All AMD R6xx chips are 65 nanometre

[Ubermann]

WHEN YOURS TRULY learned of a new respin of the chip in early January, we learned that this respin was caused by the fact that the 80 nanometre chip was bleeding current like a slaughtered pig bleeds before he or she heads to the Munchner Halle to be served up as a sausage.

However, we did not know then that AMD did not order just a respin, but the company got an attack of bravery, didn't take the R520/R580 route, and scrapped the R600 altogether.

Read the good news at The INQ (http://www.theinquirer.net/default.aspx?article=38292)

Sounds clever..it will be faaaast as a shark [Accept] :rocker:

[Cooper]

German website have similar topic:

If we assumed so far the R600 would come into 80 Nm, we were instructed in the exclusive round of a better. How it seems, AMD wants to bring the chip from the outset in 65 Nm. Reason for this are problems and/or the bad yield 80 Nm of the process, which is not to have probably insignificant difficulties with leakage currents. The R600 is manufactured by the way further with TSMC.

Can`t understand from the translation if AMD just wants or it actually will produce R600 chips using 65nm techprocess.

Original article (http://www.k-hardware.de/news.php?s=K231c50b10c20ce1df5a4d9fbcacb3e09&news_id=6395)
Google translation (http://translate.google.com/translate?u=http%3A%2F%2Fwww.k-hardware.de%2Fnews.php%3Fs%3DK231c50b10c20ce1df5a4 d9fbcacb3e09%26news_id%3D6395&langpair=de%7Cen&hl=en&ie=UTF-8&oe=UTF-8&prev=%2Flanguage_tools)

[p8ntslinger676]

That would be great but it is from the inquirer so it might or might not be true

[K404]

That would be a nice turnaround for the product and the company :D

[madcho]

i saw it a minute ago ...

incredible if this come true :D

i was thinking the R600's late was because AMD choose to not produce in 80nm :banana:

Could be very very interesting :woot:

[XeRo]

http://www.xtremesystems.org/forums/showthread.php?t=133316&page=22

Probably is true. Last couple pages of that thread.

[madcho]

German website have similar topic:



Can`t understand from the translation if AMD just wants or it actually will produce R600 chips using 65nm techprocess.

Original article (http://www.k-hardware.de/news.php?s=K231c50b10c20ce1df5a4d9fbcacb3e09&news_id=6395)
Google translation (http://translate.google.com/translate?u=http%3A%2F%2Fwww.k-hardware.de%2Fnews.php%3Fs%3DK231c50b10c20ce1df5a4 d9fbcacb3e09%26news_id%3D6395&langpair=de%7Cen&hl=en&ie=UTF-8&oe=UTF-8&prev=%2Flanguage_tools)

R600 80nm was beta, all R600 cards will be 65nm :slobber: :woot:

[w0mbat]

Old: http://www.vr-zone.com/?i=4796

[Mikesnav]

If true, I'm def buying one.

[ColdWinter]

German website have similar topic:



Can`t understand from the translation if AMD just wants or it actually will produce R600 chips using 65nm techprocess.

Original article (http://www.k-hardware.de/news.php?s=K231c50b10c20ce1df5a4d9fbcacb3e09&news_id=6395)
Google translation (http://translate.google.com/translate?u=http%3A%2F%2Fwww.k-hardware.de%2Fnews.php%3Fs%3DK231c50b10c20ce1df5a4 d9fbcacb3e09%26news_id%3D6395&langpair=de%7Cen&hl=en&ie=UTF-8&oe=UTF-8&prev=%2Flanguage_tools)

They'll be bringing the cards out using the 65nm process, not just want to. The reasons being the mentioned leakage problems. In addition the article mentions rumors of AMD still having problems with the 512 Bit memory interface.

To placate the public for not publicly presenting the new chips, it seems AMD has promised to hold a Tech-Day within the next 16 days as well as saying they'll present the cards for launch within the next 6 weeks. The article then goes to assume that this will mean a launch around the beginning of May.

Source: Your link:p:

[Frank M]

Sounds clever..it will be faaaast as a shark [Accept] :rocker:

Ok, who'll be the first to start the Shark Song? :D
[Tits] - [GTFO]
:D :D :D

Cooper:
the german article says they want to make chips on 65nm from the start as they
are having problems with 80nm (yield, leakage). I think you can also say they
will. :)

[frankR]

'Satan Clara': that's cute. I'll be there on bussiness monday.

See has already been discusessed in the 'stickyed' R600 thread.

[cantankerous]

I'm surprised going smaller in size hasn't caused more issues than when the chip was larger at 80nm. I would think it would be harder to cram all the chip has to offer in a smaller package than originally planned.

I am wondering, if going to the new process means the chip uses 1/3 less power does that mean it will take less than 1 + 8pin and 1 + 6 pin to power or will it still need a monsterous amount of power to work?

[thephenom]

It's the IQ, I'll take it with a hand full of salt.

[w0mbat]

It's the IQ, I'll take it with a hand full of salt.

No, its not the INQ!

http://www.vr-zone.com/?i=4796
http://www.techpowerup.com/?27451

[SpikeP]

Looks like it could be legit. i was looking at a camera to spend my money on, looks like i may chage my mind :toast:

- Paul

[xVeinx]

I'm surprised going smaller in size hasn't caused more issues than when the chip was larger at 80nm. I would think it would be harder to cram all the chip has to offer in a smaller package than originally planned.

