Page 2 of 3

Re: Dhrystone benchmark QL vs Amiga

Posted: Thu Apr 14, 2022 11:49 am
by XorA
Peter wrote:
tofro wrote:
Peter wrote: Why should that matter for Dhrystone? It does not involve graphics, sound, etc. I can hardly imagine the Warp 1260 spends more than a few percent of it's CPU cycles for motherboard communications during the benchmark run.
Just as it matters to the QL (or the Q68): The video circuitry accesses the RAM concurrently with the CPU, thus makes memory accesses slower because the CPU has to wait for the memory accesses of the chipset to be finished.
No, because the DDR3 RAM of the Warp 1260 is separate (unlike QL/Q68). Note that slowed down video access by the CPU is not the issue here as graphics are not benchmarked. The DDR3 RAM bandwidth of the Warp is not affected by the video refresh happening the original Amiga mainboard.
The problem is I have never managed to get QDOS4Amiga running, it always crashes.

I will have another go though, I have a variety of Amiga's from 68000->68060!

Re: Dhrystone benchmark QL vs Amiga

Posted: Thu Apr 14, 2022 11:56 am
by Peter

Re: Dhrystone benchmark QL vs Amiga

Posted: Thu Apr 14, 2022 1:16 pm
by tofro
Peter wrote:
tofro wrote:
Peter wrote: Why should that matter for Dhrystone? It does not involve graphics, sound, etc. I can hardly imagine the Warp 1260 spends more than a few percent of it's CPU cycles for motherboard communications during the benchmark run.
Just as it matters to the QL (or the Q68): The video circuitry accesses the RAM concurrently with the CPU, thus makes memory accesses slower because the CPU has to wait for the memory accesses of the chipset to be finished.
No, because the DDR3 RAM of the Warp 1260 is separate (unlike QL/Q68). Note that slowed down video access by the CPU is not the issue here as graphics are not benchmarked. The DDR3 RAM bandwidth of the Warp is not affected by the video refresh happening the original Amiga mainboard.
Yep, that's because the Warp 1260 provides fast RAM instead of chip RAM - This all comes back to where you run your program (just like in the QL with an external memory expansion). You can have a Warp 1260 in your Amiga and still experience a slowdown (because you - or the system - decided to run the program in chip mem).

Re: Dhrystone benchmark QL vs Amiga

Posted: Thu Apr 14, 2022 2:11 pm
by Nasta
Given that the test is running on a 68060, I would expect it to be heavily cached, hence not very dependant on RAM speed...

Re: Dhrystone benchmark QL vs Amiga

Posted: Thu Apr 14, 2022 9:38 pm
by bwinkel67
XorA wrote:The problem is I have never managed to get QDOS4Amiga running, it always crashes.

I will have another go though, I have a variety of Amiga's from 68000->68060!
I haven't tried QDOS4Amiga but did get QDOS Classic running under WinUAE.

https://www.youtube.com/watch?v=olHIeWHg3Aw

Once I get my hard drive sorted out on the A600 I will try and run QDOS Classic on it (real hardware :-/)

Re: Dhrystone benchmark QL vs Amiga

Posted: Mon Apr 18, 2022 10:56 pm
by Peter
Nice. Do you think it makes sense to try sysinfo dhrystone versus QL dhrystone on it? Or would the emulation (JIT compiler?) make things meaningless?

Re: Dhrystone benchmark QL vs Amiga

Posted: Mon Apr 18, 2022 11:03 pm
by Peter
tofro wrote:You can have a Warp 1260 in your Amiga and still experience a slowdown (because you - or the system - decided to run the program in chip mem).
Thanks for explaining, although that would be a weird memory allocation for a dhrystone benchmark.
The comparison sysinfo versus QL dhrystone on the same Amiga box would probably show if things are reasonable.

Re: Dhrystone benchmark QL vs Amiga

Posted: Fri Jun 03, 2022 2:43 am
by bwinkel67
I ran the Dhrystone benchmark on a BBQL and it came out at 196. I don't quite know which compiler was used as this came from the bench.zip download form Dilwyn's site. There is also a dhryst21.zip but that wouldn't run on the BBQL (neither C68 nor GCC version) though I was able to run it on a emulator...maybe it's a memory thing.

I was able to compile the original 1984 Dhrystone source from Rick Richards with Digital 'C' SE after some changes (started with 148 errors :( ) but currently it's not doing anything (stuck in a loop). I'm guessing I screwed up a pointer derefence somewhere so next I have to debug. If I get it compiled, maybe I'll add a nice interface to it. I need to check how Amiga's SysInfo computes MIPS because it doesn't divide its Dhyrstones by 1757 as stated in the Dhrystone wiki.

Re: Dhrystone benchmark QL vs Amiga

Posted: Wed Jun 08, 2022 9:32 am
by Peter
It is not so important with which compiler the QL dhrystone benchmark was generated, as long as it runs on Amiga with QDOS Classic or Amiga QDOS. What I'm looking for: A comparison to sysinfo on the same Amiga hardware. Preferrably a 68060 based Amiga hardware, but 68040 system would also give a rough indication. Unfortunatly, I do not own an Amiga, so my hope was that someone else from this forum could try.

Re: Dhrystone benchmark QL vs Amiga

Posted: Wed Jun 08, 2022 9:45 am
by XorA
Peter wrote:It is not so important with which compiler the QL dhrystone benchmark was generated, as long as it runs on Amiga with QDOS Classic or Amiga QDOS. What I'm looking for: A comparison to sysinfo on the same Amiga hardware. Preferrably a 68060 based Amiga hardware, but 68040 system would also give a rough indication. Unfortunatly, I do not own an Amiga, so my hope was that someone else from this forum could try.
My A1200 board has just come back from repair so I am just trying to get my A500 out of the build queue to build A1200 it with 060 accelerator for you.