Page 3 of 5

Re: Q68 speed vs 680X0

Posted: Tue Nov 27, 2018 10:05 pm
by Peter
My 80 MHz Q60 takes 18 seconds for 2000000.

Re: Q68 speed vs 680X0

Posted: Tue Nov 27, 2018 10:14 pm
by stephen_usher
Derek_Stewart wrote:
stephen_usher wrote:On an Atari TT030 (30MHz 68030) it takes 127 seconds.

Unfortunately SMSQ/E doesn't run in fast RAM so I can't test it with the uncontended EDO RAM only the slow RAM shared with the video sub-system.
Hi Stephen,

Are you using QVME in the TT030?
No, just the Atari SMSQ/E downloaded from the development web site.

Re: Q68 speed vs 680X0

Posted: Tue Nov 27, 2018 10:17 pm
by FrancoisLanciault
Peter wrote:My 80 MHz Q60 takes 18 seconds for 2000000.
:shock:
That's fast!

François

Re: Q68 speed vs 680X0

Posted: Tue Nov 27, 2018 11:34 pm
by Nasta
Peter wrote:My 80 MHz Q60 takes 18 seconds for 2000000.
That is very interesting.
If Q68 takes 47.8 secs at 40MHz from SRAM, let's compare this with 060@80Mhz - it is approximately 2.66 times faster, or, in other words, the Q60 would have to run at 106.2MHz to get the same result. What is interesting is, if we assume Q68 is close to 1 instruction per cycle, the 060 has dual instruction execution, so in theory if all instructions were independent of one another, it could do close to two per cycle. Instead, it 'only' manages 1.33 times Q68 performance, which effectively means it manages to do two instructions per cycle every 4 cycles roughly. This seems low, though it actually shows what one can expect from 'random' code that does not attempt to exploit this feature.

Re: Q68 speed vs 680X0

Posted: Tue Nov 27, 2018 11:58 pm
by FrancoisLanciault
stephen_usher wrote:On an Atari TT030 (30MHz 68030) it takes 127 seconds.

Unfortunately SMSQ/E doesn't run in fast RAM so I can't test it with the uncontended EDO RAM only the slow RAM shared with the video sub-system.
But if we assume the code is so small it should run from cache, the RAM speed should not matter... And if we take the ratio of 56mhz over 30mhz and apply the result to the 71 sec it takes the Amiga, your Atari is in fact performing quite well.

François

Re: Q68 speed vs 680X0

Posted: Wed Nov 28, 2018 7:31 am
by Peter
Sorry, I forgot to check in which cache mode the Q60 was. I'll report again when I get to it.

Re: Q68 speed vs 680X0

Posted: Wed Nov 28, 2018 8:17 am
by vanpeebles
I wonder what my Sharp x68000 XVI could do in 16mhz mode, would be interesting.

Re: Q68 speed vs 680X0

Posted: Wed Nov 28, 2018 7:26 pm
by Derek_Stewart
vanpeebles wrote:I wonder what my Sharp x68000 XVI could do in 16mhz mode, would be interesting.
Have you got SMSQ/E running on the Sharp X68000?

Re: Q68 speed vs 680X0

Posted: Wed Nov 28, 2018 7:28 pm
by Derek_Stewart
FrancoisLanciault wrote:
stephen_usher wrote:On an Atari TT030 (30MHz 68030) it takes 127 seconds.

Unfortunately SMSQ/E doesn't run in fast RAM so I can't test it with the uncontended EDO RAM only the slow RAM shared with the video sub-system.
But if we assume the code is so small it should run from cache, the RAM speed should not matter... And if we take the ratio of 56mhz over 30mhz and apply the result to the 71 sec it takes the Amiga, your Atari is in fact performing quite well.

François
Hi,

I have aN Atari TT030 and QVME, I would hope the use of the QVME Graphics card would speed up the TT speed, I will have look at this when time permits.

But it is looking like the Q68 is killing everything dead except for a Q60.

Re: Q68 speed vs 680X0

Posted: Wed Nov 28, 2018 8:02 pm
by vanpeebles
Derek_Stewart wrote:
vanpeebles wrote:I wonder what my Sharp x68000 XVI could do in 16mhz mode, would be interesting.
Have you got SMSQ/E running on the Sharp X68000?
No! :lol: