r 2016 09:08:20 +0100
perf/x86/mbm: Add memory bandwidth monitoring event management
Includes all the core infrastructure to measure the total_bytes and
bandwidth.
We have per socket counters for both total system wide L3 external
bytes and local socket memory-controller bytes. The OS doe
bandwidth monitoring event management
Includes all the core infrastructure to measure the total_bytes and
bandwidth.
We have per socket counters for both total system wide L3 external
bytes and local socket memory-controller bytes. The OS does MSR writes
to MSR_IA32_QM_EVTSEL and MSR_IA32_QM_CTR
Includes all the core infrastructure to measure the total_bytes and
bandwidth.
We have per socket counters for both total system wide L3 external bytes
and local socket memory-controller bytes. The OS does MSR writes to
MSR_IA32_QM_EVTSEL and MSR_IA32_QM_CTR to read the counters and uses the
Includes all the core infrastructure to measure the total_bytes and
bandwidth.
We have per socket counters for both total system wide L3 external bytes
and local socket memory-controller bytes. The OS does MSR writes to
MSR_IA32_QM_EVTSEL and MSR_IA32_QM_CTR to read the counters and uses the
From: Tony Luck
Includes all the core infrastructure to measure the total_bytes and
bandwidth.
We have per socket counters for both total system wide L3 external bytes
and local socket memory-controller bytes. The OS does MSR writes to
MSR_IA32_QM_EVTSEL and MSR_IA32_QM_CTR
From: Tony Luck
Includes all the core infrastructure to measure the total_bytes and
bandwidth.
We have per socket counters for both total system wide L3 external bytes
and local socket memory-controller bytes. The OS does MSR writes to
MSR_IA32_QM_EVTSEL and MSR_IA32_QM_CTR to read the counters
On Mon, 7 Mar 2016, Peter Zijlstra wrote:
On Tue, Mar 01, 2016 at 03:48:26PM -0800, Vikas Shivappa wrote:
Lot of the scheduling code was taken out from Tony's patch and a 3-4
lines of change were added in the intel_cqm_event_read. Since the timer
is no more added on every context switch
On Mon, 7 Mar 2016, Peter Zijlstra wrote:
On Tue, Mar 01, 2016 at 03:48:26PM -0800, Vikas Shivappa wrote:
Lot of the scheduling code was taken out from Tony's patch and a 3-4
lines of change were added in the intel_cqm_event_read. Since the timer
is no more added on every context switch
On Tue, 8 Mar 2016, Peter Zijlstra wrote:
On Mon, Mar 07, 2016 at 11:27:26PM +, Luck, Tony wrote:
+ bytes = mbm_current->interval_bytes * MSEC_PER_SEC;
+ do_div(bytes, diff_time);
+ mbm_current->bandwidth = bytes;
+
On Tue, 8 Mar 2016, Peter Zijlstra wrote:
On Mon, Mar 07, 2016 at 11:27:26PM +, Luck, Tony wrote:
+ bytes = mbm_current->interval_bytes * MSEC_PER_SEC;
+ do_div(bytes, diff_time);
+ mbm_current->bandwidth = bytes;
+
On Mon, Mar 07, 2016 at 11:27:26PM +, Luck, Tony wrote:
> >> + bytes = mbm_current->interval_bytes * MSEC_PER_SEC;
> >> + do_div(bytes, diff_time);
> >> + mbm_current->bandwidth = bytes;
> >> + mbm_current->interval_bytes = 0;
> >> +
On Mon, Mar 07, 2016 at 11:27:26PM +, Luck, Tony wrote:
> >> + bytes = mbm_current->interval_bytes * MSEC_PER_SEC;
> >> + do_div(bytes, diff_time);
> >> + mbm_current->bandwidth = bytes;
> >> + mbm_current->interval_bytes = 0;
> >> +
>> +bytes = mbm_current->interval_bytes * MSEC_PER_SEC;
>> +do_div(bytes, diff_time);
>> +mbm_current->bandwidth = bytes;
>> +mbm_current->interval_bytes = 0;
>> +mbm_current->interval_start = cur_time;
>> +}
>>> +
>> +return
>> +bytes = mbm_current->interval_bytes * MSEC_PER_SEC;
>> +do_div(bytes, diff_time);
>> +mbm_current->bandwidth = bytes;
>> +mbm_current->interval_bytes = 0;
>> +mbm_current->interval_start = cur_time;
>> +}
>>> +
>> +return
On Tue, Mar 01, 2016 at 03:48:26PM -0800, Vikas Shivappa wrote:
> Lot of the scheduling code was taken out from Tony's patch and a 3-4
> lines of change were added in the intel_cqm_event_read. Since the timer
> is no more added on every context switch this change was made.
It this here to
On Tue, Mar 01, 2016 at 03:48:26PM -0800, Vikas Shivappa wrote:
> Lot of the scheduling code was taken out from Tony's patch and a 3-4
> lines of change were added in the intel_cqm_event_read. Since the timer
> is no more added on every context switch this change was made.
It this here to
From: Tony Luck
Includes all the core infrastructure to measure the total_bytes and
bandwidth.
We have per socket counters for both total system wide L3 external bytes
and local socket memory-controller bytes. The current b/w is calculated
for a minimum diff time(time since
From: Tony Luck
Includes all the core infrastructure to measure the total_bytes and
bandwidth.
We have per socket counters for both total system wide L3 external bytes
and local socket memory-controller bytes. The current b/w is calculated
for a minimum diff time(time since it was last counted)
Vikas,
On Wed, 17 Feb 2016, Vikas Shivappa wrote:
> Was wondering if you had any feedback on the MBM patches. These are almost a
> complete rewrite (the core part 3/5 from Tony Luck). The reused parts are
> mostly the declarations of datastructures. This has tried to address all your
> comments
Vikas,
On Wed, 17 Feb 2016, Vikas Shivappa wrote:
> Was wondering if you had any feedback on the MBM patches. These are almost a
> complete rewrite (the core part 3/5 from Tony Luck). The reused parts are
> mostly the declarations of datastructures. This has tried to address all your
> comments
monitoring.
It supports both 'local bandwidth' and 'total bandwidth' monitoring for
the socket. Local bandwidth measures the amount of data sent through
the memory controller on the socket and total b/w measures the total
system bandwidth.
The tasks are associated with a Resouce Monitoring ID(RMID) just
monitoring.
It supports both 'local bandwidth' and 'total bandwidth' monitoring for
the socket. Local bandwidth measures the amount of data sent through
the memory controller on the socket and total b/w measures the total
system bandwidth.
The tasks are associated with a Resouce Monitoring ID(RMID) just
bandwidth from one level of cache to another. The current patches
support L3 external bandwitch monitoring.
It supports both 'local bandwidth' and 'total bandwidth' monitoring for
the socket. Local bandwidth measures the amount of data sent through
the memory controller on the socket and total b/w
From: Tony Luck
Includes all the core infrastructure to measure the total_bytes and
bandwidth.
We have per socket counters for both total system wide L3 external bytes
and local socket memory-controller bytes. The current b/w is calculated
for a minimum diff time(time since it was last counted)
bandwidth from one level of cache to another. The current patches
support L3 external bandwitch monitoring.
It supports both 'local bandwidth' and 'total bandwidth' monitoring for
the socket. Local bandwidth measures the amount of data sent through
the memory controller on the socket and total b/w
From: Tony Luck
Includes all the core infrastructure to measure the total_bytes and
bandwidth.
We have per socket counters for both total system wide L3 external bytes
and local socket memory-controller bytes. The current b/w is calculated
for a minimum diff time(time since
On Thu, 10 Sep, at 02:18:49PM, Kanaka Juvva wrote:
> > >
> > > > > } else {
> > > > > mbm_current = _total[vrmid];
> > > > > eventid = QOS_MBM_TOTAL_EVENT_ID;
> > > > > }
> > > > > rmid = tmp32;
> > > >
> > > > Why did you assign rmid
On Thu, 10 Sep, at 02:18:49PM, Kanaka Juvva wrote:
> > >
> > > > > } else {
> > > > > mbm_current = _total[vrmid];
> > > > > eventid = QOS_MBM_TOTAL_EVENT_ID;
> > > > > }
> > > > > rmid = tmp32;
> > > >
> > > > Why did you assign rmid
On Thu, 2015-09-10 at 14:58 +0100, Matt Fleming wrote:
> On Tue, 2015-09-08 at 18:06 +0100, Juvva, Kanaka D wrote:
> >
> > There are two aspects:
> >
> > 1) Programming MSRs
> > 2) EVENT_ATTR_STR(llc_local_bw, intel_cqm_llc_local_bw, "event=0x04");
> >
> > 1 is used for programming MSRs
> >
On Tue, 2015-09-08 at 18:06 +0100, Juvva, Kanaka D wrote:
>
> There are two aspects:
>
> 1) Programming MSRs
> 2) EVENT_ATTR_STR(llc_local_bw, intel_cqm_llc_local_bw, "event=0x04");
>
> 1 is used for programming MSRs
> 2 event attribute for perf
>
>
> For MBM_LOCAL_EVENT HW ID is 0x3.
On Tue, 2015-09-08 at 18:06 +0100, Juvva, Kanaka D wrote:
>
> There are two aspects:
>
> 1) Programming MSRs
> 2) EVENT_ATTR_STR(llc_local_bw, intel_cqm_llc_local_bw, "event=0x04");
>
> 1 is used for programming MSRs
> 2 event attribute for perf
>
>
> For MBM_LOCAL_EVENT HW ID is 0x3.
On Thu, 2015-09-10 at 14:58 +0100, Matt Fleming wrote:
> On Tue, 2015-09-08 at 18:06 +0100, Juvva, Kanaka D wrote:
> >
> > There are two aspects:
> >
> > 1) Programming MSRs
> > 2) EVENT_ATTR_STR(llc_local_bw, intel_cqm_llc_local_bw, "event=0x04");
> >
> > 1 is used for programming MSRs
> >
nesday, August 19, 2015 1:50 PM
>> > To: Kanaka Juvva
>> > Cc: Juvva, Kanaka D; Williamson, Glenn P; Fleming, Matt; Auld, Will;
>> Andi Kleen;
>> > LKML; Luck, Tony; Peter Zijlstra; Tejun Heo; x...@kernel.org; Ingo
>> Molnar; H.
>> > Peter Anvin; Shiv
leen;
> LKML; Luck, Tony; Peter Zijlstra; Tejun Heo; x...@kernel.org; Ingo Molnar; H.
> Peter Anvin; Shivappa, Vikas
> Subject: Re: [PATCH v3 1/2] perf,x86: add Intel Memory Bandwidth Monitoring
> (MBM) PMU
>
> On Mon, 2015-09-07 at 20:22 +0100, Juvva, Kanaka D wrote:
> >
Subject: Re: [PATCH v3 1/2] perf,x86: add Intel Memory Bandwidth Monitoring
> (MBM) PMU
>
> On Mon, 2015-09-07 at 20:22 +0100, Juvva, Kanaka D wrote:
> > Hi Thomas,
> >
> >I'm sending updated patch(s). I have given details for each of
> > these items be
1:50 PM
> > To: Kanaka Juvva
> > Cc: Juvva, Kanaka D; Williamson, Glenn P; Fleming, Matt; Auld, Will;
> Andi Kleen;
> > LKML; Luck, Tony; Peter Zijlstra; Tejun Heo; x...@kernel.org; Ingo
> Molnar; H.
> > Peter Anvin; Shivappa, Vikas
> > Subject: Re: [PATCH v3 1/2] perf
1:50 PM
> > To: Kanaka Juvva
> > Cc: Juvva, Kanaka D; Williamson, Glenn P; Fleming, Matt; Auld, Will;
> Andi Kleen;
> > LKML; Luck, Tony; Peter Zijlstra; Tejun Heo; x...@kernel.org; Ingo
> Molnar; H.
> > Peter Anvin; Shivappa, Vikas
> > Subject: Re: [PATCH v3 1/2] perf
nesday, August 19, 2015 1:50 PM
>> > To: Kanaka Juvva
>> > Cc: Juvva, Kanaka D; Williamson, Glenn P; Fleming, Matt; Auld, Will;
>> Andi Kleen;
>> > LKML; Luck, Tony; Peter Zijlstra; Tejun Heo; x...@kernel.org; Ingo
>> Molnar; H.
>> > Peter Anvin; Shiv
Subject: Re: [PATCH v3 1/2] perf,x86: add Intel Memory Bandwidth Monitoring
> (MBM) PMU
>
> On Mon, 2015-09-07 at 20:22 +0100, Juvva, Kanaka D wrote:
> > Hi Thomas,
> >
> >I'm sending updated patch(s). I have given details for each of
> > these items be
leen;
> LKML; Luck, Tony; Peter Zijlstra; Tejun Heo; x...@kernel.org; Ingo Molnar; H.
> Peter Anvin; Shivappa, Vikas
> Subject: Re: [PATCH v3 1/2] perf,x86: add Intel Memory Bandwidth Monitoring
> (MBM) PMU
>
> On Mon, 2015-09-07 at 20:22 +0100, Juvva, Kanaka D wrote:
> >
On Thu, 20 Aug 2015, Juvva, Kanaka D wrote:
> Hi Thomas,
Please do not top post and trim your replies proper.
> Acknowledged. Perhaps some discussions are required in terms of
> your questions and our solutions.
Fine with me.
> Mostly nothing was unresolved except some new things that are
>
On Thu, 20 Aug 2015, Juvva, Kanaka D wrote:
Hi Thomas,
Please do not top post and trim your replies proper.
Acknowledged. Perhaps some discussions are required in terms of
your questions and our solutions.
Fine with me.
Mostly nothing was unresolved except some new things that are
brought
olnar; H. Peter Anvin; Shivappa, Vikas
> Subject: Re: [PATCH v3 1/2] perf,x86: add Intel Memory Bandwidth
> Monitoring (MBM) PMU
>
> On Fri, 7 Aug 2015, Kanaka Juvva wrote:
> > +#define MBM_CNTR_MAX 0xff
> > +#define MBM_SOCKET_MAX 8
>
; Shivappa, Vikas
Subject: Re: [PATCH v3 1/2] perf,x86: add Intel Memory Bandwidth
Monitoring (MBM) PMU
On Fri, 7 Aug 2015, Kanaka Juvva wrote:
+#define MBM_CNTR_MAX 0xff
+#define MBM_SOCKET_MAX 8
+#define MBM_TIME_DELTA_MAX 1000
+#define MBM_TIME_DELTA_MIN 100
On Fri, 7 Aug 2015, Kanaka Juvva wrote:
> +#define MBM_CNTR_MAX 0xff
> +#define MBM_SOCKET_MAX 8
> +#define MBM_TIME_DELTA_MAX 1000
> +#define MBM_TIME_DELTA_MIN 100
What are these constants for and how are they determined? Pulled out
of thin air?
> +#define
On Fri, 7 Aug 2015, Kanaka Juvva wrote:
+#define MBM_CNTR_MAX 0xff
+#define MBM_SOCKET_MAX 8
+#define MBM_TIME_DELTA_MAX 1000
+#define MBM_TIME_DELTA_MIN 100
What are these constants for and how are they determined? Pulled out
of thin air?
+#define
.com; Shivappa, Vikas
> Subject: [PATCH v3 1/2] perf,x86: add Intel Memory Bandwidth Monitoring
> (MBM) PMU
>
> This patch adds Memory Bandwidth Monitoring events to the exitsing intel_cqm
> pmu in the Linux Kernel.
>
> Intel MBM builds on Cache Monitoring Technology (CMT) infrast
: [PATCH v3 1/2] perf,x86: add Intel Memory Bandwidth Monitoring
(MBM) PMU
This patch adds Memory Bandwidth Monitoring events to the exitsing intel_cqm
pmu in the Linux Kernel.
Intel MBM builds on Cache Monitoring Technology (CMT) infrastructure to allow
monitoring of bandwidth from one level
This patch adds Memory Bandwidth Monitoring events to the exitsing
intel_cqm pmu in the Linux Kernel.
Intel MBM builds on Cache Monitoring Technology (CMT) infrastructure
to allow monitoring of bandwidth from one level of the cache hierarchy
to the next - in this case focusing on the L3 cache
This patch adds Memory Bandwidth Monitoring events to the exitsing
intel_cqm pmu in the Linux Kernel.
