Cedar Mill 661 is here

[Ross]

OK, I'll skip the fun story of me physically chasing the UPS truck this afternoon after missing the delivery attempt earlier in the morning :D

First, this was purchased from TigerDirect. Sadly, it's OEM and did NOT come in an Intel retail box with HSF, packing date info or any of that other good stuff.

Cedar Mill 661
3.6GHz
SL94V (http://processorfinder.intel.com/scripts/details.asp?sspec=sl94v) Malay
Default Vc @ 1.32V
No SpeedStep, no LockFree
No VT, but HT = back in action :)

http://i10.photobucket.com/albums/a125/boostedevo/661/661.jpg

So there's no confusion, this is what it is running on right now:
-P5WD2 Premium BIOS 606 (vMCH mod @ 1.70V actual, vDDR mod, NO vdroop mod)
- Single x700 Pro (x1600XTs returned for dissatisfaction!)
- OCZ 600W dual rail (38A single rail will be here Tues.)
- BIP2 w/ 2x Evercool 120mm aluminum fans, NexXxos XP block, MCP-350 pump

Idle temps are a tad warmer than the 930, but they seem to be getting lower since I initially installed it. I might have skimped a little on the AS5 application (dab in the middle method as usual for single cores). Idle temps started at 30/32C and are now boucing around the 28-30C range for idle (~73F ambient). I haven't done any load testing yet.

Keep mind no vdroop mod = ~.03 over BIOS setting and droop on load will likely be ~.04V under BIOS setting:

http://i10.photobucket.com/albums/a125/boostedevo/661/661default.gif

And just for the heck of it, here's the CM, Presler and Prescott lined up:

http://i10.photobucket.com/albums/a125/boostedevo/661/3cpus.jpg

OC results will be coming slowly throughout the evening ;)

[MikeyLikesItSI]

Subscribed :) lots of luck ross!

[gixxer420]

slowly,

be like d94 and crank it


j/k take ur time

[Joe Camel]

somebody's going to have a fun weekend!! :D

GOOD LUCK!!!

[DanIdentity]

Sweet. Good luck. :D

Which 38A PSU did you order?

[dylskee]

Can't wait bro. :) Can't unlock the multi? :( Definitely subscribed. Clock on. :beer:

[Ross]

Thanks guys. I AM going to take my here. I was nervous ever since I ordered it that it wouldn't clock well and now that it is here, I am almost afraid to find out :D

Dan, I opted for the Silverstone Zeus ST56ZF. I would've gone with the OCZ 520 because the 600W has done me right with the exception of being enough for that 930, but I wanted to find a good PSU with a shorter depth. It's actually like 38A "MAX", but like 42A "Peak" however they come up with that ;) It's the right size, all the right connectors (24-pin/8-pin), has pots for rail adjustments, is relatively cheap ($115) and built by Etasis...it can't not be all there.

I am going to do a little OC on the default Vc here and then walk the doggie and maybe eat dinner while it burns in a little bit more....my fingers as SOOO crossed...

[Sentential]

Good luck bro!

[Alchemy1]

Can't wait to see some results. Hope you got a good one:)

[xTrEmEoVrClOcKr]

This will kind of reflect my 641 .. minus 200mhz ;).

[proth]

This will kind of reflect my 641 .. minus 200mhz ;).

If he hits 5Ghz easily on H20m Tiger's stock will sell out in 15 minutes. Glad I'm in . Good Luck Ross....

[Sentential]

Any updates?

[dylskee]

Any updates?
:) :)

[MikeyLikesItSI]

walk the doggie and maybe eat dinner

Didn't you eat yesterday? and the dog can hold it!! :p

[proth]

there are 14 people logged onto the CPU forum and 10 of them are online to this thread....

Wonder if Surfrat powered up yet....Hope he's still not on the other thread

[Ross]

Mikey: HAHAHA

OK, first let me clear this up because I made a wrong ASSumption about the Vc I was seeing on default CPU-Z screenie. I had assumed that the ~1.35Vc CPU-Z was showing was the board doing it's usual .03V overvolt from what was set. When I opened AI Booster to do a quick OC from the desktop, the Vc parameter said 1.3625V for some odd reason (was NOT running that). I turned it down to what I thought was the default Vc (1.32Vc) and applied it. CPU-Z was showing me ~1.36-1.37V after that. I turned it down further to 1.30V and am now seeing the ~1.34V that CPU-Z was when Vc was set to AUTO.

So, it looks like this CM overvolts more than any other proc I've ever used in this board (Prescott/Presler) and the default voltage really is 1.29x-1.300V. Now that's straightened out, by the same token it so far looks like the CM does not undervolt as much on load either. Granted, I am not pushing anything hard yet, but I have yet to see it go below my 1.300Vc setting on load. So the overvolt = bad news and the droop looks like good news.

NOTE that I should have mentioned in the first post that I am keeping mem at a 1:1 ratio while I OC the proc!

On to the first OC, at stock Vc it is at 4320 and *stable*. I using PCMark04 (like you all should be) for quick and pretty accurate stability testing. I have run 5 PCMark04s in a row at this clock with without any restarts or crashes and received scores on all of them, so that's all good. Max load temps I've seen so far is 40.5C, but it's a little cooler now than when I originally installed it...it's idling around 27C ATM.

http://i10.photobucket.com/albums/a125/boostedevo/661/default_vc.gif

On the stock Vc and 4.5GHz, it will hard reboot the second I start PCMark04 and at 4.4GHz it will crash to the desktop on the first test, so 4.3GHz is it on default Vc. I was really, really hoping to get 4.4-4.5GHz at default Vc, but I guess I can live with 4.3GHz. I will now see just how much Vc is required to run 4.5GHz....if it stays under a 1.35Vc setting like I am hoping it will, I will still be pretty happy ;)

More coming as the night wears on...

Where's Surfrat and his 651???

[Sentential]

Hmmm, thats better than I expected, quite a bit better infact. I expected to see roughly 4.4 on average but it seems to be closer to the 4.6 mark depending on where your's hits a wall

Where's Surfrat and his 651???
He's having BIOS issues which I am attempting to help him with (he doesnt have a 65nm bios flashed to his board)

[Vengance_01]

Nice Clocks dude! What cooling? And any new updates :santa2:

[custom90gt]

sounds like it should be a fun little chip. I hope you hit 5ghz with some voltage.

[batboy]

I'm impressed, 4.3 gig on low voltage, nice.

[Alchemy1]

Looks good so far. I guess our hopes of 5ghz aren't crushed YET.

[proth]

Not bad.. 20% with no tinkering yet

[MikeyLikesItSI]

Well, even if it scales like his presler did, it should land him real close to 5ghz

[Ross]

Sen, I am hoping the end of the world is 4.8 at least with 1.50Vc. That's still +.20Vc from where I am now, so I really have my fingers crossed. In a couple of days of running, I won't make the same mistake as the Presler and WILL try 1.55Vc this time. I might reinstall the vdroop mod sometime this weekend, but to keep the idle overvolt under control rather than the load undervolt. At 1.55Vc it's so far looking like it will idle around 1.60V and I wouldn't like that one bit.

What board is Surf using? I hope you guys can get it running soon because I feel like everyone in the room is watching me :)

vengence, 1st post! Everyone else, thanks! Off to see about 4.5GHz stable here...

[crimedog]

no 5ghz stable? :p

hurry up and slam that chip with voltage

[dylskee]

Very nice so far! :thup: Mine won't be here till Wednesday! :bang head Looks promising though.

[proth]

Ross,
If you have time, pay attention as to the rate of temp increase and decrease, not the absolute temps, in comparison to Presler on the same cooling. If my hunch is right the .65um will react and counterreact to load temps a lot faster.

[Sentential]

If yours is maxing out at 4.3 you have maybe 300-500mhz before you will hit your absolute max even with voltage.

If you really want 5ghz my prediction is that it will take almost 1.55v to make that possible :(

[nikhsub1]

Waiting with baited breath :D

[Vengance_01]

At this stage I think D/C is the way to go if you are on Intel side or AMD sides, the newer D/Cs seem to scaling very nicely :)

[proth]

I really think Cedar Mill will not only scale similar to Prescott, an additional 90um/65um ratio should add in just for the die shrink. The fact that Prescott used repeaters to reduce wire delays and Cedar Mills has the same transistor count and same repeaters, should allow a higher OC. The other side of the coin is hot spots, especially the prefetch. It's packed in a lot tighter now.

[hUMANbEATbOX]

good stuff so far! *crosses fingers* i hope we see 5ghz sometime tonight (meaning before you switch to phase tomorrow) :D

[proth]

Check out the TigerDirect review for the 661. I musta ****ed off a lotta AMD folks with that one.

