Core i7-965, Rampage 2 Extreme & 4870 X2 CrossFireX

Hi,

I did some first LN2 testing at home with Core i7-965, Asus Rampage 2 Extreme (retail) and 2 x Asus EAH4870 X2 TOP:

http://plaza.fi/s/f/editor/images/1113ln2_1.jpg

Some insulation around LGA1366 socket. Used two OCZ and one G.Skill DDR3-1600 module.

http://plaza.fi/s/f/editor/images/1113ln2_3.jpg

Scaling of this Q1CK Engineering Sample CPU seems to be very bad. I could run easily 4,4 GHz with 1,45V and -20C but 4,5 GHz needed 1,55V and -50C.. after that I hit a wall. CPU-tests crashed almost immediately.. 4,6 GHz was no go. Multipliers 34x and 35x didn't seem to work very well, lower temperature didn't help, more voltage didn't help. I got stuck.

http://plaza.fi/s/f/editor/attachmen...3ln2_multi.jpg

http://service.futuremark.com/compare?3dmv=520138

I was able to score P27491 in Performance-preset.

http://plaza.fi/s/f/editor/attachmen...ln2_single.jpg

P18530 with single card (850/990 MHz).

I'll continue testing so more coming soon :)

Nice initial results! =)

Why don't you ram up BCLK? 200 MHz should be a piece of cake. With a max multiplier of 33 you should have plenty of headroom =)

Or is the CPU really that bad?

From what people have posted here at XS, the 965XE seems to be hit and miss =/

I so feel your pain on that XE cpu sampsa, just 4 points shy of Aspstein and me, with almost identical clocks ;)

Same cpu as we used in sweden? I has the same wierd walls that we ran into also, Its frustrating for sure.. 4500Mhz is just lame :(

What bios are you running on the 4870x2 cards..? 1.35v?

Quote:

---

Originally Posted by Fr3ak

Why don't you ram up BCLK? 200 MHz should be a piece of cake. With a max multiplier of 33 you should have plenty of headroom =)

---

Just first test with CPU and tried to have as much impact on other settings as possible :)

Quote:

---

Originally Posted by dpa

Same cpu as we used in sweden? I has the same wierd walls that we ran into also, Its frustrating for sure.. 4500Mhz is just lame :(

What bios are you running on the 4870x2 cards..? 1.35v?

---

I think it's similar than we tested Sweden. I'm using 1,45V bios.

Now testing P6T Deluxe with new BIOS.. lets see..

good to see you are started to push it at home

looking forward to some damage hopefully

Funny CPU and GPU score :) Only 1 point difference..

I hope that I have more luck with my chips. Both can do 4.2GHz (Vantage CPU test) on air and 4.65GHz on dry ice (-40c) LN2 tests are coming when I have a car again..

Quote:

---

Originally Posted by SF3D

LN2 tests are coming when I have a car again..

---

:rofl: :D

I just tested P6T Deluxe and now running 155 MHz QPI and memory @ 930 MHz 8-7-7-20 (motherboard changes CAS to 8 automatically although it is set to 7 from BIOS). Maybe system is be more stable with lower multipliers 31x and 32x. Will test with LN2 tomorrow and then I need refill:

http://plaza.fi/s/f/editor/images/1023ln2.jpg

:)

gpu score 27492
cpu score 27491

1 point lol

Ahh.. look like your chip same as my second chip is i tried if 4.4G it don't need to much voltage and low temps. buy after pushed more it need alot of voltage and abit of low temps. too :)

Yah your score very interesting between CPU and GFX only 1 point different :D

:up:

Looks good Sampsa, you'll find a better cpu for sure. Would be interesting to see the Deluxe out-gunning the REII, maybe a defective board? :shrug:

Quote:

---

Originally Posted by Sampsa

:rofl: :D

I just tested P6T Deluxe and now running 155 MHz QPI and memory @ 930 MHz 8-7-7-20 (motherboard changes CAS to 8 automatically although it is set to 7 from BIOS). Maybe system is be more stable with lower multipliers 31x and 32x. Will test with LN2 tomorrow and then I need refill:

http://plaza.fi/s/f/editor/images/1023ln2.jpg

:)

---

Cool pic. :D Good stuff Sampsa.
Do you find nehalem overclocking to be sooo damn boring too? There really isn't much too it right now. I am so used to putting an e8600@ -165c/6400-6500mhz all summer, it's just 1000 times more exciting.

Quote:

---

Originally Posted by k|ngp|n

Cool pic. :D Good stuff Sampsa.
Do you find nehalem overclocking to be sooo damn boring too? There really isn't much too it right now. I am so used to putting an e8600@ -165c/6400-6500mhz all summer, it's just 1000 times more exciting.

---

I have to agree e8600 + Rampage Extreme is very exciting combo, still good for most of 3DMarks and I have a lot to do in hwbot ;)

Quote:

---

Originally Posted by G H Z

Looks good Sampsa, you'll find a better cpu for sure. Would be interesting to see the Deluxe out-gunning the REII, maybe a defective board? :shrug:

---

P6T Deluxe and it's BIOS are very mature since it has been under development longer time.

I like the RIIE a little more than the P6T, but both boards are pretty much at the same level, performance and overclocking wise. The RIIE has some more options and gimmiks, but I don't really need them. Price difference right now is 60 Euro.

What makes it kind of boring is the small margin between completely 24/7 stable and max clock. Even when going from air to extreme cooling =/

6.5 GHz looks way more impressive than 5.3 GHz ^_^

Quote:

---

Originally Posted by Fr3ak

Price difference right now is 60 Euro.

---

Should be more (normal P6T Deluxe not OC Palm)... ?

Quote:

---

Originally Posted by Fr3ak

I like the RIIE a little more than the P6T, but both boards are pretty much at the same level, performance and overclocking wise. The RIIE has some more options and gimmiks, but I don't really need them. Price difference right now is 60 Euro.

What makes it kind of boring is the small margin between completely 24/7 stable and max clock. Even when going from air to extreme cooling =/

6.5 GHz looks way more impressive than 5.3 GHz ^_^

---

why you like the RIIE a little more?

i have pre-orderd the RIIE and i hope it was not a big mistake.
german shops tell me that the RIIE will come later in november :(

Sampsa dou you have P6T's bios 901 please?

sampsa
What week is that cpu of yours?, below 3831A i'll guess?

I prefer the Rampage 2 Extreme, because it looks huge and has wider voltage ranges. If I had to buy one of the 2 boards, I would go with the P6T, because as I said, overclocking and performance wise, the 2 boards were on par. The Rampage was slightly faster for 3D, but the P6T was noticably faster in 32M. Performance might change with the latest BIOS. I only tested the OC Palm Edition, not the vanilla P6T Deluxe.


Ye, it's actually not 60 Euro:
P6T Deluxe : 252 Euro (http://preissuche.pcgameshardware.de/a365753.html)
P6T Deluxe OC Palm: 290 Euro (http://preissuche.pcgameshardware.de/a379487.html)
Rampage 2 Extreme: 325 Euro and up (not available yet) http://preissuche.pcgameshardware.de/a380449.html

I tried yesterday with P6T Deluxe and played little with higher BCLK & different multipliers. In my tests 133 MHz vs 200 MHz BCLK made ~200 points difference.

I was able to run 3DMark Vantage CPU-tests @ 4,6 GHz (27948 points) but when I started running the whole benchmark after GT1 and GT2 first CPU-test ran only like 2-4 secs. Frustrating. I have also tried this chip under my cascade (-110C idle) but it just seems to be too cold. I can boot the system ~80C but it never loads to Vista (using Hot gas bypass to lower the evap temperature)

Seems like there is alot variation in 965 Engineering Samples, would be nice to know what batch/stepping Fugger, hicookie and Shamino has. Lets hope retails are better.

What i have seen 3831a is a whole lot better then a 3830a sampsa , so i'll guess you have a 3830a :rolleyes:

Same situation here.

I test on P6T Deluxe. culd bug about -80.
But sweet point for max freq is around -65 - (-72).
Vcore higher than 1.5750 is not help me.


I have 2 batchs:
3830A408 - average
3830A417 - bad

I guess all ES CPUs are this two batches.

i've tried both...week 30 was better...but still 4.6ghz max on ln2
i just hope intel knows what the difference between the good ones and the bad ones are so they can make them all better next time.
still the cpu is proving essential for hwbot in several benchies allready so i think it has a future.

anyone tried a frozen retail sample?

Quote:

---

Originally Posted by Rol-Co

What i have seen 3831a is a whole lot better then a 3830a sampsa , so i'll guess you have a 3830a :rolleyes:

---

I need to zoom and photoshop couple pics.. should have checked before I lapped the ihs :D

what memory is in the faster cpu systems...i saw like 200mhz difference just switching memory.
maybe there are some imc things going on

Looks like the one we tested earlier in Stockholm was:

3830A417 (worst)

and this one is: 3830A432 (bad)

:down:

what is this i am hearing the RE2 is not clocking as well as the P6T?

