Should I go Athlon64 or Barton?

On Thu, 09 Sep 2004 21:18:15 +0000, borolad wrote:

....best argument yet, from a leftist-loon.

--
Keith

On Thu, 09 Sep 2004 20:16:14 -0500, willbill wrote:

gaffo wrote:

.... 64-bit will remain irrelvant for another 5 yrs. (or more).


AMD is there already, and now even the 800 pound gorilla
(Intel, now down to 700 and still losing) soon will be

....haviong lost at *least* 100lbs on Itanic.

so "irrelvant" is *not* accurate, especially
given your "5 yrs (or more)" addon

64b is relevant now, and will become more so. THe "line" really isn't
4GB, but more like 3GB where 64b addressing becomes "interesting. Add the
other enhancements (and compatability) that Intel failed to notice in
favor of attempting to lock users into Itanic...

imo min/max of 12/30 *months* is more likely for 64-bit to become the
new standard in pc's

Perhaps. Again, the real number is 2GB, not 4GB. WHen will that become
"standard"? Memory is rather expensive right now (perhaps because it can
be used?). In a year?

otoh, 32 bits *is* likely to still be usable 5+ yrs from now on 90+% of
all pc's. maybe even 100%

Hell, a P5 is good enough for 90% of the desktops. Are people going to
upgrade these antiques to a P4? I know I skipped from the K6 family to
the K8 because I didn't want to upgrade again soon.

bill

p.s.

how many "bits" was the IBM 8088 pc in 1983?

Sixteen

frankly i've forgotten, but my guess is 8 (or at most 16)

It was by any definition a 16 bit processor, though had an 8bit bus. It
was identical to the 8086, other than the bus (8bs. 16b) and prefetch
queue (4B vs. 6B). The 80186 and 80188 started out as identical
dies, with a bond-out option.

--
Keith

Look at AGP, you might have thought, a year ago that you'd
buy, for example, a Radeon 9500 video card and upgrade to a Radeon xxxx in
two years time perhaps. But, of course, AGP is now being replaced by PCI
Express, so you can't do it.

I can't imagine AGP entirely disappearing within a year or two. There
will still be a big market for AGP upgrade cards in the near future.
Heck, you can still buy a GeForce 5200 or Radeon 9200 in PCI.

*'``'*:-.,_,.-:*'``' ¸ô¶ó - Cull the O/T ****e '``'*:-.,_,.-:*'``'*

*'``'*:-.,_,.-:*'``' ¸ô¶ó - Cull the O/T ****e '``'*:-.,_,.-:*'``'*

On Wed, 08 Sep 2004 13:54:40 GMT, "John Fryatt"
wrote:

snip

Look at AGP, you might have thought, a year ago that you'd
buy, for example, a Radeon 9500 video card and upgrade to a Radeon xxxx in
two years time perhaps. But, of course, AGP is now being replaced by PCI
Express, so you can't do it. Similarly with processors, chances are you'd
need a new motherboard anyway to make use of new generations of CPU.

Not true.
according to that 2-year plan, one year has elapsed. There
are still new AGP models to come, for example nVida 6n00
series, and other lesser alternatives should remain in the
market for at least a few more years... today you can even
find a new Geforce2, why assume today's contemporary cards
will disappear any faster?

keith wrote:

On Thu, 09 Sep 2004 20:16:14 -0500, willbill wrote:

gaffo wrote:

.... 64-bit will remain irrelvant for another 5 yrs. (or more).

AMD is there already, and now even the 800 pound gorilla
(Intel, now down to 700 and still losing) soon will be


....haviong lost at *least* 100lbs on Itanic.


agreed and smiling.


snip

otoh, 32 bits *is* likely to still be usable 5+ yrs from now on 90+% of
all pc's. maybe even 100%


Hell, a P5 is good enough for 90% of the desktops. Are people going to
upgrade these antiques to a P4? I know I skipped from the K6 family

me too


to the K8 because I didn't want to upgrade again soon.

again, me too

i find it interesting that PC stuff is finally slowing down,
at least a bit

one couldn't go 5 years in the 80's/90's without having
to get a new pc. that's no longer true


p.s.

how many "bits" was the IBM 8088 pc in 1983?


Sixteen


wasn't the 8088 8 bits?

whereas the 8086 was 16 bits?


frankly i've forgotten, but my guess is 8 (or at most 16)


It was by any definition a 16 bit processor, though had an 8bit bus.

therein is **the** issue

are we talking about the bus or what?


