- Joined
- Mar 12, 2007
- Location
- Bowling Green, KY
Impressive work, very nice. If you don't mind me asking, are you an electrical engineer by trade?
-
Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!
8800 Ultra Water Cooled Modded Test Results
- Thread starter ViperJohn
- Start date
OP
- Joined
- Jul 5, 2004
- Location
- Santa Ana, Ca
- Thread Starter
- #22
Valkillmore said:
Yep along with a few other disciplines thrown in as well.
Now that I have done a few more Ultra's I can safely say the card above is definately a typical
and over the average for a modded Ultra.
Viper
OP
- Joined
- Jul 5, 2004
- Location
- Santa Ana, Ca
- Thread Starter
- #23
Deanzo said:Thanks mate
Very helpful
Deanzo if you have a GTX card and not an Ultra note that I edited the clock stepping blurp above when it comes to the 8800GTX cards. While the 8800 Ultra's memory will only step on even multiples of 27Mhz the GTX cards will step on some 1/3 and 2/3's multiple in between the evens.
Viper
- Joined
- Feb 5, 2005
- Location
- Christchurch, New Zealand
Thanks again mate,
This is a must know imo for working out the best set up of the card.
On stock cooling / bios volts anything over 729 on the core is a no go, but then again it may well be the shader clocks holding me back.
Not that Im unhappy with 729/1215
And thanks for the tip using ATItools with Rivatuner, that will save me some time
This is a must know imo for working out the best set up of the card.
On stock cooling / bios volts anything over 729 on the core is a no go, but then again it may well be the shader clocks holding me back.
Not that Im unhappy with 729/1215
And thanks for the tip using ATItools with Rivatuner, that will save me some time
- Joined
- Nov 28, 2001
- Location
- in a magical field
Wrong terminology maybe, by 'clock domain' I meant the different core/shader/memory clocks. He was specifically talking about the memory though so your specific clarification was nice. So Rivatuner's graph shows the true clocks?
OP
- Joined
- Jul 5, 2004
- Location
- Santa Ana, Ca
- Thread Starter
- #26
MadMan007 said:
Probably so. It is easy to get twisted sometimes...at least for me lol.
Yes in the case of 8800's Core ROP Domain, Core Shader Domain and Memory Domain in the graph.
If you set up RT to show all three memory domains individually then only Memory Domain (0) is truly correct.
Depending on the driver used (anything after the 97.02's but I haven't checked the 150 series drivers either)
the clocks shown for Memory Domain 1 and Memory Domain 2 are are bad and/or not used register data
being reported to RT by the drivers and moot IMO.
Viper
OP
- Joined
- Jul 5, 2004
- Location
- Santa Ana, Ca
- Thread Starter
- #27
Deanzo said:Thanks again mate,
This is a must know imo for working out the best set up of the card.
On stock cooling / bios volts anything over 729 on the core is a no go, but then again it may well be the shader clocks holding me back.
Not that Im unhappy with 729/1215
And thanks for the tip using ATItools with Rivatuner, that will save me some time
Will the card hold that 729 core clock hold in a 3DM01 Nature Only torture test (25 repeat passes
(not loop!) with the title screens disabled) and not core lock???
I would be very surprised if that 1215 memory is truly clean in ATItool SoA.
Viper
- Joined
- Feb 5, 2005
- Location
- Christchurch, New Zealand
ViperJohn said:
Im not sure about either at this point, but I can tell you it will run at that speed for all benches,
Im along way off saying for sure its 100% clean.
But I will do as above and post up my results.
I for one hope it does pass, but then I would
OP
- Joined
- Jul 5, 2004
- Location
- Santa Ana, Ca
- Thread Starter
- #30
Deanzo said:Im not sure about either at this point, but I can tell you it will run at that speed for all benches,
Im along way off saying for sure its 100% clean.
But I will do as above and post up my results.
I for one hope it does pass, but then I would
When I mod cards for customers I rate there clocks on the lowest numbers I can find in any test and I run a lot of torture loops to just hammer the crap out of the cards. I am really picky too and watch the "rug" in ATItool SoA very closely. A single yellow dot in a ten minute run will get a memory clock downgrade from me. Most of the time ATItool won't even beep but that beep is pretty unreliable with 8800's anyway.
I have often found that while a 8800's memory may not be "Viper" clean in ATItool SoA I can't spot the memory tossing any cookies in the any of benchmark test at the same or higher higher speeds though. My customers often tell me they can game at higher clocks clean than I rated their cards for. I always tell them if they are not seeing any artifacts and the card isn't locking then run the clocks higher as it is all in the eye of the final user lol.
Viper
OP
- Joined
- Jul 5, 2004
- Location
- Santa Ana, Ca
- Thread Starter
- #31
Well i'll be dipped in sheep stuff. You guys with Ultra's check for memory stepping on the 1/3 and 2/3's multiples of 27.
I have two identical eVGA 8800 Ultra's here right now that are only five cards apart in serial numbers. They both have the same bios (hell there is only one for an Ultra ATM). One card will step the memory at 1151 (27 x 42 2/3) and several other 1/3 and 2/3 multiples while the other card will only step the memory on even multiples of 27come hell or high water.
It's definitely got me scratching my head lol.
Viper
I have two identical eVGA 8800 Ultra's here right now that are only five cards apart in serial numbers. They both have the same bios (hell there is only one for an Ultra ATM). One card will step the memory at 1151 (27 x 42 2/3) and several other 1/3 and 2/3 multiples while the other card will only step the memory on even multiples of 27come hell or high water.
It's definitely got me scratching my head lol.
Viper
- Joined
- Dec 19, 2005
- Location
- New York
ViperJohn said:
John,
Is this good or bad news for my card? How is she doing in your little boot camp session?
OP
- Joined
- Jul 5, 2004
- Location
- Santa Ana, Ca
- Thread Starter
- #33
dominick32 said:John,
Is this good or bad news for my card? How is she doing in your little boot camp session?
Yours is the one that will step on some 1/3 and 2/3's memory multiples. It isn't really good or bad
just informational.
Viper
- Joined
- Feb 5, 2005
- Location
- Christchurch, New Zealand
Hey VJ
Im giving this a wee bump as I was getting a little OT on nicasoft thread and as this all got started here, best I just keep it to one thread.
The below is on a 40 repeat pass (not loop) with the title screens disabled, 3DM01 nature only torture test.
Core 720, Shader 1782, Memory 1080 (stock), Max temp 67c
Something you said in the other thread got me thinking, when I tried this last time I was running the memory at 1215 and thats why it may have locked up on the 3rd pass.
So for now I'll leave that alone and just see what the ROP/Shader will do, Im about to try 729 true, then 756, I hate how its such a big jump there.
.
Im giving this a wee bump as I was getting a little OT on nicasoft thread and as this all got started here, best I just keep it to one thread.
The below is on a 40 repeat pass (not loop) with the title screens disabled, 3DM01 nature only torture test.
Core 720, Shader 1782, Memory 1080 (stock), Max temp 67c
Something you said in the other thread got me thinking, when I tried this last time I was running the memory at 1215 and thats why it may have locked up on the 3rd pass.
So for now I'll leave that alone and just see what the ROP/Shader will do, Im about to try 729 true, then 756, I hate how its such a big jump there.
.
OP
- Joined
- Jul 5, 2004
- Location
- Santa Ana, Ca
- Thread Starter
- #35
Deanzo said:
Get some memory clock on the card. The core is spending a lot of time in wait states clocked that
high with the memory clocked that low. I haven't seen a stock Ultra yet that won't run 1152 memory.
After that try the N loop again. If ya get through that way you have one exceptional Ultra ROP clock
wise and definitely mod fodder.
Oh BTW make sure the card stays clean in ATItool Scan for Artifacts clocked up. In may toss cookies
in SoA on the higher shader clock or the memory controllers..
Viper
Last edited:
- Joined
- Feb 5, 2005
- Location
- Christchurch, New Zealand
Core 720, Shader 1782, Memory 1152, Max temp 68c
The above tests are on vista 64 and the new 162.22, Im having some real issues with this new driver and ATITool, So I'll need to get that worked out before I can scan for artifacts.
The above tests are on vista 64 and the new 162.22, Im having some real issues with this new driver and ATITool, So I'll need to get that worked out before I can scan for artifacts.
- Joined
- Feb 5, 2005
- Location
- Christchurch, New Zealand
Im using nature as a torture test,
And the best score so far with 01 is 104905 done with two Ultra's in sli 1008/1224 cpu 5310 all under ln2
And the best score so far with 01 is 104905 done with two Ultra's in sli 1008/1224 cpu 5310 all under ln2
OP
- Joined
- Jul 5, 2004
- Location
- Santa Ana, Ca
- Thread Starter
- #39
PhysX said:
It's a standard benchmark and it is stil valid however it is easily CPU/System speed bound and capped.
In this case however that is not what it is being used for. Only one test (Nature) is being run in a specific
manner to obtain a specific go/no go result.
Viper
OP
- Joined
- Jul 5, 2004
- Location
- Santa Ana, Ca
- Thread Starter
- #40
Deanzo said:
3DM01 won't SLI no matter what you do. The high scores come from the high system speed with
a fast single card (or the card in PCIe slot #1 of an SLI setup).
Viper