[dylskee]

Check out the TigerDirect review for the 661. I musta ****ed off a lotta AMD folks with that one.
LOL!!!! Man your BAAAAD! :D

[davekusa]

Has anyone hit 5Ghz (non 955)? I've been watching the different forums endlessly. Was thinking of building a new puter, But I think I can get 4Ghz+ with my current setup and some phase change.


Thanks everyone for posting their results so far. This has been great! (very informative)

[nikhsub1]

I hit 5Ghz on 875 :D but that is a different thread ;)

[Ross]

Next, 4.5GHz.

http://i10.photobucket.com/albums/a125/boostedevo/661/661_4500.gif

1.32Vc setting PCMark04 would crash to the desktop on first test
1.33Vc PCMark04 would finish with no score
1.35Vc PCMark04 completes with score 4x in a row

So I guess I am happy with that because the first wall I got to on the Presler took +.10Vc to get over. Unfortunately, it looks like 4.6GHz will take another Vc bump to be PCMark04 stable...how much I don't know. If there's a wall every 100-200MHz from here on in that takes .350-.500Vc extra to stabilize, I will be VERY lucky to crack off a 5GHz at tonight's limit of 1.50Vc, if that clock is doable at all.

Sen, yeah, I think it's gonna take 1.55Vc for 5GHz, which I am prepared to go to, but not tonight (easy there crimedog :)) and probably not until I get the idle overvolt issue resolved.

I think as the night wears on here, I will be cutting these multi PCMark04 runs down to 1 :D I am already way tired of looking at them and if it passes one, it's stable enough for me :)

These PC04 runs were the first time I saw load Vc drop below what was set, but not by much:
Set = 1.3500V
Idle = 1.39V
Load = 1.34V

Has anyone hit 5Ghz (non 955)?Yeah, for another thread, but d94 says there's a 920 on hardforums doing 5GHz on the stock cooler @ 1.50Vc :eek:

[Shelnutt2]

Check out the TigerDirect review for the 661. I musta ****ed off a lotta AMD folks with that one.

I'd buy one in a second, if that was true.

12Ghz....we will see if Ross getts there on phase.

*praise to the Intel Overclocking God*

[Ross]

LOL, nice review!!!

5GHz or not on water, this thing better take off like a rocket on phase (tomorrow) because the Presler failed me in that department :(

[proth]

Checked your ram speed and timing.?
It may help you understand things later after the dust settles, if you monitor and log temps.??

CedarMill to Prescott is like Harley to Riceburner

[davekusa]

good luck. Ross we're all cheering for you.

These threads will determin if I bite the bullet on builing new computer or not.

[Sentential]

The other side of the coin is hot spots, especially the prefetch. It's packed in a lot tighter now.
Sheesh, you're tellin me! The load delta of Presler is nothing short of incredible. Never have I seen a chip with a 25C increase in temps on load.

[proth]

Sheesh, you're tellin me! The load delta of Presler is nothing short of incredible. Never have I seen a chip with a 25C increase in temps on load.


CedarMill is much more thermally responsive, It has to be. Although absolute power has decreased, Power per square inch is the same. Cool this thing on phase and it should take off.

Ln2 would also be interesting....Hyper but may not last long, sorta like me

[dylskee]

Hey Ross, what Bios version are you running on the P5WD2-P? I'm running 519, should i flash to 606 before i install the 661?

[Sentential]

Cool this thing on phase and it should take off.
Yes and no. Ive had a long standing theory that (atleast in my mind) is gaining validity.

As these transisters keep getting smaller and smaller, phase will be counterproductive as it will turn them from conductors into super-conductors and thereby signifintally altering the chips design causing what people know as the "cold-bug" which at its root is a no POST caused by the inability to initiate the CPU at cold temps.

IMHO DSL sped that process up for AMD as SOI is already considered a superconductor. The fact that its starting to crop up on Intel now that is is on 65nm kinda proves my point (atleast to me)

[proth]

Yes and no. Ive had a long standing theory that (atleast in my mind) is gaining validity.

As these transisters keep getting smaller and smaller, phase will be counterproductive as it will turn them from conductors into super-conductors and thereby signifintally altering the chips design causing what people know as the "cold-bug" which at its root is a no POST caused by the inability to initiate the CPU at cold temps.
My knowledge of OC slang is still young, and thought the "cold bug" was a myth, If it does indeed exist, I could see your theory explaining it.

IMHO DSL sped that process up for AMD as SOI is already considered a superconductor. The fact that its starting to crop up on Intel now that is is on 65nm kinda proves my point (atleast to me)
Now that statemnt interests me...Any links to such topics would be appreciated.


Maybe Surfrat has posted.?

[Sentential]

My knowledge of OC slang is still young, and thought the "cold bug" was a myth, If it does indeed exist, I could see your theory explaining it.


Now that statemnt interests me...Any links to such topics would be appreciated.


Maybe Surfrat has posted.?
I dont want to continue to threadcrap so Im going to quikly summarize and we will take this in another thread / PM.

SOI and the following processes like it use a super-conducter between gates of each transister to reduce electrical resistence (and it works substantially well) and it reduces the charge needed to activate each gate thereby reducing needed vcore for operation.

[d94]

I think as the night wears on here, I will be cutting these multi PCMark04 runs down to 1 :D I am already way tired of looking at them and if it passes one, it's stable enough for me :)

These PC04 runs were the first time I saw load Vc drop below what was set, but not by much:
Set = 1.3500V
Idle = 1.39V
Load = 1.34V

Yeah, for another thread, but d94 says there's a 920 on hardforums doing 5GHz on the stock cooler @ 1.50Vc :eek:
very, very nice results my friend :), ill certainly be following this thread :D
btw, the cat on the other forum ~ his chip is a 930, not a 20 :) and he claims to of run superpi 1mb multiple times @ that speed :shrug:

[proth]

5 bucks says Ross is hooking up phase

[xTrEmEoVrClOcKr]

LOL ... thats what I was thinking ;).

[TechtonicPC]

Sen, I am hoping the end of the world is 4.8 at least with 1.50Vc. That's still +.20Vc from where I am now, so I really have my fingers crossed. In a couple of days of running, I won't make the same mistake as the Presler and WILL try 1.55Vc this time. I might reinstall the vdroop mod sometime this weekend, but to keep the idle overvolt under control rather than the load undervolt. At 1.55Vc it's so far looking like it will idle around 1.60V and I wouldn't like that one bit.

What board is Surf using? I hope you guys can get it running soon because I feel like everyone in the room is watching me :)

vengence, 1st post! Everyone else, thanks! Off to see about 4.5GHz stable here...

Yea I am dead in the water! I am using a P5WD2-Premium, but have the old 0509 bios flashed in. We ahve tried everything including hotflashing it and no luck!! Only way I am gonan get this thing runnign is if I can get my hands on another lga socket chip, (ie 630 or 820, etc..)

To top off my problems for the night, my sunbeam FSC just blew up and actually sparked a flame! No idea what that was all about, but everyhtign else seems to be ok in the system!

Me not a happy camper!

[Ross]

LOL, crap in this thread all you want, I don't care at this point :santa:

dylskee, 606, but after I max out this OC, I am gonna go back to 519 since I no longer have CF setup.

d94, which one was it?

NO PHASE...water is working JUST FINE for today.

OK, I figured I'd spare you the play-by-play 100MHz increases and just make notes of them...

4.6GHz completes PCMark04 with score @ 1.3875Vc set.
1.43V idle
1.38V load
27C idle
44C load

4.7GHz completes PCMark04 with score @ 1.4250Vc set.
1.46 idle
1.41 load
27C idle
45C load

4.86GHz completes PCMark04 with score @ 1.475Vc set.
1.51 idle
1.45 load
27C idle
48C load

http://i10.photobucket.com/albums/a125/boostedevo/661/661_pc04_4868.gif

http://i10.photobucket.com/albums/a125/boostedevo/661/661_4868.gif

And just a little sneak peak at what I will be working on stabilizing a little later. Note the same 1.475Vc (1.51V idle). It's not stable there yet as I only did a quick check to see if it was even in the cards...I haven't even had a 1.50Vc setting....YET.
http://i10.photobucket.com/albums/a125/boostedevo/661/661_sneak.gif

I have some stuff to do, so I will be off the OCing for a little bit. Use that time to dry yourselves off, get out your nasty work shoes/ketchup/webcam or whatever :D

If this thing scales worth a damn on phase, it will be painfully fast. I think it was definitely worth looking like an arse chasing the UPS truck down the street today :D

I would have a woody for the next 6 months if this proc had 12x LockFree on it...that would just be the ultimate and this thing on water would eat my 670 on phase for breakfast, lunch and dinner. If/when they release these in Q2 with EIST (and hopefully LockFree), I will be all over one of them.

OK, I'll be back in a little bit to see what's up and then I'll get back to stabilizing that clock!

