Right, I think that's the optimal design.. has anyone actually tried this? Ducted air all the way through their system from front to back-- at least across the processor?Or better yet, design CTX (Cooling Technology Extended) form factor (I invented it, rights reserved as of now ), which has a direct wind tunnel (=duct) and the motherboard has CPU, NB & SB lined up, and one fan in the middle of the case (and the duct, of course) sucking air in from the front and exhausting it at the back, and optionally another fan at the back to help create enough backpressure.
New MoBo/case form factor "BTX"
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Devonavar
-
- Patron of SPCR
- Posts: 946
- Joined: Sat Sep 06, 2003 9:57 pm
- Location: Berkeley, CA, USA
- Contact:
Unless you go caseless and get some sort of adapter, then no, an ATX PSU will not work. The powersupply is both a different shape and uses a different plug than a regular ATX unit.Tobias wrote:Further, It is true that to use the BTX-formfactor you will need both a mobo and a case that supports it (unless you don´t use a case:), but if you have a PSU you like you will not need to buy a new, since the ATX-PSU will work with the new standard.
Regular users have no use for a graphics interface faster than AGP 8X. Heck, the cards we already have don't even need it. The AGP 3.0 spec also allows for dual AGP slots/cards, but I haven't seen that done yet either. The BTX Reference Unit has a second expansion slot under the VGA card, but I'm guessing that's going to be used for a full-size PCI card and not a second AGP slot.What I am afraid of with this standard is that you at some point will need to buy bith a new mobo and a new GPU, since the AGP-standard is doomed as well...
PCI Express is going back to the shared-bus architecture that we've been trying so hard to move away from. Same with the new motherboards that require matched memory (SIMMs, anyone?). I don't get it...
... and AMD ?
Another question on this new format is : will it be also used by AMD ?
The worst would be that they 'compete' and make a ZTX half incompatible with intel's BTX.
An option would be to continue with ATX to take advantage of all 'ppl willing to keep their investment, which shouldn't be a problem with 30 and 55 W opterons.
But this will only concerned the ppl making and upgrading their own computers, which are far from being a majority.
Anyway, my next target for office (home) pc is a dual (single) low voltage opteron, which will be perfectly quiet in my curren setup(s), without changing anything except mobo/cpu
The worst would be that they 'compete' and make a ZTX half incompatible with intel's BTX.
An option would be to continue with ATX to take advantage of all 'ppl willing to keep their investment, which shouldn't be a problem with 30 and 55 W opterons.
But this will only concerned the ppl making and upgrading their own computers, which are far from being a majority.
Anyway, my next target for office (home) pc is a dual (single) low voltage opteron, which will be perfectly quiet in my curren setup(s), without changing anything except mobo/cpu
Does one have to be invited, or are they free sessions? I'd like to go to one of these seminars sometimes.MikeC wrote:The linked PDFs are the ones that were used in sessions at IDF. (I know, I attended them.)
Cooler_than_thou, thank You very much for providing those .PDFs. It's going to take more than a while to study them properly.
Cheers,
Jan
Motherboards for AMD will undoubtedly use BTX too. Its an excellent design; theres no reason to go against the flow with something you lack the market to sell.
However, since ATX is so proliferous, it will be a while before it is completely replaced. As long as you'll buy ATX parts they'll be selling them.
However, since ATX is so proliferous, it will be a while before it is completely replaced. As long as you'll buy ATX parts they'll be selling them.
question is wehter the moutingholes for the PSU is different. It might be that the BTX-standard added more, for use only with BTX-PSU:s but that the old standard is part of the new one.
Intels papers clearly stated that any ATX/BTX-PSU will work with any ATX/BTX-mobo. One will, however need an adapter...
Intels papers clearly stated that any ATX/BTX-PSU will work with any ATX/BTX-mobo. One will, however need an adapter...
The only reason people were moving away from shared-bus was due to there being too little bandwidth available for everything on the PCI bus. Now that they figure PCI Express will have more than they "ever" need , they're going back to that because the design becomes much simpler/cheaper that way.Syphon869 wrote: PCI Express is going back to the shared-bus architecture that we've been trying so hard to move away from. Same with the new motherboards that require matched memory (SIMMs, anyone?). I don't get it...