Specifically with memory ;)

Quote:

---

Originally Posted by Sampsa

I have to agree e8600 + Rampage Extreme is very exciting combo, still good for most of 3DMarks and I have a lot to do in hwbot ;)

---

+1 ;)

Quote:

---

Originally Posted by Sampsa

I have to agree e8600 + Rampage Extreme is very exciting combo, still good for most of 3DMarks and I have a lot to do in hwbot ;)

---

I am on my way to the post office right now for you and sf3d. More fire power on the way guys :D.

For this whole system right here, how much would it cost together, if like you are just having it as a computer, not with the OC stuff?

Dont be so quick to dismiss chips as being bad.
Multipliers 30x and above can do funny things ;) Also check that you aren't pushing too high on uncore and QPI speeds.

Quote:

---

Originally Posted by T_M

Dont be so quick to dismiss chips as being bad.
Multipliers 30x and above can do funny things ;) Also check that you aren't pushing too high on uncore and QPI speeds.

---

correct; but some chips cant do more than 29x.. (like the one we tried in Sweden)

on the 965ES chip I got now doesnt like over 28x in vantage at all! therefore: test 27x!

sweden chip: 3830A417
"norway" chip.. lol :3831A457

I saw something in another thread about 0503 BIOS for RIIE. Just a heads up in case you guys don't have it.

with new 0503 bios multiplier is not a issue, can do x42 133 easely and even boot at 45 x100, maybe its bclock related but can't test that aircooled.

3831a are a lot better like i said already, have a proper true cooler on it now and can run about everything stable @ 4200mhz 1.39v :D

like this

4.2 at just a tick below 1.4 is a beautiful thing. I hope I'm lucky enough to get a chip like that. I have water cooling I'll be putting on it too after I run and test it on stock air first. This 1.5v stuff is a little scary for me personally.

I want to join this "bad ES chip club" :D

I just finished my 4 hour test session with my other 965 CPU. It can do 4.65GHz max on LN2. It was doing the same with dry ice :(

Light benches can be done at 4.7GHz, but that's it :shrug:

Cold bug -70c
Batch: 3830A432

I wish, that my other chip is better.. the batch is the same so no high hopes here.

Kind of frustrating...

Quote:

---

Originally Posted by Tony

what is this i am hearing the RE2 is not clocking as well as the P6T?

Specifically with memory ;)

---

Tested both board here and no difference if you ask me.

sf3d
use that chip for dreamhack contest please :p: holland take this one:)

Quote:

---

Originally Posted by DeDaL

Same situation here.

I test on P6T Deluxe. culd bug about -80.
But sweet point for max freq is around -65 - (-72).
Vcore higher than 1.5750 is not help me.


I have 2 batchs:
3830A408 - average
3830A417 - bad

I guess all ES CPUs are this two batches.

---

It's not good info :shakes: I have the same batch as yours bad cpu, damm...:(

Quote:

---

Originally Posted by Rol-Co

with new 0503 bios multiplier is not a issue, can do x42 133 easely and even boot at 45 x100, maybe its bclock related but can't test that aircooled.

3831a are a lot better like i said already, have a proper true cooler on it now and can run about everything stable @ 4200mhz 1.39v :D

like this

---

thats correct, I have an 1a also.. does almost 4.85Ghz 06 cpu test.

EDIT: ok, lets get it straight. there is two "steppings" for ES it seems like. we have the 3830A and we have the 3831A

the 3830A clocks like :banana::banana::banana::banana:, while the 3831A clocks fairly good.

"intel sweden chip": 3830A stepping, did MAX 4.55Ghz vantage/06 cpu test. coldbug at about -80. didnt like multi over 29x
"usa chip": 3831A stepping, does 4.85Ghz 06 cpu test so far. coldbug at about -95. I can do 40x with this one....

summary: 3831A is the thing to get? stasj your 3830A in a daily machine on AIR!

Finnish a 3d2005 run @ 4312mhz 1.39v true cooled. pretty neat if you ask me. :) room temp 22 degrees, 1 sanyo denky 35 mm fan on it.

my best 965 so far is a 3831a460 that does 4.9ghz 32M and CPU-Z @ 5270mhz

Quote:

---

Originally Posted by T_M

my best 965 so far is a 3831a460 that does 4.9ghz 32M and CPU-Z @ 5270mhz

---

cooling and temp ??

