#LW vs GVP…BS part II
14 messages in this thread
AAAAARRRRGGGHHHH!!!
Yes, yet another NEWTEK flame!
Last November (3 MONTHS ago) I bought a GVP 040 "Impact Series II"
accellerator card for my A2000 HD.
I installed it per spec and SUPRISE! The toaster (3.0 software/1.0 board)
wouldn't come up. Well, actually, the switcher screen appeared, but
moments later, the dreaded red "software error" box appeared, with the
message:
Software Error
8000 000B Task 01035E00
And the system would reboot on pressing the left mouse button as instructed
on the screen.
Hmmm, I thought, better call NewTek Tech Support… SO, I did. I was
told by the tech that they had "had some troubles" with the GVP 040
and that an update was being released Real Soon Now to fix a bug when
used with the GVP 040 board. The tech further informed me that my
1MB of chip RAM "isn't enough". And that "The spec requirements for
1MB chip are marginal at best. You need 2MB chip RAM". After
thinking "guess it couldn't hurt", I dutifully ordered the DKB 2MB
chip RAM.
I soon found that the tech's CHIP RAM statements were BULLSHIT! The
2MB chip RAM made NO DIFFERENCE WHATSOEVER WITH THE TOASTER
CRASHING. I subsequently started doing my own Systems Analysis on
the damn thing, and found that the problem was when the GVP was
instructed to relocate system ROM into FAST RAM (a real performance
boost), the Toaster crashed on startup. After removing the FASTROM
instruction in my user-startup (leaving the MoveSSP in place), I was
able to get the toaster up and running, albeit without the humongous
speed increases I had been led by a NewTek tech to expect from the
040.
After waiting patiently for THREE MONTHS for this mythical update, I
finally received my 3.1 Update. Anticipating a fix for the GVP 040
problem, I immediately started loading it. WRONG! Disk 4 of the
update was a BAD disk! After arranging to get a replacement, and yet
ANOTHER period of NON-PRODUCTION, I received a replacement. I
FINALLY loaded the 3.1 update.
Excited with anticipation, I restored the user-startup entries for the
040, and rebooted. Then, I started the 3.1 Toaster and *** SUPRISE
*** it wouldn't come up for more that a couple of seconds as long as
the GVP FASTROM command was in place. So much for the freakin' "bug
fix" for which I had been
waiting THREE MONTHS.
I DID see SOME lightwave performance increases with the update… but STILL…
Here's the current situation:
01/27/94 20:29:25
Version 3.0 of S/W. FASTROM disabled on GVP.
SpaceFighters: Frame 195.
Camera: Trace EVERYTHING. Low Anitaliasing. Adaptive Samp. (8)
Particle Blur. Medium Resolution. Overscan.
Lights: 1 Light. Distant/RayTrace.
Objects: All shadows on on all objects.
Record: No record/No saves.
Rendering time: 15m 26s (26 seconds)
01/28/94 23:16:29
Version 3.1 of S/W. FASTROM disabled on GVP. (Toaster wouldn't run w/FASTROM)
SpaceFighters: Frame 195.
Camera: Trace EVERYTHING. Low Anitaliasing. Adaptive Samp. (8)
Particle Blur. Medium Resolution. Overscan.
Lights: 1 Light. Distant/RayTrace.
Objects: All shadows on on all objects.
Record: No record/No saves.
Render to DV1
Rendering time: 9m 26s (566 seconds)
Yeah, it WAS better with 3.1, but I can't help but think that it could be
even MUCH better if the GVP 040 was operating to its full potential.
Now, as with every situation where there is 3rd party hardware involved,
it is VERY difficult to place a blame on the 040 board. All I know
is that this whole experience has confirmed the opinions that I've heard that
the Toaster isn't REALLY a stable, professional product. <sigh, too bad>.
Although I realize, I'm not one of Speilberg's staff, and as much as I LOVE
Lightwave, something's GOT to give. I have GOT to have RELIABLE, professional
tools to run my business… I'm NOT a garage hobbyist. (10yrs TV + 10yrs
micros)
Reply comments would be GREATLY appreciated.
Luke (Pat) Montgomery "REAL" E-mail: compvid@tyrell.net
CompVid Computer Video Graphics Services CompuServe: 70274,2177
Greater Kansas City Voice: (913) 780-0222
—————————————————————————–
There's no place like home… There's no place like home… There's no…
You're blaming the wrong folks. 040 accelerators are at best
marginally reliable – if you get one that works in your machine then
you'll have years of pleasent computing – get one that doesn't like
your machine and there's little short of getting another accelerator
that will cure your problems. Toaster systems the world over work
perfectly well with 040 boards provided that the installed 040 board
works by itself. I'll bet dollars to donuts that your system would be
flakey with your fully enabled 040 even if you never entered the
Toaster software. My advice: get another 040 – there's so much
varience even within the same make and model that you might have
better luck trying another (exact same) gvp board.
I certainly don't want to get into any kind of "flame war" here, but I think
it's pretty obvious that the problem here is with GVP, not NewTek. From your
description the GVP card changes the nature the machine.
Can't see holding NewTek responsible for that. I have a Toaster with an Amiga
4000 and it's never crashed or done anything it shouldn't have that couldn't be
attributed to user error.
Jerry
Luke,
There is a specific sequence to enabling the full '040 on the gvp board:
;BEGIN 68040_FastROM
cpu nocache noburst nocopyback ; suggested by Mike GVP 5/10/93
SYS:Utilities/GVPCpuCtrl MoveSSP FASTROM
cpu cache burst copyback ; suggested by Mike GVP 5/10/93
;END 68040_FastROM
Arp040fix ; patch to handle arplib/'040 compatibility
Also, make sure you have a rev of 68040.library of 37.10 or greater. Note
that with the above, I am running 3.1 just fine. The problem with GVP
stuff on 3.0 was horizontal lines in stuff captured into TPaint from the
DV buffers or frames frozen from full motion video inputs. Also note
that the MoveSSP, Fastrom and cache burst stuff won't work if you have not
fully populated the board with memory.
wmc – via Autopilot!
REPLAY from Wayne Cole: >There is a specific sequence to enabling the full '040
on the gvp board: > >;BEGIN 68040_FastROM >cpu nocache noburst nocopyback
; suggested by Mike GVP 5/10/93 >SYS:Utilities/GVPCpuCtrl MoveSSP FASTROM >cpu
cache burst copyback ; suggested by Mike GVP 5/10/93 >;END
68040_FastROM >Arp040fix ; patch to handle arplib/'040
compatibility > >Also, make sure you have a rev of 68040.library of 37.10 or
greater. Note >that with the above, I am running 3.1 just fine. The problem
with GVP >stuff on 3.0 was horizontal lines in stuff captured into TPaint from
the >DV buffers or frames frozen from full motion video inputs. Also note
>that the MoveSSP, Fastrom and cache burst stuff won't work if you have not
>fully populated the board with memory.
OK.. Thanks for the reply, Wayne. That's the SAME sequence that I am using
except for the Arp040fix, which I have just d/L'ed and tried.
STILL no success. But the part about needing to have the board FULLY populated
is new news to me!!! That seems a bit crazy, IMHO… Is GVP gonna populate the
board FOR me so it will work like it should??!?!? >:-< Regards,
Luke Montgomery CompVid Computer Video Graphics.
Luke,
In the GVP document it says full support of 68040 burst mode with 16 or 64
MB mem configurations in several places. If you are not fully
populated, you may need to reconfigure J22 on the 040 board if its the
same as mine. Did you check/set all the jumpers before installing?
wmc – via Autopilot!
You're incorrect about MoveSSP and Fastrom requiring full memory population.
Only 040 BURST mode requires this. MoveSSP and Fastrom should work fine with
just 4MB RAM installed.
FYI: I used to work for GVP.
Regards,
Paul.
Paul,
Thanks for the correction. I knew Fastrom worked without it (sorry if I
implied otherwise) and wasn't sure about MoveSSP although there used to be
a PD utility that did the same even on 68000 – put the SSP in Fastmem. I
guess the old OS (possible still does, havent looked) put sth SSP in
chip.
I also used GPVCpuCtrl on a Combo '030 card to move SSP. Did it work or
am I just fooling myself (seems a little snappier, but not a dramatic
difference like it made on the 040).
wmc – via Autopilot!
Er, don't believe MOVESSP does anything on 030 – suggest you ask in GVP forum.
Well, I guess I could just look at the SSP before and after I run it too,
couldn't I? duuhhh! Man, it seems the older you get the more you miss
the obvious.
wmc – via Autopilot!
Or even easier type GVPCPUCTRL and see if it responds with where the stack
pointer has been moved to. I assume that if it hasn't been moved then there
should be no report available.
Jim Rooof Make that Roof
well, it says nothing -neither error nor moved…
wmc – via Autopilot!
Hmmm. I've been running two GVP 040-33 combo boards in my 2000's (one w, (one
with Toaster 2000 card, the other with T4000 card) and had zero problems. FYI,
both my computers have MegAChips and 16 MB RAM (4X4 MB SIMMS), burst enabled.
Perhaps there's a problem elsewhere?
Trent
> #: 73874 S3/NewTek [AMIGAV] > 05-Feb-94 01:58:35 > Sb: #73459-LW vs
GVP…BS part II > Fm: Trent K. Johnson 71020,1052 > To: Luke (Pat) Montgomery
70274,2177 > >Hmmm. I've been running two GVP 040-33 combo boards in my 2000's
(one w, (one >with Toaster 2000 card, the other with T4000 card) and had zero
problems. >FYI, both my computers have MegAChips and 16 MB RAM (4X4 MB SIMMS),
burst >enabled. Perhaps there's a problem elsewhere? > >Trent
Hmm.. I'm beginning to wonder…
You got 16MB, huh? Wonder if THAT's the problem… (I've got 12MB)…
<sigh>
Luke (Pat) Montgomery "REAL" E-mail: compvid@tyrell.net
CompVid Computer Video Graphics Services CompuServe: 70274,2177
Greater Kansas City Voice: (913) 780-0222
—————————————————————————–
There's no place like home… There's no place like home… There's no…