AMD 4Gz Air Cooling Club

[diekul]

4.1ghz Phenom II 965:
http://valid.canardpc.com/show_oc.php?id=776455

On 790GP-DS4H with 4x1gb team elite mem. And a CM Z600 for the cooling.

[FlanK3r]

Cinebench 4 GHz 32-bit


Superpi 32M 4080MHz

[mäggäri]

Now let's see how this finally goes, when Mugen2 lapped and CPU too + Liquid Pro. I'll be back for editing :P



1st boot after sanding and mobo change, I had M4A785TD-V EVO and now I've M4A79XTD EVO :P Fooooookin awesome cheap boards, both. Before new paste and sandings, I was driving 550BE @ X4 3919MHz 1.5V.

ps. CPU-Z is showing the "correct" VCore.


Around 1.55V on load, on air. Now running prime95 @ 4140MHz.

http://valid.canardpc.com/show_oc.php?id=803532

[Titan7171]

[akitoch]

@Titan7171 are you using water cooling system for your 965 rig? your NB is pretty awesome...

[mäggäri]

@Titan7171 are you using water cooling system for your 965 rig? your NB is pretty awesome...

On Asus M4A785TD-V and M4A79XTD boards you don't have to do anything to get 2.8GHz NB, just to set it at 2800

I'd say that OC'ing NB starts after 2800. Everything under that is not worth of mentioning.

[Titan7171]

@Titan7171 are you using water cooling system for your 965 rig? your NB is pretty awesome...

Thanks!

only the cpu is watercooled nothing else.........NB voltage is set to 1.45v though.

heres a better one

[akitoch]

@Titan7171 Waw That's amazing man! you reach 3ghz for your NB *bowing* m(-___-)m... well could you gimme some advice cuz I've struggled to make
my NB stable 2,4 ghz/1,35v. My processor Temp boost 30% by rising NB volt (70C full load @ normal cpu-vid volt).. My PC spec is on my signature... thanks

Sorry OOT lol

------- just in case if my signature doesnt appear -__- -------
Phenom II 720 BE X4 @ 3,240 Ghz Normal Vcore | NB 2,480 @1,35v
Patriot 9-7-7-18-25 1333 Mhz 1,92 Volt
MSI 770-C45
Silverstone 500w
Artic Cooling AMD Pro <-- not good cooling T_T but affordable for me

[richierich]

Well the 955/965's are the best binned chips. I know my 550's NB can't hit over 2.8GHz stable.

[FlanK3r]

Thanks!

only the cpu is watercooled nothing else.........NB voltage is set to 1.45v though.

heres a better one

its very nice TItan, with DDR2 under 17min is great. But here we watching big diferent between DDR2 and DDR3 system

[chew*]

Checkmate rolco.....win vista, seems my effeciency is fine without the voodoo but my giga's effeciency was not

[Titan7171]

@Titan7171 Waw That's amazing man! you reach 3ghz for your NB *bowing* m(-___-)m... well could you gimme some advice cuz I've struggled to make
my NB stable 2,4 ghz/1,35v. My processor Temp boost 30% by rising NB volt (70C full load @ normal cpu-vid volt).. My PC spec is on my signature... thanks

Sorry OOT lol

------- just in case if my signature doesnt appear -__- -------
Phenom II 720 BE X4 @ 3,240 Ghz Normal Vcore | NB 2,480 @1,35v
Patriot 9-7-7-18-25 1333 Mhz 1,92 Volt
MSI 770-C45
Silverstone 500w
Artic Cooling AMD Pro <-- not good cooling T_T but affordable for me

The only advice I can give is that everything I have been able to do with this cpu has come down to how cool I can get it at a certain voltage. That run at 4.2ghz could not happen at 42-45c load. at the voltages I ran. Personally Im surprised Ive been able to do what Ive done on this mobo and cpu.

its very nice TItan, with DDR2 under 17min is great. But here we watching big diferent between DDR2 and DDR3 system

Thanks its no surprise though..........we are at the top of the DDR2 list though

Chew you posted that in the superpi 1m thread here..http://www.xtremesystems.org/forums/...=228552&page=8....


thats a good run, seems theres a little less than 1 minute difference in 32m between ddr2 and ddr3.

[chew*]

The gap can be smaller........4-4-4 1100 is what you need.

Had I used XP the time would be at least 6 secs faster.

[FlanK3r]

ufff, for my Kingston si 4-4-4 not possible,only at 800 MHz :-)
Titan: yes, about 1m diference, in 1m its about 0.075s think.

[Titan7171]

The gap can be smaller........4-4-4 1100 is what you need.

Had I used XP the time would be at least 6 secs faster.

thats tight man even at 800mhz my memory auto sets to 5,5,5 mine wont do 4,4,4

Chew what you think about Asus crosshair III?? Does it have what it takes to be better than GD70 with newer bios??

ufff, for my Kingston si 4-4-4 not possible,only at 800 MHz :-)
Titan: yes, about 1m diference, in 1m its about 0.075s think.

Its not to bad, game performance does not really suffer from DDR2 over DDR3

[chew*]

thats tight man even at 800mhz my memory auto sets to 5,5,5 mine wont do 4,4,4

Chew what you think about Asus crosshair III?? Does it have what it takes to be better than GD70 with newer bios??



Its not to bad, game performance does not really suffer from DDR2 over DDR3

I think from a user point of view the MSI is better. You just have more options...

I think the asus might OC a tad higher. Does really well in some apps but the MSI scores better in game tests so likely will do better for gaming as well.

Honestly the crosshair is more of a tool for me for 32m, there is a right tool for ever app

[zagortenay]

I think I have the 2nd best 3DMark06 Cpu score!

