Chose the Q9550 instead. (Was: E8400 and video editing (using Premiere)?)

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

francisA

Member
Dec 2, 2008
153
0
0
Well, stayed up late last night to find out what the Q9550 can do. So popped her right in and it booted right up to 2.83GHz. VID is 1.250v default. Was hoping to get one with 2.125v, but 1.250v it is.

Being the lazy oc'er that I am, just used ET6 and bumped up to its max suggested oc of 3.6GHz (450x8). Booted all the way up to XP but of course voltages were way up there so went back to the BIOS and re-adjusted the voltages and DDR settings. Back in to XP and no glitches, nothing. Ran IBT for 10 passes and works fine. I initially gave it 1.30v to have enough juice to work with. Gave the RAM 1.9v too. Not much drama here.

Went back into the bios, upped multi to 8.5 for 3.830GHz and kept it at 1.3v. Still works fine. IBT 10 passes also passed. At this point, my DDR800 G.Skills are already running at DDR900 so it's getting quite a workout. OCCT ran fine for 30mins with no errors when I decided to stop it so I can keep playing.

Back to the bios again and this time went to 475x8.5. Posted but when it got to XP, bsod. So I thought the cpu must need more juice. Bumped vcore a couple notches, booted to XP but re-booted again. This time, I think now it's the RAM so I gave it 2.0v as it's now running at DDR950 spec. Now it went all the way into XP and was able to complete IBT 10 passes successfully again. Tried OCCT but came back with errors after 14mins. Upped the RAM to 2.04v and re-ran OCCT for 30mins and no errors. vcore at this point is at 1.325v I think while CPU-Z says 1.296v with LLC enabled.

Not declaring it absolutely stable yet as I have yet to do more thorough testing with my apps and some more extended Orthos and OCCT tests.
Did Super PI Mod v1.5 runs up to 32M (32M == 13m30s) and all completed successfully. Prime95 ran for 5hrs 40mins without complaining. Left it running Memtest to see if any errors come up. There was a point where OCCT crashed so am not sure yet what caused it so more testing needed.

Bottom line, I'm extremely satisfied with the Q9550. Even if at the end I just settle for 3.830GHz (450x8.5) so i can get the vcore lower than 1.3v and easier on the system, the current setup is 35% faster fsb wise than my E5200 setup. With the bonus of having enough cores to handle video editing/encoding tasks easier.
Gaming-wise, it wasn't affected much. FPS was very slightly higher and more consistent in Crysis.

With all these, temps were hitting 36-39C on idle and 54-56C on load (RealTemp) or 59-61C (CoreTemp/HWMonitor). Of course IBT was able to heat it up real good up to 65C (RealTemp). I'm expecting temps to further settle once the AS5 has broken in.

I'm really just shooting for 3.83GHz so will start lowering voltages while keeping it stable. I don't think the extra 45fsb and 208Mhz will make that much difference in whatever app I use. Bonus would be less heat and power consumption.

Some of the important settings that I copied off of someone else's posts from my research on UD3P and Q9550:

CPU Vcore [ 1.25000 ]...................[ 1.3250 ]
CPU Termination [ 1.20 ]...................[ 1.30 ]
CPU PLL [ 1.50 ]...................[ 1.50 ]
CPU Reference [ 0.760 ]...................[ 0.863 ]

MCH Core [ 1.100 ]...................[ 1.30 ]
MCH Reference [ 0.760 ]...................[ 0.863 ]
MCH/DRAM Refernce [ 0.900 ]...................[ 0.950 ]
ICH I/O [ 1.500 ]...................[ 1.500 ]
ICH Core [ 1.100 ]...................[ 1.100 ]

DRAM Voltage [ 1.800 ]...................[ 2.04 ]
DRAM Termination [ 0.900 ]...................[ 0.950 ]
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
My general rule of thumb is (and I haven't killed a chip yet) 10% max vcore boost above stock....

So you should be bale ot go as high as 1.375v and that is actual...Meaning you may have to set it to 1.4-1.425v in bios to actually get that....

Your temps look great. Push for 4ghz....

This is just further proof IMO the Yorkfield is still a better option then the slightly cheaper phenom II. based on reviews I have seen the phenom is goimg to have to be a touch above 4ghz to 4.2ghz to beat you.....and I haven't seen many hit that yet on air...

NIce job
 

francisA

Member
Dec 2, 2008
153
0
0
Thanks the tips Duvie.

Made me rather curious there now as to how high I can up the FSB. I feel that the G.Skills are limiting me now but I've read somewhere that feeding it 2.1v can force it to go DDR1000 and rock solid at that. One gave it 2.2v for DDR1066. Hmmmmm ........... guess the saga (search for the highest fsb) has just begun.
 
Dec 30, 2004
12,553
2
76
Originally posted by: Duvie
My general rule of thumb is (and I haven't killed a chip yet) 10% max vcore boost above stock....

So you should be bale ot go as high as 1.375v and that is actual...Meaning you may have to set it to 1.4-1.425v in bios to actually get that....

Your temps look great. Push for 4ghz....

This is just further proof IMO the Yorkfield is still a better option then the slightly cheaper phenom II. based on reviews I have seen the phenom is goimg to have to be a touch above 4ghz to 4.2ghz to beat you.....and I haven't seen many hit that yet on air...

NIce job

You don't need a stellar board for the Phenom, you do for this Q9550. On mine I'll be afraid of going over 450FSB for fear of wearing out the mobo components. Don't have to worry about any of that with a Phenom II. A pretty solid chance at 3.8 on Phenom II with a cheap board vs. expensive board and guaranteed 3.8....not to mention the Q9550 is $50 more expensive, and AMD takes this no prob imo.
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
Originally posted by: soccerballtux
Originally posted by: Duvie
My general rule of thumb is (and I haven't killed a chip yet) 10% max vcore boost above stock....

So you should be bale ot go as high as 1.375v and that is actual...Meaning you may have to set it to 1.4-1.425v in bios to actually get that....

Your temps look great. Push for 4ghz....

This is just further proof IMO the Yorkfield is still a better option then the slightly cheaper phenom II. based on reviews I have seen the phenom is goimg to have to be a touch above 4ghz to 4.2ghz to beat you.....and I haven't seen many hit that yet on air...

NIce job

You don't need a stellar board for the Phenom, you do for this Q9550. On mine I'll be afraid of going over 450FSB for fear of wearing out the mobo components. Don't have to worry about any of that with a Phenom II. A pretty solid chance at 3.8 on Phenom II with a cheap board vs. expensive board and guaranteed 3.8....not to mention the Q9550 is $50 more expensive, and AMD takes this no prob imo.

No you dont....you could use a sub 100 dollar DS3L mobo from the P35 or P45 chipset....

I have hit 450's with sub 100 and sub 150 boards.....I dont consider the board francisA has is stellar and he is obviously doing it....

Only time will tell on the stress on the system, but I usually have really good case cooling and actively cool my northbridge chipsets when I go plus 450's....the P35 I had had very nice chipset cooling and I didn't even actively cool it....Now with the DS3L mobo I would place a small chipset fan on it...small price we are talking about...

PLus I am not sure how stressful it is because both P35 mobos I had I didn't have to bump the chipset up in vcore much to obtain 445 fsb.


" pretty solid chance at 3.8 on Phenom II with a cheap board "

Show me some links....becuz I went around last night and saw several people with systems and PhII only at 3.6ghz (on average) and the motherboards they were running were 125+ boards....I guess they are just too lazy to try for more?....

You need to get more data of users before you can say that....

PLus an OCer knows you NEVER EVER say guaranteed...


I have seen plenty of yorkfields oc'd to 3.8-4ghz range on air with a whole gamut of boards used.....Until I see that around here with the phII I cannot say that....
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
Originally posted by: soccerballtux
Originally posted by: Duvie
My general rule of thumb is (and I haven't killed a chip yet) 10% max vcore boost above stock....

So you should be bale ot go as high as 1.375v and that is actual...Meaning you may have to set it to 1.4-1.425v in bios to actually get that....

Your temps look great. Push for 4ghz....

This is just further proof IMO the Yorkfield is still a better option then the slightly cheaper phenom II. based on reviews I have seen the phenom is goimg to have to be a touch above 4ghz to 4.2ghz to beat you.....and I haven't seen many hit that yet on air...

NIce job

You don't need a stellar board for the Phenom, you do for this Q9550. On mine I'll be afraid of going over 450FSB for fear of wearing out the mobo components. Don't have to worry about any of that with a Phenom II. A pretty solid chance at 3.8 on Phenom II with a cheap board vs. expensive board and guaranteed 3.8....not to mention the Q9550 is $50 more expensive, and AMD takes this no prob imo.

really? because all the reviews being thrown out yesterday I didn't see that...clock for clock the yorkfield appeared faster....clock for clock the Intel is more expensive...that is true...however I am seeing just as high if not higher OCs still with the Intel chips....

I saw a 920 at 4ghz last night....but it was water cooled....

I also saw a chip that could do 3.5ghz at stock volts but by the time he hit 3.7ghz he was already near 10% vcore boost and he hadn't confirmed it was stable....

I haven't seen any data that impresses me yet...I hear a lot of talk...I see reviews with decent OCs, but in most of the reviews they get their Q9400's, Q9550s, etc to hit 4ghz as well.


MarkFW900 has a 920 and a decent (I guess you would say budget board coming) motherboard ordered and we will get his unbiased opinion....We will see where he hits...

 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
I hear the BE versions may be the best because of the unlocked multis...that tells me perhaps the IMC doesn't like high fsb.....so the 920 isn't going to be helped much by that...The Deneb actually has a lower speed IMC to the agena. The Ph3 is supposed to be better...


http://firingsquad.com/hardwar..._performance/page3.asp

They could only get 200mhz with stock vcore....3.74ghz top stable with 1.55v vcore....

http://guru3d.com/article/amd-...-and-940-review-test/6

Here is another that touts "if you are going to overclock get the BE one"....They got 3.8ghz
Somehow I think the mobo they tested was a nice one....
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
sorry I want hijack the thread anymore....off to start a new thread...LETS continue the discussion there Soccerballtux
 

francisA

Member
Dec 2, 2008
153
0
0
That's the next plan yh125d. I'll delegate the DDR800s to the other rig that will house the E5200. I'm actually slapping myself now for not buying the DDR2 1066 to begin with :)
 

francisA

Member
Dec 2, 2008
153
0
0
Thanks Duvie. As soon as I get home, I'll post some screenies. I still can't believe it was able to reach 475*8.5 at 1.296v actual. But most especially, the G.Skills were able to hold up. How much more can it be stretched ..............? Hmmm, maybe I can push it some more.

I had my son check Memtest now and it's still running without errors, 272 passes and going on Test #5 and that's on DDR950. Really sweet.
 

francisA

Member
Dec 2, 2008
153
0
0
I tried getting 475*8.5 stable with OCCT but it just won't budge. 30mins into OCCT and it fails with a Code "2". I don't even know what that is. Though having gone through 20 passes of IBT and ~400 passes of Memtest #5 and almost 6hrs of Prime95, OCCT just doesn't want it.

But, I need to power down and install my HR-03 GT so need to get the rig ready for 3d testing. I reverted back to a stable 450*8.5 with the G.Skills doing DDR900. So far so good.

A screenie of OCCT declaring my rig "Stable!" after 1hr of torture

UD3P Q9550 OCCT Result

This with the following settings
vcore..........1.300v
BIOS.......... 1.284v
CPU-Z idle...1.28v
CPU-Z load..1.2640v

I'm still shooting for the big 4.0 so hopefully I can find what OCCT is complaining about.