Messages in this thread | | | Date | Mon, 18 May 1998 15:19:58 -0100 | From | André Derrick Balsa <> | Subject | Re: Cyrix 6x86MX and Centaur C6 CPUs in 2.1.102 |
| |
Hi Alan,
Alan Cox wrote: > > > IMHO this is a trade-off. If one wants to have the processor dissipating > > < 150mW when Linux is idle, one has to power down the 200MHz or so > > 64-bit TSC counter circuitry. > > The Centaur doesnt have this bug and still has a low power mode. They do > it right. So it is a bug in the Cyrix chip.
The Centaur C6, just like the Intel CPUs, enters a low power mode when the CPU executes a HALT instruction (i.e. when Linux is idle). However, it is not the same low power mode as found in the Cyrix 6x86 CPUs.
Just like the Intel and the AMD K5/K6 chips, the Centaur C6 will still dissipate around 2.5 Watts in this low power mode.
OTOH when the clock is _stopped_ (e.g. APM), the Centaur, just like the Intel and the AMD K5/K6, will dissipate around 350 mW. However, coming out of a clock stop condition is expensive in terms of clock cycles, because the CPU internal clock PLL must re-synchronize with the external clock input.
The Cyrix 6x86 is unique in its ability to power down when Halted and wake up again without a single CPU clock cycle loss.
(Context) switching to another subject: I already emailed Centaur about the workaround, and also asked about a sample CPU for Linux testing, but didn't get an answer yet. You mentionned you wanted to try some optimizations for the C6.
I have a question. The 6x86MX allows one to lock L1 cache lines. Would it be interesting to group some kernel variables and keep them in the L1 cache? Would that increase the kernel performance during context switches?
Please give me a hint of which data structures could be kept in the L1 cache, and their size.
Cheers, ------------------------ André Balsa andrebalsa@altern.org
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu
| |