From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-9.0 required=3.0 tests=BAYES_00,DKIM_ADSP_CUSTOM_MED, DKIM_INVALID,DKIM_SIGNED,FREEMAIL_FORGED_FROMDOMAIN,FREEMAIL_FROM, HEADER_FROM_DIFFERENT_DOMAINS,HTML_MESSAGE,INCLUDES_CR_TRAILER,INCLUDES_PATCH, MAILING_LIST_MULTI,NICE_REPLY_A,SPF_HELO_NONE,SPF_PASS,USER_AGENT_SANE_1 autolearn=ham autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id 3DEFBC433ED for ; Wed, 12 May 2021 06:55:22 +0000 (UTC) Received: from gabe.freedesktop.org (gabe.freedesktop.org [131.252.210.177]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by mail.kernel.org (Postfix) with ESMTPS id DCEE561177 for ; Wed, 12 May 2021 06:55:21 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org DCEE561177 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=gmail.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=amd-gfx-bounces@lists.freedesktop.org Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 7CADB6EB4B; Wed, 12 May 2021 06:55:21 +0000 (UTC) Received: from mail-ej1-x631.google.com (mail-ej1-x631.google.com [IPv6:2a00:1450:4864:20::631]) by gabe.freedesktop.org (Postfix) with ESMTPS id 474FD6EB4B for ; Wed, 12 May 2021 06:55:20 +0000 (UTC) Received: by mail-ej1-x631.google.com with SMTP id b25so33309029eju.5 for ; Tue, 11 May 2021 23:55:20 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20161025; h=subject:to:references:from:message-id:date:user-agent:mime-version :in-reply-to:content-language; bh=Wu1LrxK8RTVLOA4oysYG+Mf4FWuy+l8JooqfXCYR+zo=; b=JzCij9DXx/n/zFh9JSb7PSeQsGeCDWU2bT07LSRQ4KF6ckH2t5JcoRCQxkmwRk5S8n cURbVW1zuq1BWu5ynd+JX6JyV2GLwClhQWPuUGoA7R7JARVmi7xSpaBNeCfEM5S8m7QM Lw8wjf+qG2vPirkT5lSR3AHWlTr1xLPIEOWUkzDr8eu4AhE5y4xTdCRp+E4/hi0/bTGB EBOF41rS9hTf9QO9Uc021hSgEpQ/GF/bDT6jhO/1AYhbJYXSlHKLLjI2mpdJXQ23Qimm m78bGMc4sjSArBLx3AhovC+c3G1xEDgApz0cq+K/ZdVqCEuEgbTOCN80juVd/E3BZw8T Q1pw== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:subject:to:references:from:message-id:date :user-agent:mime-version:in-reply-to:content-language; bh=Wu1LrxK8RTVLOA4oysYG+Mf4FWuy+l8JooqfXCYR+zo=; b=qxDcVSz8MD8iP/1DwZMaUhpD56q7PDysxQHRSXa/nSGAPd81PFBpdtRcATVfvi4obf DbbqxSexUbcj+gUblBQGjA/y7b73uMiVzf5nbjtjDWbjgCLrU2XHMlyn8SFTl5YUeuxg ScZdg58HDkl/+nQ45mvwmExSPjjhj34H400BG6sBciVJGblsZYfP8eOThZwyuXoBFkzK kY7Wz2z3JM/w0iwiAoSTuTX4X4wZoSGOP+WR/8v8nF4xGK1gceimV83WUtAEyPoABnAy PvXsyLYnYrf3tEHulfcAix2g03qBMRwIZOm51hy9dYIx6GznumoVFOGzmfKd+j7iIBlo jWcQ== X-Gm-Message-State: AOAM532HI5yMplGp8mSlRnPZ5kTcXxMjPw14QXp3jZKMmt0nn2P4B91l icDA1ZhWQueEdYTSFAiyRmJAs18RzKA= X-Google-Smtp-Source: ABdhPJwXAcZ4Ha4cVKSCNlf0oY2p9GVO9oCxs+bvXdljd6448xRMDUsygWuhIl6BVsYRw+MSv+8MOQ== X-Received: by 2002:a17:906:858:: with SMTP id f24mr36508296ejd.347.1620802519040; Tue, 11 May 2021 23:55:19 -0700 (PDT) Received: from ?IPv6:2a02:908:1252:fb60:bc39:4e06:ac5e:2e31? ([2a02:908:1252:fb60:bc39:4e06:ac5e:2e31]) by smtp.gmail.com with ESMTPSA id t20sm13372934ejc.61.2021.05.11.23.55.18 (version=TLS1_3 cipher=TLS_AES_128_GCM_SHA256 bits=128/128); Tue, 11 May 2021 23:55:18 -0700 (PDT) Subject: Re: [PATCH 2/2] drm/amdgpu: fix fence calculation To: "Nieto, David M" , "amd-gfx@lists.freedesktop.org" References: <20210510202915.30502-1-david.nieto@amd.com> <20210510202915.30502-2-david.nieto@amd.com> <7579fa92-ad25-323a-0c41-ac07ac47fa42@gmail.com> From: =?UTF-8?Q?Christian_K=c3=b6nig?= Message-ID: <9b373f49-51ad-089c-2494-032b13a9a39c@gmail.com> Date: Wed, 12 May 2021 08:55:18 +0200 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101 Thunderbird/78.8.1 MIME-Version: 1.0 In-Reply-To: Content-Language: en-US X-BeenThere: amd-gfx@lists.freedesktop.