[d94]

LOL, crap in this thread all you want, I don't care at this point :santa:

dylskee, 606, but after I max out this OC, I am gonna go back to 519 since I no longer have CF setup.

d94, which one was it?

NO PHASE...water is working JUST FINE for today.

OK, I figured I'd spare you the play-by-play 100MHz increases and just make notes of them...

4.6GHz completes PCMark04 with score @ 1.3875Vc set.
1.43V idle
1.38V load
27C idle
44C load

4.7GHz completes PCMark04 with score @ 1.4250Vc set.
1.46 idle
1.41 load
27C idle
45C load

4.86GHz completes PCMark04 with score @ 1.475Vc set.
1.51 idle
1.45 load
27C idle
48C load

http://i10.photobucket.com/albums/a125/boostedevo/661/661_pc04_4868.gif

http://i10.photobucket.com/albums/a125/boostedevo/661/661_4868.gif

And just a little sneak peak at what I will be working on stabilizing a little later. Note the same 1.475Vc (1.51V idle). It's not stable there yet as I only did a quick check to see if it was even in the cards...I haven't even had a 1.50Vc setting....YET.
http://i10.photobucket.com/albums/a125/boostedevo/661/661_sneak.gif

I have some stuff to do, so I will be off the OCing for a little bit. Use that time to dry yourselves off, get out your nasty work shoes/ketchup/webcam or whatever :D

If this thing scales worth a damn on phase, it will be painfully fast. I think it was definitely worth looking like an arse chasing the UPS truck down the street today :D

I would have a woody for the next 6 months if this proc had 12x LockFree on it...that would just be the ultimate and this thing on water would eat my 670 on phase for breakfast, lunch and dinner. If/when they release these in Q2 with EIST (and hopefully LockFree), I will be all over one of them.

OK, I'll be back in a little bit to see what's up and then I'll get back to stabilizing that clock!
WOOOT WOOT!!!! THAT LOOKS GOOOOD
if you can hit 5ghz stable..ill be in AWE, not that im already not :D

[Ross]

Yea I am dead in the water! I am using a P5WD2-Premium, but have the old 0509 bios flashed in. We ahve tried everything including hotflashing it and no luck!! Only way I am gonan get this thing runnign is if I can get my hands on another lga socket chip, (ie 630 or 820, etc..)You did the ALT+F2 during boot and it wouldn't take a flash from that? I have like 3 Prescotts here for you to borrow :) OMG, find someone who's got a 775 proc and get on it!!

[TechtonicPC]

Hey Ross, what Bios version are you running on the P5WD2-P? I'm running 519, should i flash to 606 before i install the 661?

YES YES YES!! You need 0606! save yourself lots of hassle!!

[TechtonicPC]

You did the ALT+F2 during boot and it wouldn't take a flash from that? I have like 3 Prescotts here for you to borrow :) OMG, find someone who's got a 775 proc and get on it!!

Yea with no post, I dont even ahve keyboard control, so nothing is gonna work short of Sen delivering me a chip in the morning!! (I am begging him at this point)!!

Your results are lookin great btw man!! 5ghz here we come!!

[Ross]

d94, if I have to go to 1.50-1.52Vc setting I will. I am just really concerned about that idle overvolt...it's huge and 1.52V setting will be 1.57V actual already. I'll give it a shot and see if something in that range is enough to finish a PCMark04, but I won't go higher than that. If up to 1.52V setting is not enough, I will be forced to pull the board out and reinstall the vdroop mod to handle that a little better.

I am way happy to see the 5GHz screenie and I at least know that 4.9GHz will be PCMark04 stable without much work (if any). It just finished the 4868 bench and I thought the hell with it, that's close enough to see a 5 :)

Surf, 519 doesn't recognize CM???? Dang, it even does OK with Presler. OK, I'll keep 606 in then. Man, I hope you find a proc to borrow. One of your neighbors must have a Dell you can raid for 5 minutes :) Good luck and I really look forward to your result on that thing!

[TechtonicPC]

d94, if I have to go to 1.50-1.52Vc setting I will. I am just really concerned about that idle overvolt...it's huge and 1.52V setting will be 1.57V actual already. I'll give it a shot and see if something in that range is enough to finish a PCMark04, but I won't go higher than that. If up to 1.52V setting is not enough, I will be forced to pull the board out and reinstall the vdroop mod to handle that a little better.

I am way happy to see the 5GHz screenie and I at least know that 4.9GHz will be PCMark04 stable without much work (if any). It just finished the 4868 bench and I thought the hell with it, that's close enough to see a 5 :)

Surf, 519 doesn't recognize CM???? Dang, it even does OK with Presler. OK, I'll keep 606 in then. Man, I hope you find a proc to borrow. One of your neighbors must have a Dell you can raid for 5 minutes :) Good luck and I really look forward to your result on that thing!

Well it is possible 0519 will work, Asus told me no it wont, but I have 0509, and That one deffinately dosnt work!! :( Right now I am waiting for Sent to get back to me in the morning. If he has time he is gonna coem down with his 640! If not im not sure what I am gonna do!

[dylskee]

YES YES YES!! You need 0606! save yourself lots of hassle!!
I'm going to do that right now, thanks for the heads up man.

[batboy]

WOOOT WOOT!!!! THAT LOOKS GOOOOD
if you can hit 5ghz stable..ill be in AWE, not that im already not :D

Apparently you didn't catch his little sneaky hint.

http://i10.photobucket.com/albums/a125/boostedevo/661/661_sneak.gif

Looks like he already hit 5 gig.

[proth]

d94, if I have to go to 1.50-1.52Vc setting I will. I am just really concerned about that idle overvolt...it's huge and 1.52V setting will be 1.57V actual already. I'll give it a shot and see if something in that range is enough to finish a PCMark04, but I won't go higher than that. If up to 1.52V setting is not enough, I will be forced to pull the board out and reinstall the vdroop mod to handle that a little better.

I wouldn't take it past 1.55 without the vdroop mod, You'll blow that thing out of the socket when it comes off load. IMO

EDIT: Fantastic clocks so far. Thanks for sharing

[CoreGamer]

Apparently you didn't catch his little sneaky hint.

http://i10.photobucket.com/albums/a125/boostedevo/661/661_sneak.gif

Looks like he already hit 5 gig.
Im sure he caught it, just make sure you take a look again to the part where it says its is NOT stable atm :( Of course he has plenty of Vcore left (or does he?) so I think 5 Ghz will be hit and he *might* be able to go further

[xTrEmEoVrClOcKr]

Well ... 4.8ghz is still impressive ... nice Ross ... will a CedarMill work on the P5WD2 non premium w/ latest bios? 0301?

[TechtonicPC]

Well ... 4.8ghz is still impressive ... nice Ross ... will a CedarMill work on the P5WD2 non premium w/ latest bios? 0301?

I would assume so, they are essentially the same board, minus some of the extras! Asus support is down at the moment or I would verify it!

[proth]

Well ... 4.8ghz is still impressive ... nice Ross ... will a CedarMill work on the P5WD2 non premium w/ latest bios? 0301?


There's no stopping going for 5GHz on water after hitting 4.8.
This ASUS BS Core fluctuation really pizz me off. The had 8 phase power on the WS board and should have carried it on all P5WD2 variants.

PS: Nice sig. Wonder if Crimedog likes fillet of sole

[Ross]

Yes, it is not stable @ 5GHz *for PCMark04* @ 1.475Vc...it may very well run a SPi there though. I have no problem going to 1.55Vc setting if it weren't for the massive overvolt at idle. The only way I can do that is by reinstalling the vdroop mod, which I will do. If it doesn't complete a PCMark04 at 1.50-1.52Vc setting, I may very well pull the board out and do it tonight.

I want to see a solid 5GHz PCMark04 and like a 5.2+GHz suicide on water before I put the phase on tomorrow....and I would definitely reinstall the vdroop mod before that.

XOCer, I really have no idea on the non-Prem. If the BIOS was released within the last month or 1.5 months, it *probably* will. I didn't know that Surf was using a BIOS that old. 519 worked on the Presler (previous one to the 606 for the Prem) and I would imagine will work on the CM.

[xTrEmEoVrClOcKr]

I said id keep his word for it ... ;).

[xTrEmEoVrClOcKr]

Ok good, if not I will try to force flash my Non Premium to the Premium.

[proth]

Ross, Are you testing with 1:2 memory @ DDR1066 setting.?

That should make 'er fly.?

[RangerXLT8]

Just tuning in here Ross, looking good so far, good luck.

[xTrEmEoVrClOcKr]

Damn't .. Wish my CM would get here, NOW. My phase is due to come back next week ... cannot wait to test my 641 on phase :D.

[Alchemy1]

The CM's look promising!

[Ross]

OK guys, sorry I was gone so long, but it was walking a real fine line between needing more Vc, but idle overvolt being higher than I'd like. I tried everything in my power besides Vc to get it to do, but no luck...it's right on the brink. So this is how it's going to end for the night....

It looks like stable solid 5GHz for PCMark04 needs just a hair more Vc than I am willing to run right now. 1.525Vc setting, I've seen idle as high as 1.57Vc, so there's no way I am trying any higher.

At 1.525Vc, 4914 (273FSB) passes WITH score, 4950 (275FSB) finishes with NO score and 5040 (280FSB) crashes to the desktop. SOOOO close. It probably needs that full 1.55Vc setting which I can't give without running into the 1.60Vc idle range :( Vdroop is now really rearing it's ugly head and I all of sudden remember why I did it on the 670 > 5GHz to begin with :)

http://i10.photobucket.com/albums/a125/boostedevo/661/661_pc04_4900.gif

4914MHz (273FSB) 1.525Vc
1.56Vc idle
1.48 load
28C idle
50C load

And for no reason at all, SPi 1M @ 5040 because that would finish. I am really missing the 14x multi and FSB that the LockFree gave on the 670, but it's OK I guess considering it's 5GHz on water!!!!

http://i10.photobucket.com/albums/a125/boostedevo/661/spi_5GHz.gif

"Suicide shot" if you want to call it that with 1.525Vc. An even 290FSB would lock up.

http://i10.photobucket.com/albums/a125/boostedevo/661/661_5200.gif

I don't think I am going to get around to redoing the Vc droop mod tonight and this is it for OCing anymore tonight also. I'll be sleeping in late tomorrow and will do the droop mod first thing. I WILL put it back on water to get a solid 5GHz PCMark04 before I switch to phase.

If Presler did this well, it would be downright nasty.

I'll be going to bed happy tonight :)

@ proth, no, I've been running 1:1 all night for OCing the proc and just upped the ratio/timings for the SPi 1M bench above which was the last thing I did.

[d94]

OK guys, sorry I was gone so long, but it was walking a real fine line between needing more Vc, but idle overvolt being higher than I'd like. I tried everything in my power besides Vc to get it to do, but no luck...it's right on the brink. So this is how it's going to end for the night....

It looks like stable solid 5GHz for PCMark04 needs just a hair more Vc than I am willing to run right now. 1.525Vc setting, I've seen idle as high as 1.57Vc, so there's no way I am trying any higher.

At 1.525Vc, 4914 (273FSB) passes WITH score, 4950 (275FSB) finishes with NO score and 5040 (280FSB) crashes to the desktop. SOOOO close. It probably needs that full 1.55Vc setting which I can't give without running into the 1.60Vc idle range :( Vdroop is now really rearing it's ugly head and I all of sudden remember why I did it on the 670 > 5GHz to begin with :)

http://i10.photobucket.com/albums/a125/boostedevo/661/661_pc04_4900.gif

4914MHz (273FSB) 1.525Vc
1.56Vc idle
1.48 load
28C idle
50C load

And for no reason at all, SPi 1M @ 5040 because that would finish. I am really missing the 14x multi and FSB that the LockFree gave on the 670, but it's OK I guess considering it's 5GHz on water!!!!

http://i10.photobucket.com/albums/a125/boostedevo/661/spi_5GHz.gif

"Suicide shot" if you want to call it that with 1.525Vc. An even 290FSB would lock up.

http://i10.photobucket.com/albums/a125/boostedevo/661/661_5200.gif

I don't think I am going to get around to redoing the Vc droop mod tonight and this is it for OCing anymore tonight also. I'll be sleeping in late tomorrow and will do the droop mod first thing. I WILL put it back on water to get a solid 5GHz PCMark04 before I switch to phase.

If Presler did this well, it would be downright nasty.

I'll be going to bed happy tonight :)

@ proth, no, I've been running 1:1 all night for OCing the proc and just upped the ratio/timings for the SPi 1M bench above which was the last thing I did.
:drool:

[proth]

I bought my 661 10 hours after you from the same place and hope it does as well..........Thanks for the stats, tomorrow will be good too

[Ross]

No problem. This thing is looking really, really sweet right now. If I wasn't so tired, I'd do the vdroop right now, but with my luck, I'd slip and fry the board :) It's gotta be 5GHz stable in PCMark04 with 1.55Vc solid, right? Dang, maybe I can get a 5.2GHz SPi 1M on water and a 5.3 validation :)

Load temps sure did get up around the 1.50Vc/4.9GHz mark by comparison, but 50C load for that kind of clock/Vc is impressive...might be a little warm on the Intel HSF though ;)

I hope you get yours tomorrow proth and I hope Surf has some luck getting his BIOS updated. Oh man, I totally forgot to go back and try 519 that OCed better for me! Tomorrow I guess...I am just chillin' for the rest of the night @ 4.86GHz :)

[hUMANbEATbOX]

glad to see you got a nice chip after the 930 gave you troubles.

i think someone has to eat a boot? :)

[xTrEmEoVrClOcKr]

This is all on wcing too .. thats ****ing amazing.

[Ross]

beatbox, it's all good...I won't hold him to it ;)

I just installed BIOS 0519 from 606 to see if it will OC any better. I was hoping LockFree would magically be available, but no such luck, LOL. Anyway 519 is working fine, no complaints from it at all...

[Sentential]

Very nice. As I predicted it will indeed take 1.55v for 5ghz, however it is atleast attainable for you :)

[CoreGamer]

We can only wait and see, I wonder how far somewone could push this on phase?

[bobad]

Ross,
If you have time, pay attention as to the rate of temp increase and decrease, not the absolute temps, in comparison to Presler on the same cooling. If my hunch is right the .65um will react and counterreact to load temps a lot faster.

You bring up an excellent point proth. As die sizes shrink, CPU's become less like a soldering iron and more like a light bulb. With such miniscule circuitry, temperature spikes could be almost instant. I doubt that temperature sensors can resolve every spike, and only show averages. One would think that as die size shrinks, every .01v becomes more dangerous.

[proth]

Can someone go and wake Ross up..... All Kidding aside take your time, after all, once you max it out on Phase then.......well then the honeymoon is over. Get all the data you need on H2o and don't let me pressure you for quick results....... ENJOY the good times and make them last.

My rig is not yet a rig and still considered "parts" at this point. By the time I get running, all potential of CedarMill will be revealed and then it's just a matter of luck for me.

Really hope others have similar results. There's a good chance ALL Cedar Mills will hit 5GHz, regardless of published speed. Seems like initial quality was excellent and maybe they actually binned then down to satisy the market.... After all, satisfying or pizzing off overclockers are not a high priority for Intel.

[Ross]

^^ LOL, I said I was sleeping in. Had I know I was going to stay up as late as I did, I woud've done the vdroop mod last night and kept going. Ah well, I'll pull the board out in a little bit and get to it.

You bring up an excellent point proth. As die sizes shrink, CPU's become less like a soldering iron and more like a light bulb. With such miniscule circuitry, temperature spikes could be almost instant. I doubt that temperature sensors can resolve every spike, and only show averages. One would think that as die size shrinks, every .01v becomes more dangerousI missed that when proth originally posted. Yes, it seems like can hit a high temp or cool off real quick. Idle is pretty steady, but where say opening Outlook used to bump the Prescott 2-4C or so, the CM goes up like 6-8C and they both of course immediately go back to idle. I really never paid attention if the Presler was that bad or not.

If it helps, on the PCMark04s I've been running on water it would go up to the near the max load temps for the whole bench on the first tests, but between some tests it's enough time to drop almost to idle temp. I don't know if the same will hold true for air or not.

[crimedog]

c'mon ross, we wanna see the good stuff! frozen cpu! 2v! 350mhz ram 3-2-2-4!

if you dont wanna torture the cpu send it to me and i'll pick up some dry ice :p

[speed bump]

Nice job ross good luck with the getting close to 6ghz on phase hopefully.

[d94]

ross, check it out ~ bios 704!
http://www.ocforums.com/showthread.php?p=4235191#post4235191

[Ross]

d94, I already downloaded it. Did you try it yet? This is my first boot since reinstalling the vdroop mod. I changed it around a little (added a 10K resistor to my 25K pot that ran 23K max) and need to adjust it, but so far at idle it looks the same as on the Prescott/Presler = solid .07V over setting which means I can set it for 1.48V and have 1.55 without going over/under.

BTW, nothing will get your blood moving like soldering on the mobo and then plugging the power switch into the wrong header accidentally (no boot)...LOL.

[bazx]

ross you have done a first rate job i have enjoyed the read

thanks dude :)

[Ross]

bazx, anytime man :) It "should" get better from this point on...phase will going on tonight :D I am gonna recheck my Vc settings with the vdroop mod reinstalled at previous OCs, maybe give the 704 BIOS a whirl, wait a little bit for the AS5 to set in and then go for a solid 5GHz PCMark04 w/score on water...hopefully 1.525-1.55Vc will get it now. After that, dinner and then the Chilly1 SS will be in action...

