I really like what i have seen so far wrt to functionality, so i am glad to 
hear that performance has been improved! Mostly i was concerned that 
C#/.net in grpc was a bit of a "secondary" target without much 
support/optimization (for example, if you look at captn proto, the C# 
implementation is done on the side, and is a bit unsupported).


On Wednesday, February 8, 2017 at 6:44:49 PM UTC+1, Jan Tattermusch wrote:
>
> Hi,
>
> C# performance have been improved by a lot since 1.0.0 release (from 
> August 2016).
> The results on upstream/master are temporarily broken for unrelated 
> reasons, but normally you would see that the C# performance have been 
> improved drastically between 1.0.0 and 1.1.0 (which is the latest release - 
> see release notes in https://github.com/grpc/grpc/releases/tag/v1.1.0 that 
> mentions some details about the performance improvement and how to get the 
> best numbers).
>
> We'll try to provide a dashboard that corresponds to the 1.1.0 release.
> We are planning to bring more optimizations to C# in the future.
>
> Jan
>
>
>
> On Wed, Feb 8, 2017 at 9:19 AM, 'Nicolas Noble' via grpc.io <
> [email protected] <javascript:>> wrote:
>
>> Also what's the platform requirement? I think these numbers are from 
>> Linux... 
>>
>> On Wed, Feb 8, 2017, 08:53 'Carl Mastrangelo' via grpc.io <
>> [email protected] <javascript:>> wrote:
>>
>>> It is if you optimize your code.  What kind of latency / throughput 
>>> numbers are in your requirements?
>>>
>>>
>>> On Wednesday, February 8, 2017 at 2:34:41 AM UTC-8, Peter Tiedemann 
>>> wrote:
>>>>
>>>> But do you mean if i optimize *my* code, or if someone optimizes the 
>>>> *grpc* C# code base?
>>>>
>>>> I am naturally concerned with how high priority .Net support has for 
>>>> grpc, as that is our primary platform :)
>>>>
>>>> On Tuesday, February 7, 2017 at 12:48:26 AM UTC+1, Carl Mastrangelo 
>>>> wrote:
>>>>>
>>>>> It is a good indication of what you can get, if you optimize your 
>>>>> code.  I think a better idea of performance is the latency, because that 
>>>>> tends to matter a lot more for most applications.  For that they are on 
>>>>> par. 
>>>>>
>>>>> On Saturday, February 4, 2017 at 6:13:21 AM UTC-8, [email protected] 
>>>>> wrote:
>>>>>>
>>>>>> But if the benchmark is a good indication of real world throughput, 
>>>>>> then surely there is still reason to be concerned? Or do you mean that 
>>>>>> the 
>>>>>> benchmark specific Java code (as in the benchmark application ) has been 
>>>>>> optimized?
>>>>>>
>>>>>> On Saturday, February 4, 2017 at 1:30:37 AM UTC+1, Carl Mastrangelo 
>>>>>> wrote:
>>>>>>>
>>>>>>> The reason Java is fast is because there has been a lot more time 
>>>>>>> spent in making the *benchmark* fast.  Those numbers tell you what you 
>>>>>>> can 
>>>>>>> expect from an optimized gRPC server / client.  The core reason Java is 
>>>>>>> faster is likely because there was considerable time put into profiling 
>>>>>>> that benchmark code.  
>>>>>>>
>>>>>>> On Friday, February 3, 2017 at 9:57:06 AM UTC-8, Peter Tiedemann 
>>>>>>> wrote:
>>>>>>>>
>>>>>>>> I was looking over the benchmarks here (1.0.0, master does not seem 
>>>>>>>> to work):
>>>>>>>>
>>>>>>>>
>>>>>>>> https://performance-dot-grpc-testing.appspot.com/explore?dashboard=5712453606309888
>>>>>>>>
>>>>>>>> Mostly, it seems sensible enough, C++ is fastest, Java and C# 
>>>>>>>> roughly tied. Then i took a look at the throughput tests, where Java 
>>>>>>>> shows 
>>>>>>>> ~10x more QPS, leaving C# closer to Python and Node.
>>>>>>>>
>>>>>>>> Is there some performance issue with the C# implementation i need 
>>>>>>>> to be aware of?
>>>>>>>>
>>>>>>>>
>>>>>>>> -- 
>>> You received this message because you are subscribed to the Google 
>>> Groups "grpc.io" group.
>>> To unsubscribe from this group and stop receiving emails from it, send 
>>> an email to [email protected] <javascript:>.
>>> To post to this group, send email to [email protected] 
>>> <javascript:>.
>>> Visit this group at https://groups.google.com/group/grpc-io.
>>> To view this discussion on the web visit 
>>> https://groups.google.com/d/msgid/grpc-io/9d3b7f37-16b6-493b-a644-6a26f83aad35%40googlegroups.com
>>>  
>>> <https://groups.google.com/d/msgid/grpc-io/9d3b7f37-16b6-493b-a644-6a26f83aad35%40googlegroups.com?utm_medium=email&utm_source=footer>
>>> .
>>> For more options, visit https://groups.google.com/d/optout.
>>>
>> -- 
>> You received this message because you are subscribed to the Google Groups 
>> "grpc.io" group.
>> To unsubscribe from this group and stop receiving emails from it, send an 
>> email to [email protected] <javascript:>.
>> To post to this group, send email to [email protected] 
>> <javascript:>.
>> Visit this group at https://groups.google.com/group/grpc-io.
>> To view this discussion on the web visit 
>> https://groups.google.com/d/msgid/grpc-io/CAOWnRi9HSq9C-75jkc7-JrTLQQ5i9GptwkR%2B-mSHKahAbyWh7g%40mail.gmail.com
>>  
>> <https://groups.google.com/d/msgid/grpc-io/CAOWnRi9HSq9C-75jkc7-JrTLQQ5i9GptwkR%2B-mSHKahAbyWh7g%40mail.gmail.com?utm_medium=email&utm_source=footer>
>> .
>>
>> For more options, visit https://groups.google.com/d/optout.
>>
>
>

-- 
You received this message because you are subscribed to the Google Groups 
"grpc.io" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To post to this group, send email to [email protected].
Visit this group at https://groups.google.com/group/grpc-io.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/grpc-io/92b8bdb8-a8e5-428b-adb1-b1f875daae9c%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to