-
- Site Admin
- Posts: 12285
- Joined: Sun Aug 11, 2002 3:26 pm
- Location: Vancouver, BC, Canada
- Contact:
Companies pay big bucks to send their employees. It's a great biz model for Intel: charge customers to persuade them to use Intel products and be trained to use them!Jan Kivar wrote:Does one have to be invited, or are they free sessions? I'd like to go to one of these seminars sometimes.MikeC wrote:The linked PDFs are the ones that were used in sessions at IDF. (I know, I attended them.)
I get in free as an industry analyst / press.
Well, then having the .PDFs is even nicer...MikeC wrote:Companies pay big bucks to send their employees. It's a great biz model for Intel: charge customers to persuade them to use Intel products and be trained to use them!Jan Kivar wrote:Does one have to be invited, or are they free sessions? I'd like to go to one of these seminars sometimes.MikeC wrote:The linked PDFs are the ones that were used in sessions at IDF. (I know, I attended them.)
I get in free as an industry analyst / press.
Jan
-
- Site Admin
- Posts: 12285
- Joined: Sun Aug 11, 2002 3:26 pm
- Location: Vancouver, BC, Canada
- Contact:
OK, many months later, I have changed my PoV about the BTX.
What Intel REALLY did was to make it possible for them to keep their red-hot processors cool enough -- and the heck with the rest of the components, which now get air preheated by the hottest thing in the chassis. How they get cooled, and how much noise they make as they get cooled, well, that's other companies' worries.
Certainly seems like that!
It will be interesting to see who, when and how willingly BTX will be adopted. I have not seen a single BTX-specific component on the market yet (case, PSU, motherboard).
What Intel REALLY did was to make it possible for them to keep their red-hot processors cool enough -- and the heck with the rest of the components, which now get air preheated by the hottest thing in the chassis. How they get cooled, and how much noise they make as they get cooled, well, that's other companies' worries.
Certainly seems like that!
It will be interesting to see who, when and how willingly BTX will be adopted. I have not seen a single BTX-specific component on the market yet (case, PSU, motherboard).
It is very interesting to me how BTX is so controversial when we haven't seen any actual implementations of it yet. A couple observations of mine:MikeC wrote:OK, many months later, I have changed my PoV about the BTX.
What Intel REALLY did was to make it possible for them to keep their red-hot processors cool enough -- and the heck with the rest of the components, which now get air preheated by the hottest thing in the chassis. How they get cooled, and how much noise they make as they get cooled, well, that's other companies' worries.
1) The CPU produces by FAR the lionshare of heat in the system. Of COURSE Intel is going to want it to be cooled first and foremost. Nothing else NEEDS as much cooling as the CPU does. Therefore, why NOT do it that way? As far as I'm concerned, having never seen an actual BTX implementation in real life, I am not prepared to pre-judge the merits of this approach. It MIGHT actually work. If it does, I'm sure it could benefit AMD as well--their CPUs, even if they aren't quite up to the 103-watt mark yet, aren't exactly getting any cooler, either.
2) Besides, it's not like ATX is any better in this regard, either. You gotta admit, with ATX, NOTHING is really what you'd call "optimal," as it seems for the most part, ATX shows little concern for optimizing the airflow at all--or if it does, it seems more like an after-thought than a CENTRAL part of the design, as it was with the Apple G5. Just look at it: with the ATX spec, for example, they specify that the PSU should be responsible for evacuating at least some of the heat in the system. But with today's systems running as hot as they do, this idea doesn't seem quite so feasable anymore. Clearly, the ATX spec was NOT designed with optimal airflow in mind.
2) I do wonder about provisions for video card cooling with BTX--it seems to me that modern video cards are now without question the second hottest thing in the system, and they aren't exactly getting any cooler anytime soon, either. Not only that, but just look at the high-end Radeon 9800's or the NVidia GeForce FX 5950's, and you'll notice that not only are they red hot, but they also draw more power than the AGP specification allows for, forcing these cards to require a direct connection to the PSU to meet their power demands. I wonder how PCI Express plans to deal with that.
Well, i think its a waste of time, i dont really care eather way between ATX and BTX - but think if the time, money and effort that could have gone into making cooler CPUs and quiter PSU!!!
i mean look at it
they've just sawaped the Nbridge with the CPU and move the Memory slots a little... ...and then mirroed it so it looks diffrent!!!
- well, that my option
daniel
i mean look at it
they've just sawaped the Nbridge with the CPU and move the Memory slots a little... ...and then mirroed it so it looks diffrent!!!
