Buffee... It's alive!!

[lmimmfn]

Originally Posted by UberFreak

Getting really poor results trying to run WHDLoad demos.
Crashes and freezes, graphic glitches, audio glitches, demos running too slow or too fast...
I know some of these issues from my 060, but there I was able to solve most of them with changing WHDLoad slave settings.
On the V1200 I was not able to improve things, despite messing around with settings for hours.

I'm sure these things will get resolved eventually, the core is still being developed.
But for now, I still have to keep my 060 in use.

Not sure if it helps but the crashes/hanging were absolutely wrecking my head with the V1200, tried everything over several days and finally seems more stable now. I stopped using its onboard IDE and moved the CF card to mobo IDE. Played 2 hours of Apidya on Saturday vs before would hang after 2-3 mins of gameplay, loads of other games hanging at intro etc. working properly. May not be 100% but its a hell of a lot better , on latest core/coffin r58

Sorry for off topic.
Im really looking forward to seeing what Buffee can do, the more Amiga accelerator alternatives the better, especially long term.

[eXeler0]

Originally Posted by eXeler0 View Post
Given the execution speed of the 68882 it better be running at 3.2GHz
Seriously though, will we see the 68882 and 030 + MMU at the same time in the s.c. "Phase 2"? Or will that come in incremental updates?

Originally Posted by nikosidis

Why? This is emulation. The FPU will be very fast. Not like on a real 030 board.

Ok, you read that too fast. Missed the joke AND the actual question?

[britelite]

Originally Posted by nikosidis

There are no instructions in 060 that 030 with FPU do not have.

Move16

[roondar]

Originally Posted by britelite

Move16

I suppose we could create some sort of library to add the missing instructions to the 68030 through the illegal instruction trap vector. We could call it the 68060.library because it's adding the 68060's instructions to the 68030

Seriously though, I thought the MMU was also quite different between 030 and 040/060. And there may be a few more instructions that are 040+ only, though I'm not really an 040/060 expert.

[kipper2k]

[IMG]https://media.discordapp.net/attachments/836307559651606578/84424More piccies...




I had to resize it down a bit, didnt want to choke off the internet

looks pretty shiny, things are movin along



if you want to see it full size, click here... https://cdn.discordapp.com/attachmen...0/IMG_4791.jpg

Damn Fine!

[nikosidis]

Kipper2K
Whow!!! Now we are talking. TF1200 with Buffee

[Predseda]

I was watching AmigaBill's last stream

[kriz]

That looks great Kipper2k, keep on the good work!!

[eXeler0]

@kipper2k Thanx for sharing, looks great.
Are there any technical differences on this large 1200 board compared to the A500 version?

[SpeedGeek]

Originally Posted by britelite

Move16

Move16 can only offer a practical performance benefit with the Copyback cache. The 68030 uses a write-through cache. Move16 requires the source and destination addresses to have 16 byte alignment.

It's takes enough extra code just to deal with the 16 byte alignment restriction, so any performance advantage from the Move16 instruction (except for the Copyback cache) is easily negated. Not to mention, the source and destination address seldom ever do meet the alignment requirement.

[NovaCoder]

Originally Posted by nikosidis

Kipper2K
Whow!!! Now we are talking. TF1200 with Buffee

That's very cool, feels like history is being made

[eXeler0]

Originally Posted by NovaCoder

That's very cool, feels like history is being made

@NovaCoder are you (secretly) saying you are tempted to recompile/fix for 030 and see what happens on Buffee?

[britelite]

Originally Posted by SpeedGeek

Move16 can only offer a practical performance benefit with the Copyback cache. The 68030 uses a write-through cache. Move16 requires the source and destination addresses to have 16 byte alignment.

Doesn't change the fact that it's an instruction available on 040/060 but not on 030.

[SpeedGeek]

Originally Posted by britelite

Doesn't change the fact that it's an instruction available on 040/060 but not on 030.

Ok, so it doesn't change the fact that your point was still pointless. This instruction has no practical use on an 030, and that's the only point worth making.

[NovaCoder]

Originally Posted by eXeler0

@NovaCoder are you (secretly) saying you are tempted to recompile/fix for 030 and see what happens on Buffee?

I just think it's cool that the community is finally getting fast and relatively cheap accelerator options

[britelite]

Originally Posted by SpeedGeek

Ok, so it doesn't change the fact that your point was still pointless. This instruction has no practical use on an 030, and that's the only point worth making.

The original post claimed the 030+FPU has ALL the instructions that the 060 has, and my reply was only to show it was a false claim.

[frozen]

Too fast...

[sean_sk]

Wow that A1200 version looks great! So am I reading the board right that it will have fast ram? If so, how much?

[Warty]

Any new discoveries with the Buffee? Is it working? What works? What doesn't? What machines have you tried? Has anyone been able to get it working on other machines (Mac, Atari)?

[eXeler0]

Originally Posted by Warty

Any new discoveries with the Buffee? Is it working? What works? What doesn't? What machines have you tried? Has anyone been able to get it working on other machines (Mac, Atari)?

I'm not sure any beta units have actually been shipped. The first batch of alpha units only reached a select few devs.
I payed for one beta unit and I haven't really heard anything since but I see on Discord that its still an ongoing process.. (finalizing the hardware, shipping out beta units)
When the cat is out of the bag, I'm sure it will not go unnoticed.