Phenom II X4 955
Asus Crosshair III
OCZ Platinum (2 x 2GB) DDR3 DDR3 1600 7-7-7-24
TRUE Black with Noise Blocker Multiframe PWM fan
Hiper Anubis Case
OCZ ModXStream 600W PSU

At 4 GHz. I can run most benchmarks but cpu is nowhere near P95 stable. My temps are high. (I am stable around 3.8 GHz. and core temps hit 60C during 8 hour P95 Blend test.)
I will try push-pull configuration later to see if I can do any better.

[chew*]

Yah the crosshair seems to score preety high cpu scores....tradeoff is it scores lower in sm2.0 and 3.0.

[Titan7171]

Yah the crosshair seems to score preety high cpu scores....tradeoff is it scores lower in sm2.0 and 3.0.

It would be interesting to see comparisons to early GD70 results, and Im wondering if later bios will increase performance for the graphics section on crosshair.

[mav2000]

965 BE at 4Ghz.



Only issue is that it runs a far bit hotter than my previous CPU. I reseated the cooler thrice, but got only a nominal improvement. My thinking is that the readings are a bit off, as the high temps correspond with my previous experiences, but the idle temps are about 10 degrees off my old idle temps. SO am not so sure about that. Also the NB does not run too high, but I am still to try out some stuff.

The CPU does 3.4 Ghz at 1.3 and am on prime now to kind of make sure its stable. Prime has been running small fft's for about an hour now, with no issues.

Can anybody give me some ideas on the temps...running 38 degress on idle and 49 degrees on load at stock. vcore = 1.3v. Ambient is around 24-25 degrees. Earlier with my 720be unlocked at 1.325vcore and 3 ghz I used to idle at 3-4 degrees above ambient.

NB...I dont know, not able to get my ram running at 1600 yet, it will load into windows and then crash. SO am back to 1333. I think the CPU NB is a bit weak on this chip.

[Titan7171]

965 BE at 4Ghz.


Only issue is that it runs a far bit hotter than my previous CPU. I reseated the cooler thrice, but got only a nominal improvement. My thinking is that the readings are a bit off, as the high temps correspond with my previous experiences, but the idle temps are about 10 degrees off my old idle temps. SO am not so sure about that. Also the NB does not run too high, but I am still to try out some stuff.

The CPU does 3.4 Ghz at 1.3 and am on prime now to kind of make sure its stable. Prime has been running small fft's for about an hour now, with no issues.

Can anybody give me some ideas on the temps...running 38 degress on idle and 49 degrees on load at stock. vcore = 1.3v. Ambient is around 24-25 degrees. Earlier with my 720be unlocked at 1.325vcore and 3 ghz I used to idle at 3-4 degrees above ambient.

NB...I dont know, not able to get my ram running at 1600 yet, it will load into windows and then crash. SO am back to 1333. I think the CPU NB is a bit weak on this chip.

Ill tell you this, the same happened to me in 2 instances

moving from Windsor 6400x2 to 140w 9950BE Phenom 1 x4

moving from 125W Phenom 2 940BE to 140w Phenom 965 BE

The 140w jump makes a difference when trying to cool on air this is mainly the reason I went water. Its very hard to cool 140w on air and I ended up lappig my 965.

http://www.xtremesystems.org/forums/...d.php?t=234734

And it help alot but still didnt make up the difference I saw between my 940 AM2+

[FlanK3r]

"The 140w jump makes a difference when trying to cool on air this is mainly the reason I went water. Its very hard to cool 140w on air and I ended up lappig my 965."

How diferent? I dont understand good, i can good cooling x4 955 (125W) with 1.5V at AIR, max 58 C in burn test. Think, its only diferent voltages between stock 955 and 965 (1.35V vs 1.4-1.425V)

[mav2000]

The main problem is that the idle temps are about 7-10 degrees higher than my 720be, which was a 95W model. Now this happens even though my 720be was running 3.0ghz 1.3V and my 965 is running at 3.4ghz at 1.25v. SO I am a bit confused as to this whole thermal thingy. I would have guessed that reducing the voltage would have some effect on temps, but that does not seem to be the case.

[Titan7171]

"The 140w jump makes a difference when trying to cool on air this is mainly the reason I went water. Its very hard to cool 140w on air and I ended up lappig my 965."

How diferent? I dont understand good, i can good cooling x4 955 (125W) with 1.5V at AIR, max 58 C in burn test. Think, its only diferent voltages between stock 955 and 965 (1.35V vs 1.4-1.425V)

sorry its not exactly the same program but its load temps....

ambiant in house aircooling


here is 940 AM2+ 125w

Here are idle and load temps from 3 voltage settings 1.35-1.55v

1.35v idle 30c load 43c
1.45v idle 32c load 50c
1.55v idle 35c load 58c

So here we can say 1.58v will get 58c



Here 965 140w

and here 1.42v will get 54c



just a little comparison (same mobo) Not exact but a big enough difference to see you can imagine 1.58v on 140w is not possible on air in my case.

Now remember this is my system my results and does not mean it is perfect or goes for everyone but can be used as reference for compare if you want. I say though that AMDOD and cortemp show same temp at same voltage so I just happen to use coretemp one time and AMDOD another tim depending on what I was doing and monitoring but again both programs show exactly the same temp is running at the same time.

[chew*]

Yep there are rather large load temp variations among chips.

I recently tested 5 chips all 5 batch #rs in a row.......I measured up to over a 10C delta variation between all 5 chips......I also noted rather large clock variations cold.

I really think batch # means less now than ever as these were all under the 100 batch numbering scheme.