ATI Radeon HD 4890 1 GB GDDR5 in Pretty Pixels

[newtekie1]

Hopefully RV790 brings enough tweaks to an already great RV770 to warrant excitement. The already impressive clock speeds are a nice start, but I want to see real pricing and overclocking performance.

If these are just binned RV770 cards, and I think they are, then I wouldn't think they would have much headroom left, which would be a real shame.

Hopefully this is a new revision of RV770, which even though they are the same on paper, is superior in practice. Similar to the B3 to G0 move Intel made.

Originally Posted by niko084

I'm yet to have any issue with any 4850/4870..
I think people are trying to put them in their Dell's without any air movement in the case that's causing the issues.

Can't count the times I have seen 3850/3870, 8800 and 9800s in small cases with no air movement and wonder why after an hour of gaming they start artifacting.

It used to be that cards overheating in cramped cases wasn't acceptable. Video cards were expected to work, even in the crappiest of airflow situations. Which is why I sure people still expect to put the beefiest card possible in a case with no fan other than the PSU, and have it work without issue.

[DarkMatter]

Originally Posted by Assimilator

It's hardly "next generation", but it should be able to reclaim the title of world's fastest single-GPU card - for now.

I doubt it very much. 3 different cards listed so far and all of them have 850Mhz on the core, or 13% higher clock than HD4870. Without a true redesign, the max improvement they will probably take from this card is 15%-20% over the HD4870, so that's still very far from the GTX285, like another 20%.

Originally Posted by TRIPTEX_MTL

The 4870 never had the title of fastest single GPU card. The GTX280 held that title until the GTX285 was released.

His statement doesn't specify any card. If this card was faster it would reclaim title of fastest single GPU card that Ati lost after X1900XTX.

[EastCoasthandle]

Originally Posted by DarkMatter

I meant that people may not be so concerned about Nvidia clocking their 55nm cards higher, because those ran cool and quiet and consume significantly less than the Ati counterparts for the same performance. But Ati HD4870's with reference cooling, the one we see here, already run at high 80's - low 90's so it's a matter of concern. We should wait until reviews? Of course, but if we don't have to talk about the cards until reviews why post any news then??

Again, what you post makes no logical sense. People understand the arch. differences between competing video cards. And temp differences are dependent on other factors like dust accumulation, ambient room temps, temps inside the case, proper air flow, thermal compound, proper HS contact, etc. Furthermore, there are others who aren't experiencing this problem which makes that mute. As we all should know by now high temps isn't just a problem with any particular make/model.

Now, here is my original post

After reading some of the posts in this thread you would think that nvidia's video cards didn't get a bump up in GPU clocks, etc.

in which you replied

Originally Posted by DarkMatter

Maybe because Nvidia clocks were lower in the first place, which meant cooler, quieter and less consuming cards.

Having said all that you've said so far is not relevant to what I posted.
Changes in clock rates are a common thing in this example. I see no relation that one should or should not do it based on arch and board design. As I've said before we will know more about this when:
A. Insider information is provided
B. Official review

[niko084]

Originally Posted by newtekie1

It used to be that cards overheating in cramped cases wasn't acceptable. Video cards were expected to work, even in the crappiest of airflow situations.

Well yes, but video card also used to not do everything they do today.

Simple point, you don't buy a Ferrari and put 87 octane in it, and you sure don't drive it in the snow, just because it used to be all cars were driven in places without roads doesn't mean all should be today.

You want the best, you need to know how to properly take care of it, if you don't it's your problem.

Now if they were overheating in decently ventilated cases, then I would be concerned. I put them in mid tower cases with a single 120mm rear fan and they never overheat.

Maybe if anything they should add a warning label next to the power label, *This card requires AIR flow*

And at that rate, I have seen 8800's and 8600's overheat in those small cases too.. There is just some things you don't do.

[newtekie1]

I know, I'm not saying it is a standard that should be expected today, I'm just giving reason as to why some still expect it.

[DarkMatter]

Originally Posted by EastCoasthandle

Again, what you post makes no logical sense. People understand the arch. differences between competing video cards. And temp differences are dependent on other factors like dust accumulation, ambient room temps, temps inside the case, proper air flow, thermal compound, proper HS contact, etc. Furthermore, there are others who aren't experiencing this problem which makes that mute. As we all should know by now high temps isn't just a problem with any particular make/model.

Now, here is my original post

in which you replied


Having said all that you've said so far is not relevant to what I posted.
Changes in clock rates are a common thing in this example. I see no relation that one should or should not do it based on arch and board design. As I've said before we will know more about this when:
A. Insider information is provided
B. Official review

Your post comes after like 10 posts stating their concerns about heat and noise issues. So I assumed you were responding to that. Like saying that people should be accostumed to those practices. I cleared why I think they have concers about this card and not the others, based in my own opinion. It makes sense, unless you were not talking about heat/noise and in that case your own post didn't make any sense. There's ONLY one post before yours not talking about heat/noise so the plural "some posts" makes no sense and was what confused me. I hope you understand my point...

[DarkMatter]

Originally Posted by niko084

Well yes, but video card also used to not do everything they do today.

Simple point, you don't buy a Ferrari and put 87 octane in it, and you sure don't drive it in the snow, just because it used to be all cars were driven in places without roads doesn't mean all should be today.

You want the best, you need to know how to properly take care of it, if you don't it's your problem.

Now if they were overheating in decently ventilated cases, then I would be concerned. I put them in mid tower cases with a single 120mm rear fan and they never overheat.

Maybe if anything they should add a warning label next to the power label, *This card requires AIR flow*

And at that rate, I have seen 8800's and 8600's overheat in those small cases too.. There is just some things you don't do.

HD4000 cards have issues in many PCs and not only those with zero airflow. There's problems in hot areas even with relatively good airflow too. Anyway, IMHO a company that is so proud of saying they develop cards for the mainstream, should make their cards work without issues in ANY PC and not only in those of semi-enthusiasts. Almost NONE of my gamer friends clean the inside of their PCs more than once in a year, nor they have uber efficient cases with many fans.

[lemonadesoda]

This card reminds me of Pentium D. Hot. Pushing the limit. Only a % gain, but at a power/heat cost that goes up FASTER than performance. Nothing special.

Come on AMD! Out with new architecture and 40nm!

[Gzero]

Originally Posted by lemonadesoda

This card reminds me of Pentium D. Hot. Nothing special.

Come on AMD! Out with new architecture and 40nm!

Tell that to them at the next meeting with Nintendo and the billions they must be making on that deal alone.

Oh and DarkMatter stop hating on ATI. I can see your specs, I know your preference

Nothing wrong with the hd4x series, nothing wrong with this bump. It's not like Nvidia ever did this (7800 - 7900, 8800gt -> 9600/9800) *cough*

[TheMailMan78]

Anyone know the shader count?

[EastCoasthandle]

Originally Posted by DarkMatter

Your post comes after like 10 posts stating their concerns about heat and noise issues. So I assumed you were responding to that. Like saying that people should be accostumed to those practices. I cleared why I think they have concers about this card and not the others, based in my own opinion. It makes sense, unless you were not talking about heat/noise and in that case your own post didn't make any sense. There's ONLY one post before yours not talking about heat/noise so the plural "some posts" makes no sense and was what confused me. I hope you understand my point...

You have often made a decision to reply to my post having not properly understood what it is I am saying. This is why it's apparent for me that there is a communication break down. Which is why I ask that if you are not sure about the meaning of my post then you are more then welcomed to ask . But as I've stated these bumps in GPU clocks rates have been seen before. And is no cause for alarm as long as it's within proper context of the current gen model (again, IMO).


Having said that, we really don't know the heat and noise regarding this video card until:
A. Someone provides insider information
B. Official review

Regardless of what the outcome will be (and we will know soon enough) it will not stop others from buying the product if this card can produce frame rates that are close to the GTX 285 at a much lower price structure.

[DarkMatter]

Originally Posted by TheMailMan78

Anyone know the shader count?

160 x 5 = 800, just like RV770. It's just a slightly tweaked or binned RV770.

[DarkMatter]

Originally Posted by EastCoasthandle

You have often made a decision to reply to my post having not properly understood what it is I am saying. This is why it's apparent for me that there is a communication break down. Which is why I ask that if you are not sure about the meaning of my post then you are more then welcomed to ask . But as I've stated these bumps in GPU clocks rates have been seen before. And is no cause for alarm as long as it's within proper context of the current gen model (again, IMO).


Having said that, we really don't know the heat and noise regarding this video card until:
A. Someone provides insider information
B. Official review

Regardless of what the outcome will be (and we will know soon enough) it will not stop others from buying the product if this card can produce frame rates that are close to the GTX 285 at a much lower price structure.

Look, we neither know anything about the performance of the card until reviews are posted, so I suggest you stop making assunmptions about how this card will perform until then??

Man we are posting in a news thread with very little info, we are ALL speculating. You base your performance numbers in RV770 and so are we doing for noise/heat. So if you have a problem with that, I could suggest we don't speak about anything in news like this or even better, TPU stop posting any news alltogether...

Nah, we want news and we want to speculate based on past knowledge.

PD: I suggest that if you are going to talk about a topic that is not mentioned in the last 5-10 posts, you quote the one you want to reply. Because threads have a flow and in this particular one that flow went like this before your post: looks like same card -> increased clocks -> hogher clocks, higher temps/noise.

[Selene]

This card should beat the GTX260 55nm, but as normal NV has the GTX275 around the bend, to take it right back.
Its funny, how who ever has the lead sets back waits sees what the other has then bam out comes the new one thats better.

[Paintface]

No one is forced to buy this card you know, so why all the hate?

It is not an OCed 4870 for starters, 750 to 850mhz is quite a big difference , now if it was 750 to 770 like most OCed 4870s have i would agree.

while it has the same specs as a 4870 it is a revised core specificly to handle those high clocks.

We dont know how it will handle the heat, i know from experience my 4870 stock OCed handle it perfectly.

Its a not a smaller diesize, so yes it will need more power at those speeds, and the exact problem with that is? if this was a new X2 card everyone would be like GO ATI WOOOO RAW POWER without anyone mentioning the power usage which would be double of this card.

IF you dont have a card in the 4xxx series and planned to upgrade, this looks like the perfect card , if you have a 4870 already, it wont be a viable upgrade.

[TheMailMan78]

Originally Posted by DarkMatter

160 x 5 = 800, just like RV770. It's just a slightly tweaked or binned RV770.

So we are not looking at "massive" gains. Sounds like the HD2900 all over again.

[phanbuey]

I wonder if it is just a BIOS tweak... maybe one can flash a 4870 to a 4890.

[EastCoasthandle]

Originally Posted by DarkMatter

Look, we neither know anything about the performance of the card until reviews are posted, so I suggest you stop making assunmptions about how this card will perform until then??

Why so serious ? Looks like I hit a sore spot or something. I didn't make an assumptions on performance. I said "...if this card can produce frame rates that are close to the GTX 285 at a much lower price structure..."

Originally Posted by DarkMatter

Man we are posting in a news thread with very little info, we are ALL speculating.

So, you can speculate and I can't say "...if this card can produce frame rates that are close to the GTX 285 at a much lower price structure..." and you want me to stop making what you believe to be assumptions ? I think we can all post our opinions on the situation. Which is why I said that further information will become available when:
A. Insider information is made public
B. Official Review

Originally Posted by DarkMatter

You base your performance numbers in RV770 and so are we doing for noise/heat.

Actually no, that is not true. What I said is that we really don't know anything about heat and noise right now. That also goes for performance. However, it's assumed that there would be some sort of performance increase.

Originally Posted by DarkMatter

So if you have a problem with that, I could suggest we don't speak about anything in news like this or even better, TPU stop posting any news alltogether...

Huh , you are replying to my posts which attracts dialogue. If you don't want to talk about this is fine by me. However, perhaps in the future you shouldn't respond to my posts when I never ask for or made an attempt to dialogue with you .

Originally Posted by DarkMatter

Nah, we want news and we want to speculate based on past knowledge.

We really don't know what the 4890 will offer. What we do know is that the card does exist. The clock rates appear to be higher and performance is assumed better then the previous version. As far as heat/noise is concerned we really don't know at this time. However, IMO if they are using the same HSF combo I really don't see how fan noise will be any different then before. As for heat, I really don't know.

[DarkMatter]

Originally Posted by Selene

This card should beat the GTX260 55nm, but as normal NV has the GTX275 around the bend, to take it right back.
Its funny, how who ever has the lead sets back waits sees what the other has then bam out comes the new one thats better.

Yeah and this time, it seems Nvidia doesn't want to play around. It seems that with GTX275, GT212 and GT300 Nvidia will wait until Ati releases their chips before commiting the same error they did with GT200 and to little extent G92, that is, releasing the new fastest card before Ati took the crown off from them. So Ati waited and released their cards accordingly. This time Nvidia is going to wait and probably come with a surprise. They might even have two different GT300, one with a massive performance increase and one headed to be cheap and take back the crown for a few %, and depending on what strategy Ati follows, they will release one or the other.

[alexp999]

From the information we know now, the RV790 is a briefly tweaked (nothing major) version of the RV770 to allow higher clocks, this could be a case of just adding another voltage option, to sit alongside the new 5 phase power circuit, we simply don't know. It is the same fab, same shader/transistor count etc.

What we can speculate from that, is that it will be a higher performing, but consequently more power consuming, and heat producing chip. Based on the fact that the cooler looks identical, you would also assume that the chip is going to run hotter.

Now we all now the problems the 4870 faced when it started, AMD even had to release a driver that stopped furmark overheating it, to the point it would artifact and crash.

I think what we as consumers are worried about, is that AMD dont make the same make twice, but this time with a potentially hotter chip, resulting in a worse situation than the launch of the 4870's.

[TheMailMan78]

Ill say it again. HD2900 part 2

[WarEagleAU]

I dont think more shaders would help. They would be well served to go ahead and unlock shaders separately. I mean, imagine 800 shaders clocked at like 1.6ghz. Sweetness. This is a refresh with higher clocks. Not bad and I dont hate on them for that. Its a little different than what Nvidia does. Taking the 8800gt and naming it the 9800gt. then wanting to take that and name it the gts 250 or whatever ( I cant recall off the top of my head what they were gonna call it).

[DarkMatter]

Originally Posted by EastCoasthandle

Why so serious ? Looks like I hit a sore spot or something. I didn't make an assumptions on performance. I said "...if this card can produce frame rates that are close to the GTX 285 at a much lower price structure..."
So, you can speculate and I can't say "...if this card can produce frame rates that are close to the GTX 285 at a much lower price structure..." and you want me to stop making what you believe to be assumptions ? I think we can all post our opinions on the situation. Which is why I said that further information will become available when:
A. Insider information is made public
B. Official Review

It was hypothetically speaking. You are asking me not to talk about heat/noise, because we have no info about that, so wouldn't I have the right to do the same regarding performance?

EDIT: Oh and no I'm not serious there, nor didn't I take an offense. I was just hypothetically speaking, as I stated above: if you censure me, why wouldn't I censure you? I know the card will probably be much faster, and I too know it will be hotter. I just want both posibilities to be mentioned. I always want truth, the whole truth, not skewed information, something that Ati and Nvidia fans do very well.

Actually no, that is not true. What I said is that we really don't know anything about heat and noise right now. That also goes for performance. However, it's assumed that there would be some sort of performance increase.

How is it legitimally assumed that the card will be almost as fast as 285 (you said so in the other post) when GTX285 is more than 30% faster than HD4870 and the new one has only 13% higher clocks??

And at the same time how is not legitimate to do an assumption about the heat/noise, based on OCed HD4870's? Higher noise comes from higher rpms in the fan FYI.

Huh , you are replying to my posts which attracts dialogue. If you don't want to talk about this is fine by me. However, perhaps in the future you shouldn't respond to my posts when I never ask for or made an attempt to dialogue with you .

We really don't know what the 4890 will offer. What we do know is that the card does exist. The clock rates appear to be higher and performance is assumed better then the previous version. As far as heat/noise is concerned we really don't know at this time. However, IMO if they are using the same HSF combo I really don't see how fan noise will be any different then before. As for heat, I really don't know.

Once again, I was speaking hypothetically. It's not me the one who said we shouldn't speak/speculate about one aspect of the cards. It was you who pretty much censured our opinions about heat/noise because we don't know anything yet. I just said we neither know anything about performance, so we shouldn't assume/speculate either. I vote for speculating, but uncensored.

[alexp999]

Originally Posted by WarEagleAU

Its a little different than what Nvidia does. Taking the 8800gt and naming it the 9800gt. then wanting to take that and name it the gts 250 or whatever ( I cant recall off the top of my head what they were gonna call it).

Thankfully, board partners have convinced NVIDIA not to follow that through.

[tofu]

Phanbuey, I'm on the same boat heh. Flashing 4870 to 4890 sounds like the REAL DEAL.

*Stares down at my 512MB card*