MSTE vs STE TOS206

Benchmark screenshots for various boosters & machines.
User avatar
exxos
Site Admin
Site Admin
Posts: 4625
Joined: Wed Aug 16, 2017 11:19 pm
Location: UK
Contact:

Re: MSTE vs STE TOS206

Post by exxos » Tue Jul 31, 2018 8:11 pm

mstenb.jpg
mstenb.jpg (102.39 KiB) Viewed 261 times
So this shows speed with 16MHz with cache... without cache speeds are almost no improvement vs stock STFM. So I think 16MHz is only 16MHz to cache.

But blitter disabled is bad anyway, very low scores compared to STE with blitter on.

So yes, machine runs a lot faster without blitter off... but then system has huge slow down because blitter is not enabled. Enabled blitter and you are back to slow results from first post. So blitter cannot be cached as in case of blitter cache is useless.

So games will likely be faster if not using blitter, cache can give about 40% boost in ST-RAM speed. So compared to STFM , MSTE will beat easily. But for desktop or games using blitter, MSTE is not much faster than normal STE. True we have cache on ST-RAM, so in real world likely help speed up some things.

Even so, MSTE is not true 16MHz machine... disable cache and blitter and MSTE doesn't much beat normal STFM in terms of speed. MSTE likely will be better for games than STE but only because of cached ST-RAM. So I will award MSTE for that ;) But machine is not true 16MHz and my first post shows that and in case of GB6 results (desktop GEM stuff) its actually about same speed as STE.
4MB STFM 1.44 FD- VELOCE+ 020 STE - 4MB STE 32MHz - STFM 16MHz - STM - MEGA ST - Falcon 030 CT60 - Atari 2600 - Atari 7800 - Gigafile - SD Floppy Emulator - PeST - HxC - CosmosEx - Ultrasatan - various clutter

https://www.exxoshost.co.uk/atari/ All my hardware guides - mods - games - STOS
https://www.exxoshost.co.uk/atari/last/storenew/ - All my hardware mods for sale - Please help support by making a purchase.

Petari
Software Moderator
Software Moderator
Posts: 567
Joined: Tue Nov 28, 2017 1:32 pm

Re: MSTE vs STE TOS206

Post by Petari » Tue Jul 31, 2018 9:02 pm

Because blitter works always at 8 MHz no sense to cache it. I think that it was discussed here already.
Surely not true 16 MHz, but I don't think that it was ever called so - by Atari.
Compare it with Falcon: Falcon CPU can 16 MHz too, but cache is only some 512 bytes. Additionally, data bus is same - only 16 bits. So, Falcon is faster only because 68030 is more efficient. But in some cases it is actually slower. Ask Klapauzius about - he claimed that it is slower than ST in many cases.
Additionally, MSTE cache will work with much more SW than 68030 cache - what often needs to be disabled. MSTE cache maybe in max 1% cases.
I really don't care much for Desktop and GEM app speed. So much that never used NVDI. My Ataris spend most time executing diverse SW code, not TOS code. And speed gain is evident.
There is 2 kind of people: one thinking about moving to Mars after here becomes too bad, the others thinking about how to keep this planet habitable.

User avatar
exxos
Site Admin
Site Admin
Posts: 4625
Joined: Wed Aug 16, 2017 11:19 pm
Location: UK
Contact:

Re: MSTE vs STE TOS206

Post by exxos » Tue Jul 31, 2018 9:04 pm

Yeah 030 cache on falcon I saw actually can slow down things when enabled.. I wrote about that somewhere before also..

I thought of another test to try.. will post later...
4MB STFM 1.44 FD- VELOCE+ 020 STE - 4MB STE 32MHz - STFM 16MHz - STM - MEGA ST - Falcon 030 CT60 - Atari 2600 - Atari 7800 - Gigafile - SD Floppy Emulator - PeST - HxC - CosmosEx - Ultrasatan - various clutter

https://www.exxoshost.co.uk/atari/ All my hardware guides - mods - games - STOS
https://www.exxoshost.co.uk/atari/last/storenew/ - All my hardware mods for sale - Please help support by making a purchase.

User avatar
exxos
Site Admin
Site Admin
Posts: 4625
Joined: Wed Aug 16, 2017 11:19 pm
Location: UK
Contact:

Re: MSTE vs STE TOS206

Post by exxos » Tue Jul 31, 2018 9:37 pm

Using Frank's blitter demo... viewtopic.php?f=25&t=50

16MHZ CACHE ON

Code: Select all

Blitter render                $1B bobs
CPU shift  blitter active     $11 bobs  
CPU pre-shift blitter active  $1C bobs

16MHZ CACHE OFF

Code: Select all

Blitter render                $1A bobs
CPU shift  blitter active     $0B bobs  
CPU pre-shift blitter active  $13 bobs
8MHZ CACHE OFF

Code: Select all

Blitter render                $1A  bobs
CPU shift  blitter active     $0A  bobs
CPU pre-shift blitter active  $13  bobs
Will ignore 8MHz tests as basically no difference..

So cache on vs cache off..

Code: Select all

Blitter render                 +1 bob
CPU shift  blitter active      +6 bobs
CPU pre-shift blitter active   +9 bobs
So can see CPU operations are faster because of caches.

In terms of unfair test with 40MHz STE (no caches)

Code: Select all

Blitter render                $1B bobs
CPU shift  blitter active     $0C bobs  
CPU pre-shift blitter active  $16 bobs
So results faster than stock machine

Code: Select all

Blitter render                 +1 bob
CPU shift  blitter active      +2 bobs
CPU pre-shift blitter active   +3 bobs
4MB STFM 1.44 FD- VELOCE+ 020 STE - 4MB STE 32MHz - STFM 16MHz - STM - MEGA ST - Falcon 030 CT60 - Atari 2600 - Atari 7800 - Gigafile - SD Floppy Emulator - PeST - HxC - CosmosEx - Ultrasatan - various clutter

https://www.exxoshost.co.uk/atari/ All my hardware guides - mods - games - STOS
https://www.exxoshost.co.uk/atari/last/storenew/ - All my hardware mods for sale - Please help support by making a purchase.

Petari
Software Moderator
Software Moderator
Posts: 567
Joined: Tue Nov 28, 2017 1:32 pm

Re: MSTE vs STE TOS206

Post by Petari » Wed Aug 01, 2018 8:43 am

Btw. blitter is slightly slower on Mega STE - there are some extra wait states because VME bus.
And it may be that TOS works little faster from ROM than RAM - talking about case of same speed of both - because ROM must not keep dividable by 4 cycles instruction execution pattern as RAM (because of sync. with video generation) - so by instructions like dbf some extra waits may be added when running from RAM .
There is 2 kind of people: one thinking about moving to Mars after here becomes too bad, the others thinking about how to keep this planet habitable.

KyleB
Posts: 99
Joined: Tue Jul 31, 2018 7:08 pm

Re: MSTE vs STE TOS206

Post by KyleB » Thu Aug 02, 2018 11:07 am

exxos wrote:
Tue Jul 31, 2018 8:11 pm
Even so, MSTE is not true 16MHz machine... disable cache and blitter and MSTE doesn't much beat normal STFM in terms of speed. MSTE likely will be better for games than STE but only because of cached ST-RAM. So I will award MSTE for that ;) But machine is not true 16MHz and my first post shows that and in case of GB6 results (desktop GEM stuff) its actually about same speed as STE.
Well you're looking at it from your own perspective, where you've made lots of your own boosters, all no ram cache but you also made very fast ROM access.

The very serious question is how much the performance difference between your designs and MSTE w/cache changes when NVDI is running. Or some real world tests of practical applications of a computer, like calculating a massive atariworks spreadsheet or etc.

User avatar
exxos
Site Admin
Site Admin
Posts: 4625
Joined: Wed Aug 16, 2017 11:19 pm
Location: UK
Contact:

Re: MSTE vs STE TOS206

Post by exxos » Thu Aug 02, 2018 11:38 am

I'm not disputing the cache, I proved it with the bobs, no need to do more tests. Everyone knows caches are good thing. I am saying this 16MHz machine isn't running at 16MHz in all cases. If it was, it would equal the speeds of my V2.2 booster easily. If my V2.2 16MHz booster had the MSTE cache, it would run rings around the MSTE in all tests. So something isn't right somewhere with MSTE design.

As Peter mentioned. Have to turn blitter off to see the speeds. With blitter ON, something screws up badly and cripples the machine. All that "16MHz speed" is simply lost when blitter is running. Though again, in my V2.2 tests, things only get faster at 16MHz with blitter enabled, not slower.

I can only assume machine runs huge amounts of time in 8MHz when blitter is active and assume the cache is actually turned off, so the speeds equal a stock STE when using blitter. Those results show up in my first post results. GB6 doesn't test every scenario, its why I did bobs tests. Clearly shows 3x faster in bob count when cache enabled.

I'm not going to spend huge amounts of time investigating it, someone else can do that if they wish. Of course all depends on where you look for the speed. GB6 shows really bad speeds sure, but bobs show really good speeds. I maybe will add a simple "bob test" routine in GB6 at some point then it will show huge increase in speed. Currently ST-RAM shows 139% so GB6 does show caches in play there, but have no effect when blitter active.
4MB STFM 1.44 FD- VELOCE+ 020 STE - 4MB STE 32MHz - STFM 16MHz - STM - MEGA ST - Falcon 030 CT60 - Atari 2600 - Atari 7800 - Gigafile - SD Floppy Emulator - PeST - HxC - CosmosEx - Ultrasatan - various clutter

https://www.exxoshost.co.uk/atari/ All my hardware guides - mods - games - STOS
https://www.exxoshost.co.uk/atari/last/storenew/ - All my hardware mods for sale - Please help support by making a purchase.

