Polaris 10 (also codenamed Ellesmere) found on "Radeon RX 470" and "Radeon RX 480"-branded graphics cards
Polaris 11 (also codenamed Baffin) found on "Radeon RX 460"-branded graphics cards (also Radeon RX 560D)
Polaris 12 (also codenamed Lexa) found on "Radeon RX 550" and "Radeon RX 540"-branded graphics cards
Polaris 20, which is a refreshed (14 nm LPP Samsung/GloFo FinFET process) Polaris 10 with higher clocks, used for "Radeon RX 570" and "Radeon RX 580"-branded graphics cards
Polaris 21, which is a refreshed (14 nm LPP Samsung/GloFo FinFET process) Polaris 11, used for "Radeon RX 560"-branded graphics cards
Polaris 22, found on "Radeon RX Vega M GH" and "Radeon RX Vega M GL"-branded graphics cards (as part of Kaby Lake-G)
Polaris 23, which is a refreshed (14 nm LPP Samsung/GloFo FinFET process) Polaris 12, used for "Radeon Pro WX 3200" and "Radeon RX 540X"-branded graphics cards (also Radeon RX 640)
Polaris 30, which is a refreshed (12 nm LP GloFo FinFET process) Polaris 20 with higher clocks, used for "Radeon RX 590"-branded graphics cards
I think i get it. There are only three base GPU models...Polaris 10, 11 and 12
All "newer" or "revised cards" are based off of these models. The gfx driver we have is only checking for and displaying/using the underlying base GPU models. That's why later cards like my 590 show a lower GPU model.
I wonder if those "revised/updated" GPU models features stuff that is not used by the driver (but would boost them nonetheless) hence being slower than higher based GPU models albeit being later in the model range.
AMD's naming scheme has always been a bit weird and confusing. The high-end Polaris models are indeed Polaris10 (or 20/30),** and the low-end are Polaris 11 & 12. High-end cards are always released first because they want to show off the performance of the new GPU series. The low-end cards come later.
To confuse things more, the RX 590 came later. It appears to be a higher-clocked variant of the RX 580.
No idea why your RX 590 is performing so poorly. If you have the debug version of the driver, then you could record a log and send it to me (privately). I can have a look if there are any hints there.
Hans
** While wikipedia may list Polaris 10, 20, & 30, AMD's own drivers see them all as Polaris10. I'm guessing that they're all basically the same GPU series, but with slight differences (e.g., 12nm fabrication process vs 14nm).
Anything i need to perform so you might get a usable log? Or is using the debug driver enough?
Should i let it run the gfx2d benchmark?
Running the GfxBench2D benchmark might get some interesting power management info (don't upload the results, though). The rest of the info is printed during startup.
Quote:
Not part of the above, but i get a lot of these with the RX560 RadeonRX (0): Could not create a timer object, using ITimer->MicroDelay() instead
What are you running that triggers that? I've only ever got that with AudioEvolution 4. That'll be gone with the next driver release...
Having spent so much time, effort and money into these things and having even spent the money to convert the single RAM stick in the X5000 to two seperate ones etc, etc the score of the X5 seems particularly dissapointing.
Are they bad, and if so, how to improve them?
===
SDLBench using SysMon shows that each of my systems, X1 X5, are pretty much on par.
===
GPmark shows for X5: 225, 89, 101, 102, 100, 20, 23 ; with X1 half of those values apart from the last two which are the same.
===
I think that RadeonRX.chip @ 2,7 is correct for my X5000 system. I have not bought Radeon v5 for either machine (yet).
===
Doom3 runs okay on this system.
===
Perhaps I should not read into the results too much. But, if I am doing something wrong then let me know.
Edited by rjd324 on 2023/3/3 15:10:10 Edited by rjd324 on 2023/3/3 15:38:15
If liberty means anything at all, it means the right to tell people what they do not want to hear. George Orwell.
Why do you think they're appalling? What's your reference point to compare?
Quote:
Having spent so much time, effort and money into these things and having even spent the money to convert the single RAM stick in the X5000 to two seperate ones etc, etc the score of the X5 seems particularly dissapointing.
Are they bad, and if so, how to improve them?
Your X5000 is let down by the MemCopy scores. Compare them side-by-side, and you'll see that the X1000's WritePixelArray/ReadPixelArray are DMA accelerated, but the X5000's are not. That results in a very steep performance penalty (especially ReadPixelArray). This in turn hurts the random test, because each of the operations are run randomly, including WritePixelArray & ReadPixelArray.
There's nothing you can do about it. I expect an updated graphics.library will be released at some point that will include DMA accelerated WritePixelArray() & ReadPixelArray() for the X5000.
Semi-related: I noticed a few weird results at the top of the AmigaOS GfxBench2D charts. A Pegasos II with a (crappy) Silicon Motion SM502? I wasn't aware that the SM502 was available as a plug in card for the Pegasos II. It's sitting at the top due to unbelievably high MemCopy, FillRect, BlitRect & OverlappedBlitRect scores, and missing compositing scores (IIRC, the SM502 isn't capable of compositing). The MemCopy numbers efar exceeds AGP/PCI bandwidth, so I'm guessing that it's actually doing the MemCopy in RAM. FillRect may be higher due to using a 16-bit screen (which halves the bandwidth).
EDIT: Actually, those MemCopy performance test results exceed the Pegasos II's RAM bandwidth as well. What's going on?
EDIT2: Are these results from an emulated Pegasos II + SM502? The SM502 in the Sam460 is very slow (link). Emulated hardware would make more sense, given the suspect details (e.g., "Unknown" processor).
Hans
Edited by Hans on 2023/3/4 5:19:20 Edited by Hans on 2023/3/4 5:30:44 Edited by Hans on 2023/3/4 5:32:13
Compare them side-by-side, and you'll see that the X1000's WritePixelArray/ReadPixelArray are DMA accelerated, but the X5000's are not. That results in a very steep performance penalty (especially ReadPixelArray). This in turn hurts the random test, because each of the operations are run randomly, including WritePixelArray & ReadPixelArray.
X5000 do have DMA accelerated graphics.library too : this was added by Salas00 2 years ago or something (you may remember i offer 1000$ for this in beta list), but I do not remember if this graphics.library were in the latest os4 public updates through …
Yes, I know. It looks like the X5000 DMA enabled version hasn't been released yet (rjd324 clearly doesn't have it), and I have no idea when it'll be available to end-users.
I added A1222/X5000 DMA support to graphics.library in January 2022 (versions 54.250/54.251). In order to work it also needs a newer kernel for the fsldma.resource (if a too old kernel is detected the DMA support code is not used).
Scoring twice as much in WritePixelArray and about 50x more in ReadPixelArray.
But, I have no idea who is a beta tester on there, perhaps you cab tell by the library versions. I am not so perhaps there is something they have that I don't.
If liberty means anything at all, it means the right to tell people what they do not want to hear. George Orwell.
EDIT2: Are these results from an emulated Pegasos II + SM502?
All CPUs supported by the Pegasos II (750, 74xy) have an L2 cache, the results on your page have 0 bytes L2 cache. Definitely some kind of emulation, or fake results.
News feed
OS4 Feedback forum
OS4Depot Feedback forum
Amigabounty forum
OpenAmiga forum
AmiCygnix forum
ABC shell forum
AmiKit forum
Cinnamon Writer forum
CodeBench forum
Digital Universe forum
Dopus 5 forum
E-UAE forum
Gnash forum
Ibrowse forum
JAmiga forum
Odyssey forum
OWB forum
Qt forum
SmartFileSystem forum
Timberwolf forum
TouchDevice forum
TuneNet forum
Unsatisfactory Software forum
Donate
