Go Back   nV News Forums > Graphics Card Forums > NVIDIA Legacy Graphics Cards

Newegg Daily Deals

Reply
 
Thread Tools
Old 10-25-04, 06:32 PM   #517
jimmor
Registered User
 
jimmor's Avatar
 
Join Date: Mar 2004
Location: Scotland
Posts: 568
Default Re: Basic FX59xxx bios changes

Quote:
Originally Posted by Daneel Olivaw
A bit of personal experience:
2.8ns ram with tight latencies (5900xt) no cooling at all: overheats at 740-770
2.8ns ram with slacker latencies (+/-5900U) no cooling at all: overheats at 770-800
2.8ns ram with slacker latencies (+/-5900U) DHES arctic cooling: overheats at 835-845
These are very usefull numbers that show a good increase in memory bandwidth by only changing memory latencies and adding a bit of cooling to a XT card.

In fact, not bad for 2.8ns memory chips rated at 714MHz !

However, it's not always about how high you can overclock that matters, but more how you can improve on the efficiency of whatever speed there is. As an example, my Aopen 5900nu gives more performance from my current modded MSI 5950u bios running at 500/970 with latency and timing tweaks than it did running at 500/1020 with a standard 5950u bios.
__________________
The more I know, the more I know I don't know

Intel P4 3.2 @ 3.85GHz (ASUS P4C800E-Deluxe)
BFG 6800GS (16x1,6vp) @ 300/430/1200.

Driver, 82.12 :-
Aquamark3, 75039
Codecreatures1, 90.8
3Dmark05 v120, 6195
3Dmark03 v360, 14334
3Dmark06 v102, 2949
3Dmark2001se v330, 23793.
jimmor is offline   Reply With Quote
Old 10-25-04, 06:48 PM   #518
Jethro
Registered User
 
Join Date: Oct 2002
Posts: 356
Default Re: Basic FX59xxx bios changes

8 of out 10 nvidiots agree lower latency is better than higher clocks

Indeed my modded stock 5900 @ 900mhz lowlatencies had much better fillrate than 980mhz @ stock 5950 latencies.

Cant wait till yah get a 6800 series card Jimmor! your insights provided a good amount of free performance man
__________________
Abit KV7 --- XP-M @ 2.74 1.84vcore 1/2 MAze3 cooled ~Patriot XBL 512x2 @ 235x12 cas2 --- Chieftec Aluminum 601 --- Tru power 430 --- Netgear cable router RP614 --- Aud2-->Aud2 hack --- WD special 80 / Barracuda 160gb 7200.7 --- LiteON 52.32.52 --- Mitsu DiamondPRO 22inch --- Swifty cooled EVGA 6800NU @ 439/949 16p/6vs

12406 O3marks
Jethro is offline   Reply With Quote
Old 10-25-04, 07:10 PM   #519
jimmor
Registered User
 
jimmor's Avatar
 
Join Date: Mar 2004
Location: Scotland
Posts: 568
Default Re: Basic FX59xxx bios changes

Quote:
Originally Posted by Jethro
8 of out 10 nvidiots agree lower latency is better than higher clocks

Indeed my modded stock 5900 @ 900mhz lowlatencies had much better fillrate than 980mhz @ stock 5950 latencies.

Cant wait till yah get a 6800 series card Jimmor! your insights provided a good amount of free performance man
Well it's easier to have insights on 59xx types of card when you actually own one. But a 6800, not sure when that will happen ------> think I'll maybe wait untill prices and problems stabalise at much lower levels. Unless of course, a xmas or birthday etc changes things ??
__________________
The more I know, the more I know I don't know

Intel P4 3.2 @ 3.85GHz (ASUS P4C800E-Deluxe)
BFG 6800GS (16x1,6vp) @ 300/430/1200.

Driver, 82.12 :-
Aquamark3, 75039
Codecreatures1, 90.8
3Dmark05 v120, 6195
3Dmark03 v360, 14334
3Dmark06 v102, 2949
3Dmark2001se v330, 23793.
jimmor is offline   Reply With Quote
Old 10-27-04, 08:35 AM   #520
Daneel Olivaw
Northwood hugger
 
Daneel Olivaw's Avatar
 
Join Date: Jan 2004
Location: Montreal
Posts: 1,244
Default Re: Basic FX59xxx bios changes

Quote:
Originally Posted by jimmor
However, it's not always about how high you can overclock that matters, but more how you can improve on the efficiency of whatever speed there is. As an example, my Aopen 5900nu gives more performance from my current modded MSI 5950u bios running at 500/970 with latency and timing tweaks than it did running at 500/1020 with a standard 5950u bios.
That's all very true, my system is usually set up at 400/800 'cause I don't need the speed, last time I checked vid card had little impact on Btefnet TV shows... and Max Payne 1 well... seriously! I'm always putting back my current most efficient bios when I'm done tinkering.