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Discussion list for AMD gfx List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Content-Type: multipart/mixed; boundary="===============1135282661==" Errors-To: amd-gfx-bounces@lists.freedesktop.org Sender: "amd-gfx" This is a multi-part message in MIME format. --===============1135282661== Content-Type: multipart/alternative; boundary="------------3A1D30770827C865F2B76F03" Content-Language: en-US This is a multi-part message in MIME format. --------------3A1D30770827C865F2B76F03 Content-Type: text/plain; charset=windows-1252; format=flowed Content-Transfer-Encoding: 8bit In this case amdgpu_ctx_fence_time should probably be changed to initialize the variable itself. That is really bad coding style otherwise. Christian. Am 11.05.21 um 20:14 schrieb Nieto, David M: > > [AMD Official Use Only - Internal Distribution Only] > > > The local variables need to be initialized to zero, since > amdgpu_ctx_fence_time accumulates and does not initialize > > David > ------------------------------------------------------------------------ > *From:* Christian König > *Sent:* Tuesday, May 11, 2021 12:53 AM > *To:* Nieto, David M ; > amd-gfx@lists.freedesktop.org > *Subject:* Re: [PATCH 2/2] drm/amdgpu: fix fence calculation > Am 10.05.21 um 22:29 schrieb David M Nieto: > > The proper metric for fence utilization over several > > contexts is an harmonic mean, but such calculation is > > prohibitive in kernel space, so the code approximates it. > > > > Because the approximation diverges when one context has a > > very small ratio compared with the other context, this change > > filter out ratios smaller that 0.01% > > > > Signed-off-by: David M Nieto > > Change-Id: I5b6e0ce5f489a5f55855d35354a6a3653e9d613b > > --- > >   drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c | 13 ++++++++++++- > >   drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h |  1 + > >   2 files changed, 13 insertions(+), 1 deletion(-) > > > > diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c > b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c > > index 9036c93b4a0c..89ee464b9424 100644 > > --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c > > +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c > > @@ -698,16 +698,27 @@ ktime_t amdgpu_ctx_mgr_fence_usage(struct > amdgpu_ctx_mgr *mgr, uint32_t hwip, > >        struct amdgpu_ctx_entity *centity; > >        ktime_t total = 0, max = 0; > > > > + > > Unrelated white space change. > > >        if (idx >= AMDGPU_MAX_ENTITY_NUM) > >                return 0; > >        idp = &mgr->ctx_handles; > >        mutex_lock(&mgr->lock); > >        idr_for_each_entry(idp, ctx, id) { > > +             ktime_t ttotal = tmax = ktime_set(0, 0); > > There should be a blank line between decleration and code and please > don't initialize local variables if it isn't necessary. > > Christian. > > >                if (!ctx->entities[hwip][idx]) > >                        continue; > > > >                centity = ctx->entities[hwip][idx]; > > -             amdgpu_ctx_fence_time(ctx, centity, &total, &max); > > +             amdgpu_ctx_fence_time(ctx, centity, &ttotal, &tmax); > > + > > +             /* Harmonic mean approximation diverges for very small > > +              * values. If ratio < 0.01% ignore > > +              */ > > +             if (AMDGPU_CTX_FENCE_USAGE_MIN_RATIO(tmax, ttotal)) > > +                     continue; > > + > > +             total = ktime_add(total, ttotal); > > +             max = ktime_after(tmax, max) ? tmax : max; > >        } > > > >        mutex_unlock(&mgr->lock); > > diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h > b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h > > index 10dcf59a5c6b..3541dfb059ec 100644 > > --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h > > +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h > > @@ -30,6 +30,7 @@ struct drm_file; > >   struct amdgpu_fpriv; > > > >   #define AMDGPU_MAX_ENTITY_NUM 4 > > +#define AMDGPU_CTX_FENCE_USAGE_MIN_RATIO(max, total) (max > > 16384ULL*total) > > > >   struct amdgpu_ctx_entity { > >        uint64_t                sequence; > --------------3A1D30770827C865F2B76F03 Content-Type: text/html; charset=windows-1252 Content-Transfer-Encoding: 8bit In this case amdgpu_ctx_fence_time should probably be changed to initialize the variable itself.

That is really bad coding style otherwise.

Christian.

Am 11.05.21 um 20:14 schrieb Nieto, David M:

[AMD Official Use Only - Internal Distribution Only]


The local variables need to be initialized to zero, since amdgpu_ctx_fence_time accumulates and does not initialize

David
From: Christian König <ckoenig.leichtzumerken@gmail.com>
Sent: Tuesday, May 11, 2021 12:53 AM
To: Nieto, David M <David.Nieto@amd.com>; amd-gfx@lists.freedesktop.org <amd-gfx@lists.freedesktop.org>
Subject: Re: [PATCH 2/2] drm/amdgpu: fix fence calculation
 
Am 10.05.21 um 22:29 schrieb David M Nieto:
> The proper metric for fence utilization over several
> contexts is an harmonic mean, but such calculation is
> prohibitive in kernel space, so the code approximates it.
>
> Because the approximation diverges when one context has a
> very small ratio compared with the other context, this change
> filter out ratios smaller that 0.01%
>
> Signed-off-by: David M Nieto <david.nieto@amd.com>
> Change-Id: I5b6e0ce5f489a5f55855d35354a6a3653e9d613b
> ---
>   drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c | 13 ++++++++++++-
>   drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h |  1 +
>   2 files changed, 13 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c
> index 9036c93b4a0c..89ee464b9424 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c
> @@ -698,16 +698,27 @@ ktime_t amdgpu_ctx_mgr_fence_usage(struct amdgpu_ctx_mgr *mgr, uint32_t hwip,
>        struct amdgpu_ctx_entity *centity;
>        ktime_t total = 0, max = 0;
>  
> +

Unrelated white space change.

>        if (idx >= AMDGPU_MAX_ENTITY_NUM)
>                return 0;
>        idp = &mgr->ctx_handles;
>        mutex_lock(&mgr->lock);
>        idr_for_each_entry(idp, ctx, id) {
> +             ktime_t ttotal = tmax = ktime_set(0, 0);

There should be a blank line between decleration and code and please
don't initialize local variables if it isn't necessary.

Christian.

>                if (!ctx->entities[hwip][idx])
>                        continue;
>  
>                centity = ctx->entities[hwip][idx];
> -             amdgpu_ctx_fence_time(ctx, centity, &total, &max);
> +             amdgpu_ctx_fence_time(ctx, centity, &ttotal, &tmax);
> +
> +             /* Harmonic mean approximation diverges for very small
> +              * values. If ratio < 0.01% ignore
> +              */
> +             if (AMDGPU_CTX_FENCE_USAGE_MIN_RATIO(tmax, ttotal))
> +                     continue;
> +
> +             total = ktime_add(total, ttotal);
> +             max = ktime_after(tmax, max) ? tmax : max;
>        }
>  
>        mutex_unlock(&mgr->lock);
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h
> index 10dcf59a5c6b..3541dfb059ec 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h
> @@ -30,6 +30,7 @@ struct drm_file;
>   struct amdgpu_fpriv;
>  
>   #define AMDGPU_MAX_ENTITY_NUM 4
> +#define AMDGPU_CTX_FENCE_USAGE_MIN_RATIO(max, total) (max > 16384ULL*total)
>  
>   struct amdgpu_ctx_entity {
>        uint64_t                sequence;


--------------3A1D30770827C865F2B76F03-- --===============1135282661== Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline _______________________________________________ amd-gfx mailing list amd-gfx@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/amd-gfx --===============1135282661==--