vol
-
Content Count
145 -
Joined
-
Last visited
Posts posted by vol
-
-
2 hours ago, mdivancic said:I used V6, was that the latest version? The TT has a CaTTram accelerator which boost the CPU frequency to 48 mhz. I did not run PI-ST30.TOS at 32 mhz, didn’t see any reason to do so. Yes the cpu is a 68030.
The latest version is on github. However the speed must remain the same as with V6. It was not easy to find any information about the CaTTamaran accelerator - I has been able to find only this page in Polish dedicated to it - is it your card? You wrote "a CaTTram accelerator" so I have some doubts. I assume that this accelerator doesn't have TTRAM but your system may have it. Could you provide this detail too?
The reason to run PI-ST30 was in my request.
I want to understand timings of the 68030 - it is not easy. It seems I need to start a new topic about oddities of these timings. I really can't understand while PI-ST and PI-ST30 show the same time. My current guess is the MULU.W instruction may be faster than 28 cycles on the 68030 in some cases.
-
2 hours ago, mdivancic said:OK, Here my runs.
[email protected] w/normal boot (NVDI). PI-ST.TOS:
100 - 0.02
1000 - 1.34
3000 - 11.66
[email protected] w/normal boot (NVDI). PI-ST30.TOS:
100 - 0.03
1000 - 1.34
3000 - 11.67
[email protected] w/clean boot. PI-ST.TOS:
100 - 0.05
1000 - 1.63
3000 - 12.54
[email protected] w/clean boot. PI-ST30.TOS:
100 - 0.05
1000 - 1.63
3000 - 12.54
So I see no difference between the two programs.
For Reference. [email protected] w/clean boot. PI.ST.TOS:
100 - 0.06
1000 - 2.08
3000 - 16.56
Thank you very much. Please inform me what exactly is your TT system @48MHz? What card does it use?
Did you run PI-ST30.TOS on the [email protected]?
It is very strange for me that there is no difference between the two programs. Let's look at the next table.
68020 68030 moveq.l #0,d0 0-2-3 2-2 move.l d0,d1 0-2-3 2-2 lsl.l #3,d0 1-4-4 4-4 sub.l d0,d1 0-2-3 2-2 add.l d0,d0 0-2-3 2-2 sub.l d0,d1 0-2-3 2-2 sub.l d0,d1 0-2-3 2-2 lsl.l #8,d1 1-4-4 4-4 sub.l d1,d0 0-2-3 2-2 2-22-29 22-22 mulu d1,d0 25-27-28 28-28PI-ST30 uses MULU-optimization, it uses a longer sequence of instructions instead of MULU (D1=10000). Timings clearly show that we must get at least 6 cycles gain for the 68030... Does your system use the 68030?
-
On 5/19/2021 at 11:27 AM, apersson850 said:Then we have the p-code card, which in addition to having 12 K ROM (in the DSR space, >4000 - >5FFF, where the top 4 K is paged by CRU bit >1F80) also has 8 GROMS. These all have 6 K memory, but the GROM base is also inside the DSR space. Thus it's right in the ROM space, but it's fully decoded, so it only occupies the words it really uses. The code in the ROM simply jump over these addresses.
Thus a total of 60 Kbytes of memory on that card.
It seems that TI had plans to replace GPL with p-code... P-code was widely used on 8-bit systems, for example there were very good Basic and Pascal compilers into p-code.
-
Pi-pack #55 has been released and the tables are updated!
It doesn't contain more speed optimized programs for the Atari ST. So for the benchmarks the attached programs may be used.
BTW I am curious about NVDI. Is it popular among the Atari ST users? Is it compatible with all Atari software?
On 5/17/2021 at 5:02 AM, DarkLord said:Okay, here are my results (still on my STacy, equipped with a Pak 68/3 board running at 40mhz, FPU 40mhz, 4 megs of ST RAM, TOS v3.06)
Pi-ST TOS:
100 - .04
1000 - 1.59
3000 - 13.02
Pi-ST30 TOS:
100 - .04
1000 - 1.58
3000 - 12.96
Thank you very much. Your results have corrected much my estimations. PI-ST30 shows better results than PI-ST!
It is sad that there are still no other results.
It seems I am out of luck here. Maybe I should go to Atari Forum... People there helped me to get results from the Falcon. However they ran only PI-ST30 and skipped PI-ST. I am still gathering results for the Atari Falcon (PI-ST), TT (STRAM and TTRAM), MegaSTE @16Mhz, other popular Atari hardware based on 68030+ released before 1997. Any help is welcome! Thanks a million in advance.
