-Need help on a new computer...

Eric

Distinguished
Dec 31, 2007
1,373
0
19,280
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

-Hi there, I'm hoping some might be kind enough to help me out as I
know nothing about computers anymore and would rather not do 20+ hours
of research...

What I do:
Surf and play poker on-line: Not power intensive.

What I would like to do:
Play newer shoot'em up games: Very power intensive.

While I like to play games I care far more about the game play and
less about the graphics. I want to play on line against others so I
need it to be smooth but I don't care about the graphics that have to
do with ambience like reflections and how light and shadow is cast. I
would just like it to be smooth, not choppy and no lagging or bogging
down.

I would like suggestions for components that will work well together.
I definitely want AMD and can't afford top of the line so would like
to have a motherboard that will likely be able to take a good upgrade
cpu in 9-12 months.

Motherboard?
CPU?
RAM?
HD?
Video Card?

I can put it together myself and can overclock and do minor things
like that.

any other thoughts or suggestions?

thank you very much
Eric
 

keith

Distinguished
Mar 30, 2004
1,335
0
19,280
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On Thu, 23 Dec 2004 15:56:08 +0000, Eric wrote:

<snip>

> I would like suggestions for components that will work well together.
> I definitely want AMD and can't afford top of the line so would like

The all important question... What's the budget?

> to have a motherboard that will likely be able to take a good upgrade
> cpu in 9-12 months.

Forget about upgrades. Things change too fast to bother. Besides
motherboards in the arena you seem to want to play are cheap.

--
Keith
 
G

Guest

Guest
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

"Eric" <111@telus.net> wrote in message
news:chkls0luui765geu8693o9uorcp8fvo198@4ax.com...

" I definitely want AMD and can't afford top of the line so would like to
have a motherboard that will likely be able to take a good upgrade cpu in
9-12 months. I can put it together myself and can overclock and do minor
things like that. "


- DFI LANParty UT nF3 250Gb http://tinyurl.com/4eduv
- AMD Sempron 3100+ skt754 http://tinyurl.com/5nyt7
- Kingmax 2x 512MB PC3700 CL2.5 http://tinyurl.com/3km4g
- Seagate 200GB 7200RPM 8MB SATA http://tinyurl.com/42mas
- Leadtek GeForce 6600GT 128MB http://tinyurl.com/5cyko
 
G

Guest

Guest
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On Thu, 23 Dec 2004 15:56:08 GMT, Eric <111@telus.net> wrote:

>-Hi there, I'm hoping some might be kind enough to help me out as I
>know nothing about computers anymore and would rather not do 20+ hours
>of research...
>
>What I do:
>Surf and play poker on-line: Not power intensive.
>
>What I would like to do:
>Play newer shoot'em up games: Very power intensive.
>
>While I like to play games I care far more about the game play and
>less about the graphics. I want to play on line against others so I
>need it to be smooth but I don't care about the graphics that have to
>do with ambience like reflections and how light and shadow is cast. I
>would just like it to be smooth, not choppy and no lagging or bogging
>down.
>
>I would like suggestions for components that will work well together.
>I definitely want AMD and can't afford top of the line so would like
>to have a motherboard that will likely be able to take a good upgrade
>cpu in 9-12 months.

Uhh.. good luck on that one. It's VERY difficult to buy a motherboard
today expecting a good CPU upgrade more than about 6-months down the
line. That being said, a Socket 939 Athlon64 is probably your best
bet here.

>Motherboard?

As mentioned above, Socket 939 Athlon64 board. Ideally I would
recommend an nForce4 chipset, though unfortunately these are still a
bit few and far between. I see on www.newegg.com that they offer a
Chaintech board, the VNF4/Ultra, using the nForce4 Ultra chipset and
selling for a decent price ($129). They also have an Asus A8N-SLI
board using the nForce4 SLI chipset (allows for multiple video cards
to be used together, really not worth much of anything IMO unless
you've got more money than brains). Unfortunately that board sells
for a somewhat less than decent price of $269.

While I normally try to stick to one of the big name motherboard
manufacturers (mostly MSI), I did recently purchase a dirt-cheap
Chaintech board for my system the last time I needed a new board and I
was very pleasantly surprised. I actually also now have a Chaintech
sound card and video card, both of which work quite well, so I'm at
least semi-confident in tossing in a recommendation for that Chaintech
motherboard.

>CPU?

Athlon64 in socket 939. Which one you get depends on your budget.
Newegg sells an OEM Athlon64 3000+ for only $165, or you could jump up
to a retail box 3200+ for $240. The 3500+ is listed at both $274 and
$369 at the moment, and I'm guessing that one of those is a mistake.
Anything higher than that is likely to rather quickly reduce your
bang/buck ratio.

>RAM?

The CPU and motherboard combo require dual-channel memory, so you'll
need 2 of whatever you get. Probably either 2 x 256MB or 2 x 512MB,
depending on your budget. Either way it'll be DDR400 / PC3200 memory,
and ideally from a list of known-good memory for your board.
Chaintech has a link for recommended memory for the above-mentioned
motherboard, but at the moment it's blank. You might want keep an eye
on the spec page though:

http://www.chaintechusa.com/tw/eng/product_spec.asp?MPSNo=13&PISNo=318

The list for the Asus board is available at this following link:

http://usa.asus.com/products/mb/qvl/A8N-QVL.pdf


Newegg lists 2 x 256MB of Kingston ValueSelect memory (on Asus' list)
for $73, or 2 x 512MB of similar stuff for $139.

>HD?

See www.storagereview.com for the latest and greatest info on hard
drives and the like. Personally though I would probably opt for a
Seagate SATA 120GB hard drive with 8MB of cache, which Newegg has
listed at $100. Of course, if you need a larger (or smaller) hard
drive that is also available. Other manufacturers are fine as well,
though I like the low-noise aspect of Seagate drives and they seem to
be doing fairly well on the reliability front these days (though hard
drives die on a fairly regular basis regardless of who makes them, so
always back up your important data!).

>Video Card?

If you can afford one, a PCI-Express card using the nVidia GeForce
6600GT chipset seems to be the best bet for good price/performance in
the mid-range of things. They start at $186 for a Gigabyte or XFX
card. A slightly cheaper option is the non-GT version of the GeForce
6600 which starts at $119. Word of warning on that one though, some
of the 6600 cards are apparently using 64-bit memory buses, which will
rather dramatically reduce your performance.

>I can put it together myself and can overclock and do minor things
>like that.
>
>any other thoughts or suggestions?

With all the above, I would probably put together something like the
following:

Chaintech VNF4/Ultra motherboard $129
AMD Athlon64 3500+ Retail $274
2 x 512MB Kingston PC3200 memory $139
Seagate SATA 120GB hard drive $100
XFX GeForce 6600GT vid card $186

Total cost for this stuff: $826. Add in a DVD-RW drive (~$50 or $60),
a decent case + power supply (something like an Antec Sonata for $100)
and you should have a very good system coming in at right around the
$1000 mark. Of course, you can make it cheaper to suit your budget if
need be.

