Re: [WinMac] Mac processor upgrades


Daniel L. Schwartz(expresso[at]snip.net)
Tue, 25 Jan 2000 12:14:03 -0500


At 08:44 AM 1/25/00 -0500, Tom Roth wrote:
>
>
>"Daniel L. Schwartz" wrote:
>> I saw too many 6100's with blown motherboards caused by XLR8 overclocking
>> up into the 83 mHz range... I don't subscribe to the theory of turning up
>> the speed til it crashes, then backing it off one click!
>
>Why would someone even bother trying to upgrade a 6100 in the first
>place?! I wouldn't try a G3 processor upgrade with any of the first
>generation nubus PowerMacs but especially a 6100!

        Actually, I was referring to using the same 610 CPU and simply
overclocking it, i.e. increasing the CPU speed from 60 or 66 mHz up to
about 83 mHz. And when the CPU fries, you needed to toss the whole MLB
since the CPU was soldered to the board.

>> The G4 is indeed a nice CPU; but it gains much of its speed due to its
>> width. If the rest of the bus (memory & north bridge) can't handle the
>> data, then it's just a waste.

        Since someone else wrote that the G4 is such a great upgrade for the 8500
& 95 series PCI Macs, I need to respond with an emphatic NO! Just look at
the performance of the original G4/400 machines based on the Yosemite
("Yikes") motherboard - The same MLB as in the B&W G3 machines (and with
PCI video). Put one of these machines next to a G4 Sawtooth (AGP video)
with the wider bus and you'll see a BIG difference in speed... Despite the
handicap of having the cheesey AGP video.

        [Actually, Apple had it right the first time with the PCI video: AGP is a
hack that's OK as long as video RAM usage doesn't exceed the RAM on the
card. AGP falls apart as soon as it needs to "extend" itself into system
RAM with extra texture & other buffers needed for 3D performance. That's
why all of the non-Wintel workstations Sparc, Alpha, RS/6000) still use
high end PCI video cards.]

>> More importantly - in my opinion - is the I/O subsystem: You can have the
>> fastest CPU, but if it's being tied down by a slow IDE disk then it's just
>> spinning its wheels.
>
>That's about what I figured and even somewhat feared with even a G3
>upgrade.

        This brings up another landmine: When you upgrade to a G3 .AND. you have a
third party SCSI accelerator, all hell can break loose if the firmware is
flaky. Can you say "AdapCrap?!" So far, the only stable SCSI cards I've
seen are the Atto cards (although I've not tested Initio cards in PCI Macs
- Only in AlphaNT and NT/x86). At least Atto has a long history of
producing SCSI cards for the Mac, all the way back to NuBus Silicon Express
cards, and it shows with their stability in PCI Macs.

>> This is one thing I STILL can't figure out: For years Macs came with more
>> expensive SCSI disk subsystems... Even though System 6 & 7 was NOT
>> multithreaded - And couldn't take advantage of it. But today, now that
>> OS-X is about to ship, the situation has been reversed: No SCSI!
>>
>> Go figure...
>
>I did and that's why we'll be ordering our system with SCSI.
>
>Thanks to everyone who replied to this thread!
>
> ______________________________________________________________________
> Tom Roth Wake Forest University School of Medicine
> tomroth@wfubmc.edu Dept of Biomedical Communications

 -----------------------------------------------------------------

        THERE ARE NO ATTACHMENTS TO THIS MESSAGE, SO IF ONE
     APPEARS WITH IT, DO NOT OPEN OR DOWNLOAD IT!

        <mailto:expresso@snip.net>

                Webmaster for <http://www.Faulknerstudios.com>,
                                        <http://www.BrakeAndGo.com>

-> NEW! Sign up for the Mac-NT Mailing list at:
 <http://www.onelist.com/subscribe/Mac-NT>

-> REALLY New! Survey about what you administer at:
 <http://www.onelist.com/surveycenter/Mac-NT>

 -----------------------------------------------------------------

*** Windows-MacintoshOS Cooperation List ***
FAQ: http://www.darryl.com/winmacfaq/
Archive: http://www.darryl.com/winmac/

To unsubscribe, send mail to winmac-request@lists.best.com
with just the word "unsubscribe" in the body of the message.



This archive was generated by hypermail 2.0b2 on Tue Jan 25 2000 - 10:13:20 PST