Quote:

---

Originally Posted by T_M

my best 965 so far is a 3831a460 that does 4.9ghz 32M and CPU-Z @ 5270mhz

---

tested 06/vantage?

outch so many crappy CPUs

looks like KP's got the best chip so far.........not sure how far charles can push his cherries in Vantage hehehe

Have been less interested in 3D performance testing so far, ever since i saw charles killing 32M :)
But certainly for a 4.9 32M chip the multithreaded tests will be far less (4.6-4.7?).

Using Ln2 @ -57C coldboot and running at -65C to -70C benching 32M (it could run same clocks at -50C too....)

http://plaza.fi/s/f/editor/images/cascade01.jpg

I managed to add Kingpin's F1 Dragon EE mounting bracket to my dear rotary cascade's evaporator.

http://plaza.fi/s/f/editor/images/cascade02.jpg

Testing contact between evaporator and IHS.

http://plaza.fi/s/f/editor/images/cascade03.jpg

The result was VERY bad :mad:

http://plaza.fi/s/f/editor/images/cascade04.jpg

Lapping IHS

http://plaza.fi/s/f/editor/images/cascade05.jpg

Hmm... also evaporator was lapped and contact is now perfect.

http://plaza.fi/s/f/editor/images/cascade06.jpg

Insulating around Asus Rampage 2 Extreme's LGA1366 socket with Armaflex.

http://plaza.fi/s/f/editor/images/cascade07.jpg

http://plaza.fi/s/f/editor/images/cascade08.jpg

Evaporator insulated and mounted on Core i7-965.

http://plaza.fi/s/f/editor/images/cascade09.jpg

Unfortunately -110C is too cold to this 3830A batch CPU. Cold boot issues, BSOD and Vista not loading :shrug:

Anyway, I'm ready for Core i7 so now I just need to get good chip :up:

http://www.pt1t.eu/public/X58/831A.jpg

This one boot @ -100° but no as good as 3837A :D

and it needs to be lapped.

There are ALOT of issues with the chips preventing them from being run hard like yorkfield or wolfdale. The retail chips seem to be different from the few 3.2 ES's I got the chance to try, but not by much. Maybe with more bios revisions on some diff mobos, we can start to see whats holding what back. I get varying performance on the same chip depending on what mobo it is testd on atm. Some scale with volts, some don't. Some like cold, some don't. Some like high multi's, some don't...uncore, bclock etc etc.
Despite the seemingly low percentage of "good extreme overclocking" cpus, the nehalem arch. is capable of alot more than what anyone posting actively has posted, believe it. Fact: 5400mhz+ 32M run..5200mhz+ 2k6 :D.
Prolly needle in a haystack at this point with this revision, but still possible.

Quote:

---

Originally Posted by Sampsa

http://plaza.fi/s/f/editor/images/cascade01.jpg

Testing contact between evaporator and IHS.

http://plaza.fi/s/f/editor/images/cascade03.jpg

The result was VERY bad :mad:

---

Owweee. I don't wanna cast a cloud on Intel's moment of glory, but that really is bad. That is concavity right in the middle where the cores are.

I don't wanna be captain obvious here, but that is pretty bad. It would be nice if Intel would true these up before the laser ethcing process. Not even air can cool properly if it doesn;t have a proper TI.

Quote:

---

Originally Posted by k|ngp|n

There are ALOT of issues with the chips preventing them from being run hard like yorkfield or wolfdale. The retail chips seem to be different from the few 3.2 ES's I got the chance to try, but not by much. Maybe with more bios revisions on some diff mobos, we can start to see whats holding what back. I get varying performance on the same chip depending on what mobo it is testd on atm. Some scale with volts, some don't. Some like cold, some don't. Some like high multi's, some don't...uncore, bclock etc etc.
Despite the seemingly low percentage of "good extreme overclocking" cpus, the nehalem arch. is capable of alot more than what anyone posting actively has posted, believe it. Fact: 5400mhz+ 32M run..5200mhz+ 2k6 :D.
Prolly needle in a haystack at this point with this revision, but still possible.

---

:eek: post the results or it didnt happen :D hehehehe

sub-7m has definately already happened, so yeah there are some big clocks hidden out there ;)
Fun times in Vantage/'06 eh Vince :D

Quote:

---

Originally Posted by k|ngp|n

Fact: 5400mhz+ 32M run..5200mhz+ 2k6 :D.

