Hi All,
I find there have slow and fast path for IPC. I want to know the performance difference between those two. More accurate, on x86, how many cycles for slow path round-trip ipc between two address space?(on single core) And how many for fast path?
Thank you very much.
Yuxin
Hi Yuxin,
Below our our benchmark results. Figures are in cycles and do not include the cost of a the system call.
Intra-AS Call 756 Intra-AS ReplyWait 742 Inter-AS Call 980 Inter-AS ReplyWait 987 Inter-AS Call (Low to High) 1050 Inter-AS ReplyWait (High to Low) 1540 Inter-AS Call (High to Low) 1463 Inter-AS ReplyWait (Low to High) 1043 Inter-AS Send 1043 Inter-AS Call(10) 1008 Inter-AS ReplyWait(10) 1015
- Alex
________________________________________ From: Devel [devel-bounces@sel4.systems] on behalf of Yuxin Ren [ryx@gwmail.gwu.edu] Sent: Saturday, 18 October 2014 08:32 To: devel@sel4.systems Subject: [seL4] IPC performance for slowpath
Hi All,
I find there have slow and fast path for IPC. I want to know the performance difference between those two. More accurate, on x86, how many cycles for slow path round-trip ipc between two address space?(on single core) And how many for fast path?
Thank you very much.
Yuxin
________________________________
The information in this e-mail may be confidential and subject to legal professional privilege and/or copyright. National ICT Australia Limited accepts no liability for any damage caused by this email or its attachments.
Hi Alex,
Your results are great!! Thank you so much. I want to confirm something.
Inter-AS Call(10) 1008 What is meaning of "10" ?
Inter-AS Call 980 Inter-AS Call (Low to High) 1050 Inter-AS Call (High to Low) 1463 Here, "Low and High" means the priority of process, right? So without that, it means those two process have same priority. Is this true?
Thank you. Yuxin
On Sat, Oct 18, 2014 at 10:56 PM, Alexander Kroh < Alexander.Kroh@nicta.com.au> wrote:
Hi Yuxin,
Below our our benchmark results. Figures are in cycles and do not include the cost of a the system call.
Intra-AS Call 756 Intra-AS ReplyWait 742 Inter-AS Call 980 Inter-AS ReplyWait 987 Inter-AS Call (Low to High) 1050 Inter-AS ReplyWait (High to Low) 1540 Inter-AS Call (High to Low) 1463 Inter-AS ReplyWait (Low to High) 1043 Inter-AS Send 1043 Inter-AS Call(10) 1008 Inter-AS ReplyWait(10) 1015
- Alex
From: Devel [devel-bounces@sel4.systems] on behalf of Yuxin Ren [ ryx@gwmail.gwu.edu] Sent: Saturday, 18 October 2014 08:32 To: devel@sel4.systems Subject: [seL4] IPC performance for slowpath
Hi All,
I find there have slow and fast path for IPC. I want to know the performance difference between those two. More accurate, on x86, how many cycles for slow path round-trip ipc between two address space?(on single core) And how many for fast path?
Thank you very much.
Yuxin
The information in this e-mail may be confidential and subject to legal professional privilege and/or copyright. National ICT Australia Limited accepts no liability for any damage caused by this email or its attachments.
Hi Yuxin,
"10" is the number of words transferred via IPC. Otherwise, the IPC is a signal with no data component.
As you guessed, "low to High"/"High to Low" represent scheduling priority.
- Alex
________________________________________ From: Yuxin Ren [ryx@gwmail.gwu.edu] Sent: Sunday, 19 October 2014 02:36 To: Alexander Kroh Cc: devel@sel4.systems Subject: Re: [seL4] IPC performance for slowpath
Hi Alex,
Your results are great!! Thank you so much. I want to confirm something.
Inter-AS Call(10) 1008 What is meaning of "10" ?
Inter-AS Call 980 Inter-AS Call (Low to High) 1050 Inter-AS Call (High to Low) 1463 Here, "Low and High" means the priority of process, right? So without that, it means those two process have same priority. Is this true?
Thank you. Yuxin
On Sat, Oct 18, 2014 at 10:56 PM, Alexander Kroh <Alexander.Kroh@nicta.com.aumailto:Alexander.Kroh@nicta.com.au> wrote: Hi Yuxin,
Below our our benchmark results. Figures are in cycles and do not include the cost of a the system call.
Intra-AS Call 756 Intra-AS ReplyWait 742 Inter-AS Call 980 Inter-AS ReplyWait 987 Inter-AS Call (Low to High) 1050 Inter-AS ReplyWait (High to Low) 1540 Inter-AS Call (High to Low) 1463 Inter-AS ReplyWait (Low to High) 1043 Inter-AS Send 1043 Inter-AS Call(10) 1008 Inter-AS ReplyWait(10) 1015
- Alex
________________________________________ From: Devel [devel-bounces@sel4.systems] on behalf of Yuxin Ren [ryx@gwmail.gwu.edumailto:ryx@gwmail.gwu.edu] Sent: Saturday, 18 October 2014 08:32 To: devel@sel4.systems Subject: [seL4] IPC performance for slowpath
Hi All,
I find there have slow and fast path for IPC. I want to know the performance difference between those two. More accurate, on x86, how many cycles for slow path round-trip ipc between two address space?(on single core) And how many for fast path?
Thank you very much.
Yuxin
________________________________
The information in this e-mail may be confidential and subject to legal professional privilege and/or copyright. National ICT Australia Limited accepts no liability for any damage caused by this email or its attachments.
-
Alexander Kroh -
Yuxin Ren