another control API issue
the h-dsp has some 64 bit registers. the current control API doesn't offer a way of representing such values. shall i add
SNDRV_CTL_ELEM_TYPE_INTEGER64
?
_______________________________________________________________
Have big pipes? SourceForge.net is looking for download mirrors. We supply the hardware. You get the recognition. Email Us: [email protected]
Quoted 6 lines · Paul Davis
the h-dsp has some 64 bit registers. the current control API doesn't offer a way of representing such values. shall i add
SNDRV_CTL_ELEM_TYPE_INTEGER64?
Added, except for utilities.
Jaroslav
-----
Jaroslav Kysela <[email protected]>
Linux Kernel Sound Maintainer
ALSA Project http://www.alsa-project.org
SuSE Linux http://www.suse.com
_______________________________________________________________
Have big pipes? SourceForge.net is looking for download mirrors. We supply the hardware. You get the recognition. Email Us: [email protected]
Quoted 7 lines · Paul Davis
the h-dsp has some 64 bit registers. the current control API doesn't offer a way of representing such values. shall i add
SNDRV_CTL_ELEM_TYPE_INTEGER64?
Just for curiosity, what they store that cannot be represented in 32 bits?!?
Signature · 13 lines
Abramo Bagnara mailto:[email protected] Opera Unica Phone: +39.546.656023 Via Emilia Interna, 140 48014 Castel Bolognese (RA) - Italy ALSA project http://www.alsa-project.org It sounds good! _______________________________________________________________ Have big pipes? SourceForge.net is looking for download mirrors. We supply the hardware. You get the recognition. Email Us: [email protected]
Quoted 2 lines
Just for curiosity, what they store that cannot be represented in 32 bits?!?
RMS meters. It seems that the 64 bit-ness comes from it being some kind of fixed point representation - the value that you read from the registers need scaling (in user space, presumably) by a floating point transformation to actually become true RMS values. so its not so much they they need 64 bits of precision, more that the hardware needs to do the math with this type of data format. keep in mind the computation is all being done on a Xilinx FPGA, not a DSP chip, so the arithmetic options are (at this time) a bit limited.
--p
_______________________________________________________________
Have big pipes? SourceForge.net is looking for download mirrors. We supply the hardware. You get the recognition. Email Us: [email protected]
Quoted 4 lines · Paul Davis
Quoted 3 lines
Just for curiosity, what they store that cannot be represented in 32 bits?!?
Quoted 8 lines
RMS meters. It seems that the 64 bit-ness comes from it being some kind of fixed point representation - the value that you read from the registers need scaling (in user space, presumably) by a floating point transformation to actually become true RMS values. so its not so much they they need 64 bits of precision, more that the hardware needs to do the math with this type of data format. keep in mind the computation is all being done on a Xilinx FPGA, not a DSP chip, so the arithmetic options are (at this time) a bit limited.
I expected something similar. Don't you think that to have a FLOAT type is a better choice (more card independent, more accessible, more representative, etc.).
Signature · 13 lines
Abramo Bagnara mailto:[email protected] Opera Unica Phone: +39.546.656023 Via Emilia Interna, 140 48014 Castel Bolognese (RA) - Italy ALSA project http://www.alsa-project.org It sounds good! _______________________________________________________________ Have big pipes? SourceForge.net is looking for download mirrors. We supply the hardware. You get the recognition. Email Us: [email protected]
Quoted 3 lines
I expected something similar. Don't you think that to have a FLOAT type is a better choice (more card independent, more accessible, more representative, etc.).
Well, that was my initial thought. Then I remembered that we can't have floating point code in the kernel, and its therefore unsafe to even use this data type (an interrupt in kernel mode doesn't result in an FP register restore, and we may have run another task in the meantime that used the FP registers before we get back to our code).
In this particular case, even if it was safe to store the value in a float, the actual float value has to be computed with FP math, which is definitely not available in the kernel. so it wouldn't help much here.
Sound right?
--p
_______________________________________________________________
Have big pipes? SourceForge.net is looking for download mirrors. We supply the hardware. You get the recognition. Email Us: [email protected]
Quoted 11 lines · Paul Davis
Quoted 4 lines
I expected something similar. Don't you think that to have a FLOAT type is a better choice (more card independent, more accessible, more representative, etc.).
Well, that was my initial thought. Then I remembered that we can't have floating point code in the kernel, and its therefore unsafe to even use this data type (an interrupt in kernel mode doesn't result in an FP register restore, and we may have run another task in the meantime that used the FP registers before we get back to our code).
Quoted 6 lines
In this particular case, even if it was safe to store the value in a float, the actual float value has to be computed with FP math, which is definitely not available in the kernel. so it wouldn't help much here.
Sound right?
Not fully. You can use FPU in kernel space with some attention (disable preemption and save/restore FPU state for critical area).
Signature · 13 lines
Abramo Bagnara mailto:[email protected] Opera Unica Phone: +39.546.656023 Via Emilia Interna, 140 48014 Castel Bolognese (RA) - Italy ALSA project http://www.alsa-project.org It sounds good! _______________________________________________________________ Have big pipes? SourceForge.net is looking for download mirrors. We supply the hardware. You get the recognition. Email Us: [email protected]
Quoted 4 lines
Quoted 1 line
Sound right?
Not fully. You can use FPU in kernel space with some attention (disable preemption and save/restore FPU state for critical area).
Sure, you can do this, but if anyone from lkml gets a look at it, it won't last very long :)
--p
_______________________________________________________________
Have big pipes? SourceForge.net is looking for download mirrors. We supply the hardware. You get the recognition. Email Us: [email protected]