I am wondering, if going to the new process means the chip uses 1/3 less power does that mean it will take less than 1 + 8pin and 1 + 6 pin to power or will it still need a monsterous amount of power to work?

I imagine that they won't have as many problems as they now have the expertise of AMD in the mix.

[kromosto]

if what i expect from r600 80nm will came as same with 65nm then this is not good for ati. also this explains why nvidia only makes an overclocked and 65nm g80 to deal with r600 probably that will be more then enough.

[w0mbat]

if what i expect from r600 80nm will came as same with 65nm then this is not good for ati. also this explains why nvidia only makes an overclocked and 65nm g80 to deal with r600 probably that will be more then enough.

Of cause its good for ATi, cause its cheaper. And nVidia is preparing G90 cause G80 couldnt deal with R600.

[Kanten]

It's the IQ, I'll take it with a hand full of salt.

Weren't the INQ the first ones to break the AMD/ATI merger?

[[XC] gomeler]

Of cause its good for ATi, cause its cheaper. And nVidia is preparing G90 cause G80 couldnt deal with R600.

ROFL you make it sound as if R600 is already out and has been stomping on G80.

I do hope by dropping to the 65nm process we will end up with slightly cooler running cards, I'd love to be able to game w/o a LN2 pot strapped to my VGA.

[Mikesnav]

And nVidia is preparing G90 cause G80 couldnt deal with R600.

Well, let's hope so. R600 is gonna be 7 months late to the DX10 party.

[w0mbat]

gomeler;2073157']ROFL you make it sound as if R600 is already out and has been stomping on G80.

I do hope by dropping to the 65nm process we will end up with slightly cooler running cards, I'd love to be able to game w/o a LN2 pot strapped to my VGA.

And kromosto sounds like R600 an G90 are already out.

[FghtinIrshNvrDi]

Well, let's hope so. R600 is gonna be 7 months late to the DX10 party.

Depends what you mean by 'late'. I haven't seen a game yet. Is the word you're looking for 'punctual'?

Ryan

[madcho]

Well, let's hope so. R600 is gonna be 7 months late to the DX10 party.


Don't think so. nvidia don't have any driver, no games DX10 are out. Crysis could go out in april but i don't think anybody will run in dx10 mod :rolleyes:

without drivers, nvidia can't run dx10, and Ati without hadware ... same :slapass:

I think the DX10 party will begin at the end of year ;)

[Cooper]

I certainly hope that better yields will make prices lower and availability wider :)

[citat3962]

indeed... if this is in fact true and the power consumption goes down we can finally quit seeing the crybabies posting about the power consumption...

Like you'd even notice the 10$ hit on your power bill every year from gaming between 2-3 hours a day...

Honestly if your power bill goes up from a new GPU you need to get outside more..

That or you're really Extreme and you probably aren't :banana::banana::banana::banana::banana:ing like some...

[largon]

Maybe now the other PCIe power plug can be scrapped too?
No?

edit:
PCIe2 x16 slot: 100W
+
PCIe2 8pin plug: 100W
The 6 pin could actually be history...

[kromosto]

Of cause its good for ATi, cause its cheaper. And nVidia is preparing G90 cause G80 couldnt deal with R600.

yes but if 80nm to 65nm overclocked g80 is enough to beat nearly 1 year late r600 then it is certain that ati is losing dx10 ground. no matter what you call it x2800xtx or x2900xtx by reading latest news i am predicting as its end will come faster then x1800xtx.

[XS Janus]

Hmmm... in that same thread on the page 23 http://www.xtremesystems.org/forums/showthread.php?t=133316&page=23 slides, and:

quote: The California chip chipschmiede did not only confirm OPPOSITE that the R600 will come into 65 Nm, but also communicated that AMDs will have first DirectX 10 GPU a dedicated video processor. „Universal the video in such a way specified Decoding “unit should be trimmed particularly on the rendition of Full HP material (1920x1080). Thus the ATI GPU gets for the first time a dedicated video unit.

From those slides and the qoute from a few posts lower I understood that they are on track for 65nm in notebooks and that AMD confirmed that R600 will not come in 65nm.

Whats the truth now? English is not my first language so...

[xB0n3s]

If it turns out 65nm then it will be blody fast.

[deathman20]

If true that one post we saw on the 1.35Ghz core could be ture on an OC.

[w0mbat]

Hmmm... in that same thread on the page 23 http://www.xtremesystems.org/forums/showthread.php?t=133316&page=23 slides, and:

quote: The California chip chipschmiede did not only confirm OPPOSITE that the R600 will come into 65 Nm, but also communicated that AMDs will have first DirectX 10 GPU a dedicated video processor. „Universal the video in such a way specified Decoding “unit should be trimmed particularly on the rendition of Full HP material (1920x1080). Thus the ATI GPU gets for the first time a dedicated video unit.

From those slides and the qoute from a few posts lower I understood that they are on track for 65nm in notebooks and that AMD confirmed that R600 will not come in 65nm.

Whats the truth now? English is not my first language so...

U are wrong. German is my first language and they are saying: "AMD confirmed that R600 will be a 65nm product."

[XS Janus]

U are wrong. German is my first language and they are saying: "AMD confirmed that R600 will be a 65nm product."

:toast: Hope it's true! Can't accsess the original page, nor site now, but if this will be true than :woot: for normal power ussage an great performanse. I might get and ATi at last:woot: Thanks for the info:cool:

[DilTech]

Of cause its good for ATi, cause its cheaper. And nVidia is preparing G90 cause G80 couldnt deal with R600.

NVidia are prepping a new card because the shelf life of a card is ~6 months, going by the 2 high-end cards per year schedule we've seen for the past few years. The card was ready for a march release, but thanks to ATi's countless slip-ups it's been delayed until ATi finally show something.

After all, why would NVidia release a new high-end card when their current top-end beats the competition by ~2x?

[perkam]

NVidia are prepping a new card because the shelf life of a card is ~6 months, going by the 2 high-end cards per year schedule we've seen for the past few years. The card was ready for a march release, but thanks to ATi's countless slip-ups it's been delayed until ATi finally show something.

After all, why would NVidia release a new high-end card when their current top-end beats the competition by ~2x?Diltech don't stand there, I see the sun setting on that mountain :D

Perkam

[Pinnacle]

If it turns out 65nm then it will be blody fast.

IT better be after all this time!!

[xlink]

i had heard about this a long time ago, nothing new.

I hope they clock well though.

[Pinnacle]

i had heard about this a long time ago, nothing new.

I hope they clock well though.


Oh is that so mr. smarypants :)

why didnt you let us all in on it?

[perkam]

the word "smartypants" when referred to member "xlink" is like praising the penguins on the ability to fly.

Perkam

[Knight]

the word "smartypants" when referred to member "xlink" is like praising the penguins on the ability to fly.

Perkam

WOW LOLOLOLOL

Hope ATI can deliver. :(

[Ubermann]

Ubermann, lets try and not post news from the INQ, we news guys know they STINK!

stick to our guns!!

I like INQ, they make people talk =)
This news just sounded right for ones..

the word "smartypants" when referred to member "xlink" is like praising the penguins on the ability to fly.

Perkam

HAhahaha! good one :p:

[Cuthalu]

indeed... if this is in fact true and the power consumption goes down we can finally quit seeing the crybabies posting about the power consumption...

Like you'd even notice the 10$ hit on your power bill every year from gaming between 2-3 hours a day...


Crying about crybabies makes you a crybaby. :rolleyes:

10$ might not matter, less power consumption is good and a rise of a couple celsiuses in room temperature is something that I absolutely don't wan't to notice in summer.


If true that one post we saw on the 1.35Ghz core could be ture on an OC.

No, it won't be true in any case, even if it retailed at 1.35GHz, because it was based on one dudes mere speculation.

[xoqolatl]

No, it won't be true in any case, even if it retailed at 1.35GHz, because it was based on one dudes mere speculation.

And don't forget current ES R600 is probably 80nm, the retail R600 will be 65nm.

[madcho]

XTX :slobber:

:woot:

[Kanten]

indeed... if this is in fact true and the power consumption goes down we can finally quit seeing the crybabies posting about the power consumption...

Like you'd even notice the 10$ hit on your power bill every year from gaming between 2-3 hours a day...

Honestly if your power bill goes up from a new GPU you need to get outside more..


Talk to an Ameren customer.

[Shintai]

Just on alittle sidenote, with 80nm. nVidia will introduce the 8900 series. 8900GTX will have 33% more shaders, 20% higher core and 25% higher mem speed than 8800GTX. So if R600 is on 65nm, it better beat 8800GTX with a big very margin.

[Ubermann]

This is ATI on 65nm not what Nvidia will come with..arent those just rumours on inet ?

[nn_step]

ballsy move, could pay off huge if they pull it off.
Hopefully AMD can be some help in extracting extra performance from the process.

[Shintai]

This is ATI on 65nm not what Nvidia will come with..arent those just rumours on inet ?

Atm i think they hold more water than ATI 65nm rumour ;)

[Ubermann]

I dont know, in kinda lost in the Nvidia "news" what to come and what to not come lately.

[Sumanji]

Just on alittle sidenote, with 80nm. nVidia will introduce the 8900 series. 8900GTX will have 33% more shaders, 20% higher core and 25% higher mem speed than 8800GTX.

You got a source for that dude?

[NapalmV5]

65nm R600/65nm K10 april launch

at least i can dream :D

[w0mbat]

http://img250.imageshack.us/img250/6136/r600vrzone1rw5.jpg

[Sumanji]

Looks a bit... "square" :wth:

[.devastatioN]

65NM and still such a HUGE cooler....

[Hicks]

If true, I'm def buying one.

So your gonna get one, regardless of how good it is, because it has a smaller die?

[eXceededgoku]

Way to go AMD. I think this is a step in the right direction.

Both AMD and Nvidia were headed to 300+W graphics cards. And that is just insane.
lol seriously though what does it matter about the power of the graphics cards!? I'm just interested in the performance, give me a 5KW graphics card as long as it goes well I don't care. Electricity isn't expensive tbh :S...

[LOE]

Smaller die, lower head, lower power consumption...

