I have a question about Permedia 2 support. SIV reports memory clock 100 MHz for mine and core clock 90 Mhz. PowerStrip 3.90 reports 80 MHz memory clock and 53.33 MHz core clock. Reviews on the Internet mention typically 83 MHz, but there were apparently boards with 90 MHz core clock. @red-ray - can you confirm, that clocks shown by SIV for the Permedia 2 are true?

GL1zdA wrote on 2023-02-21, 23:12:

can you confirm, that clocks shown by SIV for the Permedia 2 are true?

From the information you have provided not sensibly as for starters you failed to specify the PCI Vendor ID + Device ID and say what was reported for M + N + P + S .

Post a screen shot of Menu->Machine->GPU Info and I should be able to tell what is going on.

Update: The only Permedia 2 I have is in my DS20 and when I tried it in my Tyan S2466 Tiger MPX AMD Athlon MP 1900+ (Palomino) the system would not even post so I suspect it may be an Alpha special.

When possible SIV reads the Register Map values and calculates the clocks from those values, but on my DS20 the values are zero so SIV reports the default clock speeds.

Yeah, it was not the smartest question I've ever asked 😉 It's an ELSA GLoria Synergy, I'm attaching the screenshots - looks like all zeroes. Another question - what means 100% near "Resource Usage"?

SIV 5.64 released on 14-May-2022 is rather old and I feel you should update to at least 5.68 as things may be different.

I wonder what all the Register Map values are, press [ 0 - 12 - 0 ] and post that screen shot.

GL1zdA wrote on 2023-02-22, 15:44:

what means 100% near "Resource Usage"?

On pre XP systems it's the MEMORYSTATUS->dwMemoryLoad that GlobalMemoryStatus() returns. XP and later use GlobalMemoryStatusEx().

I feel it means you should get some more RAM !

Here's the screenshot. Yes, the system is low on memory for some testing, I was just curious what metric SIV uses.

GL1zdA wrote on 2023-02-22, 20:05:

Here's the screenshot.

Thank you, it's even less interesting than what's there on my DS20.

I wonder why there is no ROM BAR.

Is there anything else I can check? I know ELSA has written customized the BIOS for most of their cards, so maybe here's a problem?

Not really, as I recall around 27-Aug-2019 I found the attached and did my best to get SIV to report it.

red-ray wrote on 2023-02-23, 12:25:

Not really, as I recall around 27-Aug-2019 I found the attached and did my best to get SIV to report it.

Do I understand it correct: the parameters are in ROM and there's no ROM Base Address, so you can't find them? FWIW, HWiNFO 6.2.2 for DOS also reports ROM Base Address as not available.

I'm not a PCI expert, but I see in the docs CFGIndirectAddress and CFGIndirectDatain the PCI Registers, and they sound like something that would allow access to Region 0 that contains the parameters, no?

GL1zdA wrote on 2023-02-23, 22:19:

I see in the docs CFGIndirectAddress and CFGIndirectDatain the PCI Registers, and they sound like something that would allow access to Region 0 that contains the parameters, no?

AFAIK these are in PCI config space, so as below

I gave the Permedia 2 another try. It looks like all of these M/N/P parameters are write-only. I poked around, and their offsets never return anything. But I've seen in previous posts you already know that.

I've searched a bit and found some Permedia 2 drivers source code. The Windows driver from the DDK has the detection routine Permedia2GetClockSpeeds. It either does take the value from the registry or from the ROM. I found images of the ELSA Synergy BIOS, and they all have 50 MHz (it looks like they use BCD to encode the MHz value). I don't really know what they do there, it looks like a convoluted way to access the Permedia's ROM, not sure if SIV can do something like that.

One usable bit SIV could return is the SClk Sel from ChipConfig (070h). It basically says how Permedia's, its memory and PCI bus clock relate. In my case SClk = MClk / 2, which means the card has 100 MHz memory.

GL1zdA wrote on 2023-02-28, 23:18:

I poked around, and their offsets never return anything. But I've seen in previous posts you already know that.

And forgotten I knew! I had a quick look at the code and noted:

The Registry does not specify an override so read the chip clock speed (in MHz) from the Video ROM BIOS (offset 0xA in the BIOS)
NB. this involves changing the aperture 2 register so aperture better be completely idle or we could be in trouble; fortunately we only call this function during a mode change and expect aperture 2 (the FrameBuffer) to be idle

Given the NB I feel SIV can't safely access offset 0xA in the BIOS as it has no way to know it's idle, what do you think?

Is there a clock value in your registry?

I have lots of versions of the code on this system and it looks like they all have exactly the same HW_DEVICE_EXTENSION which contains ULONG ChipClockSpeed; + ULONG RefClockSpeed; Given this SIV could read both from the device object extension on Windows 2000 through to Windows 2003, I have no idea if this is possible on Windows 98/Me.

