ATI Radeon HD 5890 Coming in Q1 2010

[Chrono Detector]

I've seen that page edited so many times I don't believe it anymore. I will believe it when reliable hardware sites release official information first.

[zanzabar]

i call BS u cant have a 384bit buss and 1/2GB of ram u would need to use multiple densities of chips or some magical 12 or 24bit ram chips and it just dosnt work that way. any1 who has worked with ram should know how it works and that u cannot do that, with a 384bit buss it would have 768MB or 1.5GB or 3GB

[Manicdan]

Before the 5000 series were released I do remember reading about the 5890 and that it would have 384 bit bus....

the only person i hear say that it will was me, and i was just speculating/demanding

[Tenknics]

This is some what off topic but someone mentioned GloFlo skipping 32nm and going strait to 28nm and I wanted to use this opportunity to ask a question. Is that with GPU only? Isn't bulldozer being launched on their 32nm process?

Also there was some news awhile back about the maturity and breakthroughs of GloFlo's 32nm? Why does every keep saying they are skipping it?

[tajoh111]

Before the 5000 series were released I do remember reading about the 5890 and that it would have 384 bit bus....

That whole chart is BS those, its not even something made by Donhaiber. Its worse than the Nvidia rumors specs because there are so many inconsistancies + implausibilities. The 28nm, the cards specs just not making sense. The 5990 which was never really in the pipeline. Also the specs of the 5990 don't even make sense, it has more or less twice the transisters of the 5860(which was never rumors), but has the specs of twice the 5890 which was always rumors from the beginning.

I think this is more of a counter by an AMD fan to counter the supposed specs of the gtx 380 which are almost equally as unreliable, but atleast better written and more plausible.

If AMD could make 28nm cards at this point or soon, NV would of had fermi cards long long ago. 28nm is jumping down a full node and I doubt it would be this easy. 28nm is likely to give them more headaches than 40nm because they are skipping the 32nm step.

65-55-40nm already had enough headaches and they still did the node in the middle. 55nm and 55nm B.

If AMD wants to make money forget about making videocards on 28nm, make their processor on this process and you will see AMD profit for once.

[skugpezz]

That whole chart is BS those, its not even something made by Donhaiber. Its worse than the Nvidia rumors specs because there are so many inconsistancies + implausibilities. The 28nm, the cards specs just not making sense. The 5990 which was never really in the pipeline. Also the specs of the 5990 don't even make sense, it has more or less twice the transisters of the 5860(which was never rumors), but has the specs of twice the 5890 which was always rumors from the beginning.

I think this is more of a counter by an AMD fan to counter the supposed specs of the gtx 380 which are almost equally as unreliable, but atleast better written and more plausible.

If AMD could make 28nm cards at this point or soon, NV would of had fermi cards long long ago. 28nm is jumping down a full node and I doubt it would be this easy. 28nm is likely to give them more headaches than 40nm because they are skipping the 32nm step.

65-55-40nm already had enough headaches and they still did the node in the middle. 55nm and 55nm B.

If AMD wants to make money forget about making videocards on 28nm, make their processor on this process and you will see AMD profit for once.

This is some what off topic but someone mentioned GloFlo skipping 32nm and going strait to 28nm and I wanted to use this opportunity to ask a question. Is that with GPU only? Isn't bulldozer being launched on their 32nm process?

Also there was some news awhile back about the maturity and breakthroughs of GloFlo's 32nm? Why does every keep saying they are skipping it?

for you kind sirs .
http://techreport.com/discussions.x/16758
http://www.overclockersclub.com/news/25268/
http://www.physorg.com/news170353677.html
http://www.tomshardware.com/news/28n...ries,7543.html
http://www.semiaccurate.com/2009/11/...eir-32nm-node/
http://www.pcgameshardware.com/aid,6...-process/News/

[MirageSys]

There are several technical screw ups on this chart. Its not even well made speculation. I cry BS.

[skugpezz]

There are several technical screw ups on this chart. Its not even well made speculation. I cry BS.

I cry partial BS lol

[eleeter]

Don't see a 384bit memory interface being possible, that would be a new die most likely, not gonna happen.

[eric66]

donanimhaber doesn't say anything about that chart so source and pic aren't related to each other. news only say that there is a big chance for ati releasing 5890

[Sh1tyMcGee]

i call BS u cant have a 384bit buss and 1/2GB of ram u would need to use multiple densities of chips or some magical 12 or 24bit ram chips and it just dosnt work that way. any1 who has worked with ram should know how it works and that u cannot do that, with a 384bit buss it would have 768MB or 1.5GB or 3GB

Your the only one who caught it Why no one else did is surprising 5890 will not be released, not intill ATi knows what Fermi brings to the table.

[tajoh111]

for you kind sirs .
http://techreport.com/discussions.x/16758
http://www.overclockersclub.com/news/25268/
http://www.physorg.com/news170353677.html
http://www.tomshardware.com/news/28n...ries,7543.html
http://www.semiaccurate.com/2009/11/...eir-32nm-node/
http://www.pcgameshardware.com/aid,6...-process/News/

What the heck, your own sources out do your argument.


The only scary thing that is looming now is that 28nm is a half-step from 32nm. Any problems that 32nm encountered are going to be the same or worse at 28nm, as it is basically the same thing with even smaller tolerances. If TSMC doesn't have an 'aha' moment soon, the second half of 2010 could be hard times for the stinky tofu vendors at the night markets in and around Hsinchu.S|A

Most of them say at best they might get production starting at the end of q1 2010 and that means chips are not going to be released till months later.

I already pointed out AMD is jumping to 28nm and skipping 32 which is going to make the transition harder.

Do you read the articles you posted. Most are saying latter half of the year for 28nm at best. Some of your articles even say for them to skip 32nm and go directly to 22nm for a 2012 release.

[Serpentarius]

agree. did you guys look at the load TDP for HD 5990? >300watt!! (i cannot imagine if a user has 2-3 CS of these ... can anyone imagine the heat dump)
how many rads? lol

well, atleast this table beats the Fermi Mythos.

[skugpezz]

What the heck, your own sources out do your argument.


The only scary thing that is looming now is that 28nm is a half-step from 32nm. Any problems that 32nm encountered are going to be the same or worse at 28nm, as it is basically the same thing with even smaller tolerances. If TSMC doesn't have an 'aha' moment soon, the second half of 2010 could be hard times for the stinky tofu vendors at the night markets in and around Hsinchu.S|A

Most of them say at best they might get production starting at the end of q1 2010 and that means chips are not going to be released till months later.

I already pointed out AMD is jumping to 28nm and skipping 32 which is going to make the transition harder.

Do you read the articles you posted. Most are saying latter half of the year for 28nm at best. Some of your articles even say for them to skip 32nm and go directly to 22nm for a 2012 release.

Yea I read them, I was just pointing out the jump to 28nm.

[NickF]

ATI would never use such an obscure memory link like nVidia did, simply because its an inefficient use of die area. If they were to increase bus size, they would probably try a 512-bit bus (non-ring this time), but after the HD 2900 debacle I don't see this as a possibility. I think the current 256-bit bus is here to stay. It follows ATI's bacng-for-buck silicon plan, and with the massive gDDR5 bandwidth I don't see why you would need anything wider. Even nVidia agrees, as Fermi is a 256-bit product.

Also, TSMC's head is still in the sand on current 40nm fab processes, hell 58xx products are still out of stock across the net, and nVidia's Fermi A1 and A2 silicon yields were insanely low. So I hardly see it possible that they can have any kind of real manufacturing capability on an even more complex fab process.

Just my two cents.

[skugpezz]

Lol^

[SamHughe]

[NickF]

Hahah i love you.

[jaredpace]

good laughs

[MirageSys]

My man, i think Penn and Teller have quite the position for you. They should seriously do an episode of bull on internet rumors. Hell they could dedicate an episode to moronic crap posted on fudzilla and other speculative oracles. (theres enough source material, not that they would).

In addition notice the number of transistors for the 5860 and the 5890. Are they trying to BS us into thinking amd would be developing not one but two 384 bit chips?

[Russian]

Drugs are bad, mkay?

[Mungri]

Sam makes me feel good about finding him the picture of those freaky japanese gladiator things lol.

28 nm by Q1 when 40 nm yields are terrible? 384 bit GDDR5 with 1 / 2 Gb?

Juniper XTX with 256 bit when we already know that 128 bit 5770 is meant to be the top Juniper chip?



Those specs would need brand new chips, current ones cant just be upgraded to less nm and more bits in a refresh.
5890 / 5790 will be redesigned parts with higher frequencies like the 4870 / 4890.

A 5830 type card with fewer shaders and 256 bit interface will be required to bridge the gap between 5770 and 5850, if it ever happens, because Juniper is limited to 128 bit.

[Wiggy McShades]

I know amd was saying recently in an interview that they would be moving ati's products over to global foundries as soon as they could, so could this update be on 55nm just with global foundries more advanced 40nm or 45nm process. Maybe a redesign allowed them to free up some space and get the extra memory channels? If a 5000 series comes out with a 384bit buss ill be buy that over fermi for sure

[Tenknics]

No ones answered my question about GloFlo skipping 32nm or not. Those articles listed contradict each other. I thought bulldozer was coming out on 32nm?

Edit: Answered my own question. Recent roadmaps show AMD with Bulldozer in 2011 at 32nm. TSMC isn't going to make these chips I'm assuming which leaves GloFlo. Why does everyone keep saying GloFlo is skipping 32nm?

Also what the heck, I just remembered..Llano cores are 32nm Phenom IIs. GloFlo must be skipping 32nm for GPU only, not CPU..is that what everyone means?

[god_43]

this originally came from a Turkish site, the picture was not in the article.....it was posted by a forum poster.




http://www.donanimhaber.com/ATi_Rade...emde-16991.htm