 45nm, it's the VTT (FSB voltage). I looked into this (here's Intel's documentation for 45nm) and here's what I found. Quote:
Quote:
|
Table 2-2 specifies absolute maximum and minimum ratings only and lie outside the functional limits of the processor. Within functional operation limits, functionality and long-term reliability can be expected. At conditions outside functional operation condition limits, but within absolute maximum and minimum ratings, neither functionality nor long-term reliability can be expected. If a device is returned to conditions within functional operation limits after having been subjected to conditions outside these limits, but within the absolute maximum and minimum ratings, the device may be functional, but with its lifetime degraded depending on exposure to conditions exceeding the functional operation condition limits. |
)According to that, this MAX voltage that is "outside of functional range" is 1.45 volts. Under that, there are no guarantees; over it and you're gonna break something. When I looked for this alleged "functional" voltage range, I got a chart that documents vDroop, and nothing really helpful. Anyway, what that says in a nutshell is that stay under 1.45v and essentially you're just stealing years away from its life, but it will probably work. I personally upgrade my CPU every year or two, so this doesn't bother me too much - especially since the 65nm chips were so notorious for their long life.
I decided, since I know much more about 65nm chips than these, I'd look to them for some advice. Note: I know that the two don't translate between each other very well. Intel's specs shows that the VID range goes up to 1.5v for these, yet the spec sheet for 65nm (second attachment for you lazy guys) shows that this "max voltage" outside of functional range is 1.55v! Now I had my old e6750 running at 1.55v for almost a YEAR before I replaced it with this new chip. It makes me think, maybe there's a little leeway between this MAX voltage they give, and the highest VID voltage.
So, like 1.5v was the safe limit for our previous processors, 1.3625 is assumed to be the safe limit for the new guys. But the difference between 1.3625 and the 1.45 MAX is considerable - almost twice the difference for the 65nm chips!
There are people running them above 1.3625v, and hell if you look at the records for my chip alone, people have run them as high as 1.8 volts! I'm not suggesting you try that, just pointing out that they were not instantly reduced to a pile of ash, even at a voltage I wouldn't run through my OLD processor. I know it's way too early in the game to tell what the effects of higher voltages will do to these things, but I feel like we might be able to push them harder than we tend to. While these chips have been lauded as super-sensitive to voltage, they could possibly take some abuse like their larger-die cousins.
---Stuff that people have come up with---
ericeod: for quad cores, raising NB voltage to 1.4v has a noticeable advantage, and may help attain higher overclocks without taking the vCore too high - a nice workaround if your motherboard can take some abuse!
Blameless: stay under 1.3625v for dual cores, but quad cores might be a little more resilient. Either way, you want to park your FSB voltage at 1.2-1.3. Taking it higher than that is bad news. A P35 Northbridge can easily take voltages of up to 1.65 (+0.4), so don't be afraid to take it up there for stability - as long as you keep it cool, it'll be fine.
JadeMiner: 1.3625v is the functional safe limit in the BIOS, not under load. You don't want 1.36v running under load, you want much less.
Edited by The Vector Kid - 8/18/08 at 10:39pm
Duriel
(13 items) |
|
Duriel
(13 items) |