-
An Amiga guy has been able to do an almost impossible thing and find a way to make the algorithm implementation faster a little. He also helped to prove that the 68020/30 code is actually faster than the 68000 code now. So please help me again. Run pi-st and pi-st30 on your hardware for me (100/1000/3000 digits). I am sure that further improvements are impossible so this is rather my last request.
-
10 hours ago, ParanoidLittleMan said:Hurra ! Instead Amiga vs ST war we have now TT vs Paked Stacy war 🙂 Sorry, could not resist 🤩
Serious mode on: That PAK 68/3 in Stacy - is it using regular ST video/graphic HW or there is some added graphic module, card - like Nova ?
The Atari ST is just a bit faster than the Amiga but the Amiga has its advantages too.
It is interesting that this resembles the situation with the Atari 800 and Commodore 64. History repeated itself. It is also interesting that the Apple II and first Macs were slower than the Commodore 64 and Amiga and had less impressive graphics/sound than the Atari 800 and Atari ST but Apple beat Commodore and Atari.
-
3 minutes ago, ParanoidLittleMan said:I did not mean that Atari's 50066 Hz was bad for sound quality or like - just that was not compatible.
I agree that this frequency is very unusual, I don't know of any other system which uses it.
-
6 hours ago, ParanoidLittleMan said:So, what's about this is ? About measuring performance of some computer, or how to make faster, better optimized code ?
And, this 'really interesting' talk - sorry, but this are actually trivial things. Using Falcon CT60 in 68030 mode - of course that it is necessary - otherwise it will not run big %-age of regular Falcon SW. And not always because different CPU, sometimes some SW works not just because execution speed is much higher than for what is made and tested on.
[...]
So, concrete example: 68000 and 68030 were designed before 80386, and no way that later is only 10% faster at same clock. And as I know 68030 used external cache (too), so much depends from it's size and speed (something ignored here completely by vol) .
[...]
It is just my project which can show which processor is better for the number pi computation using the easiest known algo.
The idea is quite simple - make the best code for every CPU tested. IMHO results in table #2 look quite interesting. Indeed it is not a perfect general benchmark, it is only one algorithm implementation testing.
I remember my friend had some business around Atari computers because they were popular among musicians. He had Falcons in 1993 but I never touched them.
So it is very interesting for me to get results for the π computation from this computer which has a very unusual architecture, the 68030 on the 16-bit data-bus.
The 68030 appeared in 1987 and the 80386 in 1985, so the 68030 was designed a bit later. It is still unclear what CPU is actually faster. We have very few benchmark results for both CPUs.
Every processor has its own set of advantages: the 68030 has more registers, an ability to execute several instructions at once, higher frequencies but the 80386 has its advantages too: faster memory access cycle, instant EA calculation, much faster division.
3 hours ago, DarkLord said:Atari STacy, 4 megs of ST RAM, TOS v3.06, [email protected], [email protected]
100 = .04
1000 = 1.59
3000 = 13.02
Hope this helps.
Thank you very much. Your Atari is the fastest in my tables that have just been updated. It is interesting that the 68020 and 68030 show almost identical performance.
-
14 hours ago, moulinaie said:Here are the results:
Atari TT/32MHz with ST Ram
100 = 0.05
1000 = 2.01
3000 = 15,96
Atari TT/32MHz with TT Ram
100 = 0.05
1000 = 1,89
3000 = 14,93
MegaSTE/16MHz
100 = 0.13
1000 = 8.23
3000 = 70.73
Thank you very much! The table is updated. It is really interesting, is it possible to use the Falcon CT60 in the 68030 mode?
BTW it is also interesting that your Super-Pi is much faster (35%) on the 68030 than on the 80386 but my Pi-Spigot shows that the 80386 is slightly (10%) faster. Maybe I will try to dig into your x86 code but I am not sure when.
-
Amiga people helped me to find two optimizations for my code. In particular they suggested to replace MULU with a sequence of 9 instructions which is faster than MULU for the 68020/30!
Would you like please to run for me `pi-st.tos' one last time for 100, 1000, and 3000 digits? I am seeking results from the Atari MegaSTE @16MHz, Atari TT, Atari TT + TTRAM, Atari Stacy Pak/3, Falcon and other Ataris which use 68030.
Indeed results from the 68040 and 68060 are very interesting but my project is rather for hardware released before 1995.
