On 20/04/16 00:54, Ilia Mirkin wrote:
On Tue, Apr 19, 2016 at 5:52 PM, Martin Peres <martin.pe...@free.fr> wrote:
+ result = ((s64)info.arg[0] * 15625) >>
18;
+ result += ((s64)info.arg[1] * volt->speedo
* 15625) >> 18;
+ result += ((s64)info.arg[2] * temp *
15625) >> 10;
+ result += ((s64)info.arg[3] * volt->speedo
* temp * 15625) >> 18;
+ result += ((s64)info.arg[4] * volt->speedo
* volt->speedo * 15625) >> 30;
+ result += ((s64)info.arg[5] * temp * temp
* 15625) >> 18;
+ break;
Well, I can only say that these values got us really really close to what
nvidia does... On around 10 GPUs... So... until we know better, let's stick
to them.
Hopefully, we can figure out where this 15625 comes from. This patch is:
5^6. Really it's 10^6, but it's silly to multiply by 10^6 and then
divide by powers of 2. Instead that's factored out...
That makes sense. I must have missed this on the IRC discussion.
Major congrats to Karol and you for figuring out all of this!
@Ben: I really think this is the right thing to do, it really was solid
work for Karol's part.
_______________________________________________
Nouveau mailing list
Nouveau@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/nouveau