With the reports of R600 clocking to 1Ghz and up it should be a killer GPU

[LOE]

lol seriously though what does it matter about the power of the graphics cards!? I'm just interested in the performance, give me a 5KW graphics card as long as it goes well I don't care. Electricity isn't expensive tbh :S...

Electricity comes at a much higher price than the one you pay. We all will be paying it's price the next decades. It is funny thou that the majority of USA is concetrated along the east and west coast. THe world ocean level is rising, and it's progression is not linear. Nature creates disasters to make up for our effect on the environment. Things will only get worse :)

[Shintai]

lol seriously though what does it matter about the power of the graphics cards!? I'm just interested in the performance, give me a 5KW graphics card as long as it goes well I don't care. Electricity isn't expensive tbh :S...

Thats not true, a R600 x2900XT CF setup would cost be 1000$ alone if run 24/7 at stock. Say folding. Power aint cheap all around the world. The biggest issue perhaps is that even the R630XT is dualslot cooling and 120-130W.

And 5KW GPU? The room would be a sauna on overdrive along with the jet loud volume noise.

[Spawne32]

Electricity comes at a much higher price than the one you pay. We all will be paying it's price the next decades. It is funny thou that the majority of USA is concetrated along the east and west coast. THe world ocean level is rising, and it's progression is not linear. Nature creates disasters to make up for our effect on the environment. Things will only get worse :)

a Hurricane is coming to destroy your town because you own a power sucking R600 card :lol:

[mAJORD]

lol seriously though what does it matter about the power of the graphics cards!? I'm just interested in the performance, give me a 5KW graphics card as long as it goes well I don't care. Electricity isn't expensive tbh :S...

And you're happy to buy the 5kw PSU to go with it?

Not to mention the extra case cooling. Or perhaps the extra heat in your WC loop..


You were exagerating I know, but there are many reasons why lower power is a good thing.

[Gambit_2K]

lol seriously though what does it matter about the power of the graphics cards!? I'm just interested in the performance, give me a 5KW graphics card as long as it goes well I don't care. Electricity isn't expensive tbh :S...

More Power = More Heat
More Power = A new PSU = Less money for other things
More Power = More noise
More Heat in case = Higher GPU temp = Lower overclock
More Heat in case = Higher CPU temp = Lower overclock
More Heat in case = Higher NB & SB temps = Lower overclocks
More Heat in case = Higher ram tems = Lower overclock

And with the planet in the state it's in (global warming) we all need to think about our powerconsumption.

[Poodle]

Psu hysteria... :rolleyes: Lets not go there guys.


The PSU makers are laughing all the way to the bank.

[zakelwe]

Should be more wiggle room on price as well if the war gets really tough between the two. No guarrantee with 65nm that it will overclock better though.

Nice to see a full node smaller process and a new architecture in one go :"

Regards

Andy

[FghtinIrshNvrDi]

Electricity comes at a much higher price than the one you pay. We all will be paying it's price the next decades. It is funny thou that the majority of USA is concetrated along the east and west coast. THe world ocean level is rising, and it's progression is not linear. Nature creates disasters to make up for our effect on the environment. Things will only get worse :)

you're not being serious, are you?

Ryan

[p0tter]

If I could run quad core and R600 w/ my 600watt powerstream id be a happy clocker.


Not looking forward to upping my PSU for at least another 6-8 months

[LOE]

you're not being serious, are you?

Ryan

you probably know to each action there is a reaction, same strenght, reverse direction

switch to discovery and watch a bit, quite educating, nature has its own way of bringing back the balance we manage to disturb

the more people evolve, the more they take from nature, it is quite logic nature will have to take it back from us someday

thats another reason i don't do folding, wasting power on finding cure for the few things that slow down crowding the earth, the more we get the worse it gets

but thats a bit offtopic

I've said it before and I'll say it again, GPU makers need to change their approach - CPU makers manage to stay in reasonable envelopes and still increase power, but GPUs are going way out of control

[xoqolatl]

What are you guys talking about?
Both AMD and Nvidia were headed to 300+W graphics cards.
Please show me a graphic card that has power of 300 Watts and I will work for you for free till the end of your life. Most PC's with dual core CPU and single GPU don't have 300w power even under load, let alone GPUs...

you probably know to each action there is a reaction, same strenght, reverse direction

switch to discovery and watch a bit, quite educating, nature has its own way of bringing back the balance we manage to disturb

the more people evolve, the more they take from nature, it is quite logic nature will have to take it back from us someday

Do you think your rig matters? Go check the power of train engine, or a hair dryer, or an electric oven, or whatever. If I were to reduce power consumption I would start with other things, not my PCs.

[Cuthalu]

Do you think your rig matters? Go check the power of train engine, or a hair dryer, or an electric oven, or whatever. If I were to reduce power consumption I would start with other things, not my PCs.

Why do so many think so over-simplistic? 1 rig might not matter in it self, but when 1 rig becomes hundreds and hundreds become thoudands etc., then it does matter. Add to that tens of similiar things "which don't matter" and the effect increases further.

[FghtinIrshNvrDi]

you all should start a computer boycott.

Ryan

[ramenchef]

What are you guys talking about?

Please show me a graphic card that has power of 300 Watts and I will work for you for free till the end of your life. Most PC's with dual core CPU and single GPU don't have 300w power even under load, let alone GPUs...