It
was identical to the 8086, other than the bus (8bs. 16b) and prefetch
queue (4B vs. 6B). The 80186 and 80188 started out as identical
dies, with a bond-out option.

thank you for your response

this "bit" stuff gets confusing to me.

are we talking about the bus or what?

anyhow and btw, aside from the crappy ide/floppy
connector layout on the s2875s, it's a typical/excellent
Tyan mobo (at least in my experience). i was
surprised that you spent the extra bucks for
a 144 (i got a 142)

it'll be interesting so see what we can go to when
the 9 micron Opterons show up in the near future

not to mention if Tyan will offer a bios update
that will permit using one of the new dual core
Opteron cpu's on the s2875s

bill

In article , Keith wrote:
[WillBill wrote]
how many "bits" was the IBM 8088 pc in 1983?

Sixteen

frankly i've forgotten, but my guess is 8 (or at most 16)

It was by any definition a 16 bit processor, though had an 8bit bus. It
was identical to the 8086, other than the bus (8bs. 16b) and prefetch
queue (4B vs. 6B).

It's an interesting question ... though not as interesting as it was in
1983. I used to have a rule-of-thumb for estimating the "bittedness" of
CPUs that went something like this:

The 8088 had an 8-bit external data bus, 16-bit general-purpose registers,
16-bit stack pointer, and 20-bit addressing range (being a bit generous,
considering the segmented addressing model). Taking an average of those
four gives 15 bits. Near enough.

The 8086 was the same apart from the 16-bit data bus (and, yes, the longer
instruction prefetch queue - I'd forgotten about that) so let's call it 17
bits.

Compare that with the Z80 (8+16[1]+16+16=14 bits) the 6502 beloved of
Apple II and PET user (8+8+8+16=10 bits) and the 68000 (32+32+32+24=30
bits). Considering that these chips were all available when the PC was
designed: I know which chip *I'd* have used!

Intel didn't manage to beat that 30-bit value until the '386 made it up to
32.

[1] That's counting HL as a 16-bit GP register, which some might say was
cheating. The Z80's accumulator was 8 bits, and counting that instead
would give an overall "bittedness" of 12.

Cheers,
Daniel.

willbill wrote:

keith wrote:

On Thu, 09 Sep 2004 20:16:14 -0500, willbill wrote:

gaffo wrote:

.... 64-bit will remain irrelvant for another 5 yrs. (or more).

Not quite. Expect to see great 64 bit games in '05. Imo we will
also probably see 64 bit Photoshop video editing software,
CAD, and other applications in '05.



AMD is there already, and now even the 800 pound gorilla
(Intel, now down to 700 and still losing) soon will be


....haviong lost at *least* 100lbs on Itanic.

agreed and smiling.

snip

otoh, 32 bits *is* likely to still be usable

A record player that you crank up by hand that doesn't use electricity is
still usable, but doesn't provide the same listening experience as modern hifi
equipment.

5+ yrs from now on 90+% of
all pc's. maybe even 100%

Hell, a P5 is good enough for 90% of the desktops. Are people going to
upgrade these antiques to a P4? I know I skipped from the K6 family

me too

to the K8 because I didn't want to upgrade again soon.

again, me too

i find it interesting that PC stuff is finally slowing down,
at least a bit

one couldn't go 5 years in the 80's/90's without having
to get a new pc. that's no longer true

PCs were phenomenally expensive then, especially if you
adjust for inflation. Adjusting for inflation, they were probably
8 or more times what an average pc costs now.



p.s.

how many "bits" was the IBM 8088 pc in 1983?

Sixteen

wasn't the 8088 8 bits?

whereas the 8086 was 16 bits?

frankly i've forgotten, but my guess is 8 (or at most 16)


It was by any definition a 16 bit processor, though had an 8bit bus.

therein is **the** issue

are we talking about the bus or what?

It
was identical to the 8086, other than the bus (8bs. 16b) and prefetch
queue (4B vs. 6B). The 80186 and 80188 started out as identical
dies, with a bond-out option.

thank you for your response

this "bit" stuff gets confusing to me.

are we talking about the bus or what?

anyhow and btw, aside from the crappy ide/floppy
connector layout on the s2875s, it's a typical/excellent
Tyan mobo (at least in my experience). i was
surprised that you spent the extra bucks for
a 144 (i got a 142)

it'll be interesting so see what we can go to when
the 9 micron Opterons show up in the near future

not to mention if Tyan will offer a bios update
that will permit using one of the new dual core
Opteron cpu's on the s2875s

bill