Archived from groups: alt.comp.hardware.overclocking.amd (More info?)

Lurking for a while and thought I'd ask the question now that the 939 FX's
are out. Does anyone know if a manufacturer has released or plans to
release a dual socket board for those CPU's?

Archived from groups: alt.comp.hardware.overclocking.amd (More info?)

On Tue, 15 Jun 2004 12:59:48 -0400, AJ wrote:

> Lurking for a while and thought I'd ask the question now that the 939 FX's
> are out. Does anyone know if a manufacturer has released or plans to
> release a dual socket board for those CPU's?

No. They aren't dual cpu's. If you want dual, you'll have to use the
Opteron series 2xx.

--
Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
http://mysite.verizon.net/res0exft/cpu.htm

Archived from groups: alt.comp.hardware.overclocking.amd (More info?)

"Wes Newell" <w.newell@TAKEOUTverizon.net> wrote in message
newsan.2004.06.15.20.57.08.360041@TAKEOUTverizon.net...
> On Tue, 15 Jun 2004 12:59:48 -0400, AJ wrote:
>
> > Lurking for a while and thought I'd ask the question now that the 939
FX's
> > are out. Does anyone know if a manufacturer has released or plans to
> > release a dual socket board for those CPU's?
>
> No. They aren't dual cpu's. If you want dual, you'll have to use the
> Opteron series 2xx.

That seems kind of silly. The 939 is supposed to phase out the 940, which
is dual, but the 939 isn't?


>
> --
> Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
> http://mysite.verizon.net/res0exft/cpu.htm

Archived from groups: alt.comp.hardware.overclocking.amd (More info?)

Hi AJ,
"AJ" <ajhatespammers@diespammersdie.com> wrote in message
news:tvqdnb1tBNyy9VLd4p2dnA@comcast.com...
>
> "Wes Newell" <w.newell@TAKEOUTverizon.net> wrote in message
> newsan.2004.06.15.20.57.08.360041@TAKEOUTverizon.net...
> > On Tue, 15 Jun 2004 12:59:48 -0400, AJ wrote:
> >
> > > Lurking for a while and thought I'd ask the question now that the 939
> FX's
> > > are out. Does anyone know if a manufacturer has released or plans to
> > > release a dual socket board for those CPU's?
> >
> > No. They aren't dual cpu's. If you want dual, you'll have to use the
> > Opteron series 2xx.
>
> That seems kind of silly. The 939 is supposed to phase out the 940, which
> is dual, but the 939 isn't?

According to Anandtech AMD are going to make dual cpu's on a single die
sometime next year.. But of course they will require yet another type of
skt:-(

Regards,
RD

Archived from groups: alt.comp.hardware.overclocking.amd (More info?)

AJ wrote:
> "Wes Newell" <w.newell@TAKEOUTverizon.net> wrote in message
> newsan.2004.06.15.20.57.08.360041@TAKEOUTverizon.net...
>> On Tue, 15 Jun 2004 12:59:48 -0400, AJ wrote:
>>
>>> Lurking for a while and thought I'd ask the question now that the
>>> 939 FX's are out. Does anyone know if a manufacturer has released
>>> or plans to release a dual socket board for those CPU's?
>>
>> No. They aren't dual cpu's. If you want dual, you'll have to use the
>> Opteron series 2xx.
>
> That seems kind of silly. The 939 is supposed to phase out the 940,
> which is dual, but the 939 isn't?

I doubt Socket 940 is going away any time soon for SMP machines. It costs a
huge amount of money to develop a 2-way, let alone a 4-way, motherboard
mainly due to all sorts of subtle timing and routing issues. If AMD dumped
940, then it'd be left without 2-way boards for at least a year, and 4-way
boards for probably close to 2 years, judging by the amount of time taken
for socket 940 boards to appear. This would be commercial suicide for AMD,
who is trying to get into the server (ie: SMP) market. It's also probably
the reason why AMD is not ruling out dual-cores on socket 940, as it's
unlikely to have established 2- or 4-way boards by the time dualcores come
out (2nd half 2005 by the roadmaps).

But in any case, this is irrelevant to the point made by Wes. The FX series
is not SMP capable, nor is the A64 series. The only CPUs which are SMP
capable are the 2xx or 8xx Opterons, which haven't appeared in socket 939
forms yet. This is decided by fuses set by AMD inside the core, so it's
impossible to change a FX into an Opteron (a la the XP -> MP L5
modification). Neither is it possible for chipsets/boards to override these
settings. Supposedly Opterons are supposed to come in 939 form eventually,
but the introduction date is AFAIK unknown.

--
Michael Brown
www.emboss.co.nz : OOS/RSI software and more
Add michael@ to emboss.co.nz - My inbox is always open

Archived from groups: alt.comp.hardware.overclocking.amd (More info?)

>> No. They aren't dual cpu's. If you want dual, you'll have to use the
>> Opteron series 2xx.
>
> That seems kind of silly. The 939 is supposed to phase out the 940, which
> is dual, but the 939 isn't?

On dual systems stability is more important than the last bit of
performance

Archived from groups: alt.comp.hardware.overclocking.amd (More info?)

On Tue, 15 Jun 2004 17:22:42 -0400, AJ wrote:

>
> "Wes Newell" <w.newell@TAKEOUTverizon.net> wrote in message
> newsan.2004.06.15.20.57.08.360041@TAKEOUTverizon.net...
>> On Tue, 15 Jun 2004 12:59:48 -0400, AJ wrote:
>>
>> > Lurking for a while and thought I'd ask the question now that the 939
> FX's
>> > are out. Does anyone know if a manufacturer has released or plans to
>> > release a dual socket board for those CPU's?
>>
>> No. They aren't dual cpu's. If you want dual, you'll have to use the
>> Opteron series 2xx.
>
> That seems kind of silly. The 939 is supposed to phase out the 940, which
> is dual, but the 939 isn't?
>
The 939 isn't slated to phase out the 940. It was intended to phase out
socket 754.

Before 939, you had to use a 940 for dual channel memory, and the memory
had to be registered. 939, is just an addition to the line using dual
channel unregistered memory This brings the standard Athlon 64, which was
single channel, into the dual channel arena also. Just a little confusing,
but there will be 3 boards for a while. 754, 939, and 940. I would expect
754 to byte the dust first, but you never know in this game.


--
Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
http://mysite.verizon.net/res0exft/cpu.htm

Archived from groups: alt.comp.hardware.overclocking.amd (More info?)

With AMD Sempron(Socket754) on its way, Socket 754 migh be around longer
than 940. Socket 939 performs not far behind 940 for le$$, also unregistered
memory is cheaper than registered.





"Wes Newell" <w.newell@TAKEOUTverizon.net> wrote in message
newsan.2004.06.16.04.50.20.484077@TAKEOUTverizon.net...
> On Tue, 15 Jun 2004 17:22:42 -0400, AJ wrote:
>
> >
> > "Wes Newell" <w.newell@TAKEOUTverizon.net> wrote in message
> > newsan.2004.06.15.20.57.08.360041@TAKEOUTverizon.net...
> >> On Tue, 15 Jun 2004 12:59:48 -0400, AJ wrote:
> >>
> >> > Lurking for a while and thought I'd ask the question now that the 939
> > FX's
> >> > are out. Does anyone know if a manufacturer has released or plans to
> >> > release a dual socket board for those CPU's?
> >>
> >> No. They aren't dual cpu's. If you want dual, you'll have to use the
> >> Opteron series 2xx.
> >
> > That seems kind of silly. The 939 is supposed to phase out the 940,
which
> > is dual, but the 939 isn't?
> >
> The 939 isn't slated to phase out the 940. It was intended to phase out
> socket 754.
>
> Before 939, you had to use a 940 for dual channel memory, and the memory
> had to be registered. 939, is just an addition to the line using dual
> channel unregistered memory This brings the standard Athlon 64, which was
> single channel, into the dual channel arena also. Just a little confusing,
> but there will be 3 boards for a while. 754, 939, and 940. I would expect
> 754 to byte the dust first, but you never know in this game.
>
>
> --
> Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
> http://mysite.verizon.net/res0exft/cpu.htm

Archived from groups: alt.comp.hardware.overclocking.amd (More info?)

"AJ" <ajhatespammers@diespammersdie.com> wrote in message
news:7-SdnaEngJEOt1LdRVn-uA@comcast.com...
> Lurking for a while and thought I'd ask the question now that the 939 FX's
> are out. Does anyone know if a manufacturer has released or plans to
> release a dual socket board for those CPU's?
>
If you want some 64bit dual action check out Iwill's board.
http://www.iwill.net/whats_new/SeeMore.asp?vID=84
Looks like it will be for Amd 64 what the Mpx2 was for Amd 32.
I have the Mpx2 and it was the ultimate in mainboards for Amd 32 Smp.


---
Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.708 / Virus Database: 464 - Release Date: 18/06/2004

Archived from groups: alt.comp.hardware.overclocking.amd (More info?)

boostland wrote:
> "AJ" <ajhatespammers@diespammersdie.com> wrote in message
> news:7-SdnaEngJEOt1LdRVn-uA@comcast.com...
>> Lurking for a while and thought I'd ask the question now that the
>> 939 FX's are out. Does anyone know if a manufacturer has released
>> or plans to release a dual socket board for those CPU's?
>
> If you want some 64bit dual action check out Iwill's board.
> http://www.iwill.net/whats_new/SeeMore.asp?vID=84
> Looks like it will be for Amd 64 what the Mpx2 was for Amd 32.
> I have the Mpx2 and it was the ultimate in mainboards for Amd 32 Smp.

I don't think dual A64's will take off to anywhere near the same extent as
dual A32's, solely for the reason of price. Here's a comparison of dual
modded 2500's vs dual 240's ($NZ prices):

Dual A32 (pretty much my current setup):
CPUs 145 x2 (XP2500 x2)
HSF 50 x2 (TT Silent Boost x2)
M/B 430 (Gigabyte or MSI)
Ram 140 (1x512mb generic PC2700 non-reg)
------------
Total 960

Dual A64:
CPUs 420 x2 (240 x2)
HSF 50 x2 (TT Silent Boost K8 x2)
M/B 580 (Tyan S2875ANRF, only board under $1k)
Ram 160 x2 (2x256mb generic PC3200 ecc reg)
------------
Total 2420

For a rough comparison, wages are about the same in NZ and the US without
taking into account the exchange rate (ie: a $NZ30k/yr job here is a
$US30k/yr job in the US).

For a board that has dimms for each CPU, you're looking at another ~400 on
the board cost, plus another 320 for extra RAM (add anouther 140 to the A32
setup to bring it to 1gb as well). So you'll end up with ~2940 for the dual
240's, vs ~1100 for the dual A32 system. Certainly, the A64 system is more
powerful, but not so much more powerful that it warrents costing nearly
three times as much. Using the rough price-halves-every-18-months rule, a
*dual 240* system isn't going to get to my definition of affordable (using
the A32 point as being barely affordable) for another 2 years, ie: mid 2006,
or about when the K9's start hitting the shelves.

To become even a vaugely viable option for me, the CPU costs have to come
down a lot. The cost of the two 240s alone is about the same as the cost of
the entire dual A32 upgrade ...

--
Michael Brown
www.emboss.co.nz : OOS/RSI software and more
Add michael@ to emboss.co.nz - My inbox is always open

Archived from groups: alt.comp.hardware.overclocking.amd (More info?)

> I don't think dual A64's will take off to anywhere near the same extent as
> dual A32's, solely for the reason of price. Here's a comparison of dual
> modded 2500's vs dual 240's ($NZ prices):

A64!=Opteron

If the 939 platform is someday available for dual A64 it will kick ass

Archived from groups: alt.comp.hardware.overclocking.amd (More info?)

Josef Varela wrote:
>> I don't think dual A64's will take off to anywhere near the same
>> extent as dual A32's, solely for the reason of price. Here's a
>> comparison of dual modded 2500's vs dual 240's ($NZ prices):
>
> A64!=Opteron
>
> If the 939 platform is someday available for dual A64 it will kick ass.

A64 == Opteron if you want anything but uniprocessor. Essentially the same
as with P4's and Xeons. And, to a lesser extent, XP's and MPs. Straight XP's
are incapable of doing SMP, but the disabled circuitry can be re-enabled by
rejoining the 4th L5. Only Opterons have the necessary HT links (enabled) to
do SMP, and that's the way it's staying, unless someone figures out how to
re-enable the links (very unlikely). Originally, there were dual-CPU capable
Clawhammers roadmapped in "below" the Opteron 2xx series, but these were
dropped before the first x86 CPU was ever released.

--
Michael Brown
www.emboss.co.nz : OOS/RSI software and more
Add michael@ to emboss.co.nz - My inbox is always open

Archived from groups: alt.comp.hardware.overclocking.amd (More info?)

>> A64!=Opteron
>>
>> If the 939 platform is someday available for dual A64 it will kick ass.
>
> A64 == Opteron if you want anything but uniprocessor. Essentially the same
> as with P4's and Xeons. And, to a lesser extent, XP's and MPs.

Even Celerons can do MP, it´s all about having a mainboard that supports
it...
My old BP6 with bx chipset and 2x533 Celeron was the best bord I ever had.
The VP6 can do MP new Celerons

Archived from groups: alt.comp.hardware.overclocking.amd (More info?)

Josef Varela wrote:
>>> A64!=Opteron
>>>
>>> If the 939 platform is someday available for dual A64 it will kick
>>> ass.
>>
>> A64 == Opteron if you want anything but uniprocessor. Essentially
>> the same as with P4's and Xeons. And, to a lesser extent, XP's and
>> MPs.
>
> Even Celerons can do MP, it´s all about having a mainboard that
> supports it...

Sort of ... The only Celerons that could do SMP are the Mendocino Celerons.
Try dropping a couple of Coppermine Celerons into a board. They won't do
SMP. SOME celerons can do MP, but not all. Neither the Coppermine or
Tualatin Celerons can do SMP.

Back in the Good Old Days, SMP used to be entirely dependant on the board.
This was because 486's and the like had no SMP capability build in (save for
the #LOCK pin). All the SMP stuff was done in chipsets. ServerWorks or
whoever would spends a few million dollars designing a chipset that could
run multiple 486's, and then sell it in small quantities at large prices.
This all changed with the Pentium (IIRC), which actually had some built-in
support for SMP configurations. The main part was the APIC, which suported
interprocessor interrupts, among other things. So at this point, the chipset
makers could do away with a lot of the complex stuff that used to be in the
chipset, and use the on-CPU features instead. This then gave Intel/AMD the
capability to produce SMP capable or incapable CPUs, as if these new bits
weren't enabled, then the CPU could not