Icee-"Back in the saddle again" w/ Thuban

[FlanK3r]

cele:4460MHz 32M air is brilliant. can u try lower RAms (1600+- agressive timings) and higher NB clock?

[Jethro]

Welcome back My Icee. Your post were loaded with a ton of useful info back in the day. I zapped a few things following your lead back then hahah . Look forward to your findings from brutalizing AMD's current platform!

[MrIcee]

Hi my friends

As always, thanks for the kind words.

I finally finished up my study on the 0707 bios on the Crosshair 4. I will be setting up fresh OS bench partitions in the next few days and will bench Vantage at that time for inclusion in these results. I will say though not tremendous in difference, there is certainly not any loss of points by running the higher HT as was the case on the previous 790FX boards as tested by Tony. If anything, if competitively benching IMO it is worth sync'ing NB/HT to see if a few additional points might be gained. The older benches appear to benefit more from the sync'd HT+NB, and I would presume Vantage would be closer to '06, which is really an insignificant amount and within the margin of variation that occurs in 3DMark.

________________________________________________


----------------NB 3211/ HT 1976------NB 3211/HT 3211------Difference

3DMark '06 ---------24,261---------------24,299-------------- + 38

3DMark '05 ---------29,892---------------30,264-------------- + 372

3DMark '03 ---------81,270---------------81,630---------------+ 360

Settings: CPU - 17 X 247Mhz FSB = 4199Mhz @ 1.41v
Ram: 1647Mhz 6-6-6-16-20 1T @ 1.65v
Videocard: 1x Asus EAH5870 V2 @ 1002Mhz/5005Mhz

Each test run 3 times per setting. Reboot between each test. Each test score
represents the average of 3 test runs

_________________________________________________

I've updated the bios this evening to 801 and we will see where that may lead us. So far so good, and it's now time to start fine tuning the boards bandwidth.We'll be back again soon.................

Best regards,

Randi

[Daveburt714]

Nice results MrIcee,

From my experience, keeping the IMC and HTRef in sync has always given better results (on the 965)... It seems to help system stabliity too.

If I push the IMC really hard (>2700) lowering the HTRef 1 notch can help settle things down (12x/13x), but getting them way out of whack has never shown any benefit in my testing personally,
and can cause problems when they get really skewed (10x/13x)...

It seems like your putting that chip through it's paces.

I'll be interested to see what your final conclusion is on Linked IMC/HTRef compared to High IMC vs Low HtRef.

I think that's a pretty serious question, but somehow I always seem to get distracted with other clocks to really test the theore.
Still going by gut instinct myself...

I didn't know you from the olden days bud...
But it's nice to have you back!

Dave

[Lightman]

Hi my friends

As always, thanks for the kind words.

I finally finished up my study on the 0707 bios on the Crosshair 4. I will be setting up fresh OS bench partitions in the next few days and will bench Vantage at that time for inclusion in these results. I will say though not tremendous in difference, there is certainly not any loss of points by running the higher HT as was the case on the previous 790FX boards as tested by Tony. If anything, if competitively benching IMO it is worth sync'ing NB/HT to see if a few additional points might be gained. The older benches appear to benefit more from the sync'd HT+NB, and I would presume Vantage would be closer to '06, which is really an insignificant amount and within the margin of variation that occurs in 3DMark.

________________________________________________


----------------NB 3211/ HT 1976------NB 3211/HT 3211------Difference

3DMark '06 ---------24,261---------------24,299-------------- + 38

3DMark '05 ---------29,892---------------30,264-------------- + 372

3DMark '03 ---------81,270---------------81,630---------------+ 360

Settings: CPU - 17 X 247Mhz FSB = 4199Mhz @ 1.41v
Ram: 1647Mhz 6-6-6-16-20 1T @ 1.65v
Videocard: 1x Asus EAH5870 V2 @ 1002Mhz/5005Mhz

Each test run 3 times per setting. Reboot between each test. Each test score
represents the average of 3 test runs

_________________________________________________

I've updated the bios this evening to 801 and we will see where that may lead us. So far so good, and it's now time to start fine tuning the boards bandwidth.We'll be back again soon.................

