Thanks for your bench-marc-ing. Was very informative. I am going to continue to stay on the sidelines as this matures. I have severao other Amiga ambitions to keep me occupied in the meantime. Thanks!
Kevin Saunders-NEMESIS!! :) I kid. You both are great. Thanks for the video and shout-out Doug! It really shows all this clearly. There’s another update coming soon that looks to max out chip-ram speeds. Fingers crossed.
Yes, for some reason there is an entire group of people who just never wanted to admit to a single shortcoming on the PiStorm, but instead just irrationally defend it. Inexperienced that from the first release where I mentioned that it was in fact, not a $15 accelerator and never was, to the audio issues with many, many games and the slow CHIP RAM access. All these things have been addressed one way or another, but if everybody just pretended they did not exist, they never would have been fixed.
@@10MARC totally agree - been on the receiving end of it myself. Seems all many folks want to do is run sysinfo and see an ever longer red line…rather than a genuinely faster Amiga. IMHO PiMiga offers a better experience than the PiStorm right now.
@@10MARC but the PiStorm boards, when first released, were £11 each to make. That's a fact you cannot argue with so not sure why you always say otherwise. Myself and other Discord members were making them and selling them for that much before the parts shortages bumped the prices. Yes you needed a Pi to use it, but then you also need an Amiga. Should you count the Amiga in the price too. Anyway, the PiStorm board, on its own, were available for £11 each.
@@10MARC I guess I see your point a little in that the "complete" PiStorm setup was never $15, if that's what you mean, but the actual PiStorm on its own was at one point being sold for......... £11
I think your benchmark test came too early. They are still working on the chip ram speed. I saw memory benchmarks of unreleased versions of EMU68. Writespeed to chip showed 7.1 MB!!! But the software is not ready yet.
No, my Benchmarks were not too early. If that was the case I could never do any videos at all! I do mine based on what is commonly available at the time of filming - hence the original tests based on what was available when I made my first video noting the issue last month and my new test based on the "current" release. PiStorm has always been a "well, it may have problem X now, but just you wait and see" type of product and so far many, but not all issues have been addressed. This is perfectly acceptable as it is really a labor of love, but people who use it need to know the truth about it, not some Halcyon image of what it may become.
Thank you Doug! I like your videos :) full of technical amiga knowledge - not like some other youtuber who only tests his whdload games with this high performance cards.
@@10MARC Thats true, for me it make complete no sense to review a Pistorm or a BFG or a Vampire if you only test R-Type or Turrican with it. Cards like that are not designed for this games :D But Doug, i'm glad to have found your channel with the real shit :) go on and keep the spirit alive :)
I remember playing with 1280 width mode in DPaint on the newly released A4000 and this video looks like about the same performance as I saw way back then.
The PiStorm in those Hi Res modes is absolutely slower, by quite a bit. It's not easy to notice with the naked eye in 640x480 but higher modes I can see it right away. The memory tests prove it though. PiStorm chugs when dealing with CHIP RAM, especially when stressed. The new update supposedly helps but I have not tried it yet
@@10MARC Display DMA is the absolute killer. I think that's why your benchmark numbers for bustest look low. If you ran bustest while the display was set to 640x200 4 colours, chipram access speed would improve by a lot.
Yeah chip ram has always sucked donkey balls for CPU access, theoretically the Pi 4 GPIO can do between 650Mbps and 3.25Gbps which *should* be enough overhead once the SW for the Pi storm is sorted to get closer to the limits of chip ram. Theoretical max for 1280x720x8@60hz would be ~5MB/s if I got that right off the top of my head so the 4MB/s or so on the 060 seems to confirm that... Funny to think how much slower the Chip ram speed is than the 113MB/s I can get off my internet connection :)
It is a nice improvement. 640x400x256 colors seems pretty close to the speed of my 68040 now. The higher res modes still choke on the PiStorm, but only nerds like me use those modes. They are faster than before. At 1280x720x 256 I used to be able to watch the pixels redraw, but now they refresh at a fair clip. Again, not the speed they do with a regular accelerator, but acceptable
Thanks for the update Doug 👍🏻 I still need to put together my PiStorm32 system. Good to know that this is still being worked on. Do you know if the PCMCIA issues were fixed? What’s it like with a Elbox Mediator system?
I have had reasonable luck with PCMCIA on mine. My data transfer works with it, and my CNET networking card seems OK. I exclusively use EMU68K, though... Maybe the other branch has more issues?
With Beta2 that came out yesterday I've noticed better improvement in games. Hybris now works perfectly, no more music speeding up, nothing I tried previously worked. Also, not just games, Demos are improved such as Audio - X it now displays perfectly (previously was completely glitched out) sadly music still missing channels/speed issues.
My BFG doesn't get perfect chipram speed unless running 100MHz. You will not get 7MB/s bustest results unless you run (practically) without a screen on (or at least something like 1 or 2 bitplanes 320 wide PAL/NTSC).
100% true. I tested in 640 x 480 256 color mode intentionally as I wanted to stress things out as bit., My BFG is at least 30% faster than my PiStorm in these tests, at first and about 15% faster after.
On my 1200+030 i needed to speed up AGA with FBLIT and FTEXT and the latest ICON library to make the Workbench usable. Why not use these enhancements with the PiStorm?
Because FBlit and FText are hacks that fo against the development guidelines and can cause issues with the PiStorm. Due to this they are not recommended. Even the OS3.1.4 and OS3.2 FAQs advise against using them for this reason
I have not used FBLIT in years. I do use Picasso96 with the "Native" driver to free up CHIP RAM, and it seems to work OK. These hacks do not solve the slower CHIP RAM access, though.
Agreed on that. This was a true, live test of the performance and not rehearsed in advance. I literally did the update while filming the video. Going from 32 seconds to render down to 28 was a real improvement, though. Animating an image would have been a more accurate test
th-cam.com/video/7Dv_Zx4KI70/w-d-xo.html I run BusTest Chip from RTG Workbench with AGA displaying a blank grey screen, and the benchmark reached 5.1 MB/s writem.
Sure. blanking the screen will certainly speed up the results - but we don't blank the screen when moving AGA images around. It needs to be nice and fast with the chips actually in use.
@@10MARC 5.1 MB/s is about 28% slower than the max 7.1 MB/s. I have seen bustest chip benchmark results from GVP/TekMagic A4000T T-REX II/060 chip writem 7.1 MB/s All tests from no startup-sequence, Call Setpatch, then BusTest >ram: A3000's chip ram rivals AGA's chip ram performance, it's too bad A3000 doesn't have an AGA chipset. chip writem 6.9 MB/s The poster is "thebajaguy" from abime's forum.
In many cases those are better choices. If I were using my Amiga exclusively for gaming I would not even consider a PiStorm at all, but stick with a 68030 solution. Vampire is certainly more "stable" when it comes to performance. PiStorm can be faster with some things like FPU emulation and raw speed, but slower in other things.
Thanks for your bench-marc-ing. Was very informative. I am going to continue to stay on the sidelines as this matures. I have severao other Amiga ambitions to keep me occupied in the meantime. Thanks!
It is improving, which I am happy about.
Kevin Saunders-NEMESIS!! :)
I kid. You both are great. Thanks for the video and shout-out Doug! It really shows all this clearly. There’s another update coming soon that looks to max out chip-ram speeds. Fingers crossed.
Hey! It's Kevin Bacon himself! or maybe Kevin Hart? I get confused...
@@10MARC lol! FREE KEVIN!
@@HoldandModify Not related to the British Musician “Suzie Quattro”? 🤔
@@cullmaster7361 Actually , yes. She’s a distant aunt of mine.
Thanks for keeping us updated! Guess I got some downloading and installing to do tomorrow.
Yes, the new update is totally worth it. Maybe more improvements can be made in the future. I am happier with it now.
Thanks Doug you are the man. Cheers
I hope you found the video useful!
Thanks Doug, nice to see someone doing some rational tests on this device!
Yes, for some reason there is an entire group of people who just never wanted to admit to a single shortcoming on the PiStorm, but instead just irrationally defend it. Inexperienced that from the first release where I mentioned that it was in fact, not a $15 accelerator and never was, to the audio issues with many, many games and the slow CHIP RAM access.
All these things have been addressed one way or another, but if everybody just pretended they did not exist, they never would have been fixed.
@@10MARC totally agree - been on the receiving end of it myself. Seems all many folks want to do is run sysinfo and see an ever longer red line…rather than a genuinely faster Amiga. IMHO PiMiga offers a better experience than the PiStorm right now.
@@10MARC but the PiStorm boards, when first released, were £11 each to make. That's a fact you cannot argue with so not sure why you always say otherwise. Myself and other Discord members were making them and selling them for that much before the parts shortages bumped the prices.
Yes you needed a Pi to use it, but then you also need an Amiga. Should you count the Amiga in the price too.
Anyway, the PiStorm board, on its own, were available for £11 each.
@@10MARC I guess I see your point a little in that the "complete" PiStorm setup was never $15, if that's what you mean, but the actual PiStorm on its own was at one point being sold for......... £11
I think your benchmark test came too early. They are still working on the chip ram speed. I saw memory benchmarks of unreleased versions of EMU68. Writespeed to chip showed 7.1 MB!!! But the software is not ready yet.
No, my Benchmarks were not too early. If that was the case I could never do any videos at all! I do mine based on what is commonly available at the time of filming - hence the original tests based on what was available when I made my first video noting the issue last month and my new test based on the "current" release.
PiStorm has always been a "well, it may have problem X now, but just you wait and see" type of product and so far many, but not all issues have been addressed. This is perfectly acceptable as it is really a labor of love, but people who use it need to know the truth about it, not some Halcyon image of what it may become.
Thank you Doug! I like your videos :) full of technical amiga knowledge - not like some other youtuber who only tests his whdload games with this high performance cards.
Everyone has their specialties, I suppose. Thanks fort the kind words!
@@10MARC Thats true, for me it make complete no sense to review a Pistorm or a BFG or a Vampire if you only test R-Type or Turrican with it. Cards like that are not designed for this games :D But Doug, i'm glad to have found your channel with the real shit :) go on and keep the spirit alive :)
I remember playing with 1280 width mode in DPaint on the newly released A4000 and this video looks like about the same performance as I saw way back then.
The PiStorm in those Hi Res modes is absolutely slower, by quite a bit. It's not easy to notice with the naked eye in 640x480 but higher modes I can see it right away. The memory tests prove it though. PiStorm chugs when dealing with CHIP RAM, especially when stressed. The new update supposedly helps but I have not tried it yet
@@10MARC Display DMA is the absolute killer. I think that's why your benchmark numbers for bustest look low. If you ran bustest while the display was set to 640x200 4 colours, chipram access speed would improve by a lot.
Yeah chip ram has always sucked donkey balls for CPU access, theoretically the Pi 4 GPIO can do between 650Mbps and 3.25Gbps which *should* be enough overhead once the SW for the Pi storm is sorted to get closer to the limits of chip ram. Theoretical max for 1280x720x8@60hz would be ~5MB/s if I got that right off the top of my head so the 4MB/s or so on the 060 seems to confirm that... Funny to think how much slower the Chip ram speed is than the 113MB/s I can get off my internet connection :)
Another great video.
Thanks so much! I appreciate it.
Woohoo a new nightly build ready for testing 😀 released 1 hour ago.
Thanks for the pains-taking effort to benchmark this for thousands of we subscribers!
Interesting results, really. The HAM8 gradients performed better than I thought they would.