[proth]

ross you have done a first rate job i have enjoyed the read

thanks dude :)

You're not getting off that easy Bazx... When will we see CedarMills on Ln2.???

[proth]

OFF TOPIC:
950 Presler on eBay **** UNLOCKED **** Big Bucks !!!

http://cgi.ebay.com/NIB-Intel-Pentium-D-950-Dual-Core-2x2MB-LGA775-AWESOME_W0QQitemZ6839266392QQcategoryZ3674QQssPage NameZWDVWQQrdZ1QQcmdZViewItem

[Ross]

^^ I dunno if that's the same guy, but another was on there last week in a sealed box as well. Last time I saw was a day left and bid was $450.

speed/crime, sorry guys, missed your posts! Phase will be on tonight and I am hoping for, but not expecting 6GHz. If it will just do at least 5.5GHz stable, I'll be happy.

My 670 on the SS was only 100% stable in ALL benches @ 5.0GHz, granted that was running a lot more FSB. 5.4 just for SPi and 5.6 just for screenies. Of course, that has a lot to do with the capacity limit and not being about to run over 1.55Vc for more than 2 minutes ;) The OCZ 600W might have also been nearing it's limits on that rail.

Since this CM is already running stable on water at what the 670 was good for on phase, I am already way happy with it....just wish like hell it had a 12x LockFree :D

So far, the vdroop mod looks nice and stable. Exactly .07V over setting at idle as usual and it drops just .01V on load. I dunno if I am gonna try to tune that .01V out, but it's so minor probably not.

[proth]

Ross,
Do you think "official" 5GHz stability on H2O can be claimed at this point. ? If so please post idle and load temps along with Vcore. This way I can use this for a target. Thanks

EDIT: Just ordered a OCZ PowerStream 600 and Zalman 9500. P5WD2-e is still backordered... Case will be the motherboard box for a while.

[Ross]

Yes, it is safe to say that:

http://i10.photobucket.com/albums/a125/boostedevo/661/661_pc04_5000.gif

...but it's not easy. Passing a PCMark04 WITH a score is impossible if there are ANY problems at all, which is why I use it for quick and pretty darn accurate stability tests.

It took every bit of that 1.55Vc to complete with a score. I ran PCMark04 twice and it finished with scores both times.

Idle = 28 and load = a big 55C with some pretty cool ambients blowing through the rad (~20C). For OCing though, it's looking to me at these clocks (4.9-5GHz) that temps have much less to do with it for stability than Vc does. It will start and complete PCMark04 at 1.525, but only finishes with a score at 1.55Vc...no 2 ways about it.

Once you hit the 1.50+Vc mark, temps go up noticeably. There's a pretty big disparity between what it takes for 4.7-4.8GHz stable and 5GHz stable...on air, I think it will be really tough for stable 5GHz (unless I have a poor OCer? LOL), but 4.6-4.8GHz should be relatively easy (at least on a 661 ;).

I guess in short, if you are looking for 5GHz *stable*, that's probably about the limit on water. If you are looking for just a 5GHz screenshot, that probably won't too big of a deal. Of course, if we found out down the road that say 1.60Vc is about the safe upper limit instead of 1.50-1.55Vc, I'd add a good 200MHz to all these numbers I am tossing out if cooling/PSU allow :D

I also validated a solid 5.2GHz screenie @ 1.55Vc.

http://valid.x86-secret.com/cache/64131.png

Now that 5GHz stable has come and gone, I am going to give BIOS 704 a quick run through, make a fresh pot of coffee and then slap the evap on...

[Joe Camel]

Now that 5GHz stable has come and gone, I am going to give BIOS 704 a quick run through, make a fresh pot of coffee and then slap the evap on...
:attn::attn::attn:

BEST OF LUCK!!

ill put on some coffee too and wait for the results :)

[Ross]

LOL. OK, I guess I'll skip trying the new BIOS until I get phase going. If it can get to say 4.7-4.8GHz on the default Vc with it, that will be a pretty good sign :)

[proth]

Thanks Ross, It seems 5Ghz on H20 is in the books and you'll be the first to chill the CedarMill...Quite an honor...Good luck...Wish I could buy that coffee for ya

[xTrEmEoVrClOcKr]

Definitely, cannot wait for phase results!

[hUMANbEATbOX]

anxious over here!

hoping that phase treats you a whole lot better on the 661!

all this talk of an easy 4.5+ on air really makes me want to spend a good chunk of my student loan, but my gf would rightfully have my neck in a noose. :D

i'll have to wait for my tax refund or longer :(

[Ross]

I dunno if it's anything to be excited about yet, but we'll see what happens...

Phase is on and the first freakin thing I notice is that it's idling at the exact same -16C in BIOS and SF that the Presler did. Now there's either something weird about these cores with negative temps or something really got goofy with my phase unit, though I can't see what.

I was hoping for 4.7 on default Vc, but 4.6GHz is the highest with PCMark04 score. OK, not bad. It just passed a PCMark04 with score at 4950 with 1.40Vc actual. That's pretty big jump in Vc, but 4.9-5GHz is definitely a trouble area and took .15Vc more on water to be stable there.

http://i10.photobucket.com/albums/a125/boostedevo/661/661p_pc04_4950.gif

http://i10.photobucket.com/albums/a125/boostedevo/661/661p_cpuz_4950.gif

There should be plenty of Vc left for 5.5GHz+ unless it's a battle for every single FSB from this point on. What there may not be enough of is cooling capacity. I am not even considering 6GHz obtainable after seeing what it took on water when it gets close to the limit.

I will concentrate on 5.5GHz first and see how difficult that is and what the cooling looks like there before I even think about a "6" :)

[DvBoard]

looks good. too bad we don't see these kinda numbers with dual cores.

[dylskee]

Ok, i just put a fresh 12 pack of beer in the fridge and i'm watching my Patriots lose!(Bums) So i hope you give me some good news tonight. Looks pretty good so far.

[hUMANbEATbOX]

i am also watching them lose!

that's so weird about your temps being exactly the same. have you talked to anyone else who has experience with 9xx's on phase?

[Ross]

^^ No on the temps. Most people go by evap temps and my temp probe is in the dumper.

Not totally unexpected, but it looks like 5.3xx is going to be the highest PCMark04 w/score clock @ 1.55Vc. PCMark04 is brutal and very demanding to produce a score. I would've like 5.5, but 5.3 is OK. Nice news is that it was a pretty sick score for a single core :) Of course, I screen shotted my taskbar instead of the PCMark screen without realizing it and closed PCMark before I pasted and saw what was on the clipboard :rolleyes: So no screenie of that one (but I uploaded it to ORB). It was 7996. I will do a fresh reboot and make it an official > 8000.

To put that 5.3 max stable in PCMark04 in perspective, the 661 on phase does it about 400MHz higher than the 670 on phase did, so there's a noted improvement in the CM. In fact, the 670 on phase had it's hands full running SPi 1M @ 5.4GHz, which I don't think is going to be a problem here ;)

This proc just wants voltage >5GHz whether it's cool or really cool and seems to be acting pretty much like the Presler on phase, which didn't offer much more of an OC. It's fought me every inch of the way.

I'll rerun a fresh PCMark04 and make sure to screenie the correct window. After that, stability goes to the wind and we see how high SPi 1M will run :D

[hUMANbEATbOX]

seems to me that 65nm is a water coolers sweet spot. you can do well on air, you can do really nice on water, and phase isn't giving much more.

have you tried any suicide runs on phase yet?

[nikhsub1]

Interesting that phase is not really doing it... could be the 'headroom' I was referring to in a different thread. Man I wanted you to hit 6Ghz... when I get my new CPU (whatever it may be) I want to send you my 660 which I think would do well over 6Ghz on good phase.

[proth]

I wonder if the smaller die size is affecting the phase head. My knowledge of phase is very limited. Just that the heat is probably more concentrated now. Was your 670 idleing colder, it should have been hotter

EDIT: Actually I take that back and can understand why CM runs hotter on phase. The head cooling rating is by the square inch and the same heat is focused on a smaller area. What's needed is some type of adapter to focus the cooling capacity down to a smaller area

[Ross]

nik, your proc would be *insane* on phase!!

Yeah, I don't know if:
1) I am extremely impressed how well the CMs do on air/water or
2) I am extremely disappointed how poor they do on phase

Cooling might have some to do with it here, but I've not really seen that I've hit the capacity limit. I honestly believe it's all about voltage right now (and THEN capacity might come into play). Like I said in a previous post, it doesn't seem to matter much whether it's water or phase, once you hit that wall where 100-200MHz more OC take .05-.10Vc to be stable, you are nearing the end of the line. Granted, there are good results with real extreme cooling like LN2 or cascades, but those both run WAY colder than a single stage. Single stage probably just isn't quite enough to get it past that hump to the next level.

Anyway, as for suicide shots, I am not even going to get a 5.5GHz validation right now...5490 = auto reboot @ 1.55Vc, but I guarantee if I went to 1.60Vc we'd be looking at 5.4-5.5GHz PCMark04 stable, 5.6-5.7GHz SPi stable and MAYBE a 5.8GHz suicide.

What surprises me MOST is just how stable the Presler and CM are so close to the limit. My 670 would run 5.6GHz suicide sure, but it would only bench PCMark04 w/score @ 4950 = 650MHz diff. The CM on the other hand will do PCMark04 w/score @ 5324, but only suicide at 5450ish = 125MHz difference. Same with WC even. PCMark04 @ 5004 and suicide @ 5220 = real close.

Not only that, but considering 5490 = auto reboot, it will boot to the desktop at 5400. I mean, it's like this is the line and it works perfectly stable on one side and doesn't work at all on the other....and I am talking within a 100MHz window, that's why I am almost positive it's voltage and not cooling.

Maybe once I get a little more comfortable with it I'll shove 1.60Vc right up it's arse and get it moving :D

In the meantime, here's a PCMark04 screenie with a GREAT score for a POS vid card and no RAID. I uploaded it also, but it says Unknown processor, so I can't publish it until they catch up.

http://i10.photobucket.com/albums/a125/boostedevo/661/661_pc04_5318.gif

Consider this my suicide shot for now. That wouldn't be so funny if it didn't boot to run this bench right from BIOS, LOL. SPi 1M @ 5400. This is only ~.3 away from my 670 SPi 1M record @ 5424 that was using almost 90FSB more (14x multi)! Yes, the mem here was working its butt off :) DDR1000 @ 4-3-2-4...this Mushkin just rocks!

http://i10.photobucket.com/albums/a125/boostedevo/661/661_spi_5400.gif

[bazx]

stability near the edge is very good sign for me

all i need to see now is a 3D and im sold

[davekusa]

Ross, you might want to contact Chilly1 and relay your results. I think some engineering could be done for a block that that could work better for the presler/cm cpus.

[CoreGamer]

Wow these Cedar Mills are doing just as good as we thought! People stuck with air cooling should be loving this chip :D

[solofly]

Is this new chip compatible with 915P chipset?
Thx

[batboy]

Is this new chip compatible with 915P chipset?
Thx

No. According to the Intel website, you need i945, i955, or the new i975 chipset. Make sure you have an updated BIOS that supports Cedar Mill and Presler too.

[proth]

I just ordered a 631 for "experimenting". Will take Vcore as high as it goes before she pops. Emptied the closet and dug up a few hundred bucks of stuff ready for eBay to offset the cost of this "sacrificial lamb"

Will be in D.C. next week, parts and results should be known on Friday.

[hUMANbEATbOX]

No. According to the Intel website, you need i945, i955, or the new i975 chipset. Make sure you have an updated BIOS that supports Cedar Mill and Presler too.

apparently the p5p800-se, a 865pe board, now supports bother presler and cedarmill. oh and agp and ddr1. i know what i'm upgrading to. ;)

http://www.asus.com/products4.aspx?l1=3&l2=11&l3=0&model=629&modelmenu=1

[batboy]

I had heard that rumor too, but was waiting for actual confirmation. That would be sweet if it was true. But, I have heard nothing about the i915 supporting 'em.

[davekusa]

proth, keep us up to date. I was eyeing up the 631 also. Cedarmill seems to be the way to go for me

[bobad]

apparently the p5p800-se, a 865pe board, now supports bother presler and cedarmill. oh and agp and ddr1. i know what i'm upgrading to. ;)http://www.asus.com/products4.aspx?l1=3&l2=11&l3=0&model=629&modelmenu=1

I just bought one myself. That was the deciding factor as to why I chose that board.

[hUMANbEATbOX]

you gotta let me know how it does when you get it! very interested to know, i haven't found anything online yet. someone posted a link to an asrock 865pef i think, it was running a 955XE.

[Ross]

Ross, you might want to contact Chilly1 and relay your results. I think some engineering could be done for a block that that could work better for the presler/cm cpus.I honestly do not think it's the phase unit or the evap on the procs. I really think it's a -C temp reading problem on cores themselves. There's NO WAY both the Presler and CM happen to idle at the *exact* same temp on phase and that doesn't move between default clocks/default Vc and 4-4.5GHz and +.10-.15Vc. I wish my temp probe in the evap was working and then I'd know for sure :(

In any event, I have a new beefy Chilly SS in the works already :D

[d94]

awsome results ross :D
that pcmark score is only 1k away from mine..pretty damn good for a single core cpu!
so 1.55v is okay..think i can run that through my pressie? :D
maybe get that 5ghz screenie you wanted ;)

[hUMANbEATbOX]

so 1.55v is okay..think i can run that through my pressie? :D
maybe get that 5ghz screenie you wanted ;)

pressie is already reserved for prescotts :P

you have to call it a Presley.

[Ross]

If I wasn't such a wuss, I would've done it on the 930 no problem NOW. I didn't have a clue when it was installed and after I decided to try a CM (pretty much right away), I wasn't going to beat on it any further. I fully believe that 1.55Vc on my 930 would've easily have been 4.5-4.6GHz stable on water and 4.7-4.8GHz stable on phase...MAYBE a "real" 5GHz suicide.

There will be no repeat of wussiness on the 661. It's already been to 1.55Vc and I am very much thinking of doing what it takes for a solid 5.4-5.5GHz PCMark04 score and whatever a suicide shot will return there. From the looks of the difference between PCMark04 stable and suicide so far, that may only be +100MHz, LOL.

Edit: just for reference, the 930 at no point ever saw a high Vc than 1.50V and it was 4275 PCMark stable on water there.

[xTrEmEoVrClOcKr]

Im definitely pushing more than 1.55v through my 641, im going to take mobo cooling seriously, putting sinks etc. I dont wuss out when it comes to voltage .. my 650 was @ 1.76 idle / 1.71V load on a single stage LOL.

[dylskee]

I dont wuss out when it comes to voltage .. my 650 was @ 1.76 idle / 1.71V load on a single stage LOL.
Ouch! That's not going to help your sale of your 650.... :)
That's a buttload of voltage as far as i'm concearned.

[xTrEmEoVrClOcKr]

It was @ -C when it was under that voltage, no damage was done. It was only under 1.7v for like 2 minutes anyways lol.

[dylskee]

It was @ -C when it was under that voltage, no damage was done. It was only under 1.7v for like 2 minutes anyways lol.
I know, just messing with you. :) That's an awesome chip and i can't believe it's still for sale. I was considering selling mine on the classifieds but if your's isn't selling i'll probably go to ebay with it.

[xTrEmEoVrClOcKr]

Yeah, offer wont be around very long ..I'll give it a few more days. Is your 661 in your PC yet? Mine is coming on Wednesday :(, 3 more days!

[dylskee]

Yeah, offer wont be around very long ..I'll give it a few more days. Is your 661 in your PC yet? Mine is coming on Wednesday :(, 3 more days!
LOL, mine will also be here Wednesday... It feels like i ordered this thing a month ago!

[hUMANbEATbOX]

Yeah, offer wont be around very long ..I'll give it a few more days. Is your 661 in your PC yet? Mine is coming on Wednesday :(, 3 more days!

i saw your chip there, and i would FOR SURE buy it from you, if only i had the cash in hand. its going to be another 3 weeks at least before i can even consider buying anything. :(

[Ross]

1.70Vc setting is in NO WAY unheard on Prescotts. As been mentioned talking about Vc's for Presler/CM, many people have run >2.00Vc on Prescotts without blowing them up...mobos failing (mosfets overheating) was a more common issue there.

I certainly wouldn't run 1.70Vc+ on a Prescott for 24/7 use, but it should be more or less safe for a couple of hours of benching. Worst case scenario is that you take a day or 2 off the LONG lifespan of it ;)

[fordf250]

1.70Vc setting is in NO WAY unheard on Prescotts. As been mentioned talking about Vc's for Presler/CM, many people have run >2.00Vc on Prescotts without blowing them up...mobos failing (mosfets overheating) was a more common issue there.

I certainly wouldn't run 1.70Vc+ on a Prescott for 24/7 use, but it should be more or less safe for a couple of hours of benching. Worst case scenario is that you take a day or 2 off the LONG lifespan of it ;)

I ran my 640 at 1.65V on air @4.32G with 875P-T 24/7 for 5-6 months no prob, waiting for the p5wd2-e prem to see how far it will go.I could suicide 4.5G with 1.72Vcore still on air.Lots of droop (0.1V, droop mod never worked)and notice prime running in the bottom right corner to keep the multi at 16.It would only run a 29s superpi 1.4mod.

http://community.webshots.com/photo/312249557/312256144CHylAK

http://community.webshots.com/photo/312249557/312256371lLWMca

Edit: I just noticed I should have put this at the CM/Presler safe voltage thread.

[dylskee]

Damn, that's impressive fordf250! The highest voltage I've tried on my 650 is 1.45V with air cooling. I guess i'll see how much i can push my 661 when it gets here.

[speed bump]

Man fordf250 that is abit adventerous best I ever did with my 875p-t was 1.6v and that was on water.

Well Ross probably the worst case scenario ever was SNDS but the prescotts seem to have overcame that obstacle.

[GearShift3r]

Try running 1:1 and let that PC2-8000 scream. :p

Nice clocks!

-GearShift3r

[Ross]

Well Ross probably the worst case scenario ever was SNDS but the prescotts seem to have overcame that obstacle.No doubt. I think it would take insane amount of Vc (more than the board can give) to kill one of these immediately. I am going to get a little braver with my Vc, but I might be on the border of capacity and PSU now. I wish I orderd the new PSU a day earlier :(

[Ross]

I went and got my balls out of my wife's purse :D

Just went for something more akin to a real "suicide shot" here and validated it.

http://valid.x86-secret.com/cache/64430.png

An even 5.7GHz = auto reboot at this Vc (1.58V) I'll see about getting that 5.4GHz PCMark04 score, but cooling capacity and/or the dual rail might start coming into play on long benches now...

This thing also just put the smack down my best 670 SPi 1M with just 70MHz more and 70FSB LESS!!! I am not going to submit to the SPi section yet "just in case" I can better it. If there's a better ad for the Mushkin XP6400 than this screenshot, I'd like to see it :)

http://i10.photobucket.com/albums/a125/boostedevo/661/661_spi_5500.gif

[d94]

I went and got my balls out of my wife's purse :D

Just went for something more akin to a real "suicide shot" here and validated it.

http://valid.x86-secret.com/cache/64430.png

An even 5.7GHz = auto reboot at this Vc (1.58V) I'll see about getting that 5.4GHz PCMark04 score, but cooling capacity and/or the dual rail might start coming into play on long benches now...

This thing also just put the smack down my best 670 SPi 1M with just 70MHz more and 70FSB LESS!!! I am not going to submit to the SPi section yet "just in case" I can better it. If there's a better ad for the Mushkin XP6400 than this screenshot, I'd like to see it :)

http://i10.photobucket.com/albums/a125/boostedevo/661/661_spi_5500.gif
:drool:
is that phase or h2o?

[Ross]

That's on phase man...the CM's aren't *that* good :D

I think the OCZ 600W is nearing it's limit on that CPU rail. The new PSU is still 1 day away, so I think I will wait for that and try a little higher for a 5.7-5.8GHz screenie. I will go back to water at some point and see what the higher Vc's I am braving now can do there too ;)

[hUMANbEATbOX]

hey ross, in the general section, there is a thread at the top about dual cores..

you should do the divx conversion that batboy posted there (don't know what thread it came from originally)...i would like to see how your chip at say 5.4ghz compares to a dual core at say 4.4ghz.

[pscout]

Ross, What kind of volts are you feeding the mushkin?

It turns out i have some languishing in another rig i have p5wd2-p with a 551 ... I haven't done much with it but I don't seem to have the mem timings low like your or the speed as high.

Since my 920 is producing slower (folding) than my 830 at higher fsb i am paying some attention to the mem . Maybe i should swap the muskin for the corsair 5400c4 that is with my 920.?

- Peter

[Ross]

beatbox, I'll check it out...

pscout, exactly 2.456V. No I am not kidding, I measured them today :) The DDR1100 pull in my sig took a wee bit more.

[RangerXLT8]

beatbox, I'll check it out...

pscout, exactly 2.456V. No I am not kidding, I measured them today :) The DDR1100 pull in my sig took a wee bit more.

I to have run maybe 2.4-2.5v through my mushking 6400 2gb kit(for a short time, but ross claims his handles it day in day out). Ross was talking crazy numbers earlier in a PM, numbers that are enough to melt ICs. lol

[pscout]

:eek: Thanks ... i just fed my corsair 2.2 ... it was at 2.1 for most of my high fsb runs.

maybe it will need to get some more ... i am trying to get the 920 to beat my folding throughput on the 830/p5ld2 ... when i put the 830 back in after the 920 tests i upped the speed from 4:5 to 3:4 and its folding perf went up between 5-10 % ... so even though it is only at 265 it is beating the 920 at 295.

Thanks for all the work you have done posting your presler and cm OC's ... made it much faster for me on the 920. :)
:beer:

[pscout]

I to have run maybe 2.4-2.5v through my mushking 6400 2gb kit(for a short time, but ross claims his handles it day in day out). Ross was talking crazy numbers earlier in a PM, numbers that are enough to melt ICs. lol

Thanks Ranger.

I guess i was brave enough to put 1.525 to the 920 ... so whats a little more vd to worry about! :)

[Ross]

I to have run maybe 2.4-2.5v through my mushking 6400 2gb kit(for a short time, but ross claims his handles it day in day out). Ross was talking crazy numbers earlier in a PM, numbers that are enough to melt ICs. lolLOL, those numbers in the PM were LESS than I've seen run through them by others. Where's Sen to back me up??! Yeah, they've pretty much been on 2.40-2.50V max ever since I've owned them (6-8 months?). At that Vddr, they'll run DDR1000-1020 max with the tight timings. I posted in a thread somewhere all my results. Let me dig...


|------------------------------------------------------|
| Timings |- 2.227 Vddr -|- 2.476 Vddr -|- 2.597 V ddr |
|---------|--------------|--------------|--------------|
| 3-2-2-4 | DDR675 (4:5) | DDR760 (3:4) | DDR773 (3:4) |
| 3-2-3-4 | DDR715 (3:4) | DDR780 (2:3) | DDR795 (2:3) |
| 4-2-3-4 | DDR917 (3:5) | DDR1000(1:2) | DDR1040(1:2) |
| 5-3-3-4 | DDR1000(1:2) | DDR1060(1:2) | DDR1080(1:2) |
| 5-3-3-8 | DDR1000(1:2) | DDR1060(1:2) | DDR1080(1:2) |
|------------------------------------------------------|

The DDR1100 pull took a BUNCH more though...probably the toughest 10MHz I ever got, LOL.

[pscout]

Ok ... that settles it ... i will get mine workin harder ! but they (and I ) will get the rest of tonight off!

G'night gents. :)

[Ross]

you should do the divx conversion that batboy posted thereI ran it at 5.3GHz...wasn't sure what kind of load it was going to be. I have no idea what I am looking at for the "result", but my FPS was much higher than anything in that thread? I posted there, but the result was: FPS 110.4(x3.68). If that's "good", let me know and I'll see if it will fly @ 5.4-5.5GHz...seemed to run pretty easily at 5.3GHz. Of course, being PCMark04 passable will be easy for anything else ;)

[greenmaji]

ahh.. it was to see how single vs. dual core stacked up..
I kinda thought that was the idea (its my thread and I was talking about dual cores.. its even in the title of the thread.. :/ ) but if its to sway me tword a single core it will take alot more then that. LOL
I might be asking about memory later guys (since you guys have some killer DDR2 clocks)
good luck with the cedarmill :D

[speed bump]

I used to think 2.4v was insane then I heard the stilt at XS talking about 2.8 or 2.9v:eek: through a set of corsair 8000uls.

Anyway ross is the droop mod coming back on and when is that new PSU getting here so we can see 1.65v:)

[Ross]

ahh.. it was to see how single vs. dual core stacked up..Ah, yeah, that converter app has got to be single threaded because my FPS were way higher than anyone else's and I just have a big clock (LOL) with the same or less FSB as everyone else in there. If it were multi-threaded, I would NEVER be able to keep up with a dual core that even had much lower of a clock ;)

I used to think 2.4v was insane then I heard the stilt at XS talking about 2.8 or 2.9v through a set of corsair 8000uls.

Anyway ross is the droop mod coming back on and when is that new PSU getting here so we can see 1.65vBlah, the DDR1100 took like 2.8V. There I said it :) Actually, it might have been 2.75-2.80...I'd have to dig through and see if posted what it was somewhere. In any event, it was up there. 6-8 months of 2.4-2.5V straight after that and it shows no signs of getting soft!

I reinstalled the droop yesterday...before I even put the phase on. I needed 1.55Vc to get that 5GHz PCMark04 score on water and the vdroop mod was the only way of getting it without overvolting to 1.60V @ idle.