Best regards,

Randi

You've saved me tons of testing with that little chart!
I don't know how to thank you for that

[chew*]

Nice work ICee. Now if only we could actually use that kinda HT on all cpu's when cold.....many need HT less than 2k or max clocks can't be acchieved

[theoldtimer]

It is interesting isn't it MrIcee. Very helpful work.

[PatRaceTin]

wow i love those ram

[MrIcee]

Nice work ICee. Now if only we could actually use that kinda HT on all cpu's when cold.....many need HT less than 2k or max clocks can't be acchieved

chew* my friend

Is it the cold, or the high CPU frequency or a combination the prevents the ability to clock CPU frequency with a high HT when you gents are under LN2? Or just a limitation of the chips architecture?

The other question is how cold does it have to be? I would assume the xtremes of LN2? The reason I ask is this problem is a challenge to me but not sure if phase is cold enough for me to R+D the problem and attempt to find a work around. You guys have probably tried to overcome this in your extensive testing on this platform already, but I do enjoy trying to overcome
limitation issues on the AMD platform.

Best regards,

Randi

[chew*]

chew* my friend

Is it the cold, or the high CPU frequency or a combination the prevents the ability to clock CPU frequency with a high HT when you gents are under LN2? Or just a limitation of the chips architecture?

The other question is how cold does it have to be? I would assume the xtremes of LN2? The reason I ask is this problem is a challenge to me but not sure if phase is cold enough for me to R+D the problem and attempt to find a work around. You guys have probably tried to overcome this in your extensive testing on this platform already, but I do enjoy trying to overcome
limitation issues on the AMD platform.

Best regards,

Randi

I can crank it on phase fine. It seems for some chips to be a balls to the wall max clocks/max cold problem. Alot of times if you can get it to run the score is bad.

[Cele303]

cele:4460MHz 32M air is brilliant. can u try lower RAms (1600+- agressive timings) and higher NB clock?

They always are with AMD press samples... And Noctua does it's part... I don't have it anymore, I return it to AMD after few days testing... Unfortunately 0801 came to late for me...

Sorry MrIcee for offtopic in your thread, I'm Eric btw, nice to meet you!

[MrIcee]

Just pulled off my water cooling, which was a cobbed affair as it was an old waterblock. It seems 4500ish was about it, and only benchable with a few benches. Not trying to achieve anything but stable benchability, no 24/7 stuff.

My chip is a 1013 CPMW, and I'm starting to believe its a dud The wall was 4500ish on all 6 cores, and I pushed up to 1.65v into it and it would not budge. As a matter of fact, it doesn't like voltage either as at 1.65v it would not go into Windows any longer. It's probably the heat load talking to me indeed.

I've got a great water system, but cannot achieve benchable clocks near some of our friends here at XS that are on water, and it figures..LOL My luck over the years getting a good retail chip is always a crap shoot, and this try got me crap

I was able to get into Windows at 4600Mhz on only 2 cores and run some benches, but that is a fruitless and senseless endeavor when you have a 6 core CPU.

So my bench is apart at this moment. I just checked my 404A Prommy and it is still up to snuff. I don't have one of their AM2 CPU kits and none are available any longer at their former dealers, so a major cob will be underway to get the evaporator close to fitting the CPU. Nothing has ever been easy
at the Icee test bench, so I'll be back hopefully soon with some more positive results.

Best regards,

Randi

[chew*]

Just pulled off my water cooling, which was a cobbed affair as it was an old waterblock. It seems 4500ish was about it, and only benchable with a few benches. Not trying to achieve anything but stable benchability, no 24/7 stuff.

My chip is a 1013 CPMW, and I'm starting to believe its a dud The wall was 4500ish on all 6 cores, and I pushed up to 1.65v into it and it would not budge. As a matter of fact, it doesn't like voltage either as at 1.65v it would not go into Windows any longer. It's probably the heat load talking to me indeed.

I've got a great water system, but cannot achieve benchable clocks near some of our friends here at XS that are on water, and it figures..LOL My luck over the years getting a good retail chip is always a crap shoot, and this try got me crap

I was able to get into Windows at 4600Mhz on only 2 cores and run some benches, but that is a fruitless and senseless endeavor when you have a 6 core CPU.

So my bench is apart at this moment. I just checked my 404A Prommy and it is still up to snuff. I don't have one of their AM2 CPU kits and none are available any longer at their former dealers, so a major cob will be underway to get the evaporator close to fitting the CPU. Nothing has ever been easy
at the Icee test bench, so I'll be back hopefully soon with some more positive results.

Best regards,

Randi

Randi i found retails to top around 1.70v max for average on LN2......these chips are NOT deneb. I think those running 1.60 on water be it chilled or not are gonna have problems quickly. These are made on a diff process than deneb. Alot of these guys are on sub 0 water randi as well

[Zeus]

I don't think 4.5GHz is too bad Randi, especially on normal watercooling. Mine seem to top somewhere around 4.4GHz in 64bit OS, stable enough to run multi threaded benches like cinebench 10 and 11.5. Maybe just a tad higher but i haven´t been over 1.53V yet. Pretty sure it can do 06 at 4.5GHz in a 32bit OS or close to that mark.

Yours seemed to have a nice CPU/NB too.

Looking forward to your results on phase my friend.

[charged3800z24]

I have a decent water setup, not chilled, and haven't haven't been able to bench CB 11.5 or r10 64bit above 4.2ghz core and 3000nb. So great results there.

[theoldtimer]

Don’t get discouraged Randi. It looks to me like you’re getting up to speed nicely. This Thuban I have doesn’t like volts, 1.575v is about all it wants under chilled water. Soon as I max this one out under the chilled water I’ll test it under room temp water and try to compare results with everyone else on ambient temp water.

[Zeus]

Hey Randi, here's 06 at 4.5G single card on plain water, 20ºC ambient:



Can you do that?

[MrIcee]

Zeus

I couldn't do that with my chip. I was lucky to get a run of Aquamark in as posted earlier at 4500ishMhz. I thought my issues were possibly poor pressure contact between waterblock and CPU but that was not the case.

A strange phenomena that I have never experienced before is a result of the instability I experience at high overclocks. Back in the day the system would either not go into Windows, or if you made it the least amount of stress would cause a blue screen. Now when I get into Windows, I'll run a number of benches(or one) and I get a green screen, with rolling vertical white lines.
Then thats it. I've seen similar years ago with vid cards that were pushed too far.

Is this type of green screen failure typical of these new generation of chips the last few years, or am I experiencing some weird OC failure crap?

Thanks for any insight

Best regards,

Randi

[Zeus]

That is caused by the driver for 5 series ATI videocards. It's like a BSOD but instead you will get a grey screen with vertical lines, or a different color.

With deneb cpu's instablity caused the system to spontaniously reboot under load, this platform simply hangs when unstable or so it seems. Haven't seen a BSOD yet, only one time a grey screen, at other instances it just hung.

So you're not alone my friend, just keep pushing it.

[Gautam]

Nicely done, your thoroughness shows in the results. I do wonder just how long you'll be able to last against the peer pressure in going below -100 though.

[DuraN]

Welcome back Randi, nice to see another old member return! When I looked in my inbox I saw your name on a few PM's and remembered the old Athlon XP benching days

In the process of getting back myself, but on an Intel setup for the first time! Good luck benching.

[MrIcee]

Can you do that?

I can't do that, but I did do this I had both cards in and I'm under the Prommy. Do I feel any better?? A little bit for now Gautam is right though, I'm either going to have to have a better phase setup for 24/7 benching or go LN2 someday. This is still with a normal Vista OS, this week I'll load up my tweaked OS's for benching.

Best regards,

Randi

[MrIcee]

I know Super Pi 1M is not much of a deal any longer, but this as a personal best for me @ 4705Mhz. I'll run a 32M later and Wprime.


Randi

[MrIcee]

I know this is also another useless bench these days, but I did pull it off at 4817Mhz.


Best regards,

Randi

[Zeus]

Hehehe, awesome man! 4.7G 3D will be untouchable for me, glad you made it and good to see it made you feel better.

I hope to put the second card back in soon, can't wait to see how it does. What clocks did you have on the graphic cards?