BIOS Issues Continue to Haunt PowerColor HD 4870 PCS+

[btarunr]

It was earlier reported that the non-reference high performance Radeon HD 4870 graphics card, the PowerColor Radeon HD 4870 PCS+, faced a BIOS glitch (covered here) that affected its stability. Barely days after Chinese website Expreview reported this, PowerColor posted a revised BIOS on their website. Expreview notes that the BIOS didn't quite fix the fan speed (as in up the maximum speed limit), but makes the fan operate at erratic speeds or "zipzip" as Expreview notes. The fan-speed sensor graph should give you a better idea:



Expreview notes that the first screen shot was captured when system was idle…even core temperature(DISPIO) kept 64.5° C, the fan kept "crazy changing its speed from 100% to 13% to 0%". The card continues to fail at Furmark beyond 1440×900 8xAA, leading to a driver crash during the benchmark run. The card was tested using the latest Catalyst 8.8, when it was noted that Furmark could be cleared but that the card suffers frame-rate loss compared to older drivers, in essence temperatures are lowered at the expense of performance. Expreview sought another sample from PowerColor and hopes for the best with the product.

View at TechPowerUp Main Site

 

[eidairaman1]

i remember when there was no need for temperature sensor for fan speeds on video cards, fan would run at full speed and keep stuff properly cooled, why not have a manual way of doing it, i mean CPU Fans suffer from the same thing when they are plugged into the motherboard (thats why i dont have mine on it but a molex)

And i Recall Arctic-Cooling having a 2-way toggle switch for fan speed for the AC VGA Silencer Rev 3.

To add another note, Board Makers are thinking about putting Turbo Mode back on the motherboards, same with case makers, why not have the said above (Manual Fan switching)

 

[uber_cookie]

why not have a manual way of doing it,

i agree ^_^ it's another never ending story of bulls**t attempt to automate everything

 

[VuurVOS]

Lolz I had this bug also with my normal HD4870. This issue is caused by software and bios.

I had this issue only when GPU-Z (under xp) was active with the stock bios (113-B50701-100). They fixed it with the 113-B50701-105 bios which also have powerplay. I couldn't recreate this problem with my second operation system (vista) so it could be also a part of the drivers/operating system

RBE has an option which can fixed this (when the correct bytes and section are avaible).

 

[dalekdukesboy]

i agree ^_^ it's another never ending story of bulls**t attempt to automate everything

I have 2 8800gts 512's and last radeon card (I still have but unused) is an x1900xtx so I'm unaware as to how you can deal with fan speeds on newer catalyst versions, but can't you override fan speed manually with software via catalyst or even 3rd party software or even that neat looking bios editor for ati I think wizzard came up with here? If that is the case...who the hell cares unless the bios has other issues that causes instability, if it's just a screwed up fan control/sensor issue you just pick a manual setting that cools well enough for all applications and still doesn't set your ears on fire.

 

[b1k3rdude]

Fixed...

Afternoon chaps..

Basically to fix the fan/temp issue you need to do is the following..

• GPU-Z - http://www.techpowerup.com/downloads/SysInfo/GPU-Z/
• RBE 1.14 - http://www.techpowerup.com/downloads/1177/mirrors.php
• Ati WinFLash - http://www.techpowerup.com/downloads/1122/mirrors.php
  
• Save a copy of the current bios using Gpu-z (its saves the correct 128k file size)
• Then edit the lower 2 fan speeds (60, 65%) to 6%, 12%
• Then flash the edited bios to the card.

my idle/load temps are 43/60c.

biker

 

[btarunr]

The point is, PowerColor should do that before selling you the card and making you go thru that. BIOS flash could make them reject your RMA. Besides, didn't they name this "Professional Cooling System"?

 

[chron]

why can't you just flash the regular bios to this thing? Why does it have to be some kind of special power color bios? The clocks might be different, but you can fix that...

 

[VuurVOS]

why can't you just flash the regular bios to this thing? Why does it have to be some kind of special power color bios? The clocks might be different, but you can fix that...

The problem is the HD4870 PCS+ has another part nummer in the bios. When they check the bios they could reject it because it isn't the orginal bios for the card or a newer HD4870 PCS+ bios

 

[jydie]

They are so worried about the noise level for the video card!!! They force the fan to barely run in idle, and ramp up way to slowly during gaming. I really wish it was the other way around, and they would focus on keeping the temperature at a decent level at ALL times... that way my video card would last longer. If my GPU in hitting 80+ degrees celcius, then I WANT the fan running 100%, because despite what they say I think that is to hot!!

These are high performance cards, I expect a little fan noise.

 

[b1k3rdude]

The problem is the HD4870 PCS+ has another part nummer in the bios. When they check the bios they could reject it because it isn't the orginal bios for the card or a newer HD4870 PCS+ bios

Hence why I used and edited the bios that CAME with the card, the ONLY change I made was the fan speed..! The card has now been stable for the last 2 days.

biker

 

[btarunr]

If any of you want to inspect this new BIOS using RBE or something (at your own risk):

http://www.dataland.com.cn/bios/hd4870a.rar

 

[Nkd]

this could very well be an issue with gpuz, gpu-z 0.2.7 tells me that my fan is running at 0% and nothing for rpm for my 9800gx2, just look at the temps on that screenshot, if it was at 0% those temps would not be in the 50s and 60s, I cant believe that we all jump to conclusions without proper observation

 

[EastCoasthandle]

I find it odd that they use the term "zipzip" to explain the fan speed. The graph clearly shows that they increase the fan speed and, the erratic behavior is not synonymous with the graph. After looking at the graph the fan is suppose to run at a higher RPM. Therefore, what they explain doesn't make any sense to me. Also if this keeps the card within spec we are grasping at straws here. I honestly wouldn't use furmark on that 4870 as I don't see any active cooling on the vregs. As VuurVOS said other programs can affect the card and cause similar results.

 

[WarEagleAU]

Point is, shit should work when you buy it and these should have been caught.

 

[swaaye]

Whole ignored by most, RV770 and the 780G IGP definitely have their issues. For ex, People seem to just happily accept RV770s ridiculous power use just because it's the first ATI chip in YEARS that hasn't sucked compared to the competition. Even though they advertised its "awesome" power saving features. Oh I guess it's "ok" to just let that BS slip buy. I guess it's also why I don't own RV770 in any incarnation.

My 780G mobo has a list of problems that I don't really feel like going into again. Stuff like CnQ failing after videos, 2D performance problems in XP, strange video chop, and some AHCI problems. Wow, worthy of glowing reviews indeed.

The fan speed problem here is nothing new either. How many releases now have had people seeing temp problems? And, the reason they don't just let it go full bore is because it would be DEAFENING. This things run so crazy hot that they either need giant copper heatpipe coolers or cheap ass heatsink + uber blower. You wouldn't buy these cards if they just let those blowers sound like a hair drier full time.

The answer is all-around better cooling. Bigger heatsink. But I have a big ugly feeling that they don't do that for several reasons:
1) The cards don't pump out max heat full-time so the benefit of more cooler isn't entirely there.
2) Shipping costs due to weight of a big beast cooler.
3) The simple fact that it would cost more to make you a better cooler.
4) That it probably doesn't matter if the card burns up in a year. In a year you will probably have moved on. Ha hah.

 

[EastCoasthandle]

Point is, shit should work when you buy it and these should have been caught.

GPU temps at 57C at idle and you believe it's not working? It looks like it's working to me.

My temp is 52C with the stock fan at 65%!!!! I would say that their cooling solution of 57C is working fine if it's done at a lower RPM (whatever that is).

Also, what's being posted is that the fan operates different from the actual graph. Until this is proven conclusively, I remain skeptical about this report.

Red Flags
1. No comparison between reference and non reference cooling solutions.
2. No confirmation of audio or visual inspection of fan speed with new bios.
3. No 4870 utility comparison to confirm fan speed fluctuation.
4. ZipZip noise gone in new bios but no noise heard when fan goes from 100% to 0%
5. No load temps provided in this review
6. No mention if vregs need active cooling or not under load.
7. If the fan does cycle down is it by design? Was the manufacture asked this?

The issue I've seen with Furmark has more to do with the vregs from what I've seen. One thing I notice in their cooling solution is that vregs are passively cooled. ATI/AMD's cooling solution is designed to actively cool the vregs even thought they are of a different design. But I reserve judgment until more information is provided on how hot those vregs get.

Side note:
This article appears to be stacked (ie 2 articles in one). The beginning of it appears to be the old article and page 4 appears to be a new one.

 

[eidairaman1]

I find it odd that they use the term "zipzip" to explain the fan speed. The graph clearly shows that they increase the fan speed and, the erratic behavior is not synonymous with the graph. After looking at the graph the fan is suppose to run at a higher RPM. Therefore, what they explain doesn't make any sense to me. Also if this keeps the card within spec we are grasping at straws here. I honestly wouldn't use furmark on that 4870 as I don't see any active cooling on the vregs. As VuurVOS said other programs can affect the card and cause similar results.

Instead of being a Continuous Speed Increase, its a Pulse, so it speeds up, slows down speeds up higher slows down etc it propbably also sounds like a Whirring Noise. My 1950 Pro Sapphire doesnt suffer from this, the only time i hear the Fan is during powerup (continuous RPM) then once it hits windows it silences.

 

[largon]

Ahhh...
Watercooling saves one from so many nasty headaches.

 

[Mikey242]

Am I right in thinking that this affects all Powercolor 4870 512MB cards? I have one of these and I am also experiencing high temps with it. I went to Powercolor's website but could not find any sign of a bios update. Would I be able to flash this with the bios for another card? Thanks guys, sorry if questions a bit repetitive.

 

[VuurVOS]

Am I right in thinking that this affects all Powercolor 4870 512MB cards? I have one of these and I am also experiencing high temps with it. I went to Powercolor's website but could not find any sign of a bios update. Would I be able to flash this with the bios for another card? Thanks guys, sorry if questions a bit repetitive.

Your can flash a normal bios hd4870 in the current oc cards. I already flashed the Diamond XOC and the Asus Top in my normal Sapphire 4870 card. But keep in mind you will lose your warrenty when the cards gets broken with that bios on it.

You can better use the alternative tools like ATI Tray Tools. You can still use the current bios and the program can set the fan and gpu/ram speeds for you. If you realy want to have a cooler card make a 2D profile (lower gpu/ram speed) and a 3D profile (normal gpu/ram speeds) with ATT and use it with AUTO OVERCLOCKING.

 

[Mikey242]

Thanks VuurVOS for the info. Just checking though, where was the Powercolor bios update posted? I tried to flash the newest one in the download section and it just said vendor mismatch or something to that effect (used winflash). Thanks again guys.