- well, that my option
daniel
-
- Posts: 404
- Joined: Sun Jul 04, 2004 9:18 pm
- Location: Illinois, USA
I did actually do the math, and assuming a 100-watt processor, the result I get is that the air (at 10 CFM) coming out of the processor's cooling module will be 19.2 (!) degrees hotter than the air coming into the case. Up the airflow to a painfully loud 20 CFM and you still get about 10 degrees of temperature rise.Jan Kivar wrote:No, the idea of putting the coolest air through the hottest component first is absolutely a good idea. With modest airflow (say 10 CFM), the air going through the CPU HS will be heated by the HS by approximately five degrees (no, I didn't calculate it), but the CPU HS (and thus CPU) will gain more larger temp drop.
What does this mean? Well, it has tragic implications for graphics card cooling. In your regular ATX case nowadays, the air that cools the graphics card is case air, just a few degrees hotter than ambient temperature. If a high-performance, high-heat output graphics card is fed air 19 degrees over ambient, it will require some BEEFY cooling to survive.
MikeC is right - with BTX, intel did what they needed to so that their processors could actually function - at the expense of practical cooling of the rest of the system.
BTX is not a solution to the heat crisis of 2005, just a futile attempt to delay the inevitable. Intel's in very hot water (pun intended) now that extreme edition processors have squeezed out a few percent more performance at the expense of massive power increases.
Actually it is even worse than that: Intel guidelines call for 115W cpu + 38W vreg. So the intake air will absorb 150W!!!! As I was reading the pdf, I thought NOOOOOOO!!! I did the calculation as well, for 35cfm air (Intel guideline) temp rise is 7.5C. Now this isnt tragic, I was really surprised. It is reasonable. Well lets say borderline reasonable. BUT: A 90mm fan cannot give 35cfm quietly. And according to the BTX form factor, it will be closest to you and drawing air through a non-restrictive bezel. BTX may be great for Dell and HP and others, but for SPCR, I think it might be quite a headache.swivelguy2 wrote:
I did actually do the math, and assuming a 100-watt processor, the result I get is that the air (at 10 CFM) coming out of the processor's cooling module will be 19.2 (!) degrees hotter than the air coming into the case. Up the airflow to a painfully loud 20 CFM and you still get about 10 degrees of temperature rise.
MikeC is right - with BTX, intel did what they needed to so that their processors could actually function - at the expense of practical cooling of the rest of the system.
Even if we were able to use a 120mm fan instead of 90, the nexus 120 gives 37cfm at 12volts. (I've never heard the nexus.) The question is: Is the Nexus 120 mm at 12 volts quiet enough for you?
MY thoughts:
The good: Linear airflow, low no of fans, smaller case.
The bad: Cooling design seems inflexible(Bad news for SPCR)
The ugly: Cpu heat exhausted into the case.
Notice that the BTX seems to be space optimised for a 1CDROM, 1-2HD desktop case more than anything else. Cooling the cpu is highest on the priority list and this indicates that prescott is here to stay...
I think a quieter/cooler design would be something like this: (Yes, I know about the Lian-Li v1000)
The idea is to have the cpu socket really close to the back end of the mobo, use a 120 as CPU/GPU (think NCU-2000+vm-101) cooler/exhaust fan and get the IO shield out of the way. Airflow would be much closer to the mobo, and you could even have under mobo airflow as Intel suggests in BTX. I obviously do not know the complications (if any) of placing the IO shield like that. But sice mobos sometimes include pci card type of expansion cards, I dare say its doable. The PSU would have a fan but it would be at the inner side of the psu. There would be two seperators running the whole lenght of the case to keep airflow tidy, one between PSU/IO shield, other just above vidcard (or maybe 2nd pci slot.) This way, psu fan and 120 fan would be ducted for linear airflow. Have a large filter at the front to pull air in and volia! You could even reduce the case width to the height of the pci slots by using a shorter psu ~12cm. (Just draw a line straight down the left side of the pci slots and 120 fan)
Yeah I know but I have a BTX case coming but don't like the power supply that is coming with it so I was hoping someone had some info on replacing the one that comes with the case, with an ATX type.
This is the case
http://www.in-win.us/products_case_seri ... eries_id=3
This is the case
http://www.in-win.us/products_case_seri ... eries_id=3
-
- Patron of SPCR
- Posts: 857
- Joined: Fri Dec 27, 2002 1:49 pm
- Location: Somerset, WI - USA
- Contact: