C
Chuck Dreier
I read that the ATX will soon be replaced by the BTX form factor. Is that
true, and if so, how soon is "soon"? Thanks.
true, and if so, how soon is "soon"? Thanks.
H
Harry
Soon is the second half of this year.
Having said that, ATX will be around for a few years yet. Not everyone
will be upgrading their PC in the next year or so.
HTH
Harry
D
Dorothy Bradbury
There are many changes later this year:
BTX is a form-factor driver - changes case/motherboard
o ATX/BTX compatible cases like AT/ATX are unlikely
BTX will also signal many other changes
o new CPU socket -- no longer "ZIFF" format
o new Heatsinks -- as wattages rise 100-125-150W
---- I don't think we'll exceed 150W in consumer PCs tho
o new VRM System -- over 103A required by CPUs
---- 103W CPUs are now, Tajas is projected over 125W
o new Memory Design -- DDR2 will allow 400-667Mhz, 1.8V
---- lower voltage allows more sockets & less heat
---- even so expect heat-spreaders on high-end solutions
---- 2m/sec cooling for their 30-70W heatoutput (2-16GB)
o new PCI Architecture -- PCI-X replacing AGP/PCI
You can use your old PSUs in the BTX cases tho with an adapter
plate that will be part of the specification. BTX PSUs are more an
L shape - probably re IEC connection and such like.
HD will still be S-ATA focused, eventually S-ATA-2 which will at last
allow daisy-chaining of drives per channel (like SCSI is). SCSI will move
from parallel to serial also, altho the cheap S-ATA drives are popular "as
is" re 3ware RAID controllers for data-mining SQL server boxes.
The VRM, CPU & Memory thermal output require a change in case
form-factor - you simply can't cool large amounts of DDR2 memory in
the existing ATX form-factor (at least generic cases). Junction temps
will be 90-100oC, needing 2m/sec cooling & airflow parallel to DIMMS.
Dimms that shield one another (eg, many existing 1U designs) cause a lot
of thermal shadowing to those behind - 3-4oC difference in temperature.
So a lot of changes - and a lot of obsolescence unfortunately.
Prices thro H2 are likely to drop in light of new sockets and such, which
may explain the "Prescott model" where the smaller die will allow Intel to
drop prices aggressively without losing margin. So if you can wait you may
be able to find some interesting price
erformance offerings later in 2004.
Looks like being fun anyway - PCs beginning to "grow up" re design.
Might be "intel inside = marshmallow inside" re heat output too
Burp.
BTX is a form-factor driver - changes case/motherboard
o ATX/BTX compatible cases like AT/ATX are unlikely
BTX will also signal many other changes
o new CPU socket -- no longer "ZIFF" format
o new Heatsinks -- as wattages rise 100-125-150W
---- I don't think we'll exceed 150W in consumer PCs tho
o new VRM System -- over 103A required by CPUs
---- 103W CPUs are now, Tajas is projected over 125W
o new Memory Design -- DDR2 will allow 400-667Mhz, 1.8V
---- lower voltage allows more sockets & less heat
---- even so expect heat-spreaders on high-end solutions
---- 2m/sec cooling for their 30-70W heatoutput (2-16GB)
o new PCI Architecture -- PCI-X replacing AGP/PCI
You can use your old PSUs in the BTX cases tho with an adapter
plate that will be part of the specification. BTX PSUs are more an
L shape - probably re IEC connection and such like.
HD will still be S-ATA focused, eventually S-ATA-2 which will at last
allow daisy-chaining of drives per channel (like SCSI is). SCSI will move
from parallel to serial also, altho the cheap S-ATA drives are popular "as
is" re 3ware RAID controllers for data-mining SQL server boxes.
The VRM, CPU & Memory thermal output require a change in case
form-factor - you simply can't cool large amounts of DDR2 memory in
the existing ATX form-factor (at least generic cases). Junction temps
will be 90-100oC, needing 2m/sec cooling & airflow parallel to DIMMS.
Dimms that shield one another (eg, many existing 1U designs) cause a lot
of thermal shadowing to those behind - 3-4oC difference in temperature.
So a lot of changes - and a lot of obsolescence unfortunately.
Prices thro H2 are likely to drop in light of new sockets and such, which
may explain the "Prescott model" where the smaller die will allow Intel to
drop prices aggressively without losing margin. So if you can wait you may
be able to find some interesting price
erformance offerings later in 2004.Looks like being fun anyway - PCs beginning to "grow up" re design.
Might be "intel inside = marshmallow inside" re heat output too
Burp.
C
Chuck Dreier
Wow, I was just about ready to build a new "super system" - by my standards
anyway. Now, I'm not so sure that I'll do it. Why put all that money into
a great ATX case, MB, etc.. 'Sounds like everything will truly be obsolete.
What I was hoping for was building a system that I could upgrade along the
way. This all sounds like a MAJOR change. Do most of you agree?
anyway. Now, I'm not so sure that I'll do it. Why put all that money into
a great ATX case, MB, etc.. 'Sounds like everything will truly be obsolete.
What I was hoping for was building a system that I could upgrade along the
way. This all sounds like a MAJOR change. Do most of you agree?
B
Bill L
Crikey, feel like I've just come round from a coma or something! BTX wtf is
BTX ... never heard of it!!
BillL
BTX ... never heard of it!!
BillL
M
Matt
Chuck said:
This thread deserves a name change to get people's attention.
BTX is a topic that has been neglected in this group.
The new standard is at www.formfactors.org in case you like formal
documents. Probably most people would like somebody to read between the
lines for them.
So let's hear what is important to know about BTX.
D
Dorothy Bradbury
BTX itself is just the change of motherboard form-factor:
o www.formfactors.org (may need a - in there, I'm on dialup
o The motherboard layout is "mirrored"
The PCI slots flip over to the other side, becoming PCI-X &
AGP is deleted (always reminded me of VESA personally).
No it isn't going to happen overnight - but it will happen.
If you need a machine now, eg, Dual Opteron then you need it.
Conversely if you are trying to future proof, perhaps wait.
There's an interesting argument here:
o Future proofing never works in IT due to chains-of-Moores-Law
---- Moores-Law was transistor-# doubles every 18 months
---- that doesn't necessarily mean performance, but often has
o However, we know thermal issues are putting the brakes on
---- Semi Association to even basic CFD/FEA models confirm it
---- around 2005 we see a slowdown in thermal output growth
---- once 13%, then 28% from 1995, back to 13-5% growth
o So does slowing thermal growth re limits, also slow performance?
---- if so then "some" future proofing MAY become possible
---- somewhat armageddon for Western society I might add
---- and nothing to stop them changing socket on every CPU
We are already at a point where few apps need the power we have.
o You can stick Dual Opteron or Dual G5 on a desk cheaply
o Very few applications & people need their power
---- 10-20 GigaFLOPS is here now/tomorrow economically
Sure if you are running Photoshop, or CFD/FEA, or Mathematica
then there is never enough power - but that mkt is moving into a
sub-classification of power-user which some argue is getting smaller.
Conversely, one can argue that what 1 person needs, but gets any
economic benefit from will eventually be used by everyone to gain
neglidgible economic benefit from (at least re coase economics and
marginal profit once everyone has the capability). So eventually we
will CFD/FEA our way to the fridge perhaps, with Mathematica
used to understand 1) gov't statistics and 2) anything by Greenspan.
Software will always use the h/w we have, we have 10GigaFLOPS
machines running word processors - which I find unfathomable. So
whilst performance may be slowed by heat, future proofing is not
likely since "outside the CPU" may see many changes for speed.
That brings in DDR2 memory, S-ATA-2, more disk rpm, more
graphics power, gigabit to the desktop, more cache memory. At
present CPUs move data from RAM 100x faster than from disk.
So there are many areas to "speed up" machines.
Even so, I suspect the avg CPU load on the "global PC" is 2%,
with 98% being the HALT instruction. We are trying to achieve
the power of a human, in a machine - which makes me wonder
if it's achieved when the machine is as inefficient as the human
o www.formfactors.org (may need a - in there, I'm on dialup
o The motherboard layout is "mirrored"
The PCI slots flip over to the other side, becoming PCI-X &
AGP is deleted (always reminded me of VESA personally).
No it isn't going to happen overnight - but it will happen.
If you need a machine now, eg, Dual Opteron then you need it.
Conversely if you are trying to future proof, perhaps wait.
There's an interesting argument here:
o Future proofing never works in IT due to chains-of-Moores-Law
---- Moores-Law was transistor-# doubles every 18 months
---- that doesn't necessarily mean performance, but often has
o However, we know thermal issues are putting the brakes on
---- Semi Association to even basic CFD/FEA models confirm it
---- around 2005 we see a slowdown in thermal output growth
---- once 13%, then 28% from 1995, back to 13-5% growth
o So does slowing thermal growth re limits, also slow performance?
---- if so then "some" future proofing MAY become possible
---- somewhat armageddon for Western society I might add
---- and nothing to stop them changing socket on every CPU
We are already at a point where few apps need the power we have.
o You can stick Dual Opteron or Dual G5 on a desk cheaply
o Very few applications & people need their power
---- 10-20 GigaFLOPS is here now/tomorrow economically
Sure if you are running Photoshop, or CFD/FEA, or Mathematica
then there is never enough power - but that mkt is moving into a
sub-classification of power-user which some argue is getting smaller.
Conversely, one can argue that what 1 person needs, but gets any
economic benefit from will eventually be used by everyone to gain
neglidgible economic benefit from (at least re coase economics and
marginal profit once everyone has the capability). So eventually we
will CFD/FEA our way to the fridge perhaps, with Mathematica
used to understand 1) gov't statistics and 2) anything by Greenspan.
Software will always use the h/w we have, we have 10GigaFLOPS
machines running word processors - which I find unfathomable. So
whilst performance may be slowed by heat, future proofing is not
likely since "outside the CPU" may see many changes for speed.
That brings in DDR2 memory, S-ATA-2, more disk rpm, more
graphics power, gigabit to the desktop, more cache memory. At
present CPUs move data from RAM 100x faster than from disk.
So there are many areas to "speed up" machines.
Even so, I suspect the avg CPU load on the "global PC" is 2%,
with 98% being the HALT instruction. We are trying to achieve
the power of a human, in a machine - which makes me wonder
if it's achieved when the machine is as inefficient as the human
D
Dorothy Bradbury
Forgot a bit. M-BTX is roughly 267x267mm, and the cases are generally
meant to be 3" high or 3.98" high - much smaller than existing case designs.
That said I suspect towers will continue, altho perhaps someone could come
out with something a bit different - like a cube. Just to break the mould. There
is also a Pico-BTX format which is 209x267mm (267mm is the depth, may be
269mm or that may just be my note re clearances on a Cad drawing, long day).
So basically BTX is also deeper than most present ATX designs. A pity they
didn't make Pico-BTX a 209x209mm format, that might bring broad support
for something smaller than existing largely wasteful case voluminouos designs.
Hopefully they create dual-CPU in M-BTX, rather than BTX or some E-BTX.
Quite why we need a 12"x13" board for dual CPU is beyond me, silly big.
A final note is BTX will have much bigger & heavier heatsinks, so the case
itself will be fundamentally incorporated in substantial motherboard bracing.
Hopefully an improvement from the flex-R-us of the S478 design.
Noting Telco are up to 20-way motherboards as routine now, we may also
see similar in the PC world which will be interesting - albeit expensive.
meant to be 3" high or 3.98" high - much smaller than existing case designs.
That said I suspect towers will continue, altho perhaps someone could come
out with something a bit different - like a cube. Just to break the mould. There
is also a Pico-BTX format which is 209x267mm (267mm is the depth, may be
269mm or that may just be my note re clearances on a Cad drawing, long day).
So basically BTX is also deeper than most present ATX designs. A pity they
didn't make Pico-BTX a 209x209mm format, that might bring broad support
for something smaller than existing largely wasteful case voluminouos designs.
Hopefully they create dual-CPU in M-BTX, rather than BTX or some E-BTX.
Quite why we need a 12"x13" board for dual CPU is beyond me, silly big.
A final note is BTX will have much bigger & heavier heatsinks, so the case
itself will be fundamentally incorporated in substantial motherboard bracing.
Hopefully an improvement from the flex-R-us of the S478 design.
Noting Telco are up to 20-way motherboards as routine now, we may also
see similar in the PC world which will be interesting - albeit expensive.
L
Larc
| Crikey, feel like I've just come round from a coma or something! BTX wtf is
| BTX ... never heard of it!!
It's a new plan by hardware companies such as Intel to make a shitload of money!
Larc
§§§ - Please change planet to earth to reply by e-mail - §§§
| BTX ... never heard of it!!
It's a new plan by hardware companies such as Intel to make a shitload of money!
Larc
§§§ - Please change planet to earth to reply by e-mail - §§§
C
Chuck Dreier
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...
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...
W
Wooducoodu
if you're only counting the tower in the price i'd say build a great ATX
system for about $1000 now or in the next few months. if you wait for BTX
you'll spend twice as much for a system that isn't much faster plus you'll
get to enjoy all the new bugs that crop up in the first few months with all
the new technology.
system for about $1000 now or in the next few months. if you wait for BTX
you'll spend twice as much for a system that isn't much faster plus you'll
get to enjoy all the new bugs that crop up in the first few months with all
the new technology.
D
Dave C.
OK, based on responses I've read to this so far, nobody has bothered to
actually study the new form factor before commenting on it. To sum it up
briefly, if computers were cars, here is how Chevy would make a BTX system.
They would start with an ATX, rotate the tires and then move the cupholders
a few inches. DONE!!!
If you really study BTX, you will be amazed at how similar it is to ATX. In
fact, the more you study BTX, the more you wonder why a new form factor is
needed. Eventually, you conclude a new form factor is NOT NEEDED. There is
not one element in BTX that couldn't easily be implemented in the current
ATX form factor.
The main differences in BTX . . .
1) In BTX, there is a wind tunnel built over the CPU and chipset to help
heat the chipset with super-heated air from the CPU (who's brilliant idea
was this, intel?)
2) The expansion slots are where the ports were in ATX
3) The ports are where the expansion slots were in ATX (see number 2)
4) The RAM has been moved to the opposite end of the motherboard
Someone will probably argue that BTX allows for smaller motherboards. Yes,
that's true. There is a BTX motherboard that is extremely small. It's made
by cutting off most (or all) of the expansion slots. (Gee, an ATX board
could be awfully small if you cut off all the expansion slots, also.)
Someone else might argue that new technology like PCI express is needed. I
agree. There's nothing stopping motherboard makers from putting PCI express
slots on ATX boards though. If the customers want PCI express, they will
get it . . . regardless of the form factor of the mainboard.
BTX is change that is not needed. But it's such a minor change, the only
thing it will really accomplish is to prop up falling component prices for
several months, possibly. At least on motherboards and cases, that
s. -Dave
actually study the new form factor before commenting on it. To sum it up
briefly, if computers were cars, here is how Chevy would make a BTX system.
They would start with an ATX, rotate the tires and then move the cupholders
a few inches. DONE!!!
If you really study BTX, you will be amazed at how similar it is to ATX. In
fact, the more you study BTX, the more you wonder why a new form factor is
needed. Eventually, you conclude a new form factor is NOT NEEDED. There is
not one element in BTX that couldn't easily be implemented in the current
ATX form factor.
The main differences in BTX . . .
1) In BTX, there is a wind tunnel built over the CPU and chipset to help
heat the chipset with super-heated air from the CPU (who's brilliant idea
was this, intel?)
2) The expansion slots are where the ports were in ATX
3) The ports are where the expansion slots were in ATX (see number 2)
4) The RAM has been moved to the opposite end of the motherboard
Someone will probably argue that BTX allows for smaller motherboards. Yes,
that's true. There is a BTX motherboard that is extremely small. It's made
by cutting off most (or all) of the expansion slots. (Gee, an ATX board
could be awfully small if you cut off all the expansion slots, also.)
Someone else might argue that new technology like PCI express is needed. I
agree. There's nothing stopping motherboard makers from putting PCI express
slots on ATX boards though. If the customers want PCI express, they will
get it . . . regardless of the form factor of the mainboard.
BTX is change that is not needed. But it's such a minor change, the only
thing it will really accomplish is to prop up falling component prices for
several months, possibly. At least on motherboards and cases, that
s. -Dave
D
Dave C.
Larc said:
Not really. It's a change that isn't needed, but not likely to make
hardware manufacturers of any type rich. BTX is supposed to offer better
(shorter) electrical paths and heat dissipation qualities. I find this
quite amusing, though. As if we need a whole new form factor to route a few
traces differently through a multi-layer PCB. Intel telling mainboard
makers how to cool them better is like Goodyear telling automobile
manufacturers how to make cars go faster. Like . . . ummmm, thank you . . .
now go piss off. (!) -Dave
W
Wooducoodu
i agree that btx isn't needed and everything that's going to be introduced
on btx boards like pci express could just as easily be done on atx but i
disagree that we'll ever get those things on atx. every case, psu, board,
processor, heatsink, ram, card and drive is about to be forced obsolete just
so the industry can make more money.
http://www.theinquirer.net/?article=13425
on btx boards like pci express could just as easily be done on atx but i
disagree that we'll ever get those things on atx. every case, psu, board,
processor, heatsink, ram, card and drive is about to be forced obsolete just
so the industry can make more money.
http://www.theinquirer.net/?article=13425
D
Dave C.
Wooducoodu said:
You misunderstood me. I didn't mean to say that new technology would be put
in ATX mainboards or cases. OF COURSE that's not going to happen if the
form factor is changing. BTX is the ONLY reason why you won't see new
technology like PCI express on ATX, though.
Every PSU, mainboard, processor, heatsink, ram, card and drive would be
obsolete just as fast if BTX didn't exist. Technology changes so fast that
any component purchased today will be almost obsolete or definitely obsolete
by this same day next year. It doesn't take a new form factor to make that
happen. -Dave
J
jd
Do a motherboard, processor, video card upgrade now. That should hold you (and most power users) for
the next year (or more) then see how this new format fans out.
You seem to realize what you are up against, why sweat it. Things change. Whether its needed or not
is a different topic all together but there is no one telling you to go spend $2000 now so you can
be obsolete from an upgrade stand point next year.
I don't know where you were when the switch from AT to ATX took place but it wasn't an overnight
affair. You think manufactures are going to abandoned the millions of ATX systems currently in use
as soon as something new comes along? I'm sure anything you put together today will be upgradeable
a year from now.
Now if you got money to burn buy the latest and greatest today, then again tomorrow, and the next
day........
BTW where do you live? I would love to see what is sitting on your curb after an upgrade
the next year (or more) then see how this new format fans out.
You seem to realize what you are up against, why sweat it. Things change. Whether its needed or not
is a different topic all together but there is no one telling you to go spend $2000 now so you can
be obsolete from an upgrade stand point next year.
I don't know where you were when the switch from AT to ATX took place but it wasn't an overnight
affair. You think manufactures are going to abandoned the millions of ATX systems currently in use
as soon as something new comes along? I'm sure anything you put together today will be upgradeable
a year from now.
Now if you got money to burn buy the latest and greatest today, then again tomorrow, and the next
day........
BTW where do you live? I would love to see what is sitting on your curb after an upgrade
M
Matt
Dave said:
I don't know, but probably I am willing to defer to Intel's engineering
expertise on this. This is an enormous corporation that can hire the
best engineering talent in the world. I don't think you will get
anywhere by supposing that they are incompetent.
Besides the ability, they would seem to have the motive to design the
best possible cooling strategy, since they will be wanting to build very
powerful (read: hot) processors and chipsets in the future.
Maybe they are designing an imperfect cooling system because that would
handicap AMD because the AMD chips run hotter than Intel's.
Thatsounds kooky, but if BTX is no improvement technically, I don't know how
you will explain its promotion by Intel without a theory of some
sinister conspiracy.
I doubt that this is going to affect motherboard prices much, except
that the BTX boards can be close to an inch bigger in both directions,
so maybe they will need new equipment to make them? I don't think board
makers are going to redesign an old mobo just to make it fit in a BTX
case. I would expect their mobo designs to be driven mainly as they are
now---by the introduction of new chipsets and processors. Unless I err,
each manufacturer will decide for each new chipset whether it will go
onto a BTX or an ATX board, but they will not make two versions for the
same chipset.
Unless I am wrong, Intel is not a big case maker.
Please indicate some hypotheses for Intel's motivation.
W
Wooducoodu
that's true of some parts but not all. cases and psus, can be used for
years, there arent going to be any great leaps in sound cards and when was
the last time you needed a new cd rom drive unless you wanted a writer or a
new modem unless you wanted cable or dsl?
years, there arent going to be any great leaps in sound cards and when was
the last time you needed a new cd rom drive unless you wanted a writer or a
new modem unless you wanted cable or dsl?
D
Dave C.
I have no idea why Intel would propose changes that are mostly cosmetic, and
seem to do more harm than good in some cases. What I'm most confused about
though, is why the whole form factor is being dictated by Intel. That's
like BMW dictating that all cars will have five wheels, and all other car
makers going along with it for some odd reason. -Dave
D
Dave C.
Wooducoodu said:
Yup, it's true that you can (without a form factor change, that is) recycle
one case to be used in several systems if you want to. It's unlikely your
power supply can handle more than one complete system upgrade, though . . .
unless you spent a small fortune on it, or don't care too much about
stability. Optical drives are changing fast, so I don't understand your
point on that. We started with cd-rom, then got faster and faster and
faster cd-rom and now we have dvd-rom, cdr/w, various dvd recording
technologies . . . the cd-rom drive really IS obsolete right now. -Dave