PDA

View Full Version : Undervolting and Overclocking = Nirvana

ST4
04-09-2006, 12:53 PM
One of the joys I get now out of tweaking with my system is undervolting the CPU while overclocking. YOu get the advantages of low heat, hence minimal noise, and performance all at the same time, which for me as an HTPC enthusiast, is nirvana! ;)

So, let's see some of your undervolting and OCing endeavors (especialy some of the WC and PC folks)!

I'll start with my last 2 chips:

X2 3800 @ 2465MHz 1.15V on Air (TT BT) (http://i59.photobucket.com/albums/g319/stranx44/sp24-2465-idle.jpg)

Opty 165 @ 2338 1.10V on Air (TT BT) (http://i59.photobucket.com/albums/g319/stranx44/sp04-2338.jpg)
v0dka
04-09-2006, 01:28 PM
Is the temp 29C while priming in that second screenshot? And in the first?

If youre doing this for low temps you should at least tell us what they are then. :rolleyes:
lawrywild
04-09-2006, 01:31 PM
I was running my X2 @ 2.2ghz 1.15v earlier

Load temp 30'C on XP-90C :D
Richteralan
04-09-2006, 01:40 PM
http://www.xtremesystems.org/forums/showthread.php?t=79509
;)
ST4
04-09-2006, 01:53 PM
http://www.xtremesystems.org/forums/showthread.php?t=79509
;)

AFAIK undervolting should not cause issues with the mem controller. REMEMBER that AMD actually specifies default 1.1V for Cool N Quiet spec which is publicly available on ALL BIOS and AMD processor driver itself...THe problem that Tony talks about is because of the lower core voltage of the 90nm process CPUs and folks who like to run astonically high VDIMM with the BH5s and VXs....
ST4
04-09-2006, 01:56 PM
Is the temp 29C while priming in that second screenshot? And in the first?

If youre doing this for low temps you should at least tell us what they are then. :rolleyes:

if you tak e a look at the WIndows Task Manager in each shot, it tells you the activity thereof :stick:

The x2 3800 was idle (note the big *IDLE* :fact: messages in Prime)

And the Opty was in the middle of dual priming when the screenshot took place (again note the big *GO* :clap: or the Taskman showing full CPU usage history.
MUCHO
04-09-2006, 03:35 PM
Here is a shot running at 2.6 with low voltage. I'm sure I could have gone lower but I was reading sudden CPU deaths by running too low of voltage so I stopped.

http://img138.imageshack.us/my.php?image=26lowvolts9av.jpg
uOpt
04-09-2006, 04:10 PM
I was lucky enough to get a week 0530 Opteron 170.

Unfortunately I got the worst clocking 0530 on the planet, it shut off at 2700 MHz.

However, it can run 2500 MHz undervolted 48 hours prime-stable. If you keep the temperature below 50 C - which of course gets easier with less volts.

It's about the most perfect server chip you can imagine.

Well, I wish I could say the same thing for the mainboards I put into that server but hey nothing is purrfect :)
blazin-asian
04-09-2006, 09:00 PM
I was lucky enough to get a week 0530 Opteron 170.

Unfortunately I got the worst clocking 0530 on the planet, it shut off at 2700 MHz.

However, it can run 2500 MHz undervolted 48 hours prime-stable. If you keep the temperature below 50 C - which of course gets easier with less volts.

It's about the most perfect server chip you can imagine.

Well, I wish I could say the same thing for the mainboards I put into that server but hey nothing is purrfect :)

dont mean to threadjack, but which 0530 did you get? i really hope you arent talking about the TPMW. i just got one and im waiting for it in the mail.
ST4
04-09-2006, 10:30 PM
Here is a shot running at 2.6 with low voltage. I'm sure I could have gone lower but I was reading sudden CPU deaths by running too low of voltage so I stopped.

http://img138.imageshack.us/my.php?image=26lowvolts9av.jpg

read this:

http://www.silentpcreview.com/article172-page1.html

it will tell you everything you need to know about AMD's undervolting technology, that has been in existence for quite some time (albeit in laptops).
jsriolo
04-09-2006, 11:37 PM
Actually, I find that the 0530 TPMW does very well with low volts. If you just ordered it then you shouldn't be dissapointed, overall that stepping is very good. I have run mine on air as high as 2.95ghz at ~1.5v (don't remember exactly) but now that I am at school I run it at 2.6ghz with 1.3v. It really does help keep my already hot dorm a bit cooler so I think it's worth it.
Richteralan
04-10-2006, 07:07 AM
AFAIK undervolting should not cause issues with the mem controller. REMEMBER that AMD actually specifies default 1.1V for Cool N Quiet spec which is publicly available on ALL BIOS and AMD processor driver itself...THe problem that Tony talks about is because of the lower core voltage of the 90nm process CPUs and folks who like to run astonically high VDIMM with the BH5s and VXs....

Yes, but IF you overclocking + undervolting + raising Vdimm(to get more overclocking), then you gonna:banana:
;)
uOpt
04-10-2006, 07:43 AM
dont mean to threadjack, but which 0530 did you get? i really hope you arent talking about the TPMW. i just got one and im waiting for it in the mail.

Sorry:
http://www.cons.org/tmp/0530.jpg

If you want high Vcore and frequency, it failed 2.8 GHz at 1.55 V dual-prime after 30 hours or so.

I will actually experiment with somewhat higher Vdimm and low memory clocks to get even more out of it at 1.325 or 1.30 volts.
ST4
04-10-2006, 08:49 AM
Yes, but IF you overclocking + undervolting + raising Vdimm(to get more overclocking), then you gonna:banana:
;)

Um if you read the whole thread, ANY newer 90nm core CPUs are a risk with massive VDIMM, regardless of VCORE or FREQ. Ask yourself, what's so unique if I run 1.1V at 800MHz w/ nominal 2.6V-2.8V VDIMM, versus running 1.35V @ 1800MHz w/ the same nominal VDIMM? Because that's EXACTLY how AMD has spec'd out there CnQ implementation.
Richteralan
04-10-2006, 08:50 AM
Um if you read the whole thread, ANY newer 90nm core CPUs are a risk with massive VDIMM, regardless of VCORE or FREQ. Ask yourself, what's so unique if I run 1.1V at 800MHz w/ nominal 2.6V-2.8V VDIMM, versus running 1.35V @ 1800MHz w/ the same nominal VDIMM? Because that's EXACTLY how AMD has spec'd out there CnQ implementation.

ugh I guess you missed what I've said "+ Raise Vdimm"?:)

If you didn't raise Vdimm so what's the problem?

P.S. It's not that massive Vdimm cause the problem, It's the massive DIFFERENCE between vcore and vdimm cause the problem. That's why there's an formula to calculate your lowest vcore from your vdimm.
high5
04-10-2006, 09:06 AM
It's not that massive Vdimm cause the problem, It's the massive DIFFERENCE between vcore and vdimm cause the problem. That's why there's an formula to calculate your lowest vcore from your vdimm.
again, he's unaffected by that since he isn't using some RAM like UTT.
v0dka
04-10-2006, 10:34 AM
if you tak e a look at the WIndows Task Manager in each shot, it tells you the activity thereof :stick:

The x2 3800 was idle (note the big *IDLE* :fact: messages in Prime)

And the Opty was in the middle of dual priming when the screenshot took place (again note the big *GO* :clap: or the Taskman showing full CPU usage history.

Let's make it real easy for you, since reading isnt your strong point. Being a smartass obviously is something you have mastered though - make sure you understand my question next time ok?

I'm talking about temps here, not wheter it's priming or not. The question was: "Is the temp 29C while priming in that second screenshot?"" Since you have no program open that shows temp other than a "29" in your sys tray which could aslo be your vid card for all I know. In the other you have no reference for temps so I'm asking about that one too. Not if it's priming or not...

People like you make reading forums so much more tiring.
ST4
04-10-2006, 11:00 AM
Let's make it real easy for you, since reading isnt your strong point. Being a smartass obviously is something you have mastered though - make sure you understand my question next time ok?

Stop jerking off and making yourself blind :slap: if you had any clue instead of talking out of your arse so much , you would easily see that 1) 2 LARGE SP2004 Prime programs open with either a GO or an IDLE 2) In between them are different temperature readings including a MB and CPU section depicting some number with a C after them. Hmm wonder what it could mean, huh? 3) There is a Task Manager open too to confirm any activities thereof. 4) Lastly there is a CPU-Z tabs open for the system frequencies

So from your ambiguous question, "Is the temp 29C while priming in that second screenshot", you can answer to your hearts desire the following: 1) What the 29C temp is 2) Whether prime is running 3) what frequency it is at, and 4) how much of a :banana: you are.

And just so you know that I'm trying to be helpful and not some smartass as you confess I am, I have enclosed the following pic to help guide you, appropriately named 4-the-nubs.jpg, so clean your hands and pull out your reading glasses :

http://i59.photobucket.com/albums/g319/stranx44/4-the-nubs.jpg
ST4
04-10-2006, 11:16 AM
ugh I guess you missed what I've said "+ Raise Vdimm"?:)

If you didn't raise Vdimm so what's the problem?

P.S. It's not that massive Vdimm cause the problem, It's the massive DIFFERENCE between vcore and vdimm cause the problem. That's why there's an formula to calculate your lowest vcore from your vdimm.

Again, I'd like to point out to you how CnQ works:

1) Locked FSB
2) Multiple P-State with different FIDs (Multipliers) and VIDs (VCORE)
3) Fixed VDIMM

So from your assertion, my 1.1V assigned through CnQ at stock 2.6V VDIMM will have issues because of delta-VcVd?!?! You might wanna tell AMD their designs are fubared! But seriously, please explain to me or point to me some explanation on why Cool N Quiet aka Power Now! is architectured this way if there was a caveat for delta-VcVd.

fyi - i have never been a proponent of raised VDIMM of more than +.05v-.1v MAX
Richteralan
04-10-2006, 12:18 PM
again, he's unaffected by that since he isn't using some RAM like UTT.

eh..
Which kind of RAM is not important. The important thing is huge difference between Vdimm and Vcore.

As I said UNLESS he's increasing vdimm and decreasing vcore then it will be chances that he will burn the cpu.

Of course he uses TCCD so this won't be a problem.
Richteralan
04-10-2006, 12:18 PM
Again, I'd like to point out to you how CnQ works:

1) Locked FSB
2) Multiple P-State with different FIDs (Multipliers) and VIDs (VCORE)
3) Fixed VDIMM

So from your assertion, my 1.1V assigned through CnQ at stock 2.6V VDIMM will have issues because of delta-VcVd?!?! You might wanna tell AMD their designs are fubared! But seriously, please explain to me or point to me some explanation on why Cool N Quiet aka Power Now! is architectured this way if there was a caveat for delta-VcVd.

fyi - i have never been a proponent of raised VDIMM of more than +.05v-.1v MAX

I guess you STILL didn't read my last post.

I give up:)
IFMU
04-10-2006, 02:40 PM
v0dka and st4, that is more than enough.

If I see either of you flaming anyone again you will recieve a 3 day vacation from the forums. There is no reason to flame to make a point.

Since obviouslly this thread cant be discussed like adults or even mature teenagers,

I am making it a Dead Thread

IFMU