AMD: R9xx Speculation

It seems the 6990 is not only the fastest card out there but also the loudest. :LOL:
If nvidia chooses a triple axial fan cooling for the GTX590 it will be the clear winner. People dont want a jet engine in their cases.
 
It seems the 6990 is not only the fastest card out there but also the loudest. :LOL:
If nvidia chooses a triple axial fan cooling for the GTX590 it will be the clear winner. People dont want a jet engine in their cases.

Depends on the review I think.
http://nl.hardware.info/reviews/2028/17/amd-radeon-hd-6990-review-stroomverbruik-en-geluidsproductie

Sound
107520,107519,113867,113866,82596,82274,102037,81426,80462,86545,105421,103149,107703,101862,97556,97489,115839,113521,108228,116008,117867,117872


Power
107520,107519,113867,113866,82596,82274,102037,81426,80462,86545,105421,103149,107703,101862,97556,97489,115839,113521,108228,116008,117867,117872
 
Those are quite strange numbers from netherland. If sites like anandtech or tomshardware are writing that the card is louder than gtx480 or 5970 by several decibels than its hard to believe those numbers.

The 6870 on that graph is louder than a 5970 !!!! And 5770 is close to the 6990.:LOL:
 
Last edited by a moderator:
Those are quite strange numbers from netherland. If sites like anandtech or tomshardware are writing that the card is louder than gtx480 or 5970 by several decibels than its hard to believe those numbers.

The 6870 on that graph is louder than a 5970 !!!! And 5770 is close to the 6990.:LOL:

Problem is, I played with a 6990 all last week and it was a lot cooler than anands numbers for instance as well, the same system with a 580 was a lot warmer.. but hell, it's all subjective anyway.
 
Problem is, I played with a 6990 all last week and it was a lot cooler than anands numbers for instance as well, the same system with a 580 was a lot warmer.. but hell, it's all subjective anyway.

anandtech write this :
Even the thermal paste connecting the GPUs to the vapor chambers has been changed for the 6990 – AMD has replaced traditional paste with a phase change material. Phase change materials – pastes/pads of material that melt and solidify based on temperature are nothing new, however they’re still exotic; material similar to what AMD is using is not readily available as paste is. AMD even went so far as to suggest that reviewers not directly disassemble their 6990s as it would require a new application of phase change paste in order to achieve the same efficiency as the original material.

And they did remove the heatsink it seems. http://www.anandtech.com/show/4209/amds-radeon-hd-6990-the-new-single-card-king/2

CoolerBack_575px.jpg


Or maybe that phase change material is not working on all cards as intended and makes things sometimes even worse than a basic paste. (chinese assembling quality :rolleyes:)
 
10 memory chips for 320bit bus or 12 memory chips for 384bit bus!

The GTX 590 is supposed to be based on the GF110 not GF114 right? From the pics it seems like it.:?:
Yeah, that's what you get when answering blindly in a thread about R9xx without reading closely...

Those are quite strange numbers from netherland. If sites like anandtech or tomshardware are writing that the card is louder than gtx480 or 5970 by several decibels than its hard to believe those numbers.

The 6870 on that graph is louder than a 5970 !!!! And 5770 is close to the 6990.:LOL:
I think it could depend on a lot of factors which is louder - open test bench vs. closed box for instance.
Also if you measure with a game or furmark - for the latter powertune will definitely kick in.
For instance, these numbers here http://ht4u.net/reviews/2011/amd_radeon_hd_6990_antilles_test/index11.php suggest it's louder in games than the GTX480 - but a bit quieter in furmark.
Though if you use the OC bios, it looks like it goes from vacuum cleaner noise level to jet engine noise level even...
I wonder though if nvidia will be much better in that department - that's just a lot of heat you have to get rid of in a quite small place (for that reason alone I'd think you should probably get HD6950CF instead - not to mention it will be a lot cheaper).
 
that's just a lot of heat you have to get rid of in a quite small place (for that reason alone I'd think you should probably get HD6950CF instead - not to mention it will be a lot cheaper).

Yeah the price of this thing is not very good. CF6950 or even CF6970 makes more sense. So if you clock 5970 as close to the 5870 as this is to 6970, you'll end up getting a card about 20-25% faster than 15 month old 5970. This costs 100$ more than 5970 at launch and uses more power, that's some sweet spot for you...I quess those leaked Amazon prices for 69xx cards before their release were true, but this time AMD has a proper chance to keep the prices up.

