From: Randal Whittle (rwhittle_at_usa.net)
Date: Thu Aug 05 1999 - 15:18:33 EDT
At 11:27 AM 8/5/99 -0700, you wrote:
>Actually, "Pentium-level performance" can be achieved in the IBM ThinkPad
>701 series. With the AMD5x86P75 microprocessor installed, the base clock
>rate set to 33MHz, and the 16K L1 cache write-back mode enabled, CPU
>performance is better than the P75.
>
>Various CPU benchmarks show a performance increase between 2 and 3 times
>faster than the original Intel 486. For DX4-75 models, CPU Dhrystones
>benchmarks increase from 26041 to 60975. Norton SI shows the CPU
>performance increasing from 95.4 to 287.9.
>
>Randal's 20% speed increase indicates that the 5x86 was installed, but the
>other features were not enabled.
>
>Best wishes,
>
>Douglas Hirsch
>Internet Support
>Maximum Upgrades
Ahhh....No Doug, it doesn't indicate that at all. What it *really*
indicates is that you're using phantom CPU-only benchmarks instead of
real-world ones. By that, I mean to say that the Dhrystone and Norton SI
stuff doesn't amount to squat. They measure CPU only--not overall
sytem-level, which is all that really matters in the real world.
The benchmarks I ran were far more indicative of what someone could expect
to see in their computer's (specifically a 701) overall peformance
increase. They were the ones that used real-world apps and did tests with
those apps, etc. I forgot the name...ZD Labs made them...Winstone? Also
some other stuff.
I doubt that 20 months later I still have the spreadsheet I saved all this
on, but I know I ran those same benchmarks on the pre-upgraded 701C
(DX4-75) and the post-upgraded one, and also ran them on my TP 560 (a
P-120) and a desktop Pentium Pro 200, just so I could have a real-world
comparison across some other machines. The upgraded machine performed
really quite well in CPU-only areas, but when put in context with the rest
of the system overall--that is, general overall performance that a user can
palatably see--the upgrade delivered a 20 to 25% performance increase over
the stock 486 DX4-75.
Unless you've got some new chip (other than the 5x86 I tested) that really
*does* deliver on these claims of yours, I suggest you not try to snow
anybody with claims of 2x to 3x speed increase over the original. The
reality is that just isn't the case, unless perhaps you're measuring in a
vacuum where CPU's operate alone without outside influences.
OR...here's a much better idea: You could put your money where your mouth
is, just as the good folks at PEP did 20 months ago. We on the list will
come up with an objective, neutral 3rd party 701C owner with the technical
know-how to execute "before" and "after" tests using a *legitimate*
benchmark like the ZD Winstone tests, and we'll see if indeed the upgraded
701C is 2 or 3 times faster as your claims insinuate.
Are you up to proving your claims? We're daring ya.
If you're not, then please quit making the phony claims on this list.
Jane went through the trouble of digging up the benchmarking I did 20
months ago, and she's only the tip of the iceberg. We've got plenty of
people willing to out-shout any B.S. claims until you're willing to let an
objective 3rd party that *we* choose prove it to us.
Are you game? This can be put to rest, once and for all. If the results
>from this unbiased 3rd party tester *really* show a 701C that is...oh,
we'll say 2.5X as fast as the original in overall system performance
(evenly split between 2X and 3X), I will personally make a public apology
to the entire list and endorse your upgrade.
The only caveats are (1) We (the list) chooses the tester, (2) the testing
system is otherwise unchanged, and (3) you do the upgrade on that tester's
701C free of charge. These are the exact same terms PEP did it for when I
benchmarked their upgrade. And they never pressured me to report anything
other than the truth. That truth was a 20 to 25% overall system boost over
the stock 486 DX4-75. And oh, yes--one more thing: The base 701 being
upgraded/benchmark must start from the 75 MHz version of the 486-DX4 chip,
50 Mhz one.
Are you really sure it will double or triple system performance? I'd like
to see it. Let's see.
- Randy Whittle
This archive was generated by hypermail 2.1.3 : Thu Jan 23 2003 - 09:55:13 EST