I have read that it is possible to use the Atari Falcon CT60 in the 68030 mode...
There is no need to run `pi-st30.tos' this time because it should show the same performance as `pi-st.tos'.BTW I have already updated tables but I had to use estimates and this is pretty bad for the accuracy of the data.
Thank you.
-
10 hours ago, Atariforce said:The main difference is that the ST version sucks. It requires 1 MB of RAM, is way too slow, looks like crap and is virtually unplayable without a hard disk, unless you want to swap disks every 10 seconds. But yes, beyond the technical shortcomings a great game is hiding even though I prefer the previous 2 episodes.
IMHO Ultima VI was the last true Ultima. Ultima VII has much better graphics but it is corrupted somehow. Ultima VIII became even slightly more corrupted - a man wrote a good text about it. And Ultima IX is a real horror which is almost impossible to play.
-
On 4/29/2021 at 9:54 AM, ParanoidLittleMan said:I was about to add here text about what everything has influence on benchmark results, and not only that - overall speed of running SW. But, it will be long, so better start new thread - will appear soon in programming section.
Please wait. Amiga people showed me that there are ways to make my code faster. I am working under it now...
-
I played it on the [email protected] PC without hdd but I used RAM-disk because I had 2 MB RAM. It was fast. There several fan-made recreation of Ultima VI but I didn't try to play them.
-
It is a very good game. I played on the IBM PC clone in 1991. Is there any difference between the Atari ST and IBM PC version?
Sorry I was a pirate and didn't have manual so I had to find a proper answer myself. It was easy. I just made memory dump and sought the question in it. The answer was always near the question.
-
BTW I have added the Commodore VIC-20 to my table recently. It is more than 4 times slower for the number pi computation than the TI99/4A.
-
2
-
-
6 hours ago, moulinaie said:Hello,
I made the tests again without the TT RAM (TT 32MHz, ST RAM):
pi-st.tos : 15,97 sec
pi-st30.tos : 18,95 sec
Guillaume.
Thank you. It seems that you run PI-ST v1 or 2. However the timings must be the same, only PI-ST30 must become faster. MasterMotorola's result is slightly slower, 16.61 s.
Now I only need the Falcon results and maybe several screenshots.
-
Thank you very much. However it is an unexpected surprise for me that the Atari ST OS doesn't have a standard shell.
Modern OS (Microsoft Windows, Mac OS, Linux, ...) have a wide variety of shells... TOS is so similar to MS-DOS that it was difficult to imagine that they officially ignored shells completely.
I have just checked Bconout - it doesn't work with redirection.
Gulam doc claims that this shell supports redirection but it also has a disclaimer Note that because of TOS and compiler peculiarities, not all external
commands will behave as above. I couldn't make the redirection to work for external commands under Gulam. I have checked maybe about dozen other shells and some of them are claimed as supporting redirection. But no-one actually works. What a strange impediment!
-
27 minutes ago, Elia Spallanzani fdt said:Well, in theory one could calculate the whole configuration and show only a 64x48 window, which could scroll. Just saying.
Xlife-8 has zoom in mode which is 40x24, and zoom out (hires) mode. The zoom in mode does scrolling but it is not suitable for some operations.
17 minutes ago, 9640News said:Not knowing anything about your program, speed could be different dependent upon a user's configuration. It looked like you were using the Video XOP for display. Users may be running Video Fast or Slow, the program running in slow or fast ram, and whether wait states are On or Off.
It is all not important for 3000 digit calculation when screen output takes less than 1% of all processing time. You can separate CPU and IO timings using FORMAT option for the table. This way allows me to show information about both the CPU raw power and the speed of character output procedure. The latter is not exactly correct for the TI99/4A which system just missed such a procedure and this makes it a rare exception among all computers I know.
-
4 hours ago, MasterMotorola said:Here are the stats running on my Atari TT030 @ 32Mhz...
Pi-TOSv3: 0.06, 2.09, 16.61 (100, 1000, 3000 places)
Pi-TOS30v3: 0.06, 2.09, 16.61 (100, 1000, 3000 places)
SuperPi: 0.01, 0.85, 7.51 (100, 1000, 3000 places)
There is a CaTTamaran installed which is supposed to run the 030 at 48MHz but I got the same numbers so it much be malfunctioning or something is physically out of place. 😕
Thank you very much. The table is updated. Your data is in the 11th row there. Sorry I don't have a proper screenshot to add for this entry.
