Hi All,
Does anyone know of a way to go down to just one processor running the armv8 FVP model?
This gets it down to 2 but this seems to be the minimum.
-C cluster0.NUM_CORES=1 \ -C cluster1.NUM_CORES=1 \
Trying to come up with ways to get a little more speed out the model.
Thanks!
On 02/04/2014 02:54 PM, Tom Gall wrote:
Hi All,
Does anyone know of a way to go down to just one processor running the armv8 FVP model?
This gets it down to 2 but this seems to be the minimum.
-C cluster0.NUM_CORES=1 \ -C cluster1.NUM_CORES=1 \
The VE model variant can do uniprocessor, and I would expect Foundation to as well.
Christopher
Hi Chris, All,
So how does one run uniprocessor?
If I set cluster1.NUM_CORES=0
the model will fail on startup with :
error in parameter 'cluster1.NUM_CORES' specification: value is out of range
If I remove it then I get one processor on cluster 0 and 4 processors on cluster 1!
Thanks, Tom
On Tue, Feb 4, 2014 at 2:05 PM, Christopher Covington cov@codeaurora.org wrote:
On 02/04/2014 02:54 PM, Tom Gall wrote:
Hi All,
Does anyone know of a way to go down to just one processor running the armv8 FVP model?
This gets it down to 2 but this seems to be the minimum.
-C cluster0.NUM_CORES=1 \ -C cluster1.NUM_CORES=1 \
The VE model variant can do uniprocessor, and I would expect Foundation to as well.
Christopher
-- Employee of Qualcomm Innovation Center, Inc. Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted by the Linux Foundation.
On 02/05/2014 11:04 AM, Tom Gall wrote:
Hi Chris, All,
So how does one run uniprocessor?
If I set cluster1.NUM_CORES=0
the model will fail on startup with :
error in parameter 'cluster1.NUM_CORES' specification: value is out of range
If I remove it then I get one processor on cluster 0 and 4 processors on cluster 1!
I do not know how to run uniprocessor on the Base model. I meant to suggest that you could potentially use a different model, such as the FVP_VE_AEMv8A (something of a precursor to the FVP_Base_AEMv8A_AEMv8A as I understand it) or Foundation_v8 as a work-around.
Christopher
On Tue, Feb 4, 2014 at 2:05 PM, Christopher Covington cov@codeaurora.org wrote:
On 02/04/2014 02:54 PM, Tom Gall wrote:
Hi All,
Does anyone know of a way to go down to just one processor running the armv8 FVP model?
This gets it down to 2 but this seems to be the minimum.
-C cluster0.NUM_CORES=1 \ -C cluster1.NUM_CORES=1 \
The VE model variant can do uniprocessor, and I would expect Foundation to as well.
Christopher
On 5 February 2014 08:15, Christopher Covington cov@codeaurora.org wrote:
On 02/05/2014 11:04 AM, Tom Gall wrote:
Hi Chris, All,
So how does one run uniprocessor?
If I set cluster1.NUM_CORES=0
the model will fail on startup with :
error in parameter 'cluster1.NUM_CORES' specification: value is out of range
If I remove it then I get one processor on cluster 0 and 4 processors on cluster 1!
I do not know how to run uniprocessor on the Base model. I meant to suggest that you could potentially use a different model, such as the FVP_VE_AEMv8A (something of a precursor to the FVP_Base_AEMv8A_AEMv8A as I understand it) or Foundation_v8 as a work-around.
I had somewhat similar problem but from other angle: in DS-5 for V8 I could connect only to one cluster at the time, so to do any meaningful Linux kernel debugging I had to make sure that kernel runs only on one cluster. As it was discussed on this thread I could not do anything with model command line options to disable second cluster. I ended up to setting 'maxcores=4' in Linux kernel cmd line (had to change such default in uefi and rebuild uefi). That worked for me - kernel runs only on the first cluster after that.
Tom, you could try to do 'maxcores=1' or 'maxcores=0' that would restrict kernel to run one CPU of one cluster, but I am not sure that it will really help fastmodels from speed point of view.
Thanks, Victor
Christopher
On Tue, Feb 4, 2014 at 2:05 PM, Christopher Covington cov@codeaurora.org wrote:
On 02/04/2014 02:54 PM, Tom Gall wrote:
Hi All,
Does anyone know of a way to go down to just one processor running the armv8 FVP model?
This gets it down to 2 but this seems to be the minimum.
-C cluster0.NUM_CORES=1 \ -C cluster1.NUM_CORES=1 \
The VE model variant can do uniprocessor, and I would expect Foundation to as well.
Christopher
-- Employee of Qualcomm Innovation Center, Inc. Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted by the Linux Foundation.
linaro-dev mailing list linaro-dev@lists.linaro.org http://lists.linaro.org/mailman/listinfo/linaro-dev
-
Christopher Covington -
Tom Gall -
Victor Kamensky