ATI Radeon 4890 in April?

[AmberClad]

I've removed a few posts that were straying into off-topic side discussions again.

- AmberClad (Video Mod)

 

[chizow]

Originally posted by: BFG10K

Originally posted by: chizow

As for not having to touch the BIOS or use third party tools, you do realize the launch fan problems were fixed in later editions and driver updates right? Hence the need for people to flash their BIOS to begin with.

It wasn't a fan problem as much as it was a problem with the card staying at 3D clocks when idling. Mine did the same, and it was reporting 74C idle. Had it dropped to 2D clocks, it would have been much lower than 74C.

As for relevant info, I'm not going to dig through that thread or go back to launch, if you're that interested feel free to search between 6/20 and 7/10 or so, the weeks around 4850's launch.

You made the claims, so the burden of proof is with you. I skimmed the thread you linked and couldn't find anyone claiming 90C idle temps. The fact is, even if they were, 90C idle is abnormal for either card:

http://www.techspot.com/review...-4850-4870/page10.html

4850 idles @ 74 C, 4870 idles @ 71 C, which mirrors my experiences, numerous other peoples? experiences, and multiple reviews?.

Stop trying to claim 90C idle is normal when it clearly isn?t.

But progressively the idle temp kept going up, 81, 82, 83, etc and now the card is idling at 85-86C after 6 months; while I am just in Windows, not even playing any games.

I have same prob - my 4850 (Visiontek) idle temps (@ default) went from 80 - 82 to 86 - 88. I guess I'll have to finally get the Accelero on ( thats been in my drawer for 3 months)

That's not 90 C like you claimed. You need to retract your claims.

Anyways mine was around 80C at 80% fan speed (too loud) under load. Now its about 100C at 80% fan speed and 90C at 100% fan speed.

I just RMA'd 2 Sapphire 4850 512s with the reference cooler to newegg for a refund, it got so bad since I got em in July that the cards were peaking at 103C and throttling at 100% load with the default fan profile, and even switching to 100% fan speed they'd creep up to 103C after long enough time.

Those are load temps; you were talking about idle temps. You need to retract your claims.

To quote you again: they don't distinguish which temps were used, but regardless, 90C under load overclocked 100MHz to 700MHz (16.67%) with stock fan speed of 29% is certainly much, much better than the 4850 and 4870s which were idling at stock speeds at those temperatures

You were called out on claiming idling at 90C is normal, so stop trying to obfuscate the issue by bringing load temps into it.

As for those load temps, they?re clearly faulty cards given the temps are rising with the age of the card, and exactly the same can happen to nVidia. Those temperatures are clearly above what any reviewer is reporting, and you cannot seriously be suggesting they?re normal.

Claiming they?re normal is no different to claiming the GTX280?s load temperature is 100C-105C because of multiple instances of faulty cards having the issue at launch:

http://forums.nvidia.com/lofiv.../index.php?t70478.html

Idle temp is 62C and it goes up to 105 in no time.

The guy on the phone that handles the RMA's knew about the problem.

[/quote]

There's no need for me to retract any claims, you don't get to come late to the discussion and pick and choose quotes out-of-context as you see fit, sorry. I very clearly stated numerous times the temperature thresholds I was referring to:

• Still, its obvious even their 8800GT results are much better than the 80-90C idle and 90-100C load temps of the 4850s and 4870s at launch, which couldn't even be adjusted until a few driver updates later.

Clearly those temperature thresholds of 80-90C idle and the examples I provided are closer to 90C than the 60C idle temps of the 8800GT certain people were making comparisons to, which was my point all along. You can claim they weren't normal as much as you like but the reality of it is, launch 4800s did not have Power Play to control idle temps and did not have driver-level or BIOS-level fan throttling, meaning they would see high temps in the 80-90C range without sufficient case cooling or an open air environment.

This of course is nothing like the 100C problem on launch GTX 280s as that problem was never passed off as normal as the cards did not function properly if they exhibited such defects. That's very different than every launch 4800 that did not have fans that ramped up under load, leadiing to excessively hot temps that could easily eclipse 90C under load.

 

[Creig]

Just what in the world are you trying to prove here, Chizow? Both the G92 8800GT and the 4850 originally came with small, single slot coolers and dinky fans. Both cards ran hot. Neither card could adjust the fan speed upon launch. And what does any of this have to do with the 4890? You know, the subject of this thread.

 

[chizow]

Originally posted by: Creig
Just what in the world are you trying to prove here, Chizow? Both the G92 8800GT and the 4850 originally came with small, single slot coolers and dinky fans. Both cards ran hot. Neither card could adjust the fan speed upon launch. And what does any of this have to do with the 4890? You know, the subject of this thread.

Again, the 8800GT's fan was not only better at stock, it could also be adjusted from Day 1 in either nTune or RivaTuner. The 8800GT ran hot sure, compared to a dual slot G80, but it still came nowhere close to the well-documented temps the 4800s saw at launch. Making such comparisons is a joke, really.

As for what this has to do with 4890? My point was a 4890 based on the same 55nm process as RV770 was very much in doubt due to heat and power thresholds, at which point the 8800GT was brought into the discussion as if it would somehow disprove such claims even though it shared none of the same characteristics.

 

[Creig]

Originally posted by: chizow
The 8800GT ran hot sure, compared to a dual slot G80, but it still came nowhere close to the well-documented temps the 4800s saw at launch. Making such comparisons is a joke, really.

http://www.hardwarecanucks.com...cs-card-review-22.html

G92 8800GT

Idle - 48
Load - 81

HD4850

Idle - 50
Load - 88


That looks pretty darn close. Hardly a "joke" comparison.

Originally posted by: chizow
As for what this has to do with 4890? My point was a 4890 based on the same 55nm process as RV770 was very much in doubt due to heat and power thresholds, at which point the 8800GT was brought into the discussion as if it would somehow disprove such claims even though it shared none of the same characteristics.

Well, now you see that the 8800GT and 4850 weren't all that dissimilar in their initial thermal characteristics using their launch HSFs. So we can now get back to discussing the 4890?

 

[thilanliyan]

Originally posted by: chizow
The 8800GT ran hot sure, compared to a dual slot G80, but it still came nowhere close to the well-documented temps the 4800s saw at launch. Making such comparisons is a joke, really.

Links to several reviews just from this page of the thread and what I provided on the previous page (are you suggesting ALL the reviews are wrong? And these are reviews from launch day so no fan fixes as you pointed out.) show that the cards don't even get to 90C at load. If people's cards were getting to over 100C at load, either it was faulty or they had VERY inadequate airflow in their cases. If they were "well-documented" as you put it provide links to several review sites backing up your claims.

 

[MegaWorks]

I`m using two 9800GT and they idle at 75c and load 88c and no they`re not nVidia stock heatsinks. My 4850s were much more cooler than that, they idle around 40c and 66c load with different stock coolers.

My 9800GT

My 4850

Does it make any sense to compare a G92 to a RV770?

Just what in the world are you trying to prove here, Chizow? Both the G92 8800GT and the 4850 originally came with small, single slot coolers and dinky fans. Both cards ran hot. Neither card could adjust the fan speed upon launch. And what does any of this have to do with the 4890? You know, the subject of this thread.

Isn't obvious what his trying to prove.

 

[chizow]

Originally posted by: Creig
http://www.hardwarecanucks.com...cs-card-review-22.html

G92 8800GT

Idle - 48
Load - 81

HD4850

Idle - 50
Load - 88


That looks pretty darn close. Hardly a "joke" comparison.

Well, now you see that the 8800GT and 4850 weren't all that dissimilar in their initial thermal characteristics using their launch HSFs. So we can now get back to discussing the 4890.

Originally posted by: thilan29
Links to several reviews just from this page of the thread and what I provided on the previous page (are you suggesting ALL the reviews are wrong? And these are reviews from launch day so no fan fixes as you pointed out.) show that the cards don't even get to 90C at load. If people's cards were getting to over 100C at load, either it was faulty or they had VERY inadequate airflow in their cases. If they were "well-documented" as you put it provide links to several review sites backing up your claims.

4850 90C Load
4850 87C Load
Skin Burning Hot
etc. etc.

There's no doubt the 8800GT was hotter than usual for Nvidia parts, but clearly nothing like the heat problems seen with the 4800s, a negative aspect emphasized in just about every launch review.

So lets get back to discussing that 4890, shall we?

 

[SSChevy2001]

Originally posted by: chizow
There's no doubt the 8800GT was hotter than usual for Nvidia parts, but clearly nothing like the heat problems seen with the 4800s, a negative aspect emphasized in just about every launch review.

So lets get back to discussing that 4890, shall we?

Well HardOCP actually show the 8800GT being 7c hotter under load.
http://www.hardocp.com/article...l?art=MTUzNyw4LCwxNjA=

Also if your considering buying a 4850 new. it's going to be hard to find a reference cooler anymore.

 

[josh6079]

Originally posted by: chizow
The difference is it [8800 GT] still ran within reasonable temps at both idle and load, even while overclocked, which is a stark contrast compared to similar fan speed problems with the 4850 and 4870 where idle/load temps were hitting 80-90/100C respectively.

...I very clearly stated numerous times the temperature thresholds I was referring to: Still, its obvious even their 8800GT results are much better than the 80-90C idle and 90-100C load temps of the 4850s and 4870s at launch, which couldn't even be adjusted until a few driver updates later.

...clearly nothing like the heat problems seen with the 4800s, a negative aspect emphasized in just about every launch review.

55 Celsius @ idle

63 & 65 Celsius @ idle

62 Celsius @ idle

All the above were the findings from established review sites for the 4800 series at launch. (except the Anandtech one, it was a few days before the official launch date)

Originally posted by: chizow
...launch 4800s...would see high temps in the 80-90C range without sufficient case cooling or an open air environment.

Sufficient case cooling helps temps from both vendors. Climbing GPU temps from lack of efficient case cooling doesn't mean it's the card's fault.

What I believe everyone is getting at is that the 90C idle temp you discussed was and is not a normal operating temperature for a reference 4800 series GPU. However, your comment implied it was. If someone is getting that range of idle temperature, it is either due to a faulty card or the user's PC environment.

If you disagree - if you consider the testimonies of users to be completely valid representations of the reference 4800 series - then it appears Cookie Monster is correct in your selective hearing.

Originally posted by: Cookie Monster
Yet you come out here and claim it as some "fact" when its clear that your just guessing basing your evidence from the internet users. Some of these same end users your basing your claim on have been denying your claims about "very little success overclocking" but you refuse to listen.

 

[SunnyD]

I'm not reading this whole thread, it's another one of those stupid NV vs. ATI threads.

But has anyone ever stopped to consider a possible 11th (or more) shader core and/or texture unit may have been in the RV770 die this whole time? The way AMD was pitching their development of the 4800-series (read AnandTech interview), it sounded to me like they wanted to make their chip yield as efficient as possible. It wouldn't surprise me if they've been collecting and binning chips with the extra core active... "just in case".

 

[SlowSpyder]

Originally posted by: SunnyD
I'm not reading this whole thread, it's another one of those stupid NV vs. ATI threads.

But has anyone ever stopped to consider a possible 11th (or more) shader core and/or texture unit may have been in the RV770 die this whole time? The way AMD was pitching their development of the 4800-series (read AnandTech interview), it sounded to me like they wanted to make their chip yield as efficient as possible. It wouldn't surprise me if they've been collecting and binning chips with the extra core active... "just in case".

Maybe you should read the thread. Short answer, yes. Some people said they don't beleive it's likely due to the power requirements on the 55nm process to reach the clock speeds. Some think it'll be fine as the process has matured over time. And everything in between.

 

[chizow]

Rofl its obvious some people want to keep discussing the 8800GT vs. 4850/4870 rather than the 4890, so for the last time:

4850/4870 Before and After Temps

These are actual results from users around launch, not open air test benches that clearly show temps ranging from 80-100C were quite the norm. I've linked a similar thread for the 8800GT where users were not getting anything close to those temps even overclocked. Threads and results like this were a dime a dozen at launch, heat was clearly a bigger issue with 4800 than with the 8800GT or any other recent product launch to-date.

 

[Puffnstuff]

Well I hope that ati pulls a rabbit out of their hat on this one. If they do then we all win regardless of which gpu we prefer. It will drive down the prices on existing cards and some of us will buy at that time.

 

[Creig]

Originally posted by: chizow
Rofl its obvious some people want to keep discussing the 8800GT vs. 4850/4870 rather than the 4890, just so I can get the last word in:

Fixed for you.

 

[chizow]

Originally posted by: Creig

Originally posted by: chizow
Rofl its obvious some people want to keep discussing the 8800GT vs. 4850/4870 rather than the 4890, so just so I can get the last word in:

Fixed for you.

Last word? Doubtful, people like you will continue to try and re-write history.

 

[josh6079]

Originally posted by: chizow
These are actual results from users around launch, not open air test benches...

Earlier you had claimed that the temperature "debacle" was widely backed by review sites,

Originally posted by: chizow
...clearly nothing like the heat problems seen with the 4800s, a negative aspect emphasized in just about every launch review.

but since then you haven't provided a link to a trustworthy review site to support this. Now it seems you're saying that review sites' conclusions aren't as valuable as end users because they might conduct said tests in open-air rigs.

It's just hazy on where you consider the more accurate information to originate from, review sites or end users - especially when you credit one source at one point in time then discredit it at another.

 

[chizow]

Originally posted by: josh6079

Originally posted by: chizow
These are actual results from users around launch, not open air test benches...

Earlier you had claimed that the temperature "debacle" was widely backed by review sites,

Originally posted by: chizow
...clearly nothing like the heat problems seen with the 4800s, a negative aspect emphasized in just about every launch review.

but since then you haven't provided a link to a trustworthy review site to support this. Now it seems you're saying that review sites' conclusions aren't as valuable as end users because they might conduct said tests in open-air rigs.

It's just hazy on where you consider the more accurate information to originate from, review sites or end users - especially when you credit one source at one point in time then discredit it at another.

Firing Squad
Legit Reviews
THG
Tech Report

Trustworthy enough for you? You can see the language in the articles, even the ones linked earlier do back my claim that temperatures were a major issue at launch with stock fans and profiles on the 4800s.

We can link to review sites all day, but its clear they show the 4850 and 4870 run hotter than their Nvidia counterparts and but still will not accurately reflect actual temps as the results were taken in open air test benches, which is why I instead linked to actual user results.

 

[thilanliyan]

nm

 

[josh6079]

Originally posted by: chizow
4850/4870 Before and After Temps

That doesn't support your claim that the 4800 series cards normally idle at 90C. Most users there were reporting high 70's to low 80's.

Originally posted by: chizow
Firing Squad
Legit Reviews
THG
Tech Report

Trustworthy enough for you?

Yes, and they don't support your claim that 4800 series cards normally idle at 90C. That's why BFG10K suggested you retract that statement as it's an exaggeration or abnormal.

Besides, you linked the wrong page on the Firing Squad article. The page you linked was Brandon explaining that users have reported 70-degree Celsius range at idle, further disproving your past statement.

Firing Squad
Just how hot does the Radeon HD 4850 get? Some users have reported GPU temps in the 70-degree Celsius range at idle!

What you didn't link was the page that contained Firing Squad's results here, which shows an idle temp of 64C on the reference 4850.

The other sites also did not support your claim.

Legit Reviews: 78C @ idle
THG: 81C @ idle

And Tech Report didn't even post an idle temp, but showed a load temp of 90C.

Originally posted by: chizow
You can see the language in the articles...

I can also see their facts, which don't line up with your previous claim:

Originally posted by: chizow
They don't distinguish which temps were used, but regardless, 90C under load overclocked 100MHz to 700MHz (16.67%) with stock fan speed of 29% is certainly much, much better than the 4850 and 4870s which were idling at stock speeds at those temperatures lol. :laugh:

From there, you have not offered anything conclusive indicating that's a true statement. In fact, the latest links you've provided have done the opposite (i.e., Firing Squad citing users' normal ranges as well as their own)

Originally posted by: chizow
...the ones linked earlier do back my claim that temperatures were a major issue at launch with stock fans and profiles on the 4800s.

"Issue" is a loose term considering the cards could perform fine at the temperatures reported by numerous users and reviewers alike. The later access to the fan was primarily for ease-of-mind and convenience. The heating "issue" did not incorporate an idle temp of 90C.

Originally posted by: chizow
...its clear they show the 4850 and 4870 run hotter than their Nvidia counterparts...

That much is true, but it doesn't make it 90C-@-idle hotter. As BFG10K mentioned, that previous claim of yours is simply wrong to consider as a normal operating temperature for 4800s.

 

[chizow]

Originally posted by: josh6079
Bunch of nonsense

I'll just repost what I posted earlier in reply to BFG10K, since it applies here as well:

There's no need for me to retract any claims, you don't get to come late to the discussion and pick and choose quotes out-of-context as you see fit, sorry. I very clearly stated numerous times the temperature thresholds I was referring to:

• Still, its obvious even their 8800GT results are much better than the 80-90C idle and 90-100C load temps of the 4850s and 4870s at launch, which couldn't even be adjusted until a few driver updates later.

Clearly those temperature thresholds of 80-90C idle and the examples I provided are closer to 90C than the 60C idle temps of the 8800GT certain people were making comparisons to, which was my point all along. You can claim they weren't normal as much as you like but the reality of it is, launch 4800s did not have Power Play to control idle temps and did not have driver-level or BIOS-level fan throttling, meaning they would see high temps in the 80-90C range without sufficient case cooling or an open air environment.

Now once again, do you think temps in an open air environment are going to accurately mirror actual temps from users in closed case environments? Even said reviews clearly acknowledge the potential temperature deltas. Its completely laughable that you and others are going to try and hold me to literal comments like "90C exactly!" when I clearly outlined temperature thresholds in the 80-90C range, going as high as 100C under load. Its even more laughable that you're going to point to open air benchmark results and claim there wasn't a temperature problem with the 4800s because they don't hit "90C on the dot" when they all state heat is clearly an issue.

 

[thilanliyan]

Originally posted by: chizow
Its even more laughable that you're going to point to open air benchmark results and claim there wasn't a temperature problem with the 4800s because they don't hit "90C on the dot" when they all state heat is clearly an issue.

From the initial HardOCP 8800GT review:

"As a brief overview of what we experienced we found an idle GPU temperature of 61c and a full load GPU temperature of 91c. That full load temperature is very hot for a video card, and keep in mind we are testing in an open-case environment."

8800GT heatsink roundup:
"Many of you know that NVIDIA's 8800 GT can run rather hot under load. Our GTs ran at over 90C. That spells disaster in a hot case."

8800GT OC roundup:
"As you can see above these are very hot-running video cards. They idle at above 60c and get up to around 90c at full load. The memory is in the mid 50s at idle and rises to the mid 70s under load. These temperatures were taken under the best conditions (an open case); if you have a warm case already you might see even higher temperatures and a hot case."