@anarchic_teapot

I thought this was widely accepted that it was a bug. we just don't know where it is.
we don't know if it is hardware,os,software whatever. reguardless it is not a show stopper afaik.

i guess it depends on how you look at it. if it is strictly looked at from a gx/os4 perspective it is a bug since there is a flaw somewhere we just don't know where.
yes i do know that the gx was thrown in at the last second by eyetech and hyperion was unable to prepare for this version in os4.however if looked at from the perspective of the chip families in general you can make the case it is only a compatability issue with some flavors of the cpu's. it all depends on the viewpoint of the person.

Last edited by jkirk on 13-Apr-2005 at 05:36 PM.

_________________
Win•dows: n. A thirty-two bit extension and graphical shell to a sixteen-bit patch to an eight-bit operating system originally coded for a four-bit microprocessor which was written by a two-bit company that can't stand one bit of competition.

@ikir

Quote:

Anyway it is not the point of the thread.

oooooh handbags at twelve paces

_________________
2017 Camaro 2SS

@DrBombcrater

Quote:

Do bear in mind that the 750GX is known to contain at least one very serious hardware defect, and there's been zero information from Eyetech or Hyperion as to whether the uA1 uses a buggy revision of the GX and if so what impact it has on OS4.

The GX does? The first I've heard of it. Do you have a link?

Would be interested in finding out more. The link might help out Eyetech, and Hyperion as well.

_________________
"The system no longer works " -- Young Anakin Skywalker

@Interesting

I remember reading about this to. But I think Hyperion has this info.
I think it is the official 750GX Errata from IBM.


@DrBombCrater

We/I have version 1.1 (UBoot system info) of the 750GX CPU. I seem to recall
there being a 1.2 version available from IBM. I don't know the actual difference
though.


/Harry

Last edited by Samwel on 14-Apr-2005 at 12:30 AM.

_________________
/Harry

[SOLD] µA1-C - 750GX 800MHz - 512MB - Antec Aria case

Avatar by HNL_DK!

@nicomen

Quote:

Maybe hold that thought for some time for now ;) Enjoy the spring/fall instead, take a walk outside ;)

Not for long I hope.. I want to update my Amiga!

I sleep all day and work all night (night shift).. So walking
outside is a no go

But you're right.. I should take some time to enjoy spring
some..


/Harry

_________________
/Harry

[SOLD] µA1-C - 750GX 800MHz - 512MB - Antec Aria case

Avatar by HNL_DK!

@anarchic_teapot

Quote:

Can we try and make a list of the problems, with the GR codes if possible?

I think the IBrowse team is involved also. They have alot of GR dumps on the
crashes. IBrowse is one of the software that are affected most by the GX.

There is a thread about the subject here on AW.
Let's see if I can find it.. Oh yes, here it is:
IBrowse thread


/Harry

Last edited by Samwel on 14-Apr-2005 at 01:03 AM.

_________________
/Harry

[SOLD] µA1-C - 750GX 800MHz - 512MB - Antec Aria case

Avatar by HNL_DK!

@ssolie

Quote:

P.S. Last I heard the GX problem was still being worked. We all know when it will be done...

Do we? Can you tell us once more please???


/Harry

_________________
/Harry

[SOLD] µA1-C - 750GX 800MHz - 512MB - Antec Aria case

Avatar by HNL_DK!

@Interesting

Quote:

he GX does? The first I've heard of it. Do you have a link?

There's a short news article on it here and IBM's documentation is here. Look under erratum 8.

The article only talks about the 1GHz GX, but IBM's docs make clear all v1.1 GXs are affected by this and won't work correctly unless down-clocked. Someone with a uA1 needs to take the cooler off their CPU and check if it says '800' in it.

_________________
Who do you serve, and who do you trust? - Galen

@DrBombcrater

Quote:

The article only talks about the 1GHz GX, but IBM's docs make clear all v1.1 GXs are affected by this and won't work correctly unless down-clocked. Someone with a uA1 needs to take the cooler off their CPU and check if it says '800' in it.

I hope some dealer can verify this things... GX version and speed.

(I fear I'm not able to do it without crush the thing... )

_________________
-------------------------------
AmigaOS, the last hope...

@Interesting

Quote:

The link might help out Eyetech, and Hyperion as well.

The problem mentioned is that some logical operations with condition codes don't work on the GX when it's running at it's nominal frequency. However, almost all programs use condition code arithmetic (for example when calling a vararg function), so it's _probably_ not the cause.

There are several other problems with the GX, but that's nothing new. You can check out all the errata manuals on Freescale and IBM websites; all chips have bugs of some sort, be it 750* ot 74xx.

The particular problem with the GX and iBrowse is that it is somewhat reproducable (I say somewhat because _I_ can't reproduce it).

And sorry, I don't have news for you. I had a difficult time personally for the last weeks, and I'm only slowly returning to daily routine. Besides, since I don't have a lead on this bug, there's not a lot I can do. Currently, it does look like a hardware issue with the CPU, and if it is, we can probably work around it (we could work around the other CPU's problems), if we fined the cause.

Sorry, I would have preferred better news.

_________________
Thomas, the kernel guy

"I don't have a frigging clue. I'm norwegian" -- Ole-Egil

All opinions expressed are my own and do not necessarily represent those of Hyperion Entertainment

@DrBombcrater

Found another useful link

This might be of interest/use to some: the differences between the 750FX and GX

IBM official doc (pdf format)

@EntilZha

Quote:

The particular problem with the GX and iBrowse is that it is somewhat reproducable (I say somewhat because _I_ can't reproduce it).

JFYI, Stefan now has a Micro with a GX and can reproduce the problem, so hopefully it will be tracked down and resolved soon.

@EntilZha

Quote:

I had a difficult time personally for the last weeks, and I'm only slowly returning to daily routine.

Sorry to hear that I hope your problems get solved...

@Vader

That's great

Last edited by Crumb on 14-Apr-2005 at 02:01 PM.

_________________
The only spanish amiga news web page/club: CUAZ

@anarchic_teapot

Maybe if the manufacturer and dealers were a little more forthcoming
about these issues, the "rumors" wouldn't abound. Personally, if I
had some reliable press releases to read on this and other issues
I wouldn't be reading the unofficial reports.

Whether you call it a problem, defect or bug isn't the issue. The issue
is to have it acknowleded and a projected fix and when the fix will
happen. Maybe you can answer my question and help resolve this.

Will the GX processor work with DMA using OS4 and Linux??
DMA doesn't seem to work with either OS on my MicroA1 with GX processor.
Will the fix be more than just a software adjustment to get IBrowse
to work? If IBrowse has problems then some other OS4 or Linux apps
are likely to experience it.

Your comparison to various 68xxx processors on OS3 Amiga's is a little
inaccurate. On my Amiga's, newer processors worked with the OS and
applications compiled for older processors. You just need newer software
to get some of the benifits of the newer processor.

_________________
X1000 with 2GB memory & OS4.1FE

@Xenic

Quote:

Maybe if the manufacturer and dealers were a little more forthcoming about these issues, the "rumors" wouldn't abound

I think in an earlier thread, I said everything there is to say about the GX bug.

Quote:

Will the GX processor work with DMA using OS4 and Linux??

Sigh...

Why is it always necessary to drag DMA into each and every discussion ?

FYI, DMA has absolutely NOTHING to do with the GX bug. The GX bug is probably a L2 cache flush issue.

Quote:

If IBrowse has problems then some other OS4 or Linux apps are likely to experience it.

Maybe, maybe not. Do you actually experience these issues with other apps ?

Quote:

On my Amiga's, newer processors worked with the OS and applications compiled for older processors

No, they don't. Ever wondered what the point of 68040.library and 68060.library would be ? The 68060 is not compatible to old code since it's missing some instructions. These are emulated through processor exceptions once they are executed, or with programs like cyberpatcher they are patched over after they are executed for the first time.

So no, your idea of the magically working code is completely wrong. FYI, the MMU's of the different 68k CPUs are completely different. They use different data formats, different opcodes, etc.

OS4 uses a lot more of the CPU specific features than 68k-based AmigaOS ever did. The 68k OS never bothered to use the MMU (with a few exceptions), or L2 caches. OS4 does, and there's a lot of dedicated code for different CPU models, even different submodels of the same CPU. For example, the classics use 603e and 604e for Blizard and CSPPC, respectively. The 603e can not load TLBs for it's MMU tables automatically, this has to be done in software. All other CPU's do it in hardware.

Bottom line, Rose's comparison was straight to the point.

_________________
Thomas, the kernel guy

"I don't have a frigging clue. I'm norwegian" -- Ole-Egil

All opinions expressed are my own and do not necessarily represent those of Hyperion Entertainment

@EntilZha

Quote:

Why is it always necessary to drag DMA into each and every discussion ? FYI, DMA has absolutely NOTHING to do with the GX bug. The GX bug is probably a L2 cache flush issue.

Yes, he probably thinked this because you can't activate the DMA on linux with the GX-MicroA1 (with the FX cpu maybe yes, as someone said...).

But, as you said to me in another thread... (so I will save you some hassle to repeat this a thousand times... )

Quote:

if you're using kernel 2.4.x, this one doesn't really support the GX.. only 2.6 does. There is AFAIK a patch so it support the GX, but who knows what else is still to be done to make it work. Fact is, we have no problem in OS4 with the µA1 and DMA (the IBrowse problem is definitely not related to DMA).

Bottom line : sadly the µA1-GX owners can't use linux (and os4 also, for now...) with full-dma despite having a "fixed" board...

_________________
-------------------------------
AmigaOS, the last hope...

@Zorro

BTW, do we have any official numbers of how many people have a GX instead of an FX in their µA1-C?

_________________
This weeks pet peeve:
Using "voltage" instead of "potential", which leads to inventing new words like "amperage" instead of "current" (I, measured in A) or possible "charge" (amperehours, Ah or Coulomb, C). Sometimes I don't even know what people mean.

@olegil

AFAIK, almost all...

I know of only two or three person that have the FX.

_________________
-------------------------------
AmigaOS, the last hope...

@Zorro
Just a note.

The 2.6 kernel does not support the 750GX. Like a patched
2.4 it only recognizes the revision number (pvr) of the
GX, and then builds for the FX. There is a compatability
mode but without a roadmap of the module it's difficult
to figure out which way to go.

@EntilZha

I think in an earlier thread, I said everything there is to say about the GX bug.

Sorry to say, but not every (potential) buyer is reading this website and certainly not every thread. However, I fail to see how that would be your "problem" as you are not involved in marketing or customer relations ?

Stil, I doubt it's a really big problem. Every PC has some kind of "bug" (Sorry Rose). They are either patched by a hardware revision, BIOS update or drive patch. Nothing uncommon. It's finding the cause of the problem that's difficult, the solving might be very minor, or not.

Last edited by Seer on 15-Apr-2005 at 04:05 PM.

_________________
~
Everything you say will be misquoted and used against you..
~