I'm mostly just having fun benching at different clocks, that's how I relax. Some people paint, I drill holes in my case and overclock.

Btw, is there a max_value not be used for this string:
04 03 05 17 00 1A 00 0F 00 08 08 04 05 03
and this one:
07 03 0F 43,

for example, would 07 03 0F 48 post? (As I said, I'm not trying to be efficient, just tinkering) I'm just trying to achieve one big honking clockspeed!

Is there any setting that won't post with these two strings?

from my test, 07 00 0F 40 does indeed seem to have the best performance
with 04 02 05 11 00 13 00 0B 00 06 06 03 04 04.
__________________
-Intel Core2 Duo 4300 (3.0GHz) | Asus P5N-E SLI | 6GB Corsair CVS PC2-5300 | GF8800GT 512MB | Western Digital 1TB Raid0 | Antec SLK2600AMB | Antec Earthwatts 500W
-Intel Pentium4 2.4C (3.4GHz) | Asus P4P800 Dlx | 2GB Crucial Rendition PC-3200 | GF6800LE 128MB | Seagate 200GB | Antec Overture II | Antec TruePower 2.0 480W


Mild agnosticism 葉he view that the existence or nonexistence of God or gods is currently unknown but is not necessarily unknowable, therefore one will withhold judgment until/if more evidence is available. A weak agnostic would say "I don't know, but maybe you do."

Last edited by Daneel Olivaw; 10-27-04 at 08:54 AM.
Daneel Olivaw is offline   Reply With Quote
Old 10-27-04, 07:04 PM   #521
jimmor
Registered User
 
jimmor's Avatar
 
Join Date: Mar 2004
Location: Scotland
Posts: 568
Default Re: Basic FX59xxx bios changes

Quote:
Originally Posted by Daneel Olivaw
That's all very true, my system is usually set up at 400/800 'cause I don't need the speed, last time I checked vid card had little impact on Btefnet TV shows... and Max Payne 1 well... seriously! I'm always putting back my current most efficient bios when I'm done tinkering.

I'm mostly just having fun benching at different clocks, that's how I relax. Some people paint, I drill holes in my case and overclock.

Btw, is there a max_value not be used for this string:
04 03 05 17 00 1A 00 0F 00 08 08 04 05 03
and this one:
07 03 0F 43,

for example, would 07 03 0F 48 post? (As I said, I'm not trying to be efficient, just tinkering) I'm just trying to achieve one big honking clockspeed!

Is there any setting that won't post with these two strings?

from my test, 07 00 0F 40 does indeed seem to have the best performance
with 04 02 05 11 00 13 00 0B 00 06 06 03 04 04.

So you like tinkering, just to see what is achievable !

Not sure what you are trying to achieve however by going even more relaxed in timings, but like many things in life ----> you never know until you try ?

I have never tried going this route, so obviously don't know which are best or stable values to use. However would not have thought that much will happen other that at certain values there will be artifacts. Nothing to worry about however, as I got many artifacts and sometimes lockups when finding stable tighter timings. Mind you it generally means you should keep your blind flashing floppy with a good bios ready just in case it won't post.

If you still want to play, then you will likely find interaction between some values, so initially only change the middle "17 00 1A 00 0F 00 08 08" part of timing string. And obviously you dont change any of the "00" string values.

It can take a while since you should only increase each 2 bit value by 01 at a time and reboot just in case it is very close to being unstable with artifacts or a lockup. Once this part of string is at a max, you should then do the first "04 03 05" and end "04 05 03" parts.

For the Latency string of "07 03 F0 43", you should increase the "43" value carefully, however I see no reason why "48" shouldn't work. But if I wanted to make jumps, I would probably go no more than 02 to 03 at a time. So maybe first jump should be "45" or "46". If anything you are initially looking for a relationship between step values and no artifacts. Then you can just keep stepping up, rebooting and seeing if any overclocking improvement is achieved.

I suspect however, even if you manage to improve on overclocking levels, you will be building in such delays, that outright performance has to suffer.

And for a 5900 card with 128MB of memory, as you have found, the latency string of "07 00 F0 40" should give best performance. However I am a bit surprised that you get overall best performance with the "04 02 05 11 00 13 00 0B 00 06 06 03 04 04" timing string. I would have thought that the "0301040C000F0009000505020303" string suggested in my guide would have worked better with your slower memory chips. And even if your card had 2.2ns chips then a string of "040304090013000B000506020303" should definately work better?