-
1
-
-
On 4/27/2021 at 1:18 AM, DarkLord said:Okay, I downloaded the latest version of Pi-ST and re-ran it with 100, 1000, and 3000 as the settings.
Things are still not what they should be though. Scores are pretty much identical in either mode, with the
68k version being slightly faster. That's not right.
68000 @ 100, 1000, and 3000:
Sorry for the low quality of the pictures but they should be good enough for you
to evaluate the results.
Hope this helps.
Thank you very much! The table is updated. The entry for the Atari Stacy Pak68/3 has position #8 in this table now. One of your screenshots is a part of this entry. Indeed, a bigger screen area would be a bit better there.
I believe people who send results for me, I only ask them for a screenshot because it is good illustrative material for the table.
I am sure that your result that the 68000 version is faster than the 68030 version was a kind of a random accident. The 68030 version is faster but only about fractions of a percent.
6 hours ago, DarkLord said:Not so sure about it being less efficient? According to Chris Swinson's GemBench 6, a well known and recognized benchmarking standard, the Pak 68/3 board in my
STacy is faster than a TT (as it should be).
I ran a test just now and even with the TT with Alt (FAST) RAM, the Pak 68/3 wins except in one category and I'm assuming that's because the TT has better RAM
access (is it 32 bit in ALT-RAM?). Perhaps some knowledgeable TT owners can tell us.
Indeed your Pak68/3 is faster but it uses 40 MHz clock while the TT has 32 MHz. The Pak68/3 utilizes its CPU less efficient than the TT because it is less than 25% faster. I took 25% because of 40/32=1.25. Maybe it is because the Pak68/3 doesn't have fast RAM. This is confirmed by MasterMotorola's results whose TT doesn't have fast RAM and shows itself exactly 25% slower than the Pak68/3. However moulinaie's results for his super-pi shows that the Pak68/3 @50MHz is only slightly (1%) faster than the Atari TT without fast RAM...
-
11 hours ago, 9640News said:100 0.30
1000 20.99
3000 183.69
Real hardware, Geneve.
Thank you very much. The table is updated. The Geneve 6940 has position #54 in this table now. This position has a screenshot I got from MAME...
-
14 hours ago, vol said:No, moulinaie's programs also don't work with redirection
It seems it is much more difficult problem then I could think. So I started a new topic for it.
-
I am not able to make my program to send its data to a file via redirection. My code is very simple. I put it into a file named hello.asm:
move.l #helloworld,-(A7) move #9,-(A7) trap #1 addq.l #6,A7 move #0,-(A7) trap #1 helloworld: dc.b "Hello, World!",$d,$a,0I compile it into hello.tos by vasm:
vasmm68k_mot -m68000 -Ftos -align -o hello.tos -L hello.lst -nosym helloIt works fine from CLI (emucon2 or gulam) but I failed to redirect the program output to a file.
hello >filedoesn't work.
Is it possible to write a TOS program which output can be redirected like for CLI utilities or under MS-DOS? There is a way because CLI utilities work with redirection, for instance
dir > fileor
ls > filework fine under emucon2 or gulam respectively.
-
14 hours ago, Elia Spallanzani fdt said:go with multicolor mode! 16 free colors.
Indeed it has but its resolution is too low, look XLife-8 screenshots or videos - it uses 160x200 or 320x200 raster. It is possible to make a port for 256x192 graphics but I need 4 free colors.
Like us on Facebook
Follow us on Twitter
AtariAge on Instagram
Watch us on YouTube
Chat on Discord
Benchmarking...
in Atari ST/TT/Falcon Computers
Posted
Thank you for this information. It seems that mdivancic's CaTTamaran shows rather odd results. I suspect that the CaTTamaran somehow affects normal timings. So results from your hardware would be very interesting.
No, please don't disable cache! My pi-spigot implementation is small enough to fit the internal 68020/30 cache.
Thank you. However your results are very strange. Your previous results for the TT [email protected] for PI-ST30 was 11.67 (NVDI) or 12.54s. How does it become 10.99s now?! Or you disabled TT RAM the previous time, didn't you? I also can't understand why your results from the plain [email protected] is slower than old moulinaie's results? So if it is possible please run PI-ST and PI-ST30 on the same machine configuration. I can suggest the next list of configurations:
1) [email protected];
2) [email protected] TT RAM;
3) [email protected];
4) [email protected] TT RAM.
I have also attached the newest binaries to this post but the speed must be the same as for v6.
pi-st-8-beta.zip