The release candidate should be a big improvement 🎉
It is a nice improvement. 640x400x256 colors seems pretty close to the speed of my 68040 now. The higher res modes still choke on the PiStorm, but only nerds like me use those modes. They are faster than before. At 1280x720x 256 I used to be able to watch the pixels redraw, but now they refresh at a fair clip. Again, not the speed they do with a regular accelerator, but acceptable
Thanks for the update Doug 👍🏻 I still need to put together my PiStorm32 system. Good to know that this is still being worked on. Do you know if the PCMCIA issues were fixed? What’s it like with a Elbox Mediator system?
I have had reasonable luck with PCMCIA on mine. My data transfer works with it, and my CNET networking card seems OK. I exclusively use EMU68K, though... Maybe the other branch has more issues?
With Beta2 that came out yesterday I've noticed better improvement in games. Hybris now works perfectly, no more music speeding up, nothing I tried previously worked. Also, not just games, Demos are improved such as Audio - X it now displays perfectly (previously was completely glitched out) sadly music still missing channels/speed issues.
Glad to hear about the improvements! I will be testing it out this weekend and see if it solves some of my issues.
My BFG doesn't get perfect chipram speed unless running 100MHz. You will not get 7MB/s bustest results unless you run (practically) without a screen on (or at least something like 1 or 2 bitplanes 320 wide PAL/NTSC).
100% true. I tested in 640 x 480 256 color mode intentionally as I wanted to stress things out as bit., My BFG is at least 30% faster than my PiStorm in these tests, at first and about 15% faster after.
lol @ 9:35 1.7 minutes i havnt done the math on that but as uk person .... it just sounds wired 😄
On my 1200+030 i needed to speed up AGA with FBLIT and FTEXT and the latest ICON library to make the Workbench usable. Why not use these enhancements with the PiStorm?
Because FBlit and FText are hacks that fo against the development guidelines and can cause issues with the PiStorm. Due to this they are not recommended.
Even the OS3.1.4 and OS3.2 FAQs advise against using them for this reason
I have not used FBLIT in years. I do use Picasso96 with the "Native" driver to free up CHIP RAM, and it seems to work OK. These hacks do not solve the slower CHIP RAM access, though.
I think the gradient ball in DPaint have no significance on AGA speed, the CPU power is the bottleneck in that test.
Agreed on that. This was a true, live test of the performance and not rehearsed in advance. I literally did the update while filming the video. Going from 32 seconds to render down to 28 was a real improvement, though. Animating an image would have been a more accurate test
th-cam.com/video/7Dv_Zx4KI70/w-d-xo.html
I run BusTest Chip from RTG Workbench with AGA displaying a blank grey screen, and the benchmark reached 5.1 MB/s writem.
Sure. blanking the screen will certainly speed up the results - but we don't blank the screen when moving AGA images around. It needs to be nice and fast with the chips actually in use.
@@10MARC 5.1 MB/s is about 28% slower than the max 7.1 MB/s.
I have seen bustest chip benchmark results from GVP/TekMagic A4000T T-REX II/060
chip writem 7.1 MB/s
All tests from no startup-sequence, Call Setpatch, then BusTest >ram:
A3000's chip ram rivals AGA's chip ram performance, it's too bad A3000 doesn't have an AGA chipset.
chip writem 6.9 MB/s
The poster is "thebajaguy" from abime's forum.
thanks for info now I know what I will not buy so better stick to terriblefire, vampire.
In many cases those are better choices. If I were using my Amiga exclusively for gaming I would not even consider a PiStorm at all, but stick with a 68030 solution. Vampire is certainly more "stable" when it comes to performance. PiStorm can be faster with some things like FPU emulation and raw speed, but slower in other things.
Ewww, Windows 10, I work on this every day, makes me want to vomit
Windows 11 on that laptop. I needed one to test it on, so I use it on my capture laptop. It works OK I suppose - but it is not fun like an Amiga