All this could keep you going for weeks, so good luck with your endeavours !!!
__________________
The more I know, the more I know I don't know

Intel P4 3.2 @ 3.85GHz (ASUS P4C800E-Deluxe)
BFG 6800GS (16x1,6vp) @ 300/430/1200.

Driver, 82.12 :-
Aquamark3, 75039
Codecreatures1, 90.8
3Dmark05 v120, 6195
3Dmark03 v360, 14334
3Dmark06 v102, 2949
3Dmark2001se v330, 23793.

Last edited by jimmor; 10-28-04 at 07:21 AM.
jimmor is offline   Reply With Quote
Old 10-27-04, 10:59 PM   #522
Daneel Olivaw
Northwood hugger
 
Daneel Olivaw's Avatar
 
Join Date: Jan 2004
Location: Montreal
Posts: 1,244
Default Re: Basic FX59xxx bios changes

deleted: -purposeless-
__________________
-Intel Core2 Duo 4300 (3.0GHz) | Asus P5N-E SLI | 6GB Corsair CVS PC2-5300 | GF8800GT 512MB | Western Digital 1TB Raid0 | Antec SLK2600AMB | Antec Earthwatts 500W
-Intel Pentium4 2.4C (3.4GHz) | Asus P4P800 Dlx | 2GB Crucial Rendition PC-3200 | GF6800LE 128MB | Seagate 200GB | Antec Overture II | Antec TruePower 2.0 480W


Mild agnosticism 葉he view that the existence or nonexistence of God or gods is currently unknown but is not necessarily unknowable, therefore one will withhold judgment until/if more evidence is available. A weak agnostic would say "I don't know, but maybe you do."

Last edited by Daneel Olivaw; 10-28-04 at 12:34 AM.
Daneel Olivaw is offline   Reply With Quote
Old 10-27-04, 11:51 PM   #523
Daneel Olivaw
Northwood hugger
 
Daneel Olivaw's Avatar
 
Join Date: Jan 2004
Location: Montreal
Posts: 1,244
Default Re: Basic FX59xxx bios changes

Rivatuner RC2, when testing "low power 3d setting" seems to revert to a lower voltage then the 1.4v that I specify in the bios. I suspect it tests using 1.3v and not the value I specified through nibitor (nv bios editor.)

rivatuner test say 495MHz test ok on performance 3d
but 475MHz fails on low power 3d
even though my bios is set for 1.2v 2d, 1.4v throttling, 1.4v 3d and runs 475MHz 3d+throttling fine now.

Small bug in rivatuner, but it's annoying.
__________________
-Intel Core2 Duo 4300 (3.0GHz) | Asus P5N-E SLI | 6GB Corsair CVS PC2-5300 | GF8800GT 512MB | Western Digital 1TB Raid0 | Antec SLK2600AMB | Antec Earthwatts 500W
-Intel Pentium4 2.4C (3.4GHz) | Asus P4P800 Dlx | 2GB Crucial Rendition PC-3200 | GF6800LE 128MB | Seagate 200GB | Antec Overture II | Antec TruePower 2.0 480W


Mild agnosticism 葉he view that the existence or nonexistence of God or gods is currently unknown but is not necessarily unknowable, therefore one will withhold judgment until/if more evidence is available. A weak agnostic would say "I don't know, but maybe you do."
Daneel Olivaw is offline   Reply With Quote
Old 10-28-04, 06:14 AM   #524
jimmor
Registered User
 
jimmor's Avatar
 
Join Date: Mar 2004
Location: Scotland
Posts: 568
Default Re: Basic FX59xxx bios changes

Quote:
Originally Posted by Daneel Olivaw
Rivatuner RC2, when testing "low power 3d setting" seems to revert to a lower voltage then the 1.4v that I specify in the bios. I suspect it tests using 1.3v and not the value I specified through nibitor (nv bios editor.)

rivatuner test say 495MHz test ok on performance 3d
but 475MHz fails on low power 3d
even though my bios is set for 1.2v 2d, 1.4v throttling, 1.4v 3d and runs 475MHz 3d+throttling fine now.

Small bug in rivatuner, but it's annoying.

In my 59xx bios modding guide you will see that the gpu volts lookup table has two binary values for identifying and setting volts. The first 2 bit value being only an indicator for showing volts expected. However it is this value that the driver compares to the programmed Performance Level Volts Indicator for any given operating mode. And it is the second 2 bit value at the lookup table that is actually used to set gpu volts.