-------------
Tony Hill
hilla <underscore> 20 <at> yahoo <dot> ca
 
G

Guest

Guest
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On Tue, 28 Dec 2004 02:40:13 -0500, Tony Hill <hilla_nospam_20@yahoo.ca>
wrote:

>On Thu, 23 Dec 2004 15:56:08 GMT, Eric <111@telus.net> wrote:
>
>>-Hi there, I'm hoping some might be kind enough to help me out as I
>>know nothing about computers anymore and would rather not do 20+ hours
>>of research...
>>
>>What I do:
>>Surf and play poker on-line: Not power intensive.
>>
>>What I would like to do:
>>Play newer shoot'em up games: Very power intensive.
>>
>>While I like to play games I care far more about the game play and
>>less about the graphics. I want to play on line against others so I
>>need it to be smooth but I don't care about the graphics that have to
>>do with ambience like reflections and how light and shadow is cast. I
>>would just like it to be smooth, not choppy and no lagging or bogging
>>down.
>>
>>I would like suggestions for components that will work well together.
>>I definitely want AMD and can't afford top of the line so would like
>>to have a motherboard that will likely be able to take a good upgrade
>>cpu in 9-12 months.
>
>Uhh.. good luck on that one. It's VERY difficult to buy a motherboard
>today expecting a good CPU upgrade more than about 6-months down the
>line. That being said, a Socket 939 Athlon64 is probably your best
>bet here.
>
>>Motherboard?
>
>As mentioned above, Socket 939 Athlon64 board. Ideally I would
>recommend an nForce4 chipset, though unfortunately these are still a
>bit few and far between. I see on www.newegg.com that they offer a
>Chaintech board, the VNF4/Ultra, using the nForce4 Ultra chipset and
>selling for a decent price ($129). They also have an Asus A8N-SLI
>board using the nForce4 SLI chipset (allows for multiple video cards
>to be used together, really not worth much of anything IMO unless
>you've got more money than brains). Unfortunately that board sells
>for a somewhat less than decent price of $269.

They also have a Gigabyte nForce4 but like the other two you mention, it is
also out of stock. The five new MSI PCIe boards, 2 ATI and 3 nForce4
should be coming out soon too.

>While I normally try to stick to one of the big name motherboard
>manufacturers (mostly MSI), I did recently purchase a dirt-cheap
>Chaintech board for my system the last time I needed a new board and I
>was very pleasantly surprised. I actually also now have a Chaintech
>sound card and video card, both of which work quite well, so I'm at
>least semi-confident in tossing in a recommendation for that Chaintech
>motherboard.
>
>>CPU?
>
>Athlon64 in socket 939. Which one you get depends on your budget.
>Newegg sells an OEM Athlon64 3000+ for only $165, or you could jump up
>to a retail box 3200+ for $240. The 3500+ is listed at both $274 and
>$369 at the moment, and I'm guessing that one of those is a mistake.
>Anything higher than that is likely to rather quickly reduce your
>bang/buck ratio.

Nope, that's the current price premium over a 130nm for a 90nm 3500+, which
I got for $295. end of Nov.; the MSI K8N Neo2 Plat. has also gone up by
$13. and is also out of stock. Hey, I feel lucky.:)

Makes you wonder what is going on with the 90nm Athlon64s, especially the
3500+ which goes out of stock very quickly - maybe some big OEM gobbling
them up? I wonder who?:)

>>RAM?
>
>The CPU and motherboard combo require dual-channel memory, so you'll
>need 2 of whatever you get. Probably either 2 x 256MB or 2 x 512MB,
>depending on your budget. Either way it'll be DDR400 / PC3200 memory,
>and ideally from a list of known-good memory for your board.
>Chaintech has a link for recommended memory for the above-mentioned
>motherboard, but at the moment it's blank. You might want keep an eye
>on the spec page though:
>
>http://www.chaintechusa.com/tw/eng/product_spec.asp?MPSNo=13&PISNo=318
>
>The list for the Asus board is available at this following link:
>
>http://usa.asus.com/products/mb/qvl/A8N-QVL.pdf
>
>
>Newegg lists 2 x 256MB of Kingston ValueSelect memory (on Asus' list)
>for $73, or 2 x 512MB of similar stuff for $139.

With the MSI K8N Neo2 I have two 512MB Crucial PC3200 DIMMs w. 512Mb chips
designated at www.newegg.com as 8T in the part number. They will not run
at tighter timings than 3-3-3-8 but they *do* work at 1T command rate...
which seems to be the key to Sandra giving 5.8GB/s on buffered bandwidth.
Plus, if I need to add memory I won't get into the AMD spec'd limit of
DDR333 with >2 ranks of memory per channel.

>>HD?
>
>See www.storagereview.com for the latest and greatest info on hard
>drives and the like. Personally though I would probably opt for a
>Seagate SATA 120GB hard drive with 8MB of cache, which Newegg has
>listed at $100. Of course, if you need a larger (or smaller) hard
>drive that is also available. Other manufacturers are fine as well,
>though I like the low-noise aspect of Seagate drives and they seem to
>be doing fairly well on the reliability front these days (though hard
>drives die on a fairly regular basis regardless of who makes them, so
>always back up your important data!).

The new Seagates are not as quiet as the ones from a few months ago -
Seagate got held up for licensing on the head-positioning noise reduction
patents and just dropped the feature. They are still relatively quiet
though and still my choice.

>>Video Card?
>
>If you can afford one, a PCI-Express card using the nVidia GeForce
>6600GT chipset seems to be the best bet for good price/performance in
>the mid-range of things. They start at $186 for a Gigabyte or XFX
>card. A slightly cheaper option is the non-GT version of the GeForce
>6600 which starts at $119. Word of warning on that one though, some
>of the 6600 cards are apparently using 64-bit memory buses, which will
>rather dramatically reduce your performance.

Agree on the video card and note that the XFX 6600GT is a dual DVI model
and seems to come with two VGA adapters - nice flexibility. Since the AGP
8x 6600GTs seem to be going for ~$30. more than the equivalent PCIe
version, any extra money here is better spent for a PCIe mbrd.

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??
 
G

Guest

Guest
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

In article <5nj2t0tn90kkupmls2mqf63lsmcbu2o88d@4ax.com>, fammacd=!
SPAM^nothanks@tellurian.com says...

> On Tue, 28 Dec 2004 02:40:13 -0500, Tony Hill <hilla_nospam_20@yahoo.ca> wrote:

<snip>

> Nope, that's the current price premium over a 130nm for a 90nm 3500+, which
> I got for $295. end of Nov.; the MSI K8N Neo2 Plat. has also gone up by
> $13. and is also out of stock. Hey, I feel lucky.:)
>
> Makes you wonder what is going on with the 90nm Athlon64s, especially the
> 3500+ which goes out of stock very quickly - maybe some big OEM gobbling
> them up? I wonder who?:)

Interesting. If someone has to "gobble" them up to make production,
something's wrong somewhere. Inventory costs $$.

> With the MSI K8N Neo2 I have two 512MB Crucial PC3200 DIMMs w. 512Mb chips
> designated at www.newegg.com as 8T in the part number. They will not run
> at tighter timings than 3-3-3-8 but they *do* work at 1T command rate...
> which seems to be the key to Sandra giving 5.8GB/s on buffered bandwidth.
> Plus, if I need to add memory I won't get into the AMD spec'd limit of
> DDR333 with >2 ranks of memory per channel.

Hmm, Even registered? Santa found the snow drift my new sticks of
memory were in yesterday, but I haven't had time to plug them in.

> >See www.storagereview.com for the latest and greatest info on hard
> >drives and the like. Personally though I would probably opt for a
> >Seagate SATA 120GB hard drive with 8MB of cache, which Newegg has
> >listed at $100. Of course, if you need a larger (or smaller) hard
> >drive that is also available. Other manufacturers are fine as well,
> >though I like the low-noise aspect of Seagate drives and they seem to
> >be doing fairly well on the reliability front these days (though hard
> >drives die on a fairly regular basis regardless of who makes them, so
> >always back up your important data!).

Also note that Seagate has a five year warranty.

> The new Seagates are not as quiet as the ones from a few months ago -
> Seagate got held up for licensing on the head-positioning noise reduction
> patents and just dropped the feature. They are still relatively quiet
> though and still my choice.

I have one that's a few months old (perhaps July) and it's very quiet.
IIRC that patent was all about the UI interface into the
performance/noise tradeoff feature. I believe Seagate left the feature
in the microcode, but stopped shipping the UI.

> >>Video Card?
> >
> >If you can afford one, a PCI-Express card using the nVidia GeForce
> >6600GT chipset seems to be the best bet for good price/performance in
> >the mid-range of things. They start at $186 for a Gigabyte or XFX
> >card. A slightly cheaper option is the non-GT version of the GeForce
> >6600 which starts at $119. Word of warning on that one though, some
> >of the 6600 cards are apparently using 64-bit memory buses, which will
> >rather dramatically reduce your performance.
>
> Agree on the video card and note that the XFX 6600GT is a dual DVI model
> and seems to come with two VGA adapters - nice flexibility. Since the AGP
> 8x 6600GTs seem to be going for ~$30. more than the equivalent PCIe
> version, any extra money here is better spent for a PCIe mbrd.

Yikes! I guess AGP is dead. Apparently it died in its sleep.

--
Keith
 
G

Guest

Guest
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On Tue, 28 Dec 2004 10:04:21 -0500, Keith R. Williams <krw@att.bizzzz> wrote:

>In article <5nj2t0tn90kkupmls2mqf63lsmcbu2o88d@4ax.com>, fammacd=!
>SPAM^nothanks@tellurian.com says...
[snippo]
>> Agree on the video card and note that the XFX 6600GT is a dual DVI model
>> and seems to come with two VGA adapters - nice flexibility. Since the AGP
>> 8x 6600GTs seem to be going for ~$30. more than the equivalent PCIe
>> version, any extra money here is better spent for a PCIe mbrd.
>
>Yikes! I guess AGP is dead. Apparently it died in its sleep.

Puhlease. It knew the end was near for months...

;-)
 

keith

Distinguished
Mar 30, 2004
1,335
0
19,280
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On Tue, 28 Dec 2004 18:21:32 +0000, daytripper wrote:

> On Tue, 28 Dec 2004 10:04:21 -0500, Keith R. Williams <krw@att.bizzzz> wrote:
>
>>In article <5nj2t0tn90kkupmls2mqf63lsmcbu2o88d@4ax.com>, fammacd=!
>>SPAM^nothanks@tellurian.com says...
> [snippo]
>>> Agree on the video card and note that the XFX 6600GT is a dual DVI model
>>> and seems to come with two VGA adapters - nice flexibility. Since the AGP
>>> 8x 6600GTs seem to be going for ~$30. more than the equivalent PCIe
>>> version, any extra money here is better spent for a PCIe mbrd.
>>
>>Yikes! I guess AGP is dead. Apparently it died in its sleep.
>
> Puhlease. It knew the end was near for months...
>
> ;-)

When was the last time you saw a technology disappear this fast? Ok, ok,
ok! DRDRAM. Interesting both were born of the same father at about the
same time (with the cacheless Celeron).

My point though is that there are a *lot* of AGP slots out there, and
damned few PCI-E. Since I don't have one of the latter, it does sorta
concern me. :-(

--
Keith
 
G

Guest

Guest
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On Tue, 28 Dec 2004 10:04:21 -0500, Keith R. Williams <krw@att.bizzzz>
wrote:

>In article <5nj2t0tn90kkupmls2mqf63lsmcbu2o88d@4ax.com>, fammacd=!
>SPAM^nothanks@tellurian.com says...
>
>> On Tue, 28 Dec 2004 02:40:13 -0500, Tony Hill <hilla_nospam_20@yahoo.ca> wrote:
>
><snip>
>
>> Nope, that's the current price premium over a 130nm for a 90nm 3500+, which
>> I got for $295. end of Nov.; the MSI K8N Neo2 Plat. has also gone up by
>> $13. and is also out of stock. Hey, I feel lucky.:)
>>
>> Makes you wonder what is going on with the 90nm Athlon64s, especially the
>> 3500+ which goes out of stock very quickly - maybe some big OEM gobbling
>> them up? I wonder who?:)
>
>Interesting. If someone has to "gobble" them up to make production,
>something's wrong somewhere. Inventory costs $$.

Ramp-up for roll-out?? I've seen AMD docs which say those are made in
Dresden but I though they didn't have any 90nm there yet?? It had occurred
to me that if they came from East Fishkill, maybe AMD had hit the limit on
their wafer "ration" there... for whatever reason.

>> With the MSI K8N Neo2 I have two 512MB Crucial PC3200 DIMMs w. 512Mb chips
>> designated at www.newegg.com as 8T in the part number. They will not run
>> at tighter timings than 3-3-3-8 but they *do* work at 1T command rate...
>> which seems to be the key to Sandra giving 5.8GB/s on buffered bandwidth.
>> Plus, if I need to add memory I won't get into the AMD spec'd limit of
>> DDR333 with >2 ranks of memory per channel.
>
>Hmm, Even registered? Santa found the snow drift my new sticks of
>memory were in yesterday, but I haven't had time to plug them in.

I've only looked at Athlon64 Skt939 details which are quite different from
Skt940: inverted address signals across channels?... weird. Does this mean
the mbrd designer has to re-invert at the DIMM socket for the
channel/socket? Well I was sure I'd seen an AMD doc mention this about 2
ranks per channel at 400DDR but can no longer find it - mbrd mfrs do
mention it though. Anyway it seems to be for Skt939 unbuffered operation
only... as is the 2T timing for commands.

>> >See www.storagereview.com for the latest and greatest info on hard
>> >drives and the like. Personally though I would probably opt for a
>> >Seagate SATA 120GB hard drive with 8MB of cache, which Newegg has
>> >listed at $100. Of course, if you need a larger (or smaller) hard
>> >drive that is also available. Other manufacturers are fine as well,
>> >though I like the low-noise aspect of Seagate drives and they seem to
>> >be doing fairly well on the reliability front these days (though hard
>> >drives die on a fairly regular basis regardless of who makes them, so
>> >always back up your important data!).
>
>Also note that Seagate has a five year warranty.
>
>> The new Seagates are not as quiet as the ones from a few months ago -
>> Seagate got held up for licensing on the head-positioning noise reduction
>> patents and just dropped the feature. They are still relatively quiet
>> though and still my choice.
>
>I have one that's a few months old (perhaps July) and it's very quiet.
>IIRC that patent was all about the UI interface into the
>performance/noise tradeoff feature. I believe Seagate left the feature
>in the microcode, but stopped shipping the UI.