Having said all that, nVidia is going to have its hands full with this to say the least. Can't say 40nm wasn't tapped out :smile:
 
I understand that hasn't been an issue with all the AIB cards, but if AMD wouldn't care if their official flagship is approved by PCI SIG, why do they bother getting any of their products approved by PCI SIG in the first place?
Because for the most part PCI-SIG's insane requirements don't negatively affect them, now they do they thankfully say fuck that ... rather than comply. As I've said before the patent cross licensing is done in the PCI-SIG bylaws and don't allow PCI-SIG to put any restrictions at all on anything but the "compliant portion" of the board (which is just the interface circuitry, nothing more).

PS. do the HD6990 boxes have the PCI-Express logo?
 
Or maybe that phase change material is not working on all cards as intended and makes things sometimes even worse than a basic paste. (chinese assembling quality :rolleyes:)
I recall some CPU heatsinks coming with phase-change pads a few years back.
The manufacturer stated that the material needed time under load temps to "cure" or melt its way into all the crevices. Until that happened, the performance of the material wasn't particularly impressive.
Could this be a case of varying tolerances and varying levels of curing?
 
I'd be interested to hear how 'phase change materials' are supposed to work here. Phase change for cooling basically means melting or evaporating something... It then has to be transfered somewhere cooler to recondense, as in a heat pipe. I don't see how that's a replacement for thermal paste, which is just supposed to give good condictivity between something hot and a heatsink...
 
The phase change materials I've seen were something like a wax. The phase change in that situation would be from a solid to liquid state, hence the need for a curing process that would melt the wax into the crevices of the two surfaces.
 
Because for the most part PCI-SIG's insane requirements don't negatively affect them, now they do they thankfully say fuck that ... rather than comply. As I've said before the patent cross licensing is done in the PCI-SIG bylaws and don't allow PCI-SIG to put any restrictions at all on anything but the "compliant portion" of the board (which is just the interface circuitry, nothing more).

PS. do the HD6990 boxes have the PCI-Express logo?

It's not so much them giving the PCI-SIG the middle finger as it is that this card isn't meant for nor targetted at OEM design wins. Hence the one market where it actually matters (OEMs) isn't a potential target for this card and hence there is no need for certification.

For those niche boutique lines, Alienware for example, they often go overspec anyways so they aren't going to care whether something is certified or not.

So it's not so much about it negatively affect them. As in this case PCIE certification doesn't help them in any way so why try to meet it and get certified?

Regards,
SB
 
Isn't the only advantage of phase change pads over grease that it can't be misapplied? (Well of course it can, but it takes exceptional stupidity.)
 
Yeah, that's what you get when answering blindly in a thread about R9xx without reading closely...

Sorry?

I was just wondering why there were 10 memory chips on board per gpu, which would indicate a 320bit bus, wouldn't it?

Even if that was just a proof of concept design, it still makes you wonder why didn't they go for a 12 chip/384bit design, even for that proof of concept design. I was just wondering in the light of latest news that suggested the 590 will be based on a full 384bit/1.5BG fermi. I understand that the GF114 is more power efficient now, so the end design may have changed in favor of more capable chips but we still don't know that, do we?

In any case, that picture was posted here, in the r9xx thread, where I answered.
 
Sorry?

I was just wondering why there were 10 memory chips on board per gpu, which would indicate a 320bit bus, wouldn't it?

No you were wondering why there were 8 chips per gpu. mczak responded "what did you expect?", because he didn't pay attention and thought you were talking about an AMD card since this is an AMD thread.

:)
 
No you were wondering why there were 8 chips per gpu. mczak responded "what did you expect?", because he didn't pay attention and thought you were talking about an AMD card since this is an AMD thread.

:)

Oopsy daisies!:p

I meant to say 8 yes. My initial argument was about those 8 memory chips indeed.

Ok I see now. No harm done! :) Thanks.
 
Ok I see now. No harm done! :) Thanks.
He, well was my mistake :).

In any case, I'm still wondering about the remaining rv9xx parts. What happened to Turks/Caicos??? You can even buy their mobile siblings now (at least in the macbook), the oem parts are officially out since a month yet their was no launch at all. I think at least Turks would deserve a little attention.
 
He, well was my mistake :).

In any case, I'm still wondering about the remaining rv9xx parts. What happened to Turks/Caicos??? You can even buy their mobile siblings now (at least in the macbook), the oem parts are officially out since a month yet their was no launch at all. I think at least Turks would deserve a little attention.

Patience is a Virtu (pun intended)
 
Isn't the only advantage of phase change pads over grease that it can't be misapplied? (Well of course it can, but it takes exceptional stupidity.)

From what I gather with searching around, it sounds like it's ideal for automated manufacturing. Though a lot of other companies use pre-applied thermal grease so I dunno how true that is anymore.

Most of the info I found actually said that phase change performs somewhat worse than grease.
 
Back
Top