Do you think your rig matters? Go check the power of train engine, or a hair dryer, or an electric oven, or whatever. If I were to reduce power consumption I would start with other things, not my PCs.

One, they were definitely headed in that direction. If you read through some of the posts, the 80nm version of the R600 was drawing 240w on a single card.

An 8800gtx draws about 150w, 200w if overclocked and dualcores can draw 100-150w. Throw on hdds, mobo, fans, etc and you easily hit 300w of useage.

[Quest_7F]

Electricity comes at a much higher price than the one you pay. We all will be paying it's price the next decades. It is funny thou that the majority of USA is concetrated along the east and west coast. THe world ocean level is rising, and it's progression is not linear. Nature creates disasters to make up for our effect on the environment. Things will only get worse :)

As if anyone thinks of that bs when buying ANYTHING!!!

[eXceededgoku]

Why do so many think so over-simplistic? 1 rig might not matter in it self, but when 1 rig becomes hundreds and hundreds become thoudands etc., then it does matter. Add to that tens of similiar things "which don't matter" and the effect increases further.
/offtopic
OMG are you being serious?! How many trains/vehicles/planes/etc. are there in the world? And you want to say computers have an effect?
What has this got to do with R600 anyway since we know its going to be on 65nm process and used upto 240W (dragonhead2) on 80nm so we can expect even lower. I reckon a rig with a 450W PSU will run fine with one R600 and quad core whilst 2 R600s and quad core will need a 600W PSU...

[ozzimark]

yes, for reference one horsepower is about 746 watts. think of how many horsepower a car engine is capable of producing, and keep in mind that it's wasting about 80-90% of the total energy contained within gasoline :p:
computers definitely pale in comparison :(

/end offtopic :D

[Daveb2012]

Well, let's hope so. R600 is gonna be 7 months late to the DX10 party.

Too bad no ones at the DX10 Party :slap:

[Kingcarcas]

yes, for reference one horsepower is about 746 watts. think of how many horsepower a car engine is capable of producing, and keep in mind that it's wasting about 80-90% of the total energy contained within gasoline :p:
computers definitely pale in comparison :(

/end offtopic :D
Which means we need to turbocharge our computers, brilliant! :toast:

[fhpchris]

/offtopic
OMG are you being serious?! How many trains/vehicles/planes/etc. are there in the world? And you want to say computers have an effect?
What has this got to do with R600 anyway since we know its going to be on 65nm process and used upto 240W (dragonhead2) on 80nm so we can expect even lower. I reckon a rig with a 450W PSU will run fine with one R600 and quad core whilst 2 R600s and quad core will need a 600W PSU...

Quadcore + 8800GTX will run fine on a 550-650W psu.. Something like the ST56ZF

Quadcore + 8800GTX SLI can pull 750+ watts, so a nice ~800W+ psu is needed.

450W PSU and 2x DX10 cards + QX6700 = :rofl::rotf: :ROTF::rofl::rotf: :ROTF::rofl::rotf: :ROTF:

[perkam]

Quadcore + 8800GTX will run fine on a 550-650W psu.. Something like the ST56ZF

Quadcore + 8800GTX SLI can pull 750+ watts, so a nice ~800W+ psu is needed.

450W PSU and 2x DX10 cards + QX6700 = :rofl::rotf: :ROTF::rofl::rotf: :ROTF::rofl::rotf: :ROTF:Nice call on the ST56ZF. Though...he said 1 dx10 card + quad core for 450w, and 600w for two.

The mistake there is the assumption that 600w will be enough for two...not that a 450w can handle an r600 and a quad core. A 450W Zippy will do the job, though even a 600w zippy will cannot handle quad + 2 dx10...you need at least 750w, and usually more.

Perkam

[fhpchris]

Depends on the core voltage of the quad, but mine @ 1.55v and a single 8800GTX used more than 450w

450 isnt enough for a quad and single 8800gtx.

1.6v conroe + 8800gtx = 430w for me orthos, not even 3d...

Yes, I made my roomate buy a ST56ZF because you said it was the shiznit ;)

his Xeon 3060+6600GT pulls something like 179W idle :|

I am going to measure yonah now ...

[freeloader]

Nvidia and ATI know that they can't continue down their current path with chip design. Too much power, heat, leakage, etc...

They'll more than likely be going the multicore root in the next revisions. Just my 2 cents.

[Shadowmage]

Unfortunately, multicore doesn't change anything in regards to power, heat, leakage, etc. Multicore is only a good solution when you have extra transistors to spare, in which case you can also clock lower.

Multiple dies (like Kentsfield, not Barcelona) will improve yield though, in exchange for a decrease in performance.

[nn_step]

Unfortunately, multicore doesn't change anything in regards to power, heat, leakage, etc. Multicore is only a good solution when you have extra transistors to spare, in which case you can also clock lower.

Multiple dies (like Kentsfield, not Barcelona) will improve yield though, in exchange for a decrease in performance.

one could argue that we are approaching the limits of extracting performance out of software and with current growth rates for Transistor budgets, it is becoming more and more popular to throw extra Cache or more cores at the problem. Though x86_64 offers a budget of growing room for performance but as you can tell, no one outside of AMD is pushing for 64bit Processing

[Shadowmage]

one could argue that we are approaching the limits of extracting performance out of software and with current growth rates for Transistor budgets, it is becoming more and more popular to throw extra Cache or more cores at the problem. Though x86_64 offers a budget of growing room for performance but as you can tell, no one outside of AMD is pushing for 64bit Processing

Nope, completely wrong. The industry is behind the computer architecture academic developments by quite a few years.

