10 Nov
2006
10 Nov
'06
7:47 p.m.
Harris Semiconductor made licensed 80286 CPU's.
They made them in 16 and 20mhz versions.
I used to sell systems made with them, and at the time the 386sx came
out, a 286-20 was MUCH faster than a 386 20mhz...
Al
11 Nov
11 Nov
12:32 a.m.
Al Hartman wrote:
Harris Semiconductor made licensed 80286 CPU's.
They made them in 16 and 20mhz versions.
I used to sell systems made with them, and at the time the 386sx came
out, a 286-20 was MUCH faster than a 386 20mhz...
How is that possible? The 386 improved many instructions to run in
fewer clocks...
Besides, I'd always take a 386-16 over a 286-20, because the 286-20 runs
protected mode applications at 0MHz ;-)
--
Jim Leonard (trixter at oldskool.org) http://www.oldskool.org/
Help our electronic games project: http://www.mobygames.com/
Or check out some trippy MindCandy at http://www.mindcandydvd.com/
A child borne of the home computer wars: http://trixter.wordpress.com/
1:02 a.m.
On 10 Nov 2006 at 23:32, Jim Leonard wrote:
How is that possible? The 386 improved many
instructions to run in
fewer clocks...
Besides, I'd always take a 386-16 over a 286-20, because the 286-20 runs
protected mode applications at 0MHz ;-)
Strange, but apparently true in some cases:
http://www.intersil.com/data/an/an111.pdf
I imagine that the difference would be exaggerated even more with a
386SX.
Cheers,
Chuck
1:33 a.m.
Chuck Guzis wrote:
That's not really a fair comparison, since that product note is from
1997, long after the time period we're talking about.
Doing some more thinking, it's possible that some 386 PCs are hampered
by memory wait states more than some implementations of 286 PCs.
--
Jim Leonard (trixter at oldskool.org) http://www.oldskool.org/
Help our electronic games project: http://www.mobygames.com/
Or check out some trippy MindCandy at http://www.mindcandydvd.com/
A child borne of the home computer wars: http://trixter.wordpress.com/
How is that
possible? The 386 improved many instructions to run in
fewer clocks...
Strange, but apparently true in some cases:
http://www.intersil.com/data/an/an111.pdf 
10:24 a.m.
On Sat, 11 Nov 2006, Jim Leonard wrote:
Date: Sat, 11 Nov 2006 00:33:42 -0600
From: Jim Leonard <trixter at oldskool.org>
Reply-To: "General Discussion: On-Topic and Off-Topic Posts"
<cctalk at classiccmp.org>
To: General Discussion: On-Topic and Off-Topic Posts <cctalk at classiccmp.org>
;
Subject: Re: Overclocking ATs
Chuck Guzis wrote:
That's not really a fair comparison, since that product note is from 1997,
long after the time period we're talking about.
Doing some more thinking, it's possible that some 386 PCs are hampered by
memory wait states more than some implementations of 286 PCs.
Not sure but we waited for 33 MHz 386's because our PC layout software ran
significantly faster on our 20 MHz 286's than 20 MHz 386's. It was a
noticeable difference, maybe 20%...
How is
that possible? The 386 improved many instructions to run in fewer
clocks...
Strange, but apparently true in some cases:
http://www.intersil.com/data/an/an111.pdf --
Jim Leonard (trixter at oldskool.org) http://www.oldskool.org/
Help our electronic games project: http://www.mobygames.com/
Or check out some trippy MindCandy at http://www.mindcandydvd.com/
A child borne of the home computer wars: http://trixter.wordpress.com/
Peter Wallace
Mesa Electronics
(\__/)
(='.'=) This is Bunny. Copy and paste bunny into your
(")_(") signature to help him gain world domination.
2:59 p.m.
On 11 Nov 2006 at 7:24, Peter C. Wallace wrote:
Not sure but we waited for 33 MHz 386's because
our PC layout software ran
significantly faster on our 20 MHz 286's than 20 MHz 386's. It was a
noticeable difference, maybe 20%...
There were similar anomalies throughout much of the Intel x86 line.
Early Socket 4 Pentiums could be outrun by a 486DX2 in some
circumstances and early P4s could be outrun by a P3. Or a Pentium
Pro could be substantially slower than a Socket 7 Pentium when
executing 16-bit code, and so on.
Cheers,
Chuck
12 Nov
12 Nov
2:32 a.m.
Chuck Guzis wrote:
There were similar anomalies throughout much of the
Intel x86 line.
Early Socket 4 Pentiums could be outrun by a 486DX2 in some
circumstances and early P4s could be outrun by a P3. Or a Pentium
Pro could be substantially slower than a Socket 7 Pentium when
executing 16-bit code, and so on.
The former examples are all mostly vendor implementation-specific (size
of cache, wait state of dram, first Pentium motherboard logic
unoptimized, etc.), but the Pentium Pro example was by design: Intel
deliberately sped up 32-bit protected mode at the expense of 16-bit in
the design. It was quite intentional (something they ended up taking a
small amount of flak for).
--
Jim Leonard (trixter at oldskool.org) http://www.oldskool.org/
Help our electronic games project: http://www.mobygames.com/
Or check out some trippy MindCandy at http://www.mindcandydvd.com/
A child borne of the home computer wars: http://trixter.wordpress.com/
11 Nov
11 Nov
8:54 a.m.
Nice! Reminds me of the very first 486's we saw.
We had a bunch of decent speed 386's at the time, and some upgrades came
in, in the form of 486's. The 486's were also at a higher clock rate,
so we were scratching our heads as to why our software ran much slower
on the 486's.... Turns out that we didn't order any cache chips with
them, while the 386 motherboards had the max they could support!
D'Oh! :-)
Ah, the good ol'e days of Netware 386...
A few years later, we had fun with the F00FC7C8 opcode on the Pentiums.
Chuck Guzis wrote:
Strange, but apparently true in some cases:
http://www.intersil.com/data/an/an111.pdf
I imagine that the difference would be exaggerated even more with a
386SX.
Cheers,
Chuck
3:32 p.m.
Ray Arachelian wrote:
A few years later, we had fun with the F00FC7C8 opcode
on the Pentiums.
Ah yes, the infamous F00F bug, from the people who brought you the even more
infamous FDIV bug. Officially described as an "illegal form of the CMPXCHG8B
instruction with the LOCK prefix". Also known as 'How to crash any Pentium
series chip without the messy business of having to steal ring-zero access
first'. Lovely.
--
Phil. | (\_/) This is Bunny. Copy and paste Bunny
classiccmp at philpem.me.uk | (='.'=) into your signature to help him gain
http://www.philpem.me.uk/ | (")_(") world domination.
5:40 a.m.
Jim Leonard wrote:
Al Hartman wrote:
I don't know, but I remember comparing two side-by-side and the 286
absolutely horsed the 386sx. IIRC it was about 15% faster - *really*
noticable difference.
Gordon.
Harris Semiconductor made licensed 80286
CPU's.
They made them in 16 and 20mhz versions.
I used to sell systems made with them, and at the time the 386sx came
out, a 286-20 was MUCH faster than a 386 20mhz...
How is that possible? The 386 improved many instructions to run in
fewer clocks...
Besides, I'd always take a 386-16 over a 286-20, because the 286-20 runs
protected mode applications at 0MHz ;-) 
14 Nov
14 Nov
10:02 a.m.
Gordon JC Pearce wrote:
Jim Leonard wrote:
Did the 286 have a 287? Was there floating-point code in the
application you were using?
Peace... Sridhar
Al Hartman wrote:
I don't know, but I remember comparing two side-by-side and the 286
absolutely horsed the 386sx. IIRC it was about 15% faster - *really*
noticable difference. Harris Semiconductor made licensed 80286
CPU's.
They made them in 16 and 20mhz versions.
I used to sell systems made with them, and at the time the 386sx came
out, a 286-20 was MUCH faster than a 386 20mhz...
How is that possible? The 386 improved many instructions to run in
fewer clocks...
Besides, I'd always take a 386-16 over a 286-20, because the 286-20
runs protected mode applications at 0MHz ;-)
6:06 p.m.
Sridhar Ayengar wrote:
And again, it's possible that the 286 was 0WS while the 386 was 1WS, or
interleaved vs. non-interleaved... RAM speed is a big deal, on any platform.
Another factor could have been a poor-engineered 386sx (16-bit bus)...
whereas a 386dx (32-bit bus) could have horsed the 286.
--
Jim Leonard (trixter at oldskool.org) http://www.oldskool.org/
Help our electronic games project: http://www.mobygames.com/
Or check out some trippy MindCandy at http://www.mindcandydvd.com/
A child borne of the home computer wars: http://trixter.wordpress.com/
I don't
know, but I remember comparing two side-by-side and the 286
absolutely horsed the 386sx. IIRC it was about 15% faster - *really*
noticable difference.
Did the 286 have a 287? Was there floating-point code in the
application you were using?
6466
days inactive
6469
days old
11 comments
8 participants
participants (8)
-
alhartman＠yahoo.com -
cclist＠sydex.com -
classiccmp＠philpem.me.uk -
gordon＠gjcp.net -
pcw＠mesanet.com -
ploopster＠gmail.com -
ray＠arachelian.com -
trixter＠oldskool.org