Where Rivatuner can be confused therefore is that NiBiTor only changes the second 2 bit value of lookup table, and thus kills the relationship to any Indicator values. Which actually means you can have an indicator value of "82" which normally translates to 1.3v actually giving 1.4v simply because the Bios was modded by NiBiTor.

Try therefore setting with OmniExtreme v1.5 editor, or even better change lookup table volts and Performance Level Indicators iaw my 59xx bios modding guide at first post.
__________________
The more I know, the more I know I don't know

Intel P4 3.2 @ 3.85GHz (ASUS P4C800E-Deluxe)
BFG 6800GS (16x1,6vp) @ 300/430/1200.

Driver, 82.12 :-
Aquamark3, 75039
Codecreatures1, 90.8
3Dmark05 v120, 6195
3Dmark03 v360, 14334
3Dmark06 v102, 2949
3Dmark2001se v330, 23793.
jimmor is offline   Reply With Quote

Old 10-28-04, 09:20 AM   #525
Daneel Olivaw
Northwood hugger
 
Daneel Olivaw's Avatar
 
Join Date: Jan 2004
Location: Montreal
Posts: 1,244
Default Re: Basic FX59xxx bios changes

Quote:
Originally Posted by jimmor
Where Rivatuner can be confused therefore is that NiBiTor only changes the second 2 bit value of lookup table, and thus kills the relationship to any Indicator values. Which actually means you can have an indicator value of "82" which normally translates to 1.3v actually giving 1.4v simply because the Bios was modded by NiBiTor.

Try therefore setting with OmniExtreme v1.5 editor, or even better change lookup table volts and Performance Level Indicators iaw my 59xx bios modding guide at first post.
You're right Jim, as always! I think you may even be right about those latency strings I posted as my best.
__________________
-Intel Core2 Duo 4300 (3.0GHz) | Asus P5N-E SLI | 6GB Corsair CVS PC2-5300 | GF8800GT 512MB | Western Digital 1TB Raid0 | Antec SLK2600AMB | Antec Earthwatts 500W
-Intel Pentium4 2.4C (3.4GHz) | Asus P4P800 Dlx | 2GB Crucial Rendition PC-3200 | GF6800LE 128MB | Seagate 200GB | Antec Overture II | Antec TruePower 2.0 480W


Mild agnosticism 葉he view that the existence or nonexistence of God or gods is currently unknown but is not necessarily unknowable, therefore one will withhold judgment until/if more evidence is available. A weak agnostic would say "I don't know, but maybe you do."
Daneel Olivaw is offline   Reply With Quote
Old 10-28-04, 10:09 AM   #526
Unwinder
Registered User
 
Join Date: Jul 2003
Posts: 160
Default Re: Basic FX59xxx bios changes

Jimmor, I hope you won't treat it as a flame again, but your explanation of the problem is a bit far from the reality.

Daneel Olivaw

Quote:
Rivatuner RC2, when testing "low power 3d setting" seems to revert to a lower voltage then the 1.4v that I specify in the bios. I suspect it tests using 1.3v and not the value I specified through nibitor (nv bios editor.)
Nope, it tests LP3D clocks on 2D (1.2V) voltage. It is by design feature. The ForceWare drivers provide internal clock testing for 2D and 3D modes only (there is no LP3D stress testing implementation in the driver as soon as the control panel itself don't allow you to change clocks for this performance level). So I had only 3 choices in LP3D clock frequency testing implementation in RivaTuner:

1) Don't test these clocks at all
2) Test them in 2D mode
3) Test them in 3D mode

I choosed the second one as the safest against newbies, as both 1) and 3) can allow user to set actually unstable clocks. IMHO it is better to fail test on stable clocks, than to pass it on unstable ones and cause system to throttle / freeze due to undervoltage.
If you are a power users and understand it, you may disable clock testing at all via DisableClockTesting in the Power User tab.
Unwinder is offline   Reply With Quote
Old 10-28-04, 05:42 PM   #527
Daneel Olivaw
Northwood hugger
 
Daneel Olivaw's Avatar
 
Join Date: Jan 2004
Location: Montreal
Posts: 1,244
Default Re: Basic FX59xxx bios changes

Hello Unwinder,

When I put in bios 1.4v for all three clocks the tests pass in 3DLP, when I put 1.4v on 3DLP and 3D, and 1.2v on 2D the test fail on 3DLP.

So that confirms what you say Unwinder I think. (No point doubting the guy who made Rivatuner! )

Thanks for both your help. Jim, I'm using your tighter latency 0301040C000F0009000505020303 + the looser 07000F40 at clocks of 465-794, I'm stable and closer than before to 7000 marks03! Hey, with all that you've explained to me, I'll forgive you for any mistakes you make!

Unwinder, I think you chose correctly since the only clock that needs testing is the 3D one, and 3DLP will probably be 'disabled' (read: set to same as 3D) for someone who just wants the card to freeze instead of slowing down (kinda makes knowing what's stable and whats not easier).
__________________
-Intel Core2 Duo 4300 (3.0GHz) | Asus P5N-E SLI | 6GB Corsair CVS PC2-5300 | GF8800GT 512MB | Western Digital 1TB Raid0 | Antec SLK2600AMB | Antec Earthwatts 500W
-Intel Pentium4 2.4C (3.4GHz) | Asus P4P800 Dlx | 2GB Crucial Rendition PC-3200 | GF6800LE 128MB | Seagate 200GB | Antec Overture II | Antec TruePower 2.0 480W


Mild agnosticism 葉he view that the existence or nonexistence of God or gods is currently unknown but is not necessarily unknowable, therefore one will withhold judgment until/if more evidence is available. A weak agnostic would say "I don't know, but maybe you do."
Daneel Olivaw is offline   Reply With Quote
Old 10-28-04, 05:50 PM   #528
jimmor
Registered User
 
jimmor's Avatar
 
Join Date: Mar 2004
Location: Scotland
Posts: 568
Default Re: Basic FX59xxx bios changes

Quote:
Originally Posted by Unwinder
Jimmor, I hope you won't treat it as a flame again, but your explanation of the problem is a bit far from the reality.

Daneel Olivaw



Nope, it tests LP3D clocks on 2D (1.2V) voltage. It is by design feature. The ForceWare drivers provide internal clock testing for 2D and 3D modes only (there is no LP3D stress testing implementation in the driver as soon as the control panel itself don't allow you to change clocks for this performance level). So I had only 3 choices in LP3D clock frequency testing implementation in RivaTuner:

1) Don't test these clocks at all
2) Test them in 2D mode
3) Test them in 3D mode

I choosed the second one as the safest against newbies, as both 1) and 3) can allow user to set actually unstable clocks. IMHO it is better to fail test on stable clocks, than to pass it on unstable ones and cause system to throttle / freeze due to undervoltage.
If you are a power users and understand it, you may disable clock testing at all via DisableClockTesting in the Power User tab.
Unwinder,

Thank you for your excellent explanation of how Rivatuner does LP3D testing. As only you could !

My input to Daneel Olivaw's problem was meant to be more about NiBiTor's unauthordox way of changing gpu volts possibly being the reason why the rivatuner testing was not working as expected ?

But your posting has now cleared up my misunderstanding of the position. Which is good for me and everybody else for that matter. So, why would you think that I would treat your input on your Rivatuner program as a flame ?

Now that I know, I could of course disagree or argue against the way Rivatuner does Low Power 3D testing, but it is your program, and therefore your right to do what you want with it.

And as a power user, I always disable clock testing, as I read in your Rivatuner accompanying notes !!
__________________
The more I know, the more I know I don't know

Intel P4 3.2 @ 3.85GHz (ASUS P4C800E-Deluxe)
BFG 6800GS (16x1,6vp) @ 300/430/1200.

Driver, 82.12 :-
Aquamark3, 75039
Codecreatures1, 90.8
3Dmark05 v120, 6195
3Dmark03 v360, 14334
3Dmark06 v102, 2949
3Dmark2001se v330, 23793.

Last edited by jimmor; 10-30-04 at 05:17 AM.
jimmor is offline   Reply With Quote
Reply


Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump


Similar Threads
Thread Thread Starter Forum Replies Last Post
Asus Bios bug nforceuser NVIDIA Linux 3 10-06-02 05:15 AM
new bios my motherboard and got new feature IDE PCI BUS MASTERING what is it ??? b5fan CPUs, Motherboards And Memory 12 09-22-02 06:10 PM
BIOS doesn't recognize more than 32 MB(old P1 machines) Klaus-P CPUs, Motherboards And Memory 3 09-09-02 04:33 PM
bios flashing Gunslinger68 CPUs, Motherboards And Memory 7 09-08-02 01:20 PM
PC Games, CeleronII 566, CeleronA 300, BIOS Savior, Heatsinks, NES & Sega Items +pics TekViper For Sale/Trade 5 08-07-02 11:48 PM

All times are GMT -5. The time now is 10:39 PM.


Powered by vBulletin® Version 3.7.1
Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Copyright ゥ1998 - 2014, nV News.