For example, good and accurate indirect branch prediction was invented in 1998.

K. Driesen, U. Holzle, "Accurate Indirect Branch Prediction," isca, p. 0167, 25th Annual International Symposium on Computer Architecture (ISCA'98), 1998.

There are still hundreds of ways more to improve performance and reduce power consumption. For example, you could increase issue width and end up saving transistors, reducing the critical path, and reducing power consumption by switching from a standard Tomasulo issue scheme to a Cyclone issue scheme.

http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=1207005


----

The main point was that just going to multicore wasn't going to do anything for an inherently parallel graphics core architecture.

[gdogg]

I think its funny when people talk about computers wasting power.

Honestly, atleast they have a purpose.

Everyone could use 16 r600 board in there computer.

And it still wouldn't use as much power or cause as great of an effect on global warming as what is used each year on war machines and supposed defense of countries.

Thats the true cause of global warming, no one cares about the fact global warming is caused by factories, tanks, sub marines, space travel, air planes and all the other stuff that never gets mentioned in global warming talks (because these are the governments toys for world domination, and they also don't even make an attempt to not pollute with these things, they are like 100% wasteful).

but look how off topic this has gotten. Maybe a new section should be made, called extreme power savings or something like that.
I for one know I am getting this card, after my 680i problems, I won't be buying another rushed product ever again

in the end, if you truly believe that your computer is causing global warming (or breathing or over population) then the solution is simple, get rid of them and stop breathing.

[Ubermann]

I think its funny when people talk about computers wasting power.

Honestly, atleast they have a purpose.

Everyone could use 16 r600 board in there computer.

And it still wouldn't use as much power or cause as great of an effect on global warming as what is used each year on war machines and supposed defense of countries.

Thats the true cause of global warming, no one cares about the fact global warming is caused by factories, tanks, sub marines, space travel, air planes and all the other stuff that never gets mentioned in global warming talks (because these are the governments toys for world domination, and they also don't even make an attempt to not pollute with these things, they are like 100% wasteful).

but look how off topic this has gotten. Maybe a new section should be made, called extreme power savings or something like that.
I for one know I am getting this card, after my 680i problems, I won't be buying another rushed product ever again

in the end, if you truly believe that your computer is causing global warming (or breathing or over population) then the solution is simple, get rid of them and stop breathing.

Pretty much everything cause global warming. Computers is one of many problems.

[Shintai]

Pretty much everything cause global warming. Computers is one of many problems.

Ye we gotta start one place. And things like computers and other electronics along with lightblubs and household machines is realtively easy and cane save quite abit.

Just on the Prescott vs K8 I think it was out that the world could save 1-2billion$a year for each yearly production using US powerprices. (If it was where i live it would be 3-6billion$). And looking on new GFX cards, Prescott hardly used power :P

So yes, we need to save everywhere and leave noone out of it.

[gdogg]

it was when man started farming that caused the biggest change in the temperature of earth.

With the world the way it is, saving power or the climate should be the least of our concerns.

Its too late, just look at things already, the only way we are gonna save this planet, is by wasting more power in an effort to reverse the damage we've done.

By increasing our scientific knowledge and so on, or we could go back to the stone age. Cause light blubs and so on doesn't change the fact, that the biggest cause of the global warming in 25 years, was this stupid war in iraq, and all the bombs droped.

Don't need to be a rocket scientist to connect the two.

And the more things they put in space, the more danger we are under, they keep poking holes, thats not gonna spot, these are the real things that cause global warming, they just want to blame us, because if it was there fault, we'd want them to stop.

What we need to do, is just not give a **** just like they don't and enjoy the last little bit of time we got left here.

And besides even if you believe what they'll tell you, with the industrialization of africa there is nothing we can do, that'll make any difference, it gets canceled out 10 fold.

All I am trying to say in this post is. Lets use the best of the best hardware as we always do. Overclock them to there peaks, use all the power we want, and be "extreme".
:woot: :woot: :woot:


edit

sorry for continuing the off topic with a rant.

[Metroid]

Let's put things in another way.

Example --> 1950 XTX uses 12v = 30A for a full power.90nm
Example --> 2950 XTX will use 12v = 40A for a full power. 65nm

As we can see is not going to be something huge by the fact they shrank it.

[Cuthalu]

Let's put things in another way.

Example --> 1950 XTX uses 12v = 30A for a full power.90nm
Example --> 2950 XTX will use 12v = 40A for a full power. 65nm

As we can see is not going to be something huge by the fact they shrank it.

1950XTX consumes about 11A or less. :stick:

[FghtinIrshNvrDi]

it's strange... I've been here for about a year, not quite, but you get my drift...

I've started to get to know a lot of the personalities in this place. In my own little quiet way, it seems as if I could spot the type of person that believes the global warming stuff hook, line, and sinker. Has anyone else noticed this type of personality? It's almost an innate personality trait. Personality type A & B anyone?

