T
ToolPackinMama
Wooducoodu said:
Hmm...
Hmm...
T
ToolPackinMama
Matt said:
::rolling eyes:: What would it take to convince those geniuses to agree
on a standard size screw?
M
Matt
Dave said:
Well that's just it. You ought to explain that if you want to keep your
claim that BTX is not much of a technical improvement.
M
Matt
ToolPackinMama said:
LOL. I haven't perused the BTX spec. Is there no spec for screws?
Maybe I'll retract my claim.
D
Dorothy Bradbury
BTX is a little more than that:
o Heatsinks will be bigger & heavier
---- so motherboard to case bracing will be required
---- vis., motherboard-bracing integrated to case-bracing
o Memory is turned thro 90-degrees
---- this is re 1U & desktop applications
---- later DDR2 ram will run hot, and be a lot of it re 64-bit
---- 2x 1GB DDR2 will be about 20-30W, appreciable
---- so 1-2m/sec airflow along the Dimms is required v against the Dimm
---- airflow against Dimms (eg, 1U) causing temp shadowing of 3-4oC
o CPU & Memory VRM are bigger
---- memory could run to 70W in some server boards
---- CPUs could run to 150W for single CPU eventually, 300W for dual
---- VRM thus needs to have direct cooling
o Capacitor loadings are higher
---- P4-Prescott is pushing free-air-motherboards to 235F
---- yet those capacitors are rated to 105oC, military is 125oC
---- cooling across m/b discrete components is now critical
o VRM loadings are higher
---- CPUs are already nearing the limit of existing-uptake VRM
---- Intel's last spec was for 103A VRM, we may need more soon
o CPU socket changes needed
---- ZIFF getting difficult re # of pins & current loading
---- packaging getting difficult as you scale past 103W CPUs
o 1U rack use has issues with high rear I/O backplane
---- exhaust airflow over that area is desirable re inline with dual CPU
---- if dual CPU are like Dual Prescott 103W, we need more holes
---- so BTX allows a lower profile rear I/O panel, more airflow holes
o 24-pin from 20-pin ATX change
---- marginal change, at least not another P9-II or AUX-II spaghetti
Indeed, the question is really:
o Could you make it work with ATX motherboards?
---- Yes -- heatsink weight may be an issue re shipping-weight
o Could you make it work with generic cheapy ATX cases?
---- Difficult -- they aren't so great at "thro airflow" over chassis
For one thing, I'm not convinced the BTX design has a long life:
o Graphics + CPU + VRM are cooled by CPU fan
---- RAM also, but that may be "collateral airflow"
o 75W + 103W + 15W are the figures out now = 193W
---- add RAM and potentially it's another 30W = 223W
o For One CPU it's 223W = 45cfm
o For Dual CPU it's 335W = 65cfm
o For later Tajas etc at 125-150W it's 54cfm & 84cfm
Hence I believe 125-150W will mark the heat limit of CPUs,
because we're into the "heavier car = heavier brakes etc" cycle.
BTX is also specifying smaller cases of 3.00" & 3.98" height.
The former requires 60mm fans, the latter 80/92mm fans. So
I'm not sure the chunky wide-tower using 92/120mm fan goes.
A "spanner" here is the graphics card - some will exceed 75W
thro separate power connector (v PCI-X power draw). So then
we are into cooling them in a somewhat messy inline with CPU.
BTX was aimed at creating a fanless CPU & graphics card.
o So fewer fans needed - saving money
o So quieter solution possible
Unfortunately the watt/cfm figures show We Have A Problem.
Considering Prescott heat output, exactly what will it be at 5Ghz?
The Apple Dual G5 is a better design:
o Dual-G5 -- Graphics/CPU/memory/VRM in separate cooling zones
o BTX ----- Graphics/CPU/VRM in same zone, passive heatsinks
Unfortunately the Apple uses a silly number of fans:
o Dual-G5 -- Each cooling zone gets push/pull fans re redundancy
o BTX ------ One fan, struggling to cool a very high heat density
Personally I'd have specified 2 cooling zones on "BTX"
o One for the CPU area - covering single or dual CPU
o One for the card area - covering graphics or multiple cards
o Simple standardised plastic-duct clipped to board mount posts
o Blow thro design
The hard drives (like BTX) in front of the ATX PSU air inlet re cooling.
Looks like BTX is a desktop form-factor, with the server market in dual
form left to come up with their own solution - smells of E-BTX & Case-II.
o Heatsinks will be bigger & heavier
---- so motherboard to case bracing will be required
---- vis., motherboard-bracing integrated to case-bracing
o Memory is turned thro 90-degrees
---- this is re 1U & desktop applications
---- later DDR2 ram will run hot, and be a lot of it re 64-bit
---- 2x 1GB DDR2 will be about 20-30W, appreciable
---- so 1-2m/sec airflow along the Dimms is required v against the Dimm
---- airflow against Dimms (eg, 1U) causing temp shadowing of 3-4oC
o CPU & Memory VRM are bigger
---- memory could run to 70W in some server boards
---- CPUs could run to 150W for single CPU eventually, 300W for dual
---- VRM thus needs to have direct cooling
o Capacitor loadings are higher
---- P4-Prescott is pushing free-air-motherboards to 235F
---- yet those capacitors are rated to 105oC, military is 125oC
---- cooling across m/b discrete components is now critical
o VRM loadings are higher
---- CPUs are already nearing the limit of existing-uptake VRM
---- Intel's last spec was for 103A VRM, we may need more soon
o CPU socket changes needed
---- ZIFF getting difficult re # of pins & current loading
---- packaging getting difficult as you scale past 103W CPUs
o 1U rack use has issues with high rear I/O backplane
---- exhaust airflow over that area is desirable re inline with dual CPU
---- if dual CPU are like Dual Prescott 103W, we need more holes
---- so BTX allows a lower profile rear I/O panel, more airflow holes
o 24-pin from 20-pin ATX change
---- marginal change, at least not another P9-II or AUX-II spaghetti
Indeed, the question is really:
o Could you make it work with ATX motherboards?
---- Yes -- heatsink weight may be an issue re shipping-weight
o Could you make it work with generic cheapy ATX cases?
---- Difficult -- they aren't so great at "thro airflow" over chassis
For one thing, I'm not convinced the BTX design has a long life:
o Graphics + CPU + VRM are cooled by CPU fan
---- RAM also, but that may be "collateral airflow"
o 75W + 103W + 15W are the figures out now = 193W
---- add RAM and potentially it's another 30W = 223W
o For One CPU it's 223W = 45cfm
o For Dual CPU it's 335W = 65cfm
o For later Tajas etc at 125-150W it's 54cfm & 84cfm
Hence I believe 125-150W will mark the heat limit of CPUs,
because we're into the "heavier car = heavier brakes etc" cycle.
BTX is also specifying smaller cases of 3.00" & 3.98" height.
The former requires 60mm fans, the latter 80/92mm fans. So
I'm not sure the chunky wide-tower using 92/120mm fan goes.
A "spanner" here is the graphics card - some will exceed 75W
thro separate power connector (v PCI-X power draw). So then
we are into cooling them in a somewhat messy inline with CPU.
BTX was aimed at creating a fanless CPU & graphics card.
o So fewer fans needed - saving money
o So quieter solution possible
Unfortunately the watt/cfm figures show We Have A Problem.
Considering Prescott heat output, exactly what will it be at 5Ghz?
The Apple Dual G5 is a better design:
o Dual-G5 -- Graphics/CPU/memory/VRM in separate cooling zones
o BTX ----- Graphics/CPU/VRM in same zone, passive heatsinks
Unfortunately the Apple uses a silly number of fans:
o Dual-G5 -- Each cooling zone gets push/pull fans re redundancy
o BTX ------ One fan, struggling to cool a very high heat density
Personally I'd have specified 2 cooling zones on "BTX"
o One for the CPU area - covering single or dual CPU
o One for the card area - covering graphics or multiple cards
o Simple standardised plastic-duct clipped to board mount posts
o Blow thro design
The hard drives (like BTX) in front of the ATX PSU air inlet re cooling.
Looks like BTX is a desktop form-factor, with the server market in dual
form left to come up with their own solution - smells of E-BTX & Case-II.
J
jeff findley
Chuck Dreier said:
There will surely be overlap between ATX and BTX production. I was
able to buy an AT form factor "super socket 7" motherboard long after
ATX motherboards were "standard".
Jeff
S
somebody
Maybe they are designing an imperfect cooling system because that would
handicap AMD because the AMD chips run hotter than Intel's.
No, they don't. That's just a silly assumption that a lot of people
make all the time. AMD is in fact, if anything, cooler today.
It's true that P4s allow you to be sloppy when mounting the heatsink.
It's true that undercooled P4s will not give you the heat trouble that
undercooled AthlonXPs would. But it all comes at a price. P4s will
throttle when they get warm under prolonged 100% load. That's all.
Both their 3 and 3.2 GHz chips go beyond 100W under full steam. But
thermally, they're engineered for like 70-90W. So you better cool them
really really well, if you want to get full performance out of them.
I kinda agree. I also think Intel need the cooling. Their marketing
approach of clockrates at all costs, are driving up power and heat
exponentially. They intend to clock the Prescott up to 5GHz, possibly
beyond. That's the big reason for BTX.
As for Intel engineering, they're _THE_ experts on manufacturing. But
they've always been very mediocre as designers. In fact they've always
been the _worst_ cpu designers in the business, since the 8080.
(I'm judging this on basis of features and performance/clock and
performance/transistor.)
Today, their product range is remarkably unexciting. Their P4C is the
good stuff. Excellent performance at a price that, considering the
nice HT feature, is quite reasonable.
But for the rest, they're way outclassed. P4A and P4B are poor value,
and both Celeron and Xeon perform abysmally poor against much cheaper
competition. Market buys the crap just out of habit.
Itanium is a disaster. Prescott was first much delayed and then a huge
disappointment. Certainly not an answer to K8. At such inefficiency,
even 5 GHz is not going to be enough to match future AMD 64s.
Intel is having the problems of a huge company, that gets away with
selling anything for a long time..
Personally, I don't think they're incompetent. Just as in the case of
the '286 and Itanium, ( and just like Microsoft ;-) btw) they're
building on flawed concepts.
They originally planned for the PIII to be their last '86. But the
Itanium didn't work out and they got caught unprepared by the K7. The
P4 was a simple, expedient solution. It didn't try anything fancy.
Just high clockrate, relying on ISA extension -SSE2- to deliver the
real performance increase. So it was low technical risk and a short
timetable.
Now they've got stuck on the clockrate track for too long. It's worked
very well on a clueless market, for them and Dell, but it's lousy
technology.
They need Tejas. And it's better be good.
Well, why shouldn't a mobo producer continue to offer ATX boards with
new chipsets, PCIX, DDR2 etc, if they're technically feasible and sell
well? Remember what happened to rambus. Who says you can't do better
cooling solutions on ATX as well?
ancra
L
Larc
On 27 Feb 2004 09:34:32 -0500, jeff findley pondered exceedingly, then took
quill in hand and carefully composed...
| >
| > Again, how do y'all feel about waiting for the BTX form factor? I don't see
| > putting $2,000+ in a new homebuilt (ATX) system, only to have it obsolete in
| > terms of upgrading by the end of the year. I wonder what premium there will
| > be for BTX form-factor components at the get-go??? Thanks...
|
| There will surely be overlap between ATX and BTX production. I was
| able to buy an AT form factor "super socket 7" motherboard long after
| ATX motherboards were "standard".
Of course. Remember how long ISA was still supported after the introduction of
PCI? Things in computerdom just don't tend to get replaced by something else
overnight.
The more I hear and read about BTX, the more convinced I am to go ahead and
build myself a new ATX system sometime this year. When — and if — BTX becomes
the stable standard, there'll be time enough to jump on that bandwagon.
And I can't help thinking about Intel 850 chipsets and RAMBUS memory. They were
to be the shining light of the future. But people stayed away in droves, thanks
in great part to DDR. Now, Intel is permitting the 850 to die a natural death
and RAMBUS memory isn't even available from some of the leading manufacturers.
Maybe there's at least an outside chance of BTX going the same way. It wouldn't
be the first time something has been canceled due to lack of interest.
Larc
§§§ - Please change planet to earth to reply by e-mail - §§§
quill in hand and carefully composed...
| >
| > Again, how do y'all feel about waiting for the BTX form factor? I don't see
| > putting $2,000+ in a new homebuilt (ATX) system, only to have it obsolete in
| > terms of upgrading by the end of the year. I wonder what premium there will
| > be for BTX form-factor components at the get-go??? Thanks...
|
| There will surely be overlap between ATX and BTX production. I was
| able to buy an AT form factor "super socket 7" motherboard long after
| ATX motherboards were "standard".
Of course. Remember how long ISA was still supported after the introduction of
PCI? Things in computerdom just don't tend to get replaced by something else
overnight.
The more I hear and read about BTX, the more convinced I am to go ahead and
build myself a new ATX system sometime this year. When — and if — BTX becomes
the stable standard, there'll be time enough to jump on that bandwagon.
And I can't help thinking about Intel 850 chipsets and RAMBUS memory. They were
to be the shining light of the future. But people stayed away in droves, thanks
in great part to DDR. Now, Intel is permitting the 850 to die a natural death
and RAMBUS memory isn't even available from some of the leading manufacturers.
Maybe there's at least an outside chance of BTX going the same way. It wouldn't
be the first time something has been canceled due to lack of interest.
Larc
§§§ - Please change planet to earth to reply by e-mail - §§§
D
Dorothy Bradbury
The risk of BTX is not lack of adoption:
o A joint ATX/BTX case requires two backplates
---- card slots are on the other side
---- I/O array is a different size & shape
o A joint ATX/BTX case requires two frontages
---- cooling intake is required in the front
The risk of BTX is it's a "desktop consumer" standard
o It doesn't cover dual-CPU well at all
o So server & high-end is "proprietarisation"
That's already admitted with blade servers replacing 1U, and the
BTX standard does nothing to change that trend re heat/layout etc.
So BTX defines cooling for the desktop, not the server - "as is".
So the system integrators like Dell still keep their marketing niche.
o A joint ATX/BTX case requires two backplates
---- card slots are on the other side
---- I/O array is a different size & shape
o A joint ATX/BTX case requires two frontages
---- cooling intake is required in the front
The risk of BTX is it's a "desktop consumer" standard
o It doesn't cover dual-CPU well at all
o So server & high-end is "proprietarisation"
That's already admitted with blade servers replacing 1U, and the
BTX standard does nothing to change that trend re heat/layout etc.
So BTX defines cooling for the desktop, not the server - "as is".
So the system integrators like Dell still keep their marketing niche.
T
tuta
Chuck Dreier said:
Thanks for starting this - some of the most interesting things I have read
here...
JohnH
G
Gary Tait
Not impossible, but complex.
not for a full sized ATX case, you can use an air guide to guide fresh
air into the fan unit.
D
Dorothy Bradbury
o A joint ATX/BTX case requires two frontages
Indeed, the ATX standard could have been developed over the years.
Indeed, the ATX standard could have been developed over the years.
D
Dave C.
Of course there's a spec for screws. It's under the heading Screw the
Consumer. (you're welcome) -Dave
D
Dave C.
I have no idea why Intel would propose changes that are mostly cosmetic,
and
Ummmm . . . that would seem to be it's own explanation. That is, it's not
much of an improvement because Intel did it. -Dave
and
Ummmm . . . that would seem to be it's own explanation. That is, it's not
much of an improvement because Intel did it.
-Dave
D
Dave C.
You list a lot of specific changes in BTX vs. ATX, and then you list the
number one reason why BTX isn't needed. -Dave
D
Dorothy Bradbury
You list a lot of specific changes in BTX vs. ATX, and then you list the
That was my point
o "Could you make it work with ATX motherboards?"
o "Yes"
CPU heatsink weight may be an issue - re copper v aluminium.
Copper is much heavier, aluminium much cheaper - altho bigger.
Proposed Tajas heatsinks look like a small architectural development.
That was my point
o "Could you make it work with ATX motherboards?"
o "Yes"
CPU heatsink weight may be an issue - re copper v aluminium.
Copper is much heavier, aluminium much cheaper - altho bigger.
Proposed Tajas heatsinks look like a small architectural development.
T
ToolPackinMama
jeff said:
I think that's a safe bet.
T
ToolPackinMama
You know? IMHO that's a good point.
BTW, isn't this thread OT for alt.comp.hardware.homebuilt? It's not
like you are talking about producing your own mobo at home.
T
ToolPackinMama
Dave C. said:
LOL
T
ToolPackinMama
Matt said:
It's just that I built a dozen 'puters this year, and I have at least 14
different style screws left over to show for it.
If they can have one ring to rule them all, then why can't we have one
screw to... you know...