The SATAs I've bought definitely have more head noise than the PATA models
- still not objectionable and the spindle is near-silent.

>> >>Video Card?
>> >
>> >If you can afford one, a PCI-Express card using the nVidia GeForce
>> >6600GT chipset seems to be the best bet for good price/performance in
>> >the mid-range of things. They start at $186 for a Gigabyte or XFX
>> >card. A slightly cheaper option is the non-GT version of the GeForce
>> >6600 which starts at $119. Word of warning on that one though, some
>> >of the 6600 cards are apparently using 64-bit memory buses, which will
>> >rather dramatically reduce your performance.
>>
>> Agree on the video card and note that the XFX 6600GT is a dual DVI model
>> and seems to come with two VGA adapters - nice flexibility. Since the AGP
>> 8x 6600GTs seem to be going for ~$30. more than the equivalent PCIe
>> version, any extra money here is better spent for a PCIe mbrd.
>
>Yikes! I guess AGP is dead. Apparently it died in its sleep.

Well at the top-end, like 6800GTs, which is more amenable to the gamer
upgrade market, are apparently widely available in AGP 8x. For the lower
and mid-range, yep, AGP is dying fast.

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??
 
G

Guest

Guest
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

In article <6fu3t09jt1ms31oooj7jmhtq6laiopjqi5@4ax.com>, fammacd=!
SPAM^nothanks@tellurian.com says...
> On Tue, 28 Dec 2004 10:04:21 -0500, Keith R. Williams <krw@att.bizzzz>
> wrote:
>
> >In article <5nj2t0tn90kkupmls2mqf63lsmcbu2o88d@4ax.com>, fammacd=!
> >SPAM^nothanks@tellurian.com says...
> >
> >> On Tue, 28 Dec 2004 02:40:13 -0500, Tony Hill <hilla_nospam_20@yahoo.ca> wrote:
> >
> ><snip>
> >
> >> Nope, that's the current price premium over a 130nm for a 90nm 3500+, which
> >> I got for $295. end of Nov.; the MSI K8N Neo2 Plat. has also gone up by
> >> $13. and is also out of stock. Hey, I feel lucky.:)
> >>
> >> Makes you wonder what is going on with the 90nm Athlon64s, especially the
> >> 3500+ which goes out of stock very quickly - maybe some big OEM gobbling
> >> them up? I wonder who?:)
> >
> >Interesting. If someone has to "gobble" them up to make production,
> >something's wrong somewhere. Inventory costs $$.
>
> Ramp-up for roll-out?? I've seen AMD docs which say those are made in
> Dresden but I though they didn't have any 90nm there yet?? It had occurred
> to me that if they came from East Fishkill, maybe AMD had hit the limit on
> their wafer "ration" there... for whatever reason.

Or their wafer ration is *zero*. I haven't any idea though. I make it
a point to stay away from "E. Fishkill" and anywhere or anything close.
I've been successful for the past eleven years. ;-)

> >> With the MSI K8N Neo2 I have two 512MB Crucial PC3200 DIMMs w. 512Mb chips
> >> designated at www.newegg.com as 8T in the part number. They will not run
> >> at tighter timings than 3-3-3-8 but they *do* work at 1T command rate...
> >> which seems to be the key to Sandra giving 5.8GB/s on buffered bandwidth.
> >> Plus, if I need to add memory I won't get into the AMD spec'd limit of
> >> DDR333 with >2 ranks of memory per channel.
> >
> >Hmm, Even registered? Santa found the snow drift my new sticks of
> >memory were in yesterday, but I haven't had time to plug them in.
>
> I've only looked at Athlon64 Skt939 details which are quite different from
> Skt940: inverted address signals across channels?... weird. Does this mean
> the mbrd designer has to re-invert at the DIMM socket for the
> channel/socket?

Who cares if the addresses are inverted? It simply selects a different
part of the array. They are rectangular and a power of two long.
Inverting the addresses on one side reduces the current in the rails
(thus switching current), since both sides are pumping out the "same"
address; half are now '1's and half '0's.


> Well I was sure I'd seen an AMD doc mention this about 2
> ranks per channel at 400DDR but can no longer find it - mbrd mfrs do
> mention it though. Anyway it seems to be for Skt939 unbuffered operation
> only... as is the 2T timing for commands.

I put the memory in last night and it seems to work, though I haven't
done anything to stress it. The new DIMMs are dual rank, so I'm
running three ranks on each channel. It is socket 940 so it had
*better* work.

<snip>

> >> The new Seagates are not as quiet as the ones from a few months ago -
> >> Seagate got held up for licensing on the head-positioning noise reduction
> >> patents and just dropped the feature. They are still relatively quiet
> >> though and still my choice.
> >
> >I have one that's a few months old (perhaps July) and it's very quiet.
> >IIRC that patent was all about the UI interface into the
> >performance/noise tradeoff feature. I believe Seagate left the feature
> >in the microcode, but stopped shipping the UI.
>
> The SATAs I've bought definitely have more head noise than the PATA models
> - still not objectionable and the spindle is near-silent.

I could only hear the hears move when I was backing up my old machine
to the SATA drive on the new (it's empty and useless on the new system,
so why not?) before reinstalling WinBlows on the old. At least I saved
the data. What a PITA, even so.

You're right though the spindle is silent. There is no whine at all,
just a little clicking when the heads move.

> >> >>Video Card?
> >> >
> >> >If you can afford one, a PCI-Express card using the nVidia GeForce
> >> >6600GT chipset seems to be the best bet for good price/performance in
> >> >the mid-range of things. They start at $186 for a Gigabyte or XFX
> >> >card. A slightly cheaper option is the non-GT version of the GeForce
> >> >6600 which starts at $119. Word of warning on that one though, some
> >> >of the 6600 cards are apparently using 64-bit memory buses, which will
> >> >rather dramatically reduce your performance.
> >>
> >> Agree on the video card and note that the XFX 6600GT is a dual DVI model
> >> and seems to come with two VGA adapters - nice flexibility. Since the AGP
> >> 8x 6600GTs seem to be going for ~$30. more than the equivalent PCIe
> >> version, any extra money here is better spent for a PCIe mbrd.
> >
> >Yikes! I guess AGP is dead. Apparently it died in its sleep.
>
> Well at the top-end, like 6800GTs, which is more amenable to the gamer
> upgrade market, are apparently widely available in AGP 8x. For the lower
> and mid-range, yep, AGP is dying fast.

Hmm, I guess I'm going to have to budget for a graphics card. Maybe
last years off eBay...

--
Keith
 
G

Guest

Guest
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On Wed, 29 Dec 2004 07:37:25 -0500, George Macdonald
<fammacd=!SPAM^nothanks@tellurian.com> wrote:

>On Tue, 28 Dec 2004 10:04:21 -0500, Keith R. Williams <krw@att.bizzzz>
>wrote:
>
>>Interesting. If someone has to "gobble" them up to make production,
>>something's wrong somewhere. Inventory costs $$.
>
>Ramp-up for roll-out?? I've seen AMD docs which say those are made in
>Dresden but I though they didn't have any 90nm there yet??

Dresden does indeed have 90nm production up and running and that is
exactly where all of the Athlon64, Opteron and Sempr0n chips are
coming from. I don't know what percentage of wafer starts are 90nm
vs. 130nm, though we may get some hints in a few weeks when AMD
announces their end-of-year results.

> It had occurred
>to me that if they came from East Fishkill, maybe AMD had hit the limit on
>their wafer "ration" there... for whatever reason.

To the best of my knowledge there is currently no volume production of
any Athlon64, Opteron or Sempr0n chips at E. Fishkill. Maybe a few
test wafers here and there, but I think for the time being it's mainly
just an R&D setup and maybe a backup plan if they have any major
troubles with 90nm production at Dresden.

My understanding of the deal between IBM and AMD was that it was
mainly going to be for future 65nm production.

>>> Agree on the video card and note that the XFX 6600GT is a dual DVI model
>>> and seems to come with two VGA adapters - nice flexibility. Since the AGP
>>> 8x 6600GTs seem to be going for ~$30. more than the equivalent PCIe
>>> version, any extra money here is better spent for a PCIe mbrd.
>>
>>Yikes! I guess AGP is dead. Apparently it died in its sleep.
>
>Well at the top-end, like 6800GTs, which is more amenable to the gamer
>upgrade market, are apparently widely available in AGP 8x. For the lower
>and mid-range, yep, AGP is dying fast.

I'm not sure that it's dead just yet, the 6600GT seems to be a bit
more the exception rather than the norm. For whatever reason it seems
to be in VERY tight supply for AGP, while PCI-Express is much easier
to find and cheaper. Supply and demand perhaps?

AGP is definitely on the way out though, all the new chipsets seem to
be PCI-Express native chips that require a bridge to hang off of AGP.
It definitely seems prudent to look into a board with PCI-Express if
buying a new system at this time, though sadly availability of
PCI-Express boards for Athlon64 chips are still VERY tough to come by.

-------------
Tony Hill
hilla <underscore> 20 <at> yahoo <dot> ca
 
G

Guest

Guest
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On Wed, 29 Dec 2004 08:58:42 -0500, Keith R. Williams <krw@att.bizzzz>
wrote:

>In article <6fu3t09jt1ms31oooj7jmhtq6laiopjqi5@4ax.com>, fammacd=!
>SPAM^nothanks@tellurian.com says...

>> Ramp-up for roll-out?? I've seen AMD docs which say those are made in
>> Dresden but I though they didn't have any 90nm there yet?? It had occurred
>> to me that if they came from East Fishkill, maybe AMD had hit the limit on
>> their wafer "ration" there... for whatever reason.
>
>Or their wafer ration is *zero*. I haven't any idea though. I make it
>a point to stay away from "E. Fishkill" and anywhere or anything close.
>I've been successful for the past eleven years. ;-)

Apparently I was mistaken about Dresden and 90nm - they were scheduled for
90nm production there 3Q04. After a brief bump in November, at the retail
level, the 90nm chips have gone quite scarce... just wondering why. It's
hard to separate the facts from all the analyst babble.

>> I've only looked at Athlon64 Skt939 details which are quite different from
>> Skt940: inverted address signals across channels?... weird. Does this mean
>> the mbrd designer has to re-invert at the DIMM socket for the
>> channel/socket?
>
>Who cares if the addresses are inverted? It simply selects a different
>part of the array. They are rectangular and a power of two long.
>Inverting the addresses on one side reduces the current in the rails
>(thus switching current), since both sides are pumping out the "same"
>address; half are now '1's and half '0's.

Ah OK - so the memory on the "inverted" sockets is just addressed in
reverse - top down if you like.

>> The SATAs I've bought definitely have more head noise than the PATA models
>> - still not objectionable and the spindle is near-silent.
>
>I could only hear the hears move when I was backing up my old machine
>to the SATA drive on the new (it's empty and useless on the new system,
>so why not?) before reinstalling WinBlows on the old. At least I saved
>the data. What a PITA, even so.

I used BootitNG, booted off a floppy, to just copy my partitions over from
the old PATA drive to the new SATA - worked like a charm.

>> Well at the top-end, like 6800GTs, which is more amenable to the gamer
>> upgrade market, are apparently widely available in AGP 8x. For the lower
>> and mid-range, yep, AGP is dying fast.
>
>Hmm, I guess I'm going to have to budget for a graphics card. Maybe
>last years off eBay...

Yeah well I'm not too happy with my recently purchased ATI 9600XT - came
with the memory clocked at DDR500 instead of DDR600 so it performs like the
lower rated 9600 Pro. I'm not too happy with GeCube on this - they're
listed as Info-Tek at NewEgg BTW... so maybe I'll be looking at a 6600GT
for my next present. The extra cost on the AGP cards may be due to the
extra power connector - ugly that having to plug a drive power connector
into a video card.

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??
 

keith

Distinguished
Mar 30, 2004
1,335
0
19,280
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On Wed, 29 Dec 2004 20:13:17 -0500, George Macdonald wrote:

> On Wed, 29 Dec 2004 08:58:42 -0500, Keith R. Williams <krw@att.bizzzz>
> wrote:
>
>>In article <6fu3t09jt1ms31oooj7jmhtq6laiopjqi5@4ax.com>, fammacd=!
>>SPAM^nothanks@tellurian.com says...
>
>>> Ramp-up for roll-out?? I've seen AMD docs which say those are made in
>>> Dresden but I though they didn't have any 90nm there yet?? It had occurred
>>> to me that if they came from East Fishkill, maybe AMD had hit the limit on
>>> their wafer "ration" there... for whatever reason.
>>
>>Or their wafer ration is *zero*. I haven't any idea though. I make it
>>a point to stay away from "E. Fishkill" and anywhere or anything close.
>>I've been successful for the past eleven years. ;-)
>
> Apparently I was mistaken about Dresden and 90nm - they were scheduled for
> 90nm production there 3Q04. After a brief bump in November, at the retail
> level, the 90nm chips have gone quite scarce... just wondering why. It's
> hard to separate the facts from all the analyst babble.

I read analyst's babble for entertainment at lunch. This makes me really
scared to venture into markets I don't understand. ...and don't. Reading
this one has done my wife's IRA rather well. She's made about 3X over
the last five years, which isn't bad considering which years these were.
She would have done better had she listened to me a little more on AMD
(here mot was made) or a little less on AAPL (where I had her bail
premturely, though still doubled her money). The point is that I sorta
know what's going on here, and laugh at the "analysts". What would happen
if I trusted them (as in another market I have no clue about)?

>>> aw>
>>> I've only looked at Athlon64 Skt939 details which are quite different
>>> from Skt940: inverted address signals across channels?... weird. Does
>>> this mean the mbrd designer has to re-invert at the DIMM socket for
>>> the channel/socket?
>>
>>Who cares if the addresses are inverted? It simply selects a different
>>part of the array. They are rectangular and a power of two long.
>>Inverting the addresses on one side reduces the current in the rails
>>(thus switching current), since both sides are pumping out the "same"
>>address; half are now '1's and half '0's.
>
> Ah OK - so the memory on the "inverted" sockets is just addressed in
> reverse - top down if you like.

Sure. Scramble the address lines (as long as the CAS/Banks/... are all
straight) and see who cares. I hadn't considered the inversions before
you mentioned it, but it makes perfect sense from a circuit standpoint.
We go to great lengths to balance 0->1 and 1->0 transistions to minimize
noise. This is a freebie.


>>> The SATAs I've bought definitely have more head noise than the PATA
>>> models - still not objectionable and the spindle is near-silent.
>>
>>I could only hear the hears move when I was backing up my old machine to
>>the SATA drive on the new (it's empty and useless on the new system, so
>>why not?) before reinstalling WinBlows on the old. At least I saved the
>>data. What a PITA, even so.
>
> I used BootitNG, booted off a floppy, to just copy my partitions over
> from the old PATA drive to the new SATA - worked like a charm.

I just copied the data over the network to the empty drive, reinstalled
(still don't know what happened) and coppied it all back. It worked quite
well, though took a few hours.

I thought my work laptop died yesterday and was in a panic. The thing was
unciontrollable with windows flickering all over. The backups at
work don't. Becasue of some assinine restrictions (they only backup "my
data" on the root drive) I didn'thave a decent backup. I had the thing
at home over the week (I was off all week), but didn't bother. I could
have easily backed it up. Damn! Well, the problem was that my keyboard
tray pushed up against the platform the laptop was on and was
intermittently pressing the ESC key. *whew*.

The PC police (the other kind) now want to replace my A21p with a new
machine (likely an R50). I'm likely in for some kicking and screaming.

>>> Well at the top-end, like 6800GTs, which is more amenable to the gamer
>>> upgrade market, are apparently widely available in AGP 8x. For the
>>> lower and mid-range, yep, AGP is dying fast.
>>
>>Hmm, I guess I'm going to have to budget for a graphics card. Maybe
>>last years off eBay...
>
> Yeah well I'm not too happy with my recently purchased ATI 9600XT - came
> with the memory clocked at DDR500 instead of DDR600 so it performs like
> the lower rated 9600 Pro. I'm not too happy with GeCube on this -
> they're listed as Info-Tek at NewEgg BTW... so maybe I'll be looking at
> a 6600GT for my next present. The extra cost on the AGP cards may be
> due to the extra power connector - ugly that having to plug a drive
> power connector into a video card.

I thought the same about plugging a drive connector into my motherboard.
"they can't be serious?* *UUUGGLLY*!

I was thinking abotu *something* in case I want to do 3D. I have no
reason to dump my G550, but if AGP is dead, I'd like to cover my bases.

--
Keith
 
G

Guest

Guest
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On Wed, 29 Dec 2004 19:47:29 -0500, Tony Hill <hilla_nospam_20@yahoo.ca>
wrote:

>On Wed, 29 Dec 2004 07:37:25 -0500, George Macdonald
><fammacd=!SPAM^nothanks@tellurian.com> wrote:
>
>>On Tue, 28 Dec 2004 10:04:21 -0500, Keith R. Williams <krw@att.bizzzz>
>>wrote:
>>
>>>Interesting. If someone has to "gobble" them up to make production,
>>>something's wrong somewhere. Inventory costs $$.
>>
>>Ramp-up for roll-out?? I've seen AMD docs which say those are made in
>>Dresden but I though they didn't have any 90nm there yet??
>
>Dresden does indeed have 90nm production up and running and that is
>exactly where all of the Athlon64, Opteron and Sempr0n chips are
>coming from. I don't know what percentage of wafer starts are 90nm
>vs. 130nm, though we may get some hints in a few weeks when AMD
>announces their end-of-year results.

"All"??... Where??:) The 90nm Athlon64s are drying up - since they first
appeared, availability has been spotty at best, I've seem nothing >3500+
and you saw the price premium at NewEgg... and it's just gone out of stock
again. When I got mine end Nov, it was out of stock within hours of my
order.

As mentioned elsewhere, I realized I was wrong in thinking 90nm wasn't
being done in Dresden.

>> It had occurred
>>to me that if they came from East Fishkill, maybe AMD had hit the limit on
>>their wafer "ration" there... for whatever reason.
>
>To the best of my knowledge there is currently no volume production of
>any Athlon64, Opteron or Sempr0n chips at E. Fishkill. Maybe a few
>test wafers here and there, but I think for the time being it's mainly
>just an R&D setup and maybe a backup plan if they have any major
>troubles with 90nm production at Dresden.
>
>My understanding of the deal between IBM and AMD was that it was
>mainly going to be for future 65nm production.

I guess it depends on what deal you mean and whether the current one is
considered new or a continuation of the one which got 130nm working on SOI
~2 years ago. It was my impression that AMD had been working closely with
IBM ever since, continuing the joint effort on 90nm development, though
they have increased their presence at E. Fishkill during the past year and
formally extended the relationship.

>>>> Agree on the video card and note that the XFX 6600GT is a dual DVI model
>>>> and seems to come with two VGA adapters - nice flexibility. Since the AGP
>>>> 8x 6600GTs seem to be going for ~$30. more than the equivalent PCIe
>>>> version, any extra money here is better spent for a PCIe mbrd.
>>>
>>>Yikes! I guess AGP is dead. Apparently it died in its sleep.
>>
>>Well at the top-end, like 6800GTs, which is more amenable to the gamer
>>upgrade market, are apparently widely available in AGP 8x. For the lower
>>and mid-range, yep, AGP is dying fast.
>
>I'm not sure that it's dead just yet, the 6600GT seems to be a bit
>more the exception rather than the norm. For whatever reason it seems
>to be in VERY tight supply for AGP, while PCI-Express is much easier
>to find and cheaper. Supply and demand perhaps?

Doesn't look that way to me - all the new GPU card are predominantly
PCI-Express, apart from the very top end X800s and 6800s... in the
$400-500. range, which is always a strong upgrade gamer market. Try to
find an ATI X600 or X700 with AGP; until a couple of weeks ago, there was
only one 6600GT with AGP, from XFX, available at NewEgg and it was priced
at ~$50. more than the equivalent PCI-Express version. Some of the mfrs
have promised 6600s in AGP but I haven't seen one for sale yet.

>AGP is definitely on the way out though, all the new chipsets seem to
>be PCI-Express native chips that require a bridge to hang off of AGP.

Yep and the AGP 6600GTs need a drive connector for 12V power supply.

>It definitely seems prudent to look into a board with PCI-Express if
>buying a new system at this time, though sadly availability of
>PCI-Express boards for Athlon64 chips are still VERY tough to come by.

Yes that's what I've been telling people here but too late for me - I have
my K8N Neo2 PLat. MSI is showing 5 new Athon64 PCIe mbrds at their global
site - 2 ATI and 3 nForce4 - which have already been reviewed so... any day
now?:)

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??
 
G

Guest

Guest
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On Thu, 30 Dec 2004 01:36:23 -0500, George Macdonald
<fammacd=!SPAM^nothanks@tellurian.com> wrote:

>On Wed, 29 Dec 2004 19:47:29 -0500, Tony Hill <hilla_nospam_20@yahoo.ca>
>wrote:
>
>>Dresden does indeed have 90nm production up and running and that is
>>exactly where all of the Athlon64, Opteron and Sempr0n chips are
>>coming from. I don't know what percentage of wafer starts are 90nm
>>vs. 130nm, though we may get some hints in a few weeks when AMD
>>announces their end-of-year results.
>
>"All"??... Where??:)

Why, in Dresden of course! :>

> The 90nm Athlon64s are drying up - since they first
>appeared, availability has been spotty at best, I've seem nothing >3500+
>and you saw the price premium at NewEgg... and it's just gone out of stock
>again. When I got mine end Nov, it was out of stock within hours of my
>order.

Yeah, I don't know if that indicates that the 90nm ramp up is still
rather slow at Dresden and most of their output is still 130nm parts,
or if some big OEM (HPaq?) is grabbing all of the 90nm parts they can
get their hands on and no one else can get any stock. As best as I
can tell though, HPaq isn't selling any 90nm parts (or more
specifically, they aren't selling any Socket 939 parts), so the latter
doesn't seem likely.

>>To the best of my knowledge there is currently no volume production of
>>any Athlon64, Opteron or Sempr0n chips at E. Fishkill. Maybe a few
>>test wafers here and there, but I think for the time being it's mainly
>>just an R&D setup and maybe a backup plan if they have any major
>>troubles with 90nm production at Dresden.
>>
>>My understanding of the deal between IBM and AMD was that it was
>>mainly going to be for future 65nm production.
>
>I guess it depends on what deal you mean and whether the current one is
>considered new or a continuation of the one which got 130nm working on SOI
>~2 years ago. It was my impression that AMD had been working closely with
>IBM ever since, continuing the joint effort on 90nm development, though
>they have increased their presence at E. Fishkill during the past year and
>formally extended the relationship.

They have, but mostly it seems like an R&D venture more than an actual
"Have IBM make the chips" venture. I could be wrong though, I
certainly have no inside info on this one.

>>AGP is definitely on the way out though, all the new chipsets seem to
>>be PCI-Express native chips that require a bridge to hang off of AGP.
>
>Yep and the AGP 6600GTs need a drive connector for 12V power supply.

Am I the only one that doesn't like having a video chipset that
consumes so much power it needs a drive connector to power it? If
nothing else that is yet another extremely unreliable fan in the
system just waiting to fail the day after the (1 year) warranty
expires. It's not even like you can buy easily buy replacement fans
for most of these cards since they're all customer jobs.

>>It definitely seems prudent to look into a board with PCI-Express if
>>buying a new system at this time, though sadly availability of
>>PCI-Express boards for Athlon64 chips are still VERY tough to come by.
>
>Yes that's what I've been telling people here but too late for me - I have
>my K8N Neo2 PLat. MSI is showing 5 new Athon64 PCIe mbrds at their global
>site - 2 ATI and 3 nForce4 - which have already been reviewed so... any day
>now?:)

Like all exciting products I'm sure they'll be here Real Soon Now.

-------------
Tony Hill
hilla <underscore> 20 <at> yahoo <dot> ca
 
G

Guest

Guest
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On Fri, 31 Dec 2004 00:53:21 -0500, Tony Hill <hilla_nospam_20@yahoo.ca>
wrote:

>On Thu, 30 Dec 2004 01:36:23 -0500, George Macdonald
><fammacd=!SPAM^nothanks@tellurian.com> wrote:

>>I guess it depends on what deal you mean and whether the current one is
>>considered new or a continuation of the one which got 130nm working on SOI
>>~2 years ago. It was my impression that AMD had been working closely with
>>IBM ever since, continuing the joint effort on 90nm development, though
>>they have increased their presence at E. Fishkill during the past year and
>>formally extended the relationship.
>
>They have, but mostly it seems like an R&D venture more than an actual
>"Have IBM make the chips" venture. I could be wrong though, I
>certainly have no inside info on this one.

I probably read some analyst babble on this.

>>>AGP is definitely on the way out though, all the new chipsets seem to
>>>be PCI-Express native chips that require a bridge to hang off of AGP.
>>
>>Yep and the AGP 6600GTs need a drive connector for 12V power supply.
>
>Am I the only one that doesn't like having a video chipset that
>consumes so much power it needs a drive connector to power it? If
>nothing else that is yet another extremely unreliable fan in the
>system just waiting to fail the day after the (1 year) warranty
>expires. It's not even like you can buy easily buy replacement fans
>for most of these cards since they're all customer jobs.

Even the PCI-Express mbrds have the extra power but of course it's
integrated into the 24-pin power header on the mbrd - something else to
worry about in future: is the power supply really putting out more or did
they just add a 20-24-pin adapter somewhere?

Plugging a drive power connector into a video card, any card, is
ugly...ugly... ugly.

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??
 
G

Guest

Guest
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On Fri, 31 Dec 2004 13:21:27 -0500, George Macdonald
<fammacd=!SPAM^nothanks@tellurian.com> wrote:

>Even the PCI-Express mbrds have the extra power but of course it's
>integrated into the 24-pin power header on the mbrd - something else to
>worry about in future: is the power supply really putting out more or did
>they just add a 20-24-pin adapter somewhere?

Nope, they just added 4 more wires in parallel to the existing ones in
most cases :ppPP

--
L.Angel: I'm looking for web design work.
If you need basic to med complexity webpages at affordable rates, email me :)
Standard HTML, SHTML, MySQL + PHP or ASP, Javascript.
If you really want, FrontPage & DreamWeaver too.
But keep in mind you pay extra bandwidth for their bloated code
 
G

Guest

Guest
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On Fri, 31 Dec 2004 13:21:27 -0500, George Macdonald
<fammacd=!SPAM^nothanks@tellurian.com> wrote:

>On Fri, 31 Dec 2004 00:53:21 -0500, Tony Hill <hilla_nospam_20@yahoo.ca>
>wrote:
>>Am I the only one that doesn't like having a video chipset that
>>consumes so much power it needs a drive connector to power it? If
>>nothing else that is yet another extremely unreliable fan in the
>>system just waiting to fail the day after the (1 year) warranty
>>expires. It's not even like you can buy easily buy replacement fans
>>for most of these cards since they're all customer jobs.
>
>Even the PCI-Express mbrds have the extra power but of course it's
>integrated into the 24-pin power header on the mbrd - something else to
>worry about in future: is the power supply really putting out more or did
>they just add a 20-24-pin adapter somewhere?

Well, it doesn't make too much of a difference here, the power all
comes from the same place regardless of whether it's going through an
extra 4 pins on the main power header or a separate connector.

>Plugging a drive power connector into a video card, any card, is
>ugly...ugly... ugly.

Agreed. It doesn't really make a difference electrically speaking
where it gets the power from (so long as it can get a good, clean and
steady supply), but the simple idea of plugging a hard drive connector
into a card just seems somehow wrong.

-------------
Tony Hill
hilla <underscore> 20 <at> yahoo <dot> ca
 
G

Guest

Guest
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On Fri, 31 Dec 2004 20:47:23 GMT, a?n?g?e?l@lovergirl.lrigrevol.moc.com
(The little lost angel) wrote:

>On Fri, 31 Dec 2004 13:21:27 -0500, George Macdonald
><fammacd=!SPAM^nothanks@tellurian.com> wrote:
>
>>Even the PCI-Express mbrds have the extra power but of course it's
>>integrated into the 24-pin power header on the mbrd - something else to
>>worry about in future: is the power supply really putting out more or did
>>they just add a 20-24-pin adapter somewhere?
>
>Nope, they just added 4 more wires in parallel to the existing ones in
>most cases :ppPP

What I thought... but those extra pins *are* specifically for PCI-Express
video if I'm understanding things right. IOW if you have a PCI-Express
mbrd, one of those higher power video cards and an ATX12V 1.1 PS, you
really need the 20-->24 pin adapter or you don't get the extra juice to the
card.

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??
 
G

Guest

Guest
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On Sat, 01 Jan 2005 01:35:25 -0500, Tony Hill <hilla_nospam_20@yahoo.ca>
wrote:

>On Fri, 31 Dec 2004 13:21:27 -0500, George Macdonald
><fammacd=!SPAM^nothanks@tellurian.com> wrote:
>
>>On Fri, 31 Dec 2004 00:53:21 -0500, Tony Hill <hilla_nospam_20@yahoo.ca>
>>wrote:
>>>Am I the only one that doesn't like having a video chipset that
>>>consumes so much power it needs a drive connector to power it? If
>>>nothing else that is yet another extremely unreliable fan in the
>>>system just waiting to fail the day after the (1 year) warranty
>>>expires. It's not even like you can buy easily buy replacement fans
>>>for most of these cards since they're all customer jobs.
>>
>>Even the PCI-Express mbrds have the extra power but of course it's
>>integrated into the 24-pin power header on the mbrd - something else to
>>worry about in future: is the power supply really putting out more or did
>>they just add a 20-24-pin adapter somewhere?
>
>Well, it doesn't make too much of a difference here, the power all
>comes from the same place regardless of whether it's going through an
>extra 4 pins on the main power header or a separate connector.

Yes, apparently for the main connector it's the same 12V1 supply rail --
the 4-pin CPU 12V supply is now spec'd as a separate 12V2 rail -- but it's
not clear to me if the extra 12V main connection goes to a separate mbrd
power plane plane for the PCI-Express - wouldn't make sense but then is the
extra wire required because combined current is too high for a single
connector point... e.g. melting of solder?:)

>>Plugging a drive power connector into a video card, any card, is
>>ugly...ugly... ugly.
>
>Agreed. It doesn't really make a difference electrically speaking
>where it gets the power from (so long as it can get a good, clean and
>steady supply), but the simple idea of plugging a hard drive connector
>into a card just seems somehow wrong.

Apparently, even with the new 24-pin main connector Intel has defined an
add-on 6-pin power connector for PCI Express video cards which need >75W.
The really dumb one on all this is AMD's ATXGES, with a completely
different arrangement of pins.

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??
 

keith

Distinguished
Mar 30, 2004
1,335
0
19,280
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On Fri, 31 Dec 2004 13:21:27 -0500, George Macdonald wrote:

>
> Plugging a drive power connector into a video card, any card, is
> ugly...ugly... ugly.

I thought the same when I had to plug a drive connector into my
*motherboard*, since I "only" had an ATX12V supply.

--
Keith
 

keith

Distinguished
Mar 30, 2004
1,335
0
19,280
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On Sat, 01 Jan 2005 01:35:25 -0500, Tony Hill wrote:

> On Fri, 31 Dec 2004 13:21:27 -0500, George Macdonald
> <fammacd=!SPAM^nothanks@tellurian.com> wrote:
>
>>On Fri, 31 Dec 2004 00:53:21 -0500, Tony Hill <hilla_nospam_20@yahoo.ca>
>>wrote:
>>>Am I the only one that doesn't like having a video chipset that
>>>consumes so much power it needs a drive connector to power it? If
>>>nothing else that is yet another extremely unreliable fan in the
>>>system just waiting to fail the day after the (1 year) warranty
>>>expires. It's not even like you can buy easily buy replacement fans
>>>for most of these cards since they're all customer jobs.
>>
>>Even the PCI-Express mbrds have the extra power but of course it's
>>integrated into the 24-pin power header on the mbrd - something else to
>>worry about in future: is the power supply really putting out more or did
>>they just add a 20-24-pin adapter somewhere?
>
> Well, it doesn't make too much of a difference here, the power all
> comes from the same place regardless of whether it's going through an
> extra 4 pins on the main power header or a separate connector.
>
>>Plugging a drive power connector into a video card, any card, is
>>ugly...ugly... ugly.
>
> Agreed. It doesn't really make a difference electrically speaking
> where it gets the power from (so long as it can get a good, clean and
> steady supply), but the simple idea of plugging a hard drive connector
> into a card just seems somehow wrong.

Actually it does make a difference. The smaller the loop (one can argue
the drive supplies are a short loop) the better the supply and lower
EMI. Grounds and power in long loops isn't generally what one tries
to do. Long wires also increase the impedance of the supply. Since it's
not possible ot model all combinations of supplies, cables, and do-dads
hanging off the cables, I'd say it was a rather weak situation.

--
Keith
 
G

Guest

Guest
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On Sat, 01 Jan 2005 23:22:04 -0500, keith <krw@att.bizzzz> wrote:

>On Fri, 31 Dec 2004 13:21:27 -0500, George Macdonald wrote:
>
>>
>> Plugging a drive power connector into a video card, any card, is
>> ugly...ugly... ugly.
>
>I thought the same when I had to plug a drive connector into my
>*motherboard*, since I "only" had an ATX12V supply.

In poking around for info I came across this:
http://www.intel.com/cd/channel/reseller/asmo-na/eng/tech_reference/box_processors/int_inst_info/proc_tested_source_lists/109616.htm
I dunno what's worse... using a drive connector or defining a new
connector... which nobody has yet?? I guess people who need >75watts for
their video card can just pony up the $$ for a new PS to match it.:)

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??
 
G

Guest

Guest
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On Sat, 01 Jan 2005 18:44:50 -0500, George Macdonald
<fammacd=!SPAM^nothanks@tellurian.com> wrote:

>The really dumb one on all this is AMD's ATXGES, with a completely
>different arrangement of pins.

It's pretty much a dead standard since the Opteron came out. Very few
boards ever used the ATXGES connectors.

--
L.Angel: I'm looking for web design work.
If you need basic to med complexity webpages at affordable rates, email me :)
Standard HTML, SHTML, MySQL + PHP or ASP, Javascript.
If you really want, FrontPage & DreamWeaver too.
But keep in mind you pay extra bandwidth for their bloated code