I'm not trying to be mean or break anyone's sensitive compassionate heart, but you should do more reading on the critisisms of global warming before you just assume it's really happening. Not only can you argue that it's not man's fault, but on top of that, you can argue that it's not even happening. Scientifically, global warming is not even close to being indisputable.

Anyway, all you type A personalities, take a chill pill, read the other side's facts and data. Nobody likes a caring, in-your-face alarmist attitude.

Ryan

[Cuthalu]

I'm not trying to be mean or break anyone's sensitive compassionate heart, but you should do more reading on the critisisms of global warming before you just assume it's really happening. Not only can you argue that it's not man's fault, but on top of that, you can argue that it's not even happening. Scientifically, global warming is not even close to being indisputable.

Why do you assume that this "your type" haven't read criticism? I've read a lot of it. And I also know that a lot of it is B$, and some have good points to consider.

Here's one example of very biased criticism on IPCC's & co. climate change theories: http://news.independent.co.uk/environment/climate_change/article2355956.ece

Of course you can argue anything, like there's no gravity or you don't exist or whatever. There's uncontroversial evidence that climate is warming atm, and there's always warming before ice ages - climate works in cycles. But the defining factor on if humans have big impact on climate or not can be observed watching the pace how fast climate changes compared to previous data from past thousands of years.

[FghtinIrshNvrDi]

we don't have previous data from past thousands of years... We have rough ideas of what the global temp might have been from sediment samples and things of the like.

Ryan

[Metroid]

1950XTX consumes about 11A or less. :stick:

That is the recommended by ATI :p:

[generics_user]

we don't have previous data from past thousands of years... We have rough ideas of what the global temp might have been from sediment samples and things of the like.

Ryan

we have that's why there are axpeditions to antarctic ;)

[largon]

Discussion about global warming and antarctic ends here. That's unless someone took an ES R600 to the antarctic for some xtreme OC'ing.

Back on topic.

We know how big the 80nm R600 chip is. How much smaller should the 65nm die end up?

[Helmore]

We know how big the 80nm R600 chip is. How much smaller should the 65nm die end up?
That's something I would really like to know.
I would also like to know to what that stated 30% reduction in power consumption is compared? Is it compared with an 80nm R600 with both of them running at the same clocks (750MHz where the rumours), or is that reduction still there after the rise in clocks (over 1 GHz as the rumours go :p:)?

[MaxxxRacer]

I SO hope this is true.

[Shintai]

That's something I would really like to know.
I would also like to know to what that stated 30% reduction in power consumption is compared? Is it compared with an 80nm R600 with both of them running at the same clocks (750MHz where the rumours), or is that reduction still there after the rise in clocks (over 1 GHz as the rumours go :p:)?

It wont be rised clocks aswell. Its one of them only. And looking on the cooler solutions its not the power saving.

[nn_step]

It wont be rised clocks aswell. Its one of them only. And looking on the cooler solutions its not the power saving.

the more power the system has, the more resources that are wasted

[Kanten]

With the world the way it is, saving power or the climate should be the least of our concerns.


Eh....climate changes have caused extinction. I'd say that'd be a pretty big concern. :confused:

[Shintai]

the more power the system has, the more resources that are wasted

Yes thats mighty fine. But we are talking 80->65nm performance/power consumption.

[gdogg]

Eh....climate changes have caused extinction. I'd say that'd be a pretty big concern. :confused:

I am already using 2 7800GT's in sli.
I am sure the extra 100W isn't gonna make a difference, since I am gonna go quad and octa core too.

I got 6 hard drives in raid 0, sucking back the juice too.

This x2900xtx is gonna be my dream card. Why you might ask? Because for the little extra power usage, I'll get more than 2X the performance of my current sli setup. So in performance vs power usage, this will use less then be getting about 2 more 7800gt's in an attempt to match this 1 card in performance.

[Jamesrt2004]

I am already using 2 7800GT's in sli.
I am sure the extra 100W isn't gonna make a difference, since I am gonna go quad and octa core too.

I got 6 hard drives in raid 0, sucking back the juice too.

This x2900xtx is gonna be my dream card. Why you might ask? Because for the little extra power usage, I'll get more than 2X the performance of my current sli setup. So in performance vs power usage, this will use less then be getting about 2 more 7800gt's in an attempt to match this 1 card in performance.

You, My Friend are the cause of Global warming :D

jk lol

[Revv23]

I am already using 2 7800GT's in sli.
I am sure the extra 100W isn't gonna make a difference, since I am gonna go quad and octa core too.

I got 6 hard drives in raid 0, sucking back the juice too.

This x2900xtx is gonna be my dream card. Why you might ask? Because for the little extra power usage, I'll get more than 2X the performance of my current sli setup. So in performance vs power usage, this will use less then be getting about 2 more 7800gt's in an attempt to match this 1 card in performance.


Thats the spirit now lets vmod those suckers! :)

[Jamesrt2004]

I wonder how the Ati will OC if they keep the same clocks with 65nm should be Great oc'er depends on the cooling I guess though

[Ubermann]

"We pushed out the launch of the R600 and people thought is must be a silicon or software problem…it's got to be a bug," said Dave Orton, president and chief executiveof ATI. "In fact, our mainstream chips are in 65nm and are coming out extremely fast. Because of that configuration, we have an interesting opportunity to come to market with a broader range of products"

Thats what Dave Orton said about the delay.
So it fits kinda well into this news =)

(Found it at elitebastards.com...)