Petari
Software Moderator
Software Moderator
Posts: 567
Joined: Tue Nov 28, 2017 1:32 pm

Re: MSTE vs STE TOS206

Post by Petari » Thu Aug 02, 2018 12:16 pm

Folks, I pointed here in some errors done in approach. Tests with blitter on have no sense, because blitter runs not faster than 8 MHz. And actually it is even little slower than in STE, in case of MSTE. Where MSTE is faster is pure CPU power. But main RAM and other chips are same as in STE, only section CPU-cache is what works at 16 MHz, when cache hit occurs. I think that CPU runs all time at 16 MHz when is set, only that adds wait states when accessing main RAM or some chip. That's easy to check. And not really relevant how concrete slowdown happens, what matters is performance with usual SW, as is pointed.
Saying that runs most time at 8 MHz contradicts to test results, where MSTE at 16 MHz is about 60-70% faster. That means that it runs actually some 65% time at 16MHz . Of course, all depends from running SW - it may be more or less .

And something interesting about speed of TOS functions. There is thread with Ijor at Atariage where it came up.
ST design with shared RAM access for CPU and video means that CPU cycle count for instructions must be always dividable with 4. And most of instructions are such. dbf is not such, and then 2 wait states are added after execution - from RAM. But if it runs from TOS ROM it must be not added - for sure not if there is no RAM access soon. So, I came on idea to compare speed with GemBench6 for same TOS v. running from RAM and ROM.
TOS 2.06. Comparing with TOS 1.04, blitter NP - well, I did not see ref. TOS 2.06 with blitter off in GB6, and blitter must be off, of course.

VDI txt: ROM 100%, RAM 99%
GEM Window: ROM 111%, RAM 111% *
Justified txt: ROM 100%, RAM 99%
VDI scroll: ROM 100%, RAM 99%
GEM Dialog: ROM 103% , RAM 103% *
VDI gr. pr. : ROM 100%, RAM 97%

As may see in most running from RAM is 1% slower. Some are same speed (may be just because even in case of ROM TOS it runs actually from RAM) . One is 3% slower. In average it is about 1% - and that's of course not noticeable.
Here must to remind on speed comparing of TOS 1.04 and 2.06, done some months ago - but I can not find now that thread, where 2.06 was declared as slower in almost all tests. Here, it is faster in GEM operations *, and not slower in any test I done. Of course in normal usage - from ROM.
I will do same test with 1.04 from ROM and RAM after program EPROMs with 1.04 ...
There is 2 kind of people: one thinking about moving to Mars after here becomes too bad, the others thinking about how to keep this planet habitable.

User avatar
exxos
Site Admin
Site Admin
Posts: 4625
Joined: Wed Aug 16, 2017 11:19 pm
Location: UK
Contact:

Re: MSTE vs STE TOS206

Post by exxos » Thu Aug 02, 2018 12:41 pm

You probably thinking of this thread ? viewtopic.php?f=7&t=45&p=3956#p3956

I mention VDI is slower in TOS206 than TOS104.
4MB STFM 1.44 FD- VELOCE+ 020 STE - 4MB STE 32MHz - STFM 16MHz - STM - MEGA ST - Falcon 030 CT60 - Atari 2600 - Atari 7800 - Gigafile - SD Floppy Emulator - PeST - HxC - CosmosEx - Ultrasatan - various clutter

https://www.exxoshost.co.uk/atari/ All my hardware guides - mods - games - STOS
https://www.exxoshost.co.uk/atari/last/storenew/ - All my hardware mods for sale - Please help support by making a purchase.

Petari
Software Moderator
Software Moderator
Posts: 567
Joined: Tue Nov 28, 2017 1:32 pm

Re: MSTE vs STE TOS206

Post by Petari » Thu Aug 02, 2018 12:51 pm

exxos wrote:
Thu Aug 02, 2018 12:41 pm
You probably thinking of this thread ? viewtopic.php?f=7&t=45&p=3956#p3956

I mention VDI is slower in TOS206 than TOS104.
With blitter on only, to be precise. And not only VDI is what is little slower. Vblank takes more time in 2.06 .
Someone should do concrete blitter speed comparison test(s) - MSTE vs. STE vs. MST (or some ST with blitter) .
There is 2 kind of people: one thinking about moving to Mars after here becomes too bad, the others thinking about how to keep this planet habitable.

Post Reply