Intel MBM builds on Cache Monitoring Technology (CMT) infrastructure
to allow monitoring of bandwidth from one level of the cache hierarchy
to the next - in this case focusing on the L3 cache
> mi...@redhat.com; H. Peter Anvin; Shivappa, Vikas
> Subject: Re: [PATCH v2] perf,x86: add Intel Memory Bandwidth Monitoring
> (MBM) PMU
>
> On Tue, 21 Jul 2015, Kanaka Juvva wrote:
> > diff --git a/arch/x86/kernel/cpu/perf_event_intel_cqm.c
> > b/arch/x86/kernel/cpu/p
Anvin; Shivappa, Vikas
Subject: Re: [PATCH v2] perf,x86: add Intel Memory Bandwidth Monitoring
(MBM) PMU
On Tue, 21 Jul 2015, Kanaka Juvva wrote:
diff --git a/arch/x86/kernel/cpu/perf_event_intel_cqm.c
b/arch/x86/kernel/cpu/perf_event_intel_cqm.c
index 1880761..dc1ce58 100644
--- a/arch
On Tue, 21 Jul 2015, Kanaka Juvva wrote:
> diff --git a/arch/x86/kernel/cpu/perf_event_intel_cqm.c
> b/arch/x86/kernel/cpu/perf_event_intel_cqm.c
> index 1880761..dc1ce58 100644
> --- a/arch/x86/kernel/cpu/perf_event_intel_cqm.c
> +++ b/arch/x86/kernel/cpu/perf_event_intel_cqm.c
> @@ -12,10
On Tue, 21 Jul 2015, Kanaka Juvva wrote:
diff --git a/arch/x86/kernel/cpu/perf_event_intel_cqm.c
b/arch/x86/kernel/cpu/perf_event_intel_cqm.c
index 1880761..dc1ce58 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_cqm.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_cqm.c
@@ -12,10 +12,20 @@
This patch adds Memory Bandwidth Monitoring events to the exitsing
intel_cqm pmu in the Linux Kernel.
Intel MBM builds on Cache Monitoring Technology (CMT) infrastructure
to allow monitoring of bandwidth from one level of the cache hierarchy
to the next - in this case focusing on the L3 cache
This patch adds Memory Bandwidth Monitoring events to the exitsing
intel_cqm pmu in the Linux Kernel.
Intel MBM builds on Cache Monitoring Technology (CMT) infrastructure
to allow monitoring of bandwidth from one level of the cache hierarchy
to the next - in this case focusing on the L3 cache
This patch adds Memory Bandwidth Monitoring events to the exitsing
intel_cqm pmu in the Linux Kernel.
Intel MBM builds on Cache Monitoring Technology (CMT) infrastructure
to allow monitoring of bandwidth from one level of the cache hierarchy
to the next - in this case focusing on the L3 cache
This patch adds Memory Bandwidth Monitoring events to the exitsing
intel_cqm pmu in the Linux Kernel.
Intel MBM builds on Cache Monitoring Technology (CMT) infrastructure
to allow monitoring of bandwidth from one level of the cache hierarchy
to the next - in this case focusing on the L3 cache
.com;
> jacob.w.s...@gmail.com; Brandewie, Dirk J; Shivappa, Vikas; Verplanke, Edwin;
> Kantecki, Tomasz
> Subject: [PATCH v1 1/2] perf,x86: add Intel Memory Bandwidth Monitoring
> (MBM) PMU
>
> This patch adds Memory Bandwidth Monitoring events to the exitsing intel_cqm
> pmu
This patch adds Memory Bandwidth Monitoring events to the exitsing
intel_cqm pmu in the Linux Kernel.
Intel MBM builds on Cache Monitoring Technology (CMT) infrastructure
to allow monitoring of bandwidth from one level of the cache hierarchy
to the next - in this case focusing on the L3 cache
This patch adds Memory Bandwidth Monitoring events to the exitsing
intel_cqm pmu in the Linux Kernel.
Intel MBM builds on Cache Monitoring Technology (CMT) infrastructure
to allow monitoring of bandwidth from one level of the cache hierarchy
to the next - in this case focusing on the L3 cache
, Dirk J; Shivappa, Vikas; Verplanke, Edwin;
Kantecki, Tomasz
Subject: [PATCH v1 1/2] perf,x86: add Intel Memory Bandwidth Monitoring
(MBM) PMU
This patch adds Memory Bandwidth Monitoring events to the exitsing intel_cqm
pmu in the Linux Kernel.
Intel MBM builds on Cache Monitoring
On Sun, Oct 15, 2000 at 04:35:13AM -0500, Brian Parris wrote:
> I apologise if this is the wrong place for this, but i've been searching for
Well... the list is about kernel develoment, so it is a bit off-topic.
> 3 weeks for a way to record the bandwidth used by each user under 2.2.17, i
>
I apologise if this is the wrong place for this, but i've been searching for
3 weeks for a way to record the bandwidth used by each user under 2.2.17, i
know there are ways to record bandwidth used by webpage hits but i also want
to monitor bandwidth used by any program the user has running, if
I apologise if this is the wrong place for this, but i've been searching for
3 weeks for a way to record the bandwidth used by each user under 2.2.17, i
know there are ways to record bandwidth used by webpage hits but i also want
to monitor bandwidth used by any program the user has running, if
65 matches
Mail list logo