---

Not quite fact, a little embellished there.

hahaha show offs heheheh

http://www-personal.umich.edu/~gautamb/sig.JPG

Quote:

---

Originally Posted by dinos22

hahaha show offs heheheh

http://www-personal.umich.edu/~gautamb/sig.JPG

---

Now that is a nice signature! :yepp: Awaiting to watch a good show :D

Quote:

---

Originally Posted by T_Flight

Owweee. I don't wanna cast a cloud on Intel's moment of glory, but that really is bad. That is concavity right in the middle where the cores are.

I don't wanna be captain obvious here, but that is pretty bad. It would be nice if Intel would true these up before the laser ethcing process. Not even air can cool properly if it doesn;t have a proper TI.

---

Part of this was also my evaporator's fault.. now IHS and evap are both lapped and contact is perfect.

Pity its too cold for most chips heheh

Quote:

---

Originally Posted by T_M

Pity its too cold for most chips heheh

---

Lets see.. sooner or later there will be suitable chip for it and I only need one ;)

Quote:

---

Originally Posted by k|ngp|n

Fact: 5400mhz+ 32M run..5200mhz+ 2k6 :D.
Prolly needle in a haystack at this point with this revision, but still possible.

---

:eek2: Yeah, indeed a needle in a haystack cpu ... :yepp::eek:

Quote:

---

Originally Posted by dinos22

:eek: post the results or it didnt happen :D hehehehe

---

Haha DINO, it's not me that is running those sick speeds. But it's not too hard to figure out who is though :D

Sham OR The BOSS aka Charles??? :D
But seriously those speeds.... *need to talk to Intel for another chip* :p:

its Andre for sure ;)

It's definitely NOT the guy with the results in his sig...not a chance :rolleyes:

Quote:

---

Originally Posted by k|ngp|n

Haha DINO, it's not me that is running those sick speeds. But it's not too hard to figure out who is though :D

---

lol i'm sure he will end up sending you this thing as well

:rofl:

not fair not fair :D

Quote:

---

Originally Posted by dpa

its Andre for sure ;)

---

lol.....:rofl:

Whatever these alleged scores and speeds are...won't really matter in the long run. Prelease results always look like minced meat when put against what happens several months into the product cycle, when the parts start getting really pushed by the community. This is the case with all the prerelease scores that we're seeing. Just the way it is...with every new generation of CPU. Just remember when we were marvelling at 3.6GHz Conroe under single stage for example, and then look at how far it wound up.

Well.. someone is testing his 965 at 5.4GHz and I'm only 650MHz behind him :rofl:

I quickly tested my max 4 core 8 thread clocks yesterday. My other chip is average clocker and other one is bad clocker. Nice! :rolleyes:

3DMark06 (vag's are not maxed in the screen of course)
http://www.aijaa.com/img/b/00542/3104537.jpg

At least this can take -110C and it can do high multi and high BCLK on low temperatures. There might be more in it, when I have time to test different multi/BCLK/voltage combinations.

Like T_M said, there is bad multipliers and good multipliers.

Quote:

---

Originally Posted by Gautam

Whatever these alleged scores and speeds are...won't really matter in the long run. Prelease results always look like minced meat when put against what happens several months into the product cycle, when the parts start getting really pushed by the community. This is the case with all the prerelease scores that we're seeing. Just the way it is...with every new generation of CPU. Just remember when we were marvelling at 3.6GHz Conroe under single stage for example, and then look at how far it wound up.

---

just show it to us!! :D

hell, iam just at 4.83.. right now I want to drive over my 965 and kill it :D

Quote:

---

Originally Posted by Sampsa

Unfortunately -110C is too cold to this 3830A batch CPU. Cold boot issues, BSOD and Vista not loading :shrug:

Anyway, I'm ready for Core i7 so now I just need to get good chip :up:

---

Aw man...good luck in the hunt.

Quote:

---

Originally Posted by Gautam

Whatever these alleged scores and speeds are...won't really matter in the long run. Prelease results always look like minced meat when put against what happens several months into the product cycle.

---

Yeah but we don't care about that right now. This is pretty fun too though so whatever :up:

Hehehe, nice usertitle Gautam

Did you luck out on your very first sample? Or was there a bit of searching?

Nice to see 25 world famous overclockers commenting in on one thread here at XS. I might learn sumin'

I think its time for yet another build. Nice GTX score SF3D:clap: