RBE and 4870 stuttering

[2DividedbyZero]

I have flashed my Powecolor HD 4870VBIOS to 830/1000 following all the useful info found here, on doing so I have found a problem with my system!

First, the BIOS flash gives good benchmark results and is stable

Second, the problem is that it causes benchmarks programs like 3DM06, Vantage and Furmark to stutter/lag every 3-4 seconds like it is dropping a whole bunch of frames - the whole scene jumps ahead and it is throughout the whole of the benchmarks!??!?

I have a suspicion(?) purely my own without any proof (which is why I am here) that it may have to do with the "fanfix" side of it...

On looking at the default BIOSes and the ASUS TOP BIOS floating around, on the fan page PWM ramp on and PWM ramp(%) are not selected, yet when I try and save an edited BIOS it always selects them On/Enabled when I change the fan.
If I create a BIOS without changing the fan settings, my card no longer stutters!

any further help on this fellas????

system screenshot


benchmark screenies:-
3DM06 3DM V Furmark

RBE BIOS

 

[H82LUZ73]

change your 500 to 1000 2d clocks down to 900 on the memory.

 

[BAGZZlash]

On looking at the default BIOSes and the ASUS TOP BIOS floating around, on the fan page PWM ramp on and PWM ramp(%) are not selected, yet when I try and save an edited BIOS it always selects them On/Enabled when I change the fan.
If I create a BIOS without changing the fan settings, my card no longer stutters!

Interesting. The stuttering problem seems to be a not-so-rare issue. How could the PWM-ramp have anything to do with it, though? However, you never know. Regarding ATI's BIOSes, it's hard to surprise me after all...
So: Can anyone having trouble with stuttering acknowlegde the stuttering to disappear after switching off the PWM ramp?

 

[2DividedbyZero]

to re-itrate my problem in a clearer way

I will display in the order of CORE CLOCK : MEM CLOCK : FAN FIX : RESULT.....
(Def = Default settings)

1. Def : Def : Def : NO STUTTER
2. 830 : Def : Def : NO STUTTER
3. Def : 1000 : Def : NO STUTTER
4. Def : Def : FIX : STUTTER
5. 830 : 1000 : Def : NO STUTTER
6. 830 : 1000 : FIX : STUTTER

etc etc and in the sense of "stutter" i mean as described in my initial post

to be honest, any clocks set in the core or mem are fine - it IS the FAN FIX that causes the problem.

change your 500 to 1000

you want me to change my CORE to 1000????????????? are you craaaaaaaazy?

change your 500 to 1000 2d clocks down to 900 on the memory.

Can you (or anyone) explain to me why 'technically' how this will help in 2D when it is 3D benchmarking that causes the problem!


the problem doesn't really bother me as it is ONLY in the above benchmark programs, I am curious as to why it happens (fresh install etct etc taken into account Cat 8.6 hotfix and 8.7)

 

[2DividedbyZero]

Interesting. The stuttering problem seems to be a not-so-rare issue. How could the PWM-ramp have anything to do with it, though? However, you never know. Regarding ATI's BIOSes, it's hard to surprise me after all...
So: Can anyone having trouble with stuttering acknowlegde the stuttering to disappear after switching off the PWM ramp?

the PWN ramp cannot be switched OFF if you change the fan. Other pre-set BIOSes including the default one from my card have it OFF. when you 'Save BIOS' in RBE, the PWN switches default on automatically when fan is changed - ho hum... my reply above was to the previous post.

 

[BAGZZlash]

the PWN ramp cannot be switched OFF if you change the fan.

If you're using the look up table mode, you're right. You are, aren't you? If you use the transfer function, you can switch if off for sure, aren't you?

 

[2DividedbyZero]

look up table yes

transfer function didnt make any diference to the fan on my 1st attempt but haven't tested it for stuttering.

 

[2DividedbyZero]

Just been playing and.... I CAN CONFIRM that the stuttering happens with both look up table AND transfer function fan speed adjustments.... leave that one with ya BAGZZlash

would be nice to get some more input from those that have BIOS flashed their 4850/70 cards....

 

[BAGZZlash]

would be nice to get some more input from those that have BIOS flashed their 4850/70 cards....

Indeed!

@all: Please tell us you experiences.
@2DividedbyZero: Do you use XP or Vista?

 

[2DividedbyZero]

Vista64

 

[FR@NK]

If you could upload one of your bios that cause the stuttering I can test it and see if I get the same result.

 

[2DividedbyZero]

If you could upload one of your bios that cause the stuttering I can test it and see if I get the same result.

i take it that yes you have modded your 4870 BIOS and no you don't have the reported problem with benchmarking?

 

[H82LUZ73]

I have the ASUS Top bios ,And it was stuttering in Vista64 Ultimate like crazy,But in XP it is fine.

Also 2Zero try setting the clock info00 to your 3d clocks bet ya your stutter free after.

 

[ViciousXUSMC]

All I can say is that my 4850's came out fine when I rip and edited my own Visiontek bios. I noticed that I do not have the look up table at all when editing my Visiontek bios, but when I was looking at other bios from our database like the MSI bios because it was popular with the lower 2d clocks and the Asus Bios because it was popular due to the Asus program that let you set clock speeds and fan speeds. Well both of those bios did have the look up table on there.

I also played it safe and edited every single value pretty much to my high3d setting except for the 2 values that I knew were my 2d clocks and boot clocks. This was to prevent any issue with flickering with ram changing speeds or issues if clocks dont match.

I also will share again that I had the worst time ever with my bios flashing not becuase of of RBE or not know what I was doing but because of those damned beta 8.7 drivers. I dont know why but they give you a black screen and wont let you boot. the 8.6 hotfix or the new official 8.7's worked fine.

 

[2DividedbyZero]

I have the ASUS Top bios ,And it was stuttering in Vista64 Ultimate like crazy,But in XP it is fine.

Also 2Zero try setting the clock info00 to your 3d clocks bet ya your stutter free after.

nope

(and yes i tried - it's what i have running currently, can't be arsed to flash again)

 

[campdude]

i flashed my bios with success..
i have a gigabyte Radeon 4870.....

i do have stuttering in a games like NEED FOR SPEED MOST WANTED and CARBON,,,,,and PROSTREET (just a bit)

but i believe in those titles its a cpu problem.(hyper threading vs actual dual core optimized)

i just did exactly what someone else did with his directions.
I cant find his step by step bios flash guide for fan speed anymore.. but heres a screenshot attached..
i TRIED bios flashing using the look up table but i alwayz got a winflash error.


the screenshot sucks on here.... you cant really tell BUT i put in Tmin 38 and Tmax 81...

i found those settings by someone else on the net and i like them.

 

[campdude]

http://forums.whirlpool.net.au/forum-replies-archive.cfm/1020578.html


i found his guide!!!!!

 

[Humble]

Hello everyone...
I'm using PowerColor HD4870 PCS+ with Zerotherm cooling and I have a similar problem with TS.

Please take a look at the BIOS' fan speed setting:

And with that default setting, it results in high temp like this, which I think is not too good for the card and my internal casing temp:

It seems that PowerColor intentionally sets the BIOS fan speed setting to this card to maintain the card operational temp at min. 70C!

I wish to change a little bit of the BIOS fan setting by using RBE 1.1.4 and editing the "Look Up Table" a bit. But, everytime when I want to save the edited BIOS, the "PWM Ramp On" options is automatically turned ON by itself. Meanwhile the stock BIOS come with the "Look Up Table" method with "PWM Ramp On" unchecked.
But the option "PWM Ramp On" isn't turned ON when saving, if I use the "Transfer Function" method.

Will my card be work fine if I flash the edited BIOS with "Look Up Table" method and "PWM Ramp On" enabled? I'm not sure about this, since my card's fan only use 3 power wires (not 4 wires as the HD4870's stock cooler fan) and the missing wires is the blue one.
Please help...

One more thing, I already read that RBE offers BIOS fix for the HD4870 spin-up bug. And I can confirm that my card also suffer from the same problem. Will this BIOS fix work fine for my custom cooler HD4870 card?

Thanks in advance.

 

[LuxZg]

If someone is willing to confirm, I think I know what causes stuttering

Actualy, has nothing to do with RBE. I had it on stock BIOS as well.

Ingredients:
- Vista (I have SP1 64bit, not sure about if onyl this version or any Vista)
- any GPU sensor polling application (GPUZ, Everest etc)

Regarding first post by 2DividedByZero:
If you close GPU-Z, you will not have stutters. If you disable that option on bottom of sensors tab-you get no stutter (talking about "Continue refreshing this screen.." option)

Also, if you use Everest, you can choose how fast it polls sensors. If you put it on 5 seconds, you'll get only one stutter every 5s, if you put it on 1s you get stutters all the time (every 1s).

Please people, try it out. I am 99,99999% certain it has nothing to do with RBE, but I'd like to have it confirmed by more than one person (had one confirmation about month ago, forgot in which thread/forum, sorry )

 

[VuurVOS]

I tried to reproduce it with my card and vista64 sp1. Its proberly caused by the latest catalyst 8.8 (BETA) because I dont have it with my catalyst 8.7 and a bios with the fan fixed installed. I also dont run any sensoring software because it always slowing down your pc.

On a dutch forum I saw someone complaining about having stuttering every 3 seconds. He had also Vista and a HD4870 with the latest catalyst drivers installed. His solution was to downgrade the driver to catalyst 8.5.

 

[LuxZg]

Could you just try downloading GPU-Z? Take 5 seconds, and doesn't install or leave any traces or whatever. Just put it on sensors tab, and enable that option on the bottom, than start 3dmark or something..
I have had that issue with 8.7 final driver, haven't really checked 8.8 cos I've just made it work a day ago in Vista In XP I had no problems at all, with any driver, and any BIOS, original or RBE 1.13/1.14 modded ones..

 

[VuurVOS]

I know what GPU-Z is, I dont need it for normale use but I could use it to test the issue ^^

 

[LuxZg]

I know you know what it is I'm just asking you to try to reproduce the bug And tnx for deciding to try

EDIT: Nevermind, seems like I got a real detailed answer from the last person I've expected it to happen
Unwinder replied to me when I've asked if these stutters will be seen with Rivatuner as well.. Well, did I get an answer Have a look here - http://forums.guru3d.com/showpost.php?p=2822647&postcount=74

Thing is (in short) that polls cost CPU cycles, RV770's sensors must be polled one after another (can't be done at once), and when you try with GPUZ that has 7 sensor readings - you get a lot of CPU cycles lost for that one moment. That's also why polling with just CCC (only single temperature reading) shows no obvious stutters, and probably why Vista (as a real resource hog) exibits this in more obvious way than XP does.

He also gave a tip why we used to have fans spinning to 84% now and than when using BIOS with fan tables. It happened only with GPU-Z running in background (as a lot of people noticed), and thing is that probably GPU-Z while reading sensors switches to a higher-temperature sensor (like VRM) in the same moment when fan controller checks for temperature to decide how fast the fan should be spinning. So it spins the fan way up for that one moment..

Man, did I learn a lot from just one post That's why I'm posting it here as well, so maybe more people learn the same things Enjoy reading! (btw it's not that long, my post is now probably as long as his )

 

[VuurVOS]

You saved 15 minutes of my life by doing it ^^ Serious I actually forgot after I post it

 

[LuxZg]

Tsk tsk, to forget such an important thing I'll still be testing this a bit.. but it's less important to me now that I'm pretty certain of a reasons why it happens..