One step at a time, LOL. I've got my eye on 1.60Vc for sure...I'll need to see what happens there and then have a few drinks before thinking about 1.65Vc :beer: PSU will be here Tues.

[hUMANbEATbOX]

thanks for running that ross. i was more curious than anything, just wanted to see how quick something at 5+ghz could run that test.

i wonder how your 661 at 4.4 would compare to batboys. i know its single threaded, but i wonder if the extra cache helps him along much (or would it have any impact at all). hmmm...

[jmsandrsn]

I'm excited. UPS just delivered my P4 641. I set it straight to 266FSB (4.26GHZ) using default voltage (1.30v). It is somewhat early but she's in XP and no problems so far. I'm running Prime95 and threw a round of 3DMark01 at her. It passed 3DMark01 just fine and has run about 3 instances of Prime so far (large FFT's).

Darn it...I have to go to work in a few mintues. More fun will be coming after Midnight when I get home.

I'm using a somewhat crappy Gigabyte 945P motherboard and a retail Intel HSF (with ceramique rather than thermal pad). With the case open and sitting on it's side, the temperature under load is maxing out around 102-104F but this is only after 10 minutes or so. The voltage on this baord droops pretty bad so even though it is set at 1.30v it's only at about 1.21v under load as measured by Easy Tune. In comparison, a P4 630 @ 4.0GHz droops the voltage anywehre from 0.1 - 0.12v and temperature under load is around 125-130F(with same Intel HSF with ceramique).

I'm impressed so far. I think I'll leave her at 266 FSB while at work and see how she does. Sorry...but no more results until late tongiht.

PS. It does not appear to have the problem with flatness that Surfrat's did but then again I haven't pulled it off yet after the first mounting.

[hUMANbEATbOX]

i'm sorry, but fahrenheit confuses me. :D

Celcius is more commonly used for computer temps.

[jmsandrsn]

Temperature is now around 107-109F (42-43C).

Here is a quick screenshot.

http://img.villagephotos.com/p/2003-7/271271/P46414.26GHz.JPG

[hUMANbEATbOX]

Temperature is now around 107-109F (42-43C).

Here is a quick screenshot.

http://img.villagephotos.com/p/2003-7/271271/P46414.26GHz.JPG


holy cow! how stable is she at that speed/vcore?

very very nice!

[bobad]

Temperature is now around 107-109F (42-43C).

Here is a quick screenshot.

http://img.villagephotos.com/p/2003-7/271271/P46414.26GHz.JPG

Better be careful!

I think your temp is incorrect. Your load temps should be at least 50C under load even with premium air cooling.

[Joe Camel]

ehh, @ 1.2 Vc i can believe fairly low temps would be possible...

[bobad]

ehh, @ 1.2 Vc i can believe fairly low temps would be possible...

Oh yes, I agree it's possible. But,,, it would be the first time I ever saw a Prescott of any kind running 43C under load on any voltage on any air cooling. It made me a little suspicious that the temperature sensor was not reporting correctly, but I could be wrong.

[hUMANbEATbOX]

it would be the first time I ever saw a Prescott of any kind running 43C under load on any voltage on any air cooling.

well that's because this is NOT a prescott of any kind. prescott = 90nm. presler and cedarmill = 65nm.

look around at the other threads involving 920's, you'll find many people are getting great results on air.

and i don't think he specified whether 43c is at idle or at load.

[jmsandrsn]

well that's because this is NOT a prescott of any kind. prescott = 90nm. presler and cedarmill = 65nm.

look around at the other threads involving 920's, you'll find many people are getting great results on air.

and i don't think he specified whether 43c is at idle or at load.

Those are actually load temps with dual Prime95. The temp measurement may be wrong but that's all I have to go by. That is using Gygabyte's program called Easy Tune.

In comparison, I've had a P4 630 @ 4.0GHz with about 1.35v (actual) in the same system with same heatsink and the temps were about 10C higher.

I'm at work now so I can only hope that Prime95 is still going strong. As of the time I had left it was going for about 30 minutes or so and no problems.

[hUMANbEATbOX]

ohh i always hated being at work while priming....getting home was such a letdown while i was trying for 3.6 stable. GL!

[bobad]

well that's because this is NOT a prescott of any kind. prescott = 90nm. presler and cedarmill = 65nm.

look around at the other threads involving 920's, you'll find many people are getting great results on air.

and i don't think he specified whether 43c is at idle or at load.

My bad. I should have said "any sub 130nm Intel chip". ***

He did say "under load",,, your bad. :)

Why don't you think it's a good idea to double check temperature sensing accuracy? That's all I was suggesting.

***If it turns out that CM and/or Presler can manage to O/C well on 1.2v and 43C on stock cooling under load, I'll run out and sell my cat and speed up my plans to buy one.

[hUMANbEATbOX]

i do believe Ross got to or over 4ghz with the stock cooler on his 930. i think it was the stock cooler, with a 120mm fan pointed in its direction.

oh and ya, i didn't see that about the load, but judging by others results, it doesn't surprise me one bit. he is using such low vcore!

[bobad]

i do believe Ross got to or over 4ghz with the stock cooler on his 930. i think it was the stock cooler, with a 120mm fan pointed in its direction.

oh and ya, i didn't see that about the load, but judging by others results, it doesn't surprise me one bit. he is using such low vcore!

Well, the ultra-low voltages went right over my head, so thanks for that!

Hooooooly schmoly! I see what all the buzz is about now! I'll be watching the CM/P threads hungrily for new posts. Make that greedily. :drool:

[Ross]

Yeah, right around 4GHz on the 930 and with a 120mm fan pointed towards the Intel HSF, but I would expect the CM to much higher. 661 is studpid fast, but some 631/641s step up hardcore (looking great so far), I'll grab a lower one just to get my FSB back :)

jm, good luck with it. Unless you hit a MAJOR wall, that will easily be a 5+GHz proc...

[proth]

Mikey: HAHAHA

OK, first let me clear this up because I made a wrong ASSumption about the Vc I was seeing on default CPU-Z screenie. I had assumed that the ~1.35Vc CPU-Z was showing was the board doing it's usual .03V overvolt from what was set. When I opened AI Booster to do a quick OC from the desktop, the Vc parameter said 1.3625V for some odd reason (was NOT running that). I turned it down to what I thought was the default Vc (1.32Vc) and applied it. CPU-Z was showing me ~1.36-1.37V after that. I turned it down further to 1.30V and am now seeing the ~1.34V that CPU-Z was when Vc was set to AUTO.

So, it looks like this CM overvolts more than any other proc I've ever used in this board (Prescott/Presler) and the default voltage really is 1.29x-1.300V. Now that's straightened out, by the same token it so far looks like the CM does not undervolt as much on load either. Granted, I am not pushing anything hard yet, but I have yet to see it go below my 1.300Vc setting on load. So the overvolt = bad news and the droop looks like good news.

NOTE that I should have mentioned in the first post that I am keeping mem at a 1:1 ratio while I OC the proc!

On to the first OC, at stock Vc it is at 4320 and *stable*. I using PCMark04 (like you all should be) for quick and pretty accurate stability testing. I have run 5 PCMark04s in a row at this clock with without any restarts or crashes and received scores on all of them, so that's all good. Max load temps I've seen so far is 40.5C, but it's a little cooler now than when I originally installed it...it's idling around 27C ATM.

http://i10.photobucket.com/albums/a125/boostedevo/661/default_vc.gif

On the stock Vc and 4.5GHz, it will hard reboot the second I start PCMark04 and at 4.4GHz it will crash to the desktop on the first test, so 4.3GHz is it on default Vc. I was really, really hoping to get 4.4-4.5GHz at default Vc, but I guess I can live with 4.3GHz. I will now see just how much Vc is required to run 4.5GHz....if it stays under a 1.35Vc setting like I am hoping it will, I will still be pretty happy ;)

More coming as the night wears on...

Where's Surfrat and his 651???


Ross, What type of HSF did you run for this shot?. My temps are signifigantly higher on the 661.

[Ross]

The only air cooler I ever used on it was the stock Intel HSF that came with it....with the stock Intel TIM stuff too (not AS5). At some point I put a 120mm blowing on that HSF though because it was getting a tad warm.

If yours is a lot warmer, make sure that it is seated properly. It could be just the reading on the board or the ambients too...

[proth]

I was running 50c with the stock HSF and AS5, switched to Zalman with AS5 and only got 2c cooler. Something is wrong because the Zalman dropped my other rig by 15c but with some other TIM. This is my first time working with AS5 and maybe I applied it wrong, It seems a lot thicker than others, applied it about 1/2 the size of a peanut....um, I mean 1/2 the size of a piece of rice. The "spot" on the IHS looked perfect though.

Will try the Scythe tonight along with loading XP SP2......Let the games begin

Thanks