[bloodbanger]

I don't care that they can come to market with a broader range,i only care WHEN...this delay stuff is getting really on my nerves:)

[Ubermann]

Do like me, give it up =)

[bloodbanger]

lol,mmaybe that's the best...i hoped to finally get my new pc after the pricecuts from Intel in April,and i hoped to include a R600 with it...Seems that all my hope is in vain :(

[Shintai]

Last info is the yield sucks and they wont hardly be able to make 20000 cards at launch...

[MaxxxRacer]

Last info is the yield sucks and they wont hardly be able to make 20000 cards at launch...

Got a source for this info?

[Jamesrt2004]

I have to agree with Shintai there is a link somewhere in the AMD R600 thread :)

[Jamesrt2004]

Over at CeBIT, we overheard some production problems with R600. The yield at 65nm is very low at this point of time. It is projected by middle of April, AMD is only able to churn out some 20,000 Radeon X2900 XTX graphics cards and XT SKU might not even make it for the launch in May.

http://www.vr-zone.com/index.php?i=4828

Sorry for double post I clicked edit put it in n my post disapeared put this in and then .. uhh confusing

[Dimitriman]

I was hoping DAAMIT wouldn't screw it up with R600, this is finally very good news.
Hopefully R600 < 150W and not sounding like a hairdryer

[perkam]

I was hoping DAAMIT wouldn't screw it up with R600, this is finally very good news.
Hopefully R600 < 150W and not sounding like a hairdryerTechnically, even if they do, they know they have some headroom for a slow launch seeing as Nvidia's 65NM offerings:

A. Won't come till '08
B. Current 80nm won't clocks as much as 65NM
C. Will come only once Nvidia is satisfied it has cleared the channel of 80nm versions.

Perkam

[Dimitriman]

Technically, even if they do, they know they have some headroom for a slow launch seeing as Nvidia's 65NM offerings:

A. Won't come till '08
B. Current 80nm won't clocks as much as 65NM
C. Will come only once Nvidia is satisfied it has cleared the channel of 80nm versions.

Perkam

I'm confident part of the concern on the part of AMD was with G90, and if they can release R600 @ 65nm it would give them enough headroom to clock it high enough to beat even the G90 or at least be on par. If R600 came out sucking 200W and actually got outbenched by G90 just a couple of months later, it would be R520 all over again. Seems like this time they've taken their time to do it right.

[citat3962]

you all should start a computer boycott.

Ryan

I know... All of XS should turn off all their lights and all their PCs and light candles and sing Kumbaya for lower power consumption...

Maybe instead of tracking your Folding stats we should all count how much power we've saved by turning off our PCs..

Laff.. forget it hippie.:rolleyes:

[FghtinIrshNvrDi]

I know you're not calling me a hippie...

I just KNOW you're talking about the other guys on here, the alarmists.

Ryan

[Shadowmage]

the more power the system has, the more resources that are wasted

Nope. For a graphics card, due to the massively parallel workload, very little of the available resources are wasted. That's why increasing the number of pipelines gives roughly linear performance scaling in the latest games.

That's also why we're seeing all these heat problems and huge dies: it's worth it performance-wise to make them. This is not true for a CPU.

[eXceededgoku]

I know... All of XS should turn off all their lights and all their PCs and light candles and sing Kumbaya for lower power consumption...

Maybe instead of tracking your Folding stats we should all count how much power we've saved by turning off our PCs..

Laff.. forget it hippie.:rolleyes:
Hmmm candles are a pollutant so no candles thanks :rolleyes:

[[XC] Lead Head]

Hmmm candles are a pollutant so no candles thanks :rolleyes:

Oh yes, lets all build static generators and use the pulsating sparks to light our houses!

[eXceededgoku]

R600 is 80 nanometre
We spoke with dozen of contacts and we can now confirm that R600 is a 80 nanometre chip. It is not 60 nanometre and it won't be 65 nanometre as going from 80 nanometre to 65 requires complete redesign of chip.


ATI doesn't have the time for it and we don't believe that AMD / ATI could get away deceiving dozens of its partners and OEM's with news that R600 is 80 nanometre, until it suits them.


R600 was and is 80 nanometre and when it launches it will be 80 nanometre. Remember G80, Geforce 8800 series is 90 nanometre and it still runs great. So if you do you homework well even the big chip works well.


Sorry to disappoint you but R600 will be re branded and probably cheaper than everyone expects but its very unlikely that we are waiting for G80 killer.
source (http://www.fudzilla.com/index.php?option=com_content&task=view&id=221&Itemid=1)
NO!!!! :(

[biohead]

"news" changes every :banana:ing day, it's rediculous.

[VulgarHandle]

hmm, techreport says AMD stated it's 65nm... of course, AMD could have just been eluding to 610 adn 630...

[Ubermann]

Lol that was a fast turnaround =)

[cantankerous]

Sigh

[Shintai]

hmm, techreport says AMD stated it's 65nm... of course, AMD could have just been eluding to 610 adn 630...

Techreport is theinq copy. And seriously, 80nm seems alot more likely.

[LOE]

anyone to confirm/deny this info?

or we will have another 3490832 posts of speculation before some more info comes out