I just spotted as below so I suspect SIV could just read the speed from the registry, is the value there?

1    //
2    // record the final chip clock in the registry
3    //
4
5    SystemClock *= 100;
6    VideoPortSetRegistryParameters(HwDeviceExtension,
7                                   L"HardwareInformation.CurrentChipClockSpeed",
8                                   &SystemClock,
9                                   sizeof(ULONG));

I just searched and found:

L:\WINDDK\2128\src\video\miniport\3dlabs\video.c(931): L"HardwareInformation.CurrentChipClockSpeed",
L:\WINDDK\2195\src\video\miniport\3dlabs\VIDEO.C(931): L"HardwareInformation.CurrentChipClockSpeed",
L:\WINDDK\3790.1830\src\video\displays\3dlabs\perm2\mini\video.c(922): L"HardwareInformation.CurrentChipClockSpeed",
L:\WINDDK\3790.1830\src\video\displays\3dlabs\perm3\mini\video.c(482): L"HardwareInformation.CurrentChipClockSpeed",

So it's also there on the Permedia 3, does anyone have one? When I checked it's there on my DS20 Windows 2000 RC2 system with the value being 0x04C57DF4 (80,051,700 Hz).

Unfortunately, there's no value in the registry for the 98. I've forgotten, 9x uses a totally different driver than 2000 🙁

Another question. Does SIV report anywhere maximum sample rate for playback and capture of sound cards in Windows 9x? I was hoping MS System Info, DxDiag or DxCapsViewer would show me it, but no, there's no information. EDIT: Never mind, found what I was looking for in Devices/Audio Devices.

GL1zdA wrote on 2023-03-03, 21:25:

EDIT: Never mind, found what I was looking for in Devices/Audio Devices.

Good as I don't think I have ever fully tested it on Windows 9x as none of my test systems have sound cards!

I just looked at the panel on this system, felt it should be width resizable, so Beta-07 will be.

Back when I implemented the panel SIV did not do width resizing so I am adjusting panels as I spot them.

Can you check, whether the sample rates are correctly reported? SIV reports for the YMF718/YMF719 only: 11250 Hz, 22500 Hz and 44100 Hz. These cards also support 8, 16, 32 and 48 kHz. I'm attaching a screenshot from Cool Edit Pro with all the reported "modes".

GL1zdA wrote on 2023-03-04, 21:44:

Can you check, whether the sample rates are correctly reported?

By SIV or by Windows 98? Looking at the screen shot from my system you should see 48 KHz + 96 KHz are reported, so SIV will report them.

Below are the defined values from 1999 + 2003 R2 + what in the .H files.

As 48 KHz + 96 KHz did not exist in 1999 I suspect Windows 98 does not report them, further 48 KHz is not even in the latest on-line definition!

I assume the ones not defined are extensions specific to the YMF718/YMF719, is there an YMF718/YMF719 API ?

GL1zdA wrote on 2023-03-04, 21:44:

These cards also support 8, 16, 32 and 48 kHz.

What does the attached SIV32L 5.69 Wave-07 test SIV report please?

I can't seem to find an Enum function, do you know if there is one?

It looks exactly like what Cool Edit Pro reported, and it's what I've expected. I've tested it with WAVE files with each of these sample rates and all of them worked.

I was hoping, Yamaha driver is smart enough to not report the 8/16/32/48 kHz sample rates if the card doesn't have the second oscillator (I have several YMF71x based cards with small differences in implementation), but no, instead of refusing playback it will just play the files back with wrong pitch.

No, I don't know, what function would enumerate this. But I assume you've found the right one (or at least the same that Cool Edit Pro uses). I haven't found other programs reporting this, Sandra 99 only reports the maximum frequency and in this case it's 44.1, which would match the maximum frequency SIV reports in "Audio Devices".

GL1zdA wrote on 2023-03-05, 19:48:

But I assume you've found the right one.

Thank you for checking and no, I have not found the enumerator function, currently SIV has a table of rates and does wave opens specifying WAVE_FORMAT_QUERY for these.

It looks like it works, but I feel the current implementation is sub-optimal. I found https://microsoft.public.win32.programmer.mme … e-audio-devices and will ponder it.

It's also strange that WAVEOUTCAPS->dwFormats (the first rate) reports formats that can't be opened.

GL1zdA wrote on 2023-03-05, 19:48:

It looks exactly like what Cool Edit Pro reported

I tidied up the code and have now released SIV 5.69 Beta-07, does it still report as it should on your system please?

Basically it no longer reports the WAVEINCAPS->dwFormats or WAVEOUTCAPS->dwFormats information and on this system I now get: