Rocksolid Light

Welcome to novaBBS (click a section below)

mail  files  register  newsreader  groups  login

Message-ID:  

Conquest is easy. Control is not. -- Kirk, "Mirror, Mirror", stardate unknown


devel / comp.arch / My experience with Apple M1 chip

SubjectAuthor
* My experience with Apple M1 chipBranimir Maksimovic
+* Re: My experience with Apple M1 chipBranimir Maksimovic
|`* Re: My experience with Apple M1 chipBranimir Maksimovic
| `- Re: My experience with Apple M1 chipBranimir Maksimovic
+* Re: My experience with Apple M1 chipChris M. Thomasson
|`* Re: My experience with Apple M1 chipBranimir Maksimovic
| +* Re: My experience with Apple M1 chipThomas Koenig
| |`* Re: My experience with Apple M1 chipBranimir Maksimovic
| | `- Re: My experience with Apple M1 chipQuadibloc
| `* Re: My experience with Apple M1 chipaph
|  `* Re: My experience with Apple M1 chipBranimir Maksimovic
|   +* Re: My experience with Apple M1 chipaph
|   |`* Re: My experience with Apple M1 chipBranimir Maksimovic
|   | `* Re: My experience with Apple M1 chipBranimir Maksimovic
|   |  `- Re: My experience with Apple M1 chipBranimir Maksimovic
|   `* Re: My experience with Apple M1 chipJohn Dallman
|    +- Re: My experience with Apple M1 chipBranimir Maksimovic
|    `* Re: My experience with Apple M1 chipMichael S
|     +- Re: My experience with Apple M1 chipTheo
|     +- Re: My experience with Apple M1 chipJohn Levine
|     `* Re: My experience with Apple M1 chipBranimir Maksimovic
|      `* Re: My experience with Apple M1 chipMichael S
|       +* Re: My experience with Apple M1 chipKent Dickey
|       |+* Re: My experience with Apple M1 chipBranimir Maksimovic
|       ||`* Re: My experience with Apple M1 chipMichael S
|       || `* Re: My experience with Apple M1 chipBranimir Maksimovic
|       ||  +* Re: My experience with Apple M1 chipMichael S
|       ||  |+* Re: My experience with Apple M1 chipBranimir Maksimovic
|       ||  ||`* Re: My experience with Apple M1 chipMichael S
|       ||  || `- Re: My experience with Apple M1 chipBranimir Maksimovic
|       ||  |`* Re: My experience with Apple M1 chipBranimir Maksimovic
|       ||  | +- Re: My experience with Apple M1 chipBranimir Maksimovic
|       ||  | `* Re: My experience with Apple M1 chipMichael S
|       ||  |  `* Re: My experience with Apple M1 chipBranimir Maksimovic
|       ||  |   `- Re: My experience with Apple M1 chipMichael S
|       ||  `* Re: My experience with Apple M1 chipantispam
|       ||   `- Re: My experience with Apple M1 chipBranimir Maksimovic
|       |`* Re: My experience with Apple M1 chipMichael S
|       | +* Re: My experience with Apple M1 chipBranimir Maksimovic
|       | |+* Re: My experience with Apple M1 chipMichael S
|       | ||`- Re: My experience with Apple M1 chipBranimir Maksimovic
|       | |`* Re: My experience with Apple M1 chipAnton Ertl
|       | | +* Re: My experience with Apple M1 chipBranimir Maksimovic
|       | | |+* Re: My experience with Apple M1 chipKent Dickey
|       | | ||`- Re: My experience with Apple M1 chipBranimir Maksimovic
|       | | |`* Re: My experience with Apple M1 chipAnton Ertl
|       | | | +* Re: My experience with Apple M1 chipBranimir Maksimovic
|       | | | |`* Re: My experience with Apple M1 chipAnton Ertl
|       | | | | `* Re: My experience with Apple M1 chipJohn Dallman
|       | | | |  +- Re: My experience with Apple M1 chipBranimir Maksimovic
|       | | | |  `- Re: My experience with Apple M1 chipAnton Ertl
|       | | | `* Re: My experience with Apple M1 chipStefan Monnier
|       | | |  `- Re: My experience with Apple M1 chipMarcus
|       | | `* Re: My experience with Apple M1 chipMichael S
|       | |  `* Re: My experience with Apple M1 chipAnton Ertl
|       | |   `- Re: My experience with Apple M1 chipMichael S
|       | +* Re: My experience with Apple M1 chipKent Dickey
|       | |`* Re: My experience with Apple M1 chipMichael S
|       | | +- Re: My experience with Apple M1 chipBranimir Maksimovic
|       | | `* Re: My experience with Apple M1 chipJohn Dallman
|       | |  `- Re: My experience with Apple M1 chipBranimir Maksimovic
|       | `* Re: My experience with Apple M1 chipMichael S
|       |  `* Re: My experience with Apple M1 chipKent Dickey
|       |   `* Re: My experience with Apple M1 chipKent Dickey
|       |    `* Re: My experience with Apple M1 chipMichael S
|       |     `* Re: My experience with Apple M1 chipBranimir Maksimovic
|       |      `* Re: My experience with Apple M1 chipMichael S
|       |       `- Re: My experience with Apple M1 chipBranimir Maksimovic
|       `- Re: My experience with Apple M1 chipBranimir Maksimovic
`* Re: My experience with Apple M1 chipChris M. Thomasson
 +* Re: My experience with Apple M1 chipBranimir Maksimovic
 |`- Re: My experience with Apple M1 chipChris M. Thomasson
 `* Re: My experience with Apple M1 chipKent Dickey
  `- Re: My experience with Apple M1 chipChris M. Thomasson

Pages:123
My experience with Apple M1 chip

<muhEI.26121$P64.11471@fx47.iad>

 copy mid

https://www.novabbs.com/devel/article-flat.php?id=18357&group=comp.arch#18357

 copy link   Newsgroups: comp.arch
Path: i2pn2.org!i2pn.org!weretis.net!feeder6.news.weretis.net!4.us.feeder.erje.net!2.eu.feeder.erje.net!feeder.erje.net!feeder1.feed.usenet.farm!feed.usenet.farm!newsreader4.netcologne.de!news.netcologne.de!peer02.ams1!peer.ams1.xlned.com!news.xlned.com!peer02.iad!feed-me.highwinds-media.com!news.highwinds-media.com!fx47.iad.POSTED!not-for-mail
Newsgroups: comp.arch
From: branimir...@gmail.com (Branimir Maksimovic)
Subject: My experience with Apple M1 chip
User-Agent: slrn/1.0.3 (Darwin)
Lines: 10
Message-ID: <muhEI.26121$P64.11471@fx47.iad>
X-Complaints-To: abuse@usenet-news.net
NNTP-Posting-Date: Sun, 04 Jul 2021 12:05:38 UTC
Organization: usenet-news.net
Date: Sun, 04 Jul 2021 12:05:38 GMT
X-Received-Bytes: 1013
 by: Branimir Maksimovic - Sun, 4 Jul 2021 12:05 UTC

Fantastic chip, blows away my 2700X in single thread.
3950 scimark2 score vs 3050 2700X.
But, chip scales badly. Tried WCG and when running 2 threads no slowdown.
However with 3 threads slowdown iz 25%, with 4 threads slowdown is ~50%.
So chip is actually like i3 with two cores+HT.
But it is much faster as single thread perfomance is fantastic.
What is puzzling is that 5-7 thread does not have such slowdown,
that is on low power cores performance loss per thread is more like 10-15%.

Re: My experience with Apple M1 chip

<wapEI.21065$5J1.10275@fx08.iad>

 copy mid

https://www.novabbs.com/devel/article-flat.php?id=18363&group=comp.arch#18363

 copy link   Newsgroups: comp.arch
Path: i2pn2.org!i2pn.org!weretis.net!feeder8.news.weretis.net!newsreader4.netcologne.de!news.netcologne.de!peer03.ams1!peer.ams1.xlned.com!news.xlned.com!peer02.iad!feed-me.highwinds-media.com!news.highwinds-media.com!fx08.iad.POSTED!not-for-mail
Newsgroups: comp.arch
From: branimir...@gmail.com (Branimir Maksimovic)
Subject: Re: My experience with Apple M1 chip
References: <muhEI.26121$P64.11471@fx47.iad>
User-Agent: slrn/1.0.3 (Darwin)
Lines: 13
Message-ID: <wapEI.21065$5J1.10275@fx08.iad>
X-Complaints-To: abuse@usenet-news.net
NNTP-Posting-Date: Sun, 04 Jul 2021 20:50:36 UTC
Organization: usenet-news.net
Date: Sun, 04 Jul 2021 20:50:36 GMT
X-Received-Bytes: 1288
 by: Branimir Maksimovic - Sun, 4 Jul 2021 20:50 UTC

On 2021-07-04, Branimir Maksimovic <branimir.maksimovic@gmail.com> wrote:
>
> Fantastic chip, blows away my 2700X in single thread.
> 3950 scimark2 score vs 3050 2700X.
> But, chip scales badly. Tried WCG and when running 2 threads no slowdown.
> However with 3 threads slowdown iz 25%, with 4 threads slowdown is ~50%.
> So chip is actually like i3 with two cores+HT.
> But it is much faster as single thread perfomance is fantastic.
> What is puzzling is that 5-7 thread does not have such slowdown,
> that is on low power cores performance loss per thread is more like 10-15%.
>
Oh completely forgot to account turbo boost. perhaps macOS lowers freq with
each core running? How can I monitor freq on macOS?

Re: My experience with Apple M1 chip

<GSpEI.6041$5S.1356@fx02.iad>

 copy mid

https://www.novabbs.com/devel/article-flat.php?id=18366&group=comp.arch#18366

 copy link   Newsgroups: comp.arch
Path: i2pn2.org!i2pn.org!weretis.net!feeder8.news.weretis.net!newsreader4.netcologne.de!news.netcologne.de!peer01.ams1!peer.ams1.xlned.com!news.xlned.com!peer03.iad!feed-me.highwinds-media.com!news.highwinds-media.com!fx02.iad.POSTED!not-for-mail
Newsgroups: comp.arch
From: branimir...@gmail.com (Branimir Maksimovic)
Subject: Re: My experience with Apple M1 chip
References: <muhEI.26121$P64.11471@fx47.iad> <wapEI.21065$5J1.10275@fx08.iad>
User-Agent: slrn/1.0.3 (Darwin)
Lines: 19
Message-ID: <GSpEI.6041$5S.1356@fx02.iad>
X-Complaints-To: abuse@usenet-news.net
NNTP-Posting-Date: Sun, 04 Jul 2021 21:37:42 UTC
Organization: usenet-news.net
Date: Sun, 04 Jul 2021 21:37:42 GMT
X-Received-Bytes: 1679
 by: Branimir Maksimovic - Sun, 4 Jul 2021 21:37 UTC

On 2021-07-04, Branimir Maksimovic <branimir.maksimovic@gmail.com> wrote:
> On 2021-07-04, Branimir Maksimovic <branimir.maksimovic@gmail.com> wrote:
>>
>> Fantastic chip, blows away my 2700X in single thread.
>> 3950 scimark2 score vs 3050 2700X.
>> But, chip scales badly. Tried WCG and when running 2 threads no slowdown.
>> However with 3 threads slowdown iz 25%, with 4 threads slowdown is ~50%.
>> So chip is actually like i3 with two cores+HT.
>> But it is much faster as single thread perfomance is fantastic.
>> What is puzzling is that 5-7 thread does not have such slowdown,
>> that is on low power cores performance loss per thread is more like 10-15%.
>>
> Oh completely forgot to account turbo boost. perhaps macOS lowers freq with
> each core running? How can I monitor freq on macOS?

Confirmed, it lowers down clock, slowdown is not momental rather after
some time, perhaps triggered by temp. Apple didn't account that someone
will run WCG on laptop without fan :P
I can't find out freq neither control it and Google is useless for new M1 chips :(

Re: My experience with Apple M1 chip

<WcsEI.33014$NP.30428@fx42.iad>

 copy mid

https://www.novabbs.com/devel/article-flat.php?id=18375&group=comp.arch#18375

 copy link   Newsgroups: comp.arch
Path: i2pn2.org!i2pn.org!aioe.org!news.uzoreto.com!newsfeed.xs4all.nl!newsfeed7.news.xs4all.nl!news-out.netnews.com!news.alt.net!fdc2.netnews.com!peer03.ams1!peer.ams1.xlned.com!news.xlned.com!peer01.iad!feed-me.highwinds-media.com!news.highwinds-media.com!fx42.iad.POSTED!not-for-mail
Newsgroups: comp.arch
From: branimir...@gmail.com (Branimir Maksimovic)
Subject: Re: My experience with Apple M1 chip
References: <muhEI.26121$P64.11471@fx47.iad>
<wapEI.21065$5J1.10275@fx08.iad> <GSpEI.6041$5S.1356@fx02.iad>
User-Agent: slrn/1.0.3 (Darwin)
Lines: 24
Message-ID: <WcsEI.33014$NP.30428@fx42.iad>
X-Complaints-To: abuse@usenet-news.net
NNTP-Posting-Date: Mon, 05 Jul 2021 00:17:58 UTC
Organization: usenet-news.net
Date: Mon, 05 Jul 2021 00:17:58 GMT
X-Received-Bytes: 1942
 by: Branimir Maksimovic - Mon, 5 Jul 2021 00:17 UTC

On 2021-07-04, Branimir Maksimovic <branimir.maksimovic@gmail.com> wrote:
> On 2021-07-04, Branimir Maksimovic <branimir.maksimovic@gmail.com> wrote:
>> On 2021-07-04, Branimir Maksimovic <branimir.maksimovic@gmail.com> wrote:
>>>
>>> Fantastic chip, blows away my 2700X in single thread.
>>> 3950 scimark2 score vs 3050 2700X.
>>> But, chip scales badly. Tried WCG and when running 2 threads no slowdown.
>>> However with 3 threads slowdown iz 25%, with 4 threads slowdown is ~50%.
>>> So chip is actually like i3 with two cores+HT.
>>> But it is much faster as single thread perfomance is fantastic.
>>> What is puzzling is that 5-7 thread does not have such slowdown,
>>> that is on low power cores performance loss per thread is more like 10-15%.
>>>
>> Oh completely forgot to account turbo boost. perhaps macOS lowers freq with
>> each core running? How can I monitor freq on macOS?
>
> Confirmed, it lowers down clock, slowdown is not momental rather after
> some time, perhaps triggered by temp. Apple didn't account that someone
> will run WCG on laptop without fan :P
> I can't find out freq neither control it and Google is useless for new M1 chips :(

Yes, that is, found command line tool for monitoring frequency. It lowers freq greatly
when more then 2 cores are loaded :P

Re: My experience with Apple M1 chip

<sbtkp3$1i4$1@gioia.aioe.org>

 copy mid

https://www.novabbs.com/devel/article-flat.php?id=18376&group=comp.arch#18376

 copy link   Newsgroups: comp.arch
Path: i2pn2.org!i2pn.org!aioe.org!NBiuIU74OKL7NpIOsbuNjQ.user.gioia.aioe.org.POSTED!not-for-mail
From: chris.m....@gmail.com (Chris M. Thomasson)
Newsgroups: comp.arch
Subject: Re: My experience with Apple M1 chip
Date: Sun, 4 Jul 2021 17:46:30 -0700
Organization: Aioe.org NNTP Server
Lines: 12
Message-ID: <sbtkp3$1i4$1@gioia.aioe.org>
References: <muhEI.26121$P64.11471@fx47.iad>
NNTP-Posting-Host: NBiuIU74OKL7NpIOsbuNjQ.user.gioia.aioe.org
Mime-Version: 1.0
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 7bit
X-Complaints-To: abuse@aioe.org
User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:78.0) Gecko/20100101
Thunderbird/78.11.0
X-Notice: Filtered by postfilter v. 0.9.2
Content-Language: en-US
 by: Chris M. Thomasson - Mon, 5 Jul 2021 00:46 UTC

On 7/4/2021 5:05 AM, Branimir Maksimovic wrote:
> Fantastic chip, blows away my 2700X in single thread.
> 3950 scimark2 score vs 3050 2700X.
> But, chip scales badly. Tried WCG and when running 2 threads no slowdown.
> However with 3 threads slowdown iz 25%, with 4 threads slowdown is ~50%.
> So chip is actually like i3 with two cores+HT.
> But it is much faster as single thread perfomance is fantastic.
> What is puzzling is that 5-7 thread does not have such slowdown,
> that is on low power cores performance loss per thread is more like 10-15%.
>

Were you making sure to properly pad and align your data structures?

Re: My experience with Apple M1 chip

<13uEI.28078$Vj7.1721@fx46.iad>

 copy mid

https://www.novabbs.com/devel/article-flat.php?id=18377&group=comp.arch#18377

 copy link   Newsgroups: comp.arch
Path: i2pn2.org!i2pn.org!aioe.org!news.uzoreto.com!feeder5.feed.usenet.farm!feeder1.feed.usenet.farm!feed.usenet.farm!peer02.ams4!peer.am4.highwinds-media.com!peer01.iad!feed-me.highwinds-media.com!news.highwinds-media.com!fx46.iad.POSTED!not-for-mail
Newsgroups: comp.arch
From: branimir...@gmail.com (Branimir Maksimovic)
Subject: Re: My experience with Apple M1 chip
References: <muhEI.26121$P64.11471@fx47.iad> <sbtkp3$1i4$1@gioia.aioe.org>
User-Agent: slrn/1.0.3 (Darwin)
Lines: 19
Message-ID: <13uEI.28078$Vj7.1721@fx46.iad>
X-Complaints-To: abuse@usenet-news.net
NNTP-Posting-Date: Mon, 05 Jul 2021 02:23:57 UTC
Organization: usenet-news.net
Date: Mon, 05 Jul 2021 02:23:57 GMT
X-Received-Bytes: 1557
 by: Branimir Maksimovic - Mon, 5 Jul 2021 02:23 UTC

On 2021-07-05, Chris M. Thomasson <chris.m.thomasson.1@gmail.com> wrote:
> On 7/4/2021 5:05 AM, Branimir Maksimovic wrote:
>> Fantastic chip, blows away my 2700X in single thread.
>> 3950 scimark2 score vs 3050 2700X.
>> But, chip scales badly. Tried WCG and when running 2 threads no slowdown.
>> However with 3 threads slowdown iz 25%, with 4 threads slowdown is ~50%.
>> So chip is actually like i3 with two cores+HT.
>> But it is much faster as single thread perfomance is fantastic.
>> What is puzzling is that 5-7 thread does not have such slowdown,
>> that is on low power cores performance loss per thread is more like 10-15%.
>>
>
> Were you making sure to properly pad and align your data structures?

I just run WCG :P
Problem is solved Apple lowers down frequency to 1200-2000Mhz when running on more
then 2 cores :P
You know default freq iz 3200Mhz :P
I guess this laptop without fan would melt with WCG running on all 8 cores :P

Re: My experience with Apple M1 chip

<sbuc61$7ip$1@newsreader4.netcologne.de>

 copy mid

https://www.novabbs.com/devel/article-flat.php?id=18389&group=comp.arch#18389

 copy link   Newsgroups: comp.arch
Path: i2pn2.org!i2pn.org!paganini.bofh.team!news.dns-netz.com!news.freedyn.net!newsreader4.netcologne.de!news.netcologne.de!.POSTED.2001-4dd7-30c8-0-7285-c2ff-fe6c-992d.ipv6dyn.netcologne.de!not-for-mail
From: tkoe...@netcologne.de (Thomas Koenig)
Newsgroups: comp.arch
Subject: Re: My experience with Apple M1 chip
Date: Mon, 5 Jul 2021 07:25:53 -0000 (UTC)
Organization: news.netcologne.de
Distribution: world
Message-ID: <sbuc61$7ip$1@newsreader4.netcologne.de>
References: <muhEI.26121$P64.11471@fx47.iad> <sbtkp3$1i4$1@gioia.aioe.org>
<13uEI.28078$Vj7.1721@fx46.iad>
Injection-Date: Mon, 5 Jul 2021 07:25:53 -0000 (UTC)
Injection-Info: newsreader4.netcologne.de; posting-host="2001-4dd7-30c8-0-7285-c2ff-fe6c-992d.ipv6dyn.netcologne.de:2001:4dd7:30c8:0:7285:c2ff:fe6c:992d";
logging-data="7769"; mail-complaints-to="abuse@netcologne.de"
User-Agent: slrn/1.0.3 (Linux)
 by: Thomas Koenig - Mon, 5 Jul 2021 07:25 UTC

Branimir Maksimovic <branimir.maksimovic@gmail.com> schrieb:

> I just run WCG :P

Wikipedia has to offer:

World Choir Games

World Cyber Games, an international e-sports event

World Combat Games, an international multi-sport event featuring
combat sports and martial arts

Wide Color Gamut, a certain complete subset of colors

World Community Grid, a public computing grid for scientific
research projects

Worldwide Church of God, which changed its name to Grace Communion
International in 2009

WCG (firm), an advertising agency

WCG (College), a group of colleges based in the United Kingdom

WCG (Wide DC electric goods), a classification of Indian locomotives

Which of these, if any, do you mean?

Re: My experience with Apple M1 chip

<PODEI.9021$t64.2085@fx13.iad>

 copy mid

https://www.novabbs.com/devel/article-flat.php?id=18393&group=comp.arch#18393

 copy link   Newsgroups: comp.arch
Path: i2pn2.org!i2pn.org!weretis.net!feeder8.news.weretis.net!feeds.phibee-telecom.net!newsfeed.xs4all.nl!newsfeed7.news.xs4all.nl!news-out.netnews.com!news.alt.net!fdc2.netnews.com!peer02.ams1!peer.ams1.xlned.com!news.xlned.com!peer02.iad!feed-me.highwinds-media.com!news.highwinds-media.com!fx13.iad.POSTED!not-for-mail
Newsgroups: comp.arch
From: branimir...@gmail.com (Branimir Maksimovic)
Subject: Re: My experience with Apple M1 chip
References: <muhEI.26121$P64.11471@fx47.iad> <sbtkp3$1i4$1@gioia.aioe.org>
<13uEI.28078$Vj7.1721@fx46.iad> <sbuc61$7ip$1@newsreader4.netcologne.de>
User-Agent: slrn/1.0.3 (Darwin)
Lines: 34
Message-ID: <PODEI.9021$t64.2085@fx13.iad>
X-Complaints-To: abuse@usenet-news.net
NNTP-Posting-Date: Mon, 05 Jul 2021 13:29:19 UTC
Organization: usenet-news.net
Date: Mon, 05 Jul 2021 13:29:19 GMT
X-Received-Bytes: 1643
 by: Branimir Maksimovic - Mon, 5 Jul 2021 13:29 UTC

On 2021-07-05, Thomas Koenig <tkoenig@netcologne.de> wrote:
> Branimir Maksimovic <branimir.maksimovic@gmail.com> schrieb:
>
>> I just run WCG :P
>
> Wikipedia has to offer:
>
> World Choir Games
>
> World Cyber Games, an international e-sports event
>
> World Combat Games, an international multi-sport event featuring
> combat sports and martial arts
>
> Wide Color Gamut, a certain complete subset of colors
>
> World Community Grid, a public computing grid for scientific
> research projects
>
> Worldwide Church of God, which changed its name to Grace Communion
> International in 2009
>
> WCG (firm), an advertising agency
>
> WCG (College), a group of colleges based in the United Kingdom
>
> WCG (Wide DC electric goods), a classification of Indian locomotives
>
> Which of these, if any, do you mean?

Oh, I thought IBM WCG is widely known ;p
https://www.worldcommunitygrid.org/index.jsp
it puts stress on CPU more then prime95 :P

Re: My experience with Apple M1 chip

<93b7685f-2b1a-4ad1-941a-00a44e89b7c3n@googlegroups.com>

 copy mid

https://www.novabbs.com/devel/article-flat.php?id=18398&group=comp.arch#18398

 copy link   Newsgroups: comp.arch
X-Received: by 2002:a37:bb44:: with SMTP id l65mr16879668qkf.70.1625531305209;
Mon, 05 Jul 2021 17:28:25 -0700 (PDT)
X-Received: by 2002:a05:6808:107:: with SMTP id b7mr401631oie.44.1625531304962;
Mon, 05 Jul 2021 17:28:24 -0700 (PDT)
Path: i2pn2.org!i2pn.org!paganini.bofh.team!usenet.pasdenom.info!usenet-fr.net!proxad.net!feeder1-2.proxad.net!209.85.160.216.MISMATCH!news-out.google.com!nntp.google.com!postnews.google.com!google-groups.googlegroups.com!not-for-mail
Newsgroups: comp.arch
Date: Mon, 5 Jul 2021 17:28:24 -0700 (PDT)
In-Reply-To: <PODEI.9021$t64.2085@fx13.iad>
Injection-Info: google-groups.googlegroups.com; posting-host=2001:56a:fa3c:a000:4863:27bf:6a4b:5558;
posting-account=1nOeKQkAAABD2jxp4Pzmx9Hx5g9miO8y
NNTP-Posting-Host: 2001:56a:fa3c:a000:4863:27bf:6a4b:5558
References: <muhEI.26121$P64.11471@fx47.iad> <sbtkp3$1i4$1@gioia.aioe.org>
<13uEI.28078$Vj7.1721@fx46.iad> <sbuc61$7ip$1@newsreader4.netcologne.de> <PODEI.9021$t64.2085@fx13.iad>
User-Agent: G2/1.0
MIME-Version: 1.0
Message-ID: <93b7685f-2b1a-4ad1-941a-00a44e89b7c3n@googlegroups.com>
Subject: Re: My experience with Apple M1 chip
From: jsav...@ecn.ab.ca (Quadibloc)
Injection-Date: Tue, 06 Jul 2021 00:28:25 +0000
Content-Type: text/plain; charset="UTF-8"
 by: Quadibloc - Tue, 6 Jul 2021 00:28 UTC

On Monday, July 5, 2021 at 7:29:22 AM UTC-6, Branimir Maksimovic wrote:

> Oh, I thought IBM WCG is widely known ;p
> https://www.worldcommunitygrid.org/index.jsp
> it puts stress on CPU more then prime95 :P

I had absolutely no idea that IBM was out there competing
with Folding@Home and the like.

I *had* heard of Garner Ted Armstrong, his father Herbert W. Armstrong,
and the hijacking of his church by traditional Protestants after his death.

John Savard

Re: My experience with Apple M1 chip

<rtudnVCjNr63hnn9nZ2dnUU78cWdnZ2d@supernews.com>

 copy mid

https://www.novabbs.com/devel/article-flat.php?id=18404&group=comp.arch#18404

 copy link   Newsgroups: comp.arch
Path: i2pn2.org!i2pn.org!weretis.net!feeder6.news.weretis.net!4.us.feeder.erje.net!2.eu.feeder.erje.net!feeder.erje.net!border1.nntp.ams1.giganews.com!nntp.giganews.com!buffer1.nntp.ams1.giganews.com!nntp.supernews.com!news.supernews.com.POSTED!not-for-mail
NNTP-Posting-Date: Tue, 06 Jul 2021 03:58:18 -0500
Sender: Andrew Haley <aph@zarquon.pink>
From: aph...@littlepinkcloud.invalid
Subject: Re: My experience with Apple M1 chip
Newsgroups: comp.arch
References: <muhEI.26121$P64.11471@fx47.iad> <sbtkp3$1i4$1@gioia.aioe.org> <13uEI.28078$Vj7.1721@fx46.iad>
User-Agent: tin/1.9.2-20070201 ("Dalaruan") (UNIX) (Linux/4.18.0-240.8.1.el8_3.x86_64 (x86_64))
Message-ID: <rtudnVCjNr63hnn9nZ2dnUU78cWdnZ2d@supernews.com>
Date: Tue, 06 Jul 2021 03:58:18 -0500
Lines: 26
X-Trace: sv3-6Sfau1nzAF4vm5I7R4eKKxxppPMSLMqqAN9Qf0s7mIG0y6h0y9DfNWLLMu1T4voW0NRzn0Y58Z0CyI2!0AVhNJE5PxFCEBfpkccHuJk7mFZNSe4P3FRg+eugFXn6b7l6pe0EftSBL5rEHHoff8ed1RleTvG5!+o1UkG5A
X-Complaints-To: www.supernews.com/docs/abuse.html
X-DMCA-Complaints-To: www.supernews.com/docs/dmca.html
X-Abuse-and-DMCA-Info: Please be sure to forward a copy of ALL headers
X-Abuse-and-DMCA-Info: Otherwise we will be unable to process your complaint properly
X-Postfilter: 1.3.40
X-Original-Bytes: 2322
 by: aph...@littlepinkcloud.invalid - Tue, 6 Jul 2021 08:58 UTC

Branimir Maksimovic <branimir.maksimovic@gmail.com> wrote:
> On 2021-07-05, Chris M. Thomasson <chris.m.thomasson.1@gmail.com> wrote:
>> On 7/4/2021 5:05 AM, Branimir Maksimovic wrote:
>>> Fantastic chip, blows away my 2700X in single thread.
>>> 3950 scimark2 score vs 3050 2700X.
>>> But, chip scales badly. Tried WCG and when running 2 threads no slowdown.
>>> However with 3 threads slowdown iz 25%, with 4 threads slowdown is ~50%.
>>> So chip is actually like i3 with two cores+HT.
>>> But it is much faster as single thread perfomance is fantastic.
>>> What is puzzling is that 5-7 thread does not have such slowdown,
>>> that is on low power cores performance loss per thread is more like 10-15%.
>>>
>>
>> Were you making sure to properly pad and align your data structures?
>
> I just run WCG :P
> Problem is solved Apple lowers down frequency to 1200-2000Mhz when running on more
> then 2 cores :P
> You know default freq iz 3200Mhz :P
> I guess this laptop without fan would melt with WCG running on all 8 cores :P

Yes, exactly: marginal thermal properties on Macbook Air. Macbook Pro
seems to be fine. By the way, did you even get a M1 release of WCG? I
can't find one, just Intel binaries.

Andrew.

Re: My experience with Apple M1 chip

<T6VEI.159$VU3.17@fx46.iad>

 copy mid

https://www.novabbs.com/devel/article-flat.php?id=18405&group=comp.arch#18405

 copy link   Newsgroups: comp.arch
Path: i2pn2.org!i2pn.org!weretis.net!feeder8.news.weretis.net!newsreader4.netcologne.de!news.netcologne.de!peer02.ams1!peer.ams1.xlned.com!news.xlned.com!peer02.iad!feed-me.highwinds-media.com!news.highwinds-media.com!fx46.iad.POSTED!not-for-mail
Newsgroups: comp.arch
From: branimir...@gmail.com (Branimir Maksimovic)
Subject: Re: My experience with Apple M1 chip
References: <muhEI.26121$P64.11471@fx47.iad> <sbtkp3$1i4$1@gioia.aioe.org>
<13uEI.28078$Vj7.1721@fx46.iad>
<rtudnVCjNr63hnn9nZ2dnUU78cWdnZ2d@supernews.com>
User-Agent: slrn/1.0.3 (Darwin)
Lines: 38
Message-ID: <T6VEI.159$VU3.17@fx46.iad>
X-Complaints-To: abuse@usenet-news.net
NNTP-Posting-Date: Tue, 06 Jul 2021 09:11:15 UTC
Organization: usenet-news.net
Date: Tue, 06 Jul 2021 09:11:15 GMT
X-Received-Bytes: 2234
 by: Branimir Maksimovic - Tue, 6 Jul 2021 09:11 UTC

On 2021-07-06, aph@littlepinkcloud.invalid <aph@littlepinkcloud.invalid> wrote:
> Branimir Maksimovic <branimir.maksimovic@gmail.com> wrote:
>> On 2021-07-05, Chris M. Thomasson <chris.m.thomasson.1@gmail.com> wrote:
>>> On 7/4/2021 5:05 AM, Branimir Maksimovic wrote:
>>>> Fantastic chip, blows away my 2700X in single thread.
>>>> 3950 scimark2 score vs 3050 2700X.
>>>> But, chip scales badly. Tried WCG and when running 2 threads no slowdown.
>>>> However with 3 threads slowdown iz 25%, with 4 threads slowdown is ~50%.
>>>> So chip is actually like i3 with two cores+HT.
>>>> But it is much faster as single thread perfomance is fantastic.
>>>> What is puzzling is that 5-7 thread does not have such slowdown,
>>>> that is on low power cores performance loss per thread is more like 10-15%.
>>>>
>>>
>>> Were you making sure to properly pad and align your data structures?
>>
>> I just run WCG :P
>> Problem is solved Apple lowers down frequency to 1200-2000Mhz when running on more
>> then 2 cores :P
>> You know default freq iz 3200Mhz :P
>> I guess this laptop without fan would melt with WCG running on all 8 cores :P
>
> Yes, exactly: marginal thermal properties on Macbook Air. Macbook Pro
> seems to be fine.
Fine in what way? It has fan?

By the way, did you even get a M1 release of WCG? I
> can't find one, just Intel binaries.

Rosetta takes care of that...
Binaries are tweaked for M1 as Rosetta doesn't support AVX.
There is no performance loss, for what I tested...
>
> Andrew.

--
something dumb

Re: My experience with Apple M1 chip

<oJWdnayjuspXr3n9nZ2dnUU78SXNnZ2d@supernews.com>

 copy mid

https://www.novabbs.com/devel/article-flat.php?id=18407&group=comp.arch#18407

 copy link   Newsgroups: comp.arch
Path: i2pn2.org!i2pn.org!aioe.org!news.mixmin.net!border2.nntp.ams1.giganews.com!nntp.giganews.com!buffer2.nntp.ams1.giganews.com!nntp.supernews.com!news.supernews.com.POSTED!not-for-mail
NNTP-Posting-Date: Tue, 06 Jul 2021 05:39:06 -0500
Sender: Andrew Haley <aph@zarquon.pink>
From: aph...@littlepinkcloud.invalid
Subject: Re: My experience with Apple M1 chip
Newsgroups: comp.arch
References: <muhEI.26121$P64.11471@fx47.iad> <sbtkp3$1i4$1@gioia.aioe.org> <13uEI.28078$Vj7.1721@fx46.iad> <rtudnVCjNr63hnn9nZ2dnUU78cWdnZ2d@supernews.com> <T6VEI.159$VU3.17@fx46.iad>
User-Agent: tin/1.9.2-20070201 ("Dalaruan") (UNIX) (Linux/4.18.0-240.8.1.el8_3.x86_64 (x86_64))
Message-ID: <oJWdnayjuspXr3n9nZ2dnUU78SXNnZ2d@supernews.com>
Date: Tue, 06 Jul 2021 05:39:06 -0500
Lines: 23
X-Trace: sv3-NPXvT3P0Vd1ST+YTEGII/F3Wm/suj4LkWxey+XG/d3UwPsCeoOye3FLOiSNu5iJJIMFE9TvvZ/+Nzyl!eevxdNlBU1uKGDPolEanaclEwQfTDUvAXsXwaHeq8HRHg8L9yNlC8US4Oqc7kuLZaofrl9jqrhq0!pwsw83G/
X-Complaints-To: www.supernews.com/docs/abuse.html
X-DMCA-Complaints-To: www.supernews.com/docs/dmca.html
X-Abuse-and-DMCA-Info: Please be sure to forward a copy of ALL headers
X-Abuse-and-DMCA-Info: Otherwise we will be unable to process your complaint properly
X-Postfilter: 1.3.40
X-Original-Bytes: 1974
 by: aph...@littlepinkcloud.invalid - Tue, 6 Jul 2021 10:39 UTC

Branimir Maksimovic <branimir.maksimovic@gmail.com> wrote:
> On 2021-07-06, aph@littlepinkcloud.invalid <aph@littlepinkcloud.invalid> wrote:
>>> I guess this laptop without fan would melt with WCG running on all 8 cores :P
>>
>> Yes, exactly: marginal thermal properties on Macbook Air. Macbook Pro
>> seems to be fine.
> Fine in what way? It has fan?

Yes. That's one of the important differences.

>> By the way, did you even get a M1 release of WCG? I
>> can't find one, just Intel binaries.
>
> Rosetta takes care of that...

Well, sort of. Rosetta works very well, but it's not native code.

> Binaries are tweaked for M1 as Rosetta doesn't support AVX.
> There is no performance loss, for what I tested...

How can you tell what the performance loss is without native binaries?

Andrew.

Re: My experience with Apple M1 chip

<memo.20210706143219.6784R@jgd.cix.co.uk>

 copy mid

https://www.novabbs.com/devel/article-flat.php?id=18413&group=comp.arch#18413

 copy link   Newsgroups: comp.arch
Path: i2pn2.org!i2pn.org!eternal-september.org!reader02.eternal-september.org!.POSTED!not-for-mail
From: jgd...@cix.co.uk (John Dallman)
Newsgroups: comp.arch
Subject: Re: My experience with Apple M1 chip
Date: Tue, 6 Jul 2021 14:32 +0100 (BST)
Organization: A noiseless patient Spider
Lines: 16
Message-ID: <memo.20210706143219.6784R@jgd.cix.co.uk>
References: <T6VEI.159$VU3.17@fx46.iad>
Reply-To: jgd@cix.co.uk
Injection-Info: reader02.eternal-september.org; posting-host="98fd09aedc2764985b29175facddca86";
logging-data="32664"; mail-complaints-to="abuse@eternal-september.org"; posting-account="U2FsdGVkX19mxnXNnZTbmhJv3VVDkSpUyopE/1V82vc="
Cancel-Lock: sha1:VFSYbYPt6q2Bg8KSaby0ytsreJQ=
 by: John Dallman - Tue, 6 Jul 2021 13:32 UTC

In article <T6VEI.159$VU3.17@fx46.iad>, branimir.maksimovic@gmail.com
(Branimir Maksimovic) wrote:

> Rosetta takes care of that...
> Binaries are tweaked for M1 as Rosetta doesn't support AVX.
> There is no performance loss, for what I tested...

Rosetta typically runs Intel macOS binaries on an M1 as fast as an Intel
Mac. However, native M1 binaries are usually considerably faster than
that. The M1 is really quite a good microprocessor.

I'm building the same software and running the same tests: an Intel-based
Macmini8,1 takes about 12.5 hours for the job that an M1-based Macmini9,1
does in a little over 7 hours.

John

Re: My experience with Apple M1 chip

<koZEI.3417$rr3.341@fx34.iad>

 copy mid

https://www.novabbs.com/devel/article-flat.php?id=18415&group=comp.arch#18415

 copy link   Newsgroups: comp.arch
Path: i2pn2.org!i2pn.org!weretis.net!feeder8.news.weretis.net!feeder5.feed.usenet.farm!feeder1.feed.usenet.farm!feed.usenet.farm!tr1.eu1.usenetexpress.com!feeder.usenetexpress.com!tr2.iad1.usenetexpress.com!peer03.iad!feed-me.highwinds-media.com!news.highwinds-media.com!fx34.iad.POSTED!not-for-mail
Newsgroups: comp.arch
From: branimir...@gmail.com (Branimir Maksimovic)
Subject: Re: My experience with Apple M1 chip
References: <muhEI.26121$P64.11471@fx47.iad> <sbtkp3$1i4$1@gioia.aioe.org> <13uEI.28078$Vj7.1721@fx46.iad> <rtudnVCjNr63hnn9nZ2dnUU78cWdnZ2d@supernews.com> <T6VEI.159$VU3.17@fx46.iad> <oJWdnayjuspXr3n9nZ2dnUU78SXNnZ2d@supernews.com>
User-Agent: slrn/1.0.3 (Darwin)
Lines: 37
Message-ID: <koZEI.3417$rr3.341@fx34.iad>
X-Complaints-To: abuse@usenet-news.net
NNTP-Posting-Date: Tue, 06 Jul 2021 14:02:56 UTC
Organization: usenet-news.net
Date: Tue, 06 Jul 2021 14:02:56 GMT
X-Received-Bytes: 1912
 by: Branimir Maksimovic - Tue, 6 Jul 2021 14:02 UTC

On 2021-07-06, aph@littlepinkcloud.invalid <aph@littlepinkcloud.invalid> wrote:
> Branimir Maksimovic <branimir.maksimovic@gmail.com> wrote:
>> On 2021-07-06, aph@littlepinkcloud.invalid <aph@littlepinkcloud.invalid> wrote:
>>>> I guess this laptop without fan would melt with WCG running on all 8 cores :P
>>>
>>> Yes, exactly: marginal thermal properties on Macbook Air. Macbook Pro
>>> seems to be fine.
>> Fine in what way? It has fan?
>
> Yes. That's one of the important differences.

It's same chip, with fan it can sustain higher
clock with heavy load, then.
>
>>> By the way, did you even get a M1 release of WCG? I
>>> can't find one, just Intel binaries.
>>
>> Rosetta takes care of that...
>
> Well, sort of. Rosetta works very well, but it's not native code.
>
>> Binaries are tweaked for M1 as Rosetta doesn't support AVX.
>> There is no performance loss, for what I tested...
>
> How can you tell what the performance loss is without native binaries?

Compiled native and x86 some benchmarks, same some even fasterthroug rosetta :P.
Also comparable with x86 task times.
M1 @ 1600Mhz produces same times as 2700X @ 2200Mhz for WCG tasks.
Some even faster.

>
> Andrew.

--
something dumb

Re: My experience with Apple M1 chip

<PpZEI.3566$rr3.2561@fx34.iad>

 copy mid

https://www.novabbs.com/devel/article-flat.php?id=18416&group=comp.arch#18416

 copy link   Newsgroups: comp.arch
Path: i2pn2.org!i2pn.org!aioe.org!feeder1.feed.usenet.farm!feed.usenet.farm!news-out.netnews.com!news.alt.net!fdc3.netnews.com!peer03.ams1!peer.ams1.xlned.com!news.xlned.com!peer02.iad!feed-me.highwinds-media.com!news.highwinds-media.com!fx34.iad.POSTED!not-for-mail
Newsgroups: comp.arch
From: branimir...@gmail.com (Branimir Maksimovic)
Subject: Re: My experience with Apple M1 chip
References: <T6VEI.159$VU3.17@fx46.iad>
<memo.20210706143219.6784R@jgd.cix.co.uk>
User-Agent: slrn/1.0.3 (Darwin)
Lines: 15
Message-ID: <PpZEI.3566$rr3.2561@fx34.iad>
X-Complaints-To: abuse@usenet-news.net
NNTP-Posting-Date: Tue, 06 Jul 2021 14:04:31 UTC
Organization: usenet-news.net
Date: Tue, 06 Jul 2021 14:04:31 GMT
X-Received-Bytes: 1148
 by: Branimir Maksimovic - Tue, 6 Jul 2021 14:04 UTC

On 2021-07-06, John Dallman <jgd@cix.co.uk> wrote:
> In article <T6VEI.159$VU3.17@fx46.iad>, branimir.maksimovic@gmail.com
> (Branimir Maksimovic) wrote:
>
>> Rosetta takes care of that...
>> Binaries are tweaked for M1 as Rosetta doesn't support AVX.
>> There is no performance loss, for what I tested...
>
> Rosetta typically runs Intel macOS binaries on an M1 as fast as an Intel
> Mac. However, native M1 binaries are usually considerably faster than
> that. Or x86 binaries faster then native, depends :P

--
something dumb

Re: My experience with Apple M1 chip

<ZBZEI.5$o_5.3@fx29.iad>

 copy mid

https://www.novabbs.com/devel/article-flat.php?id=18418&group=comp.arch#18418

 copy link   Newsgroups: comp.arch
Path: i2pn2.org!i2pn.org!aioe.org!feeder5.feed.usenet.farm!feeder1.feed.usenet.farm!feed.usenet.farm!news-out.netnews.com!news.alt.net!fdc3.netnews.com!peer01.ams1!peer.ams1.xlned.com!news.xlned.com!peer01.iad!feed-me.highwinds-media.com!news.highwinds-media.com!fx29.iad.POSTED!not-for-mail
Newsgroups: comp.arch
From: branimir...@gmail.com (Branimir Maksimovic)
Subject: Re: My experience with Apple M1 chip
References: <muhEI.26121$P64.11471@fx47.iad> <sbtkp3$1i4$1@gioia.aioe.org>
<13uEI.28078$Vj7.1721@fx46.iad>
<rtudnVCjNr63hnn9nZ2dnUU78cWdnZ2d@supernews.com>
<T6VEI.159$VU3.17@fx46.iad>
<oJWdnayjuspXr3n9nZ2dnUU78SXNnZ2d@supernews.com>
<koZEI.3417$rr3.341@fx34.iad>
User-Agent: slrn/1.0.3 (Darwin)
Lines: 84
Message-ID: <ZBZEI.5$o_5.3@fx29.iad>
X-Complaints-To: abuse@usenet-news.net
NNTP-Posting-Date: Tue, 06 Jul 2021 14:17:29 UTC
Organization: usenet-news.net
Date: Tue, 06 Jul 2021 14:17:29 GMT
X-Received-Bytes: 3931
 by: Branimir Maksimovic - Tue, 6 Jul 2021 14:17 UTC

On 2021-07-06, Branimir Maksimovic <branimir.maksimovic@gmail.com> wrote:
> On 2021-07-06, aph@littlepinkcloud.invalid <aph@littlepinkcloud.invalid> wrote:
>> Branimir Maksimovic <branimir.maksimovic@gmail.com> wrote:
>>> On 2021-07-06, aph@littlepinkcloud.invalid <aph@littlepinkcloud.invalid> wrote:
>>>>> I guess this laptop without fan would melt with WCG running on all 8 cores :P
>>>>
>>>> Yes, exactly: marginal thermal properties on Macbook Air. Macbook Pro
>>>> seems to be fine.
>>> Fine in what way? It has fan?
>>
>> Yes. That's one of the important differences.
>
> It's same chip, with fan it can sustain higher
> clock with heavy load, then.
>>
>>>> By the way, did you even get a M1 release of WCG? I
>>>> can't find one, just Intel binaries.
>>>
>>> Rosetta takes care of that...
>>
>> Well, sort of. Rosetta works very well, but it's not native code.
>>
>>> Binaries are tweaked for M1 as Rosetta doesn't support AVX.
>>> There is no performance loss, for what I tested...
>>
>> How can you tell what the performance loss is without native binaries?
>
> Compiled native and x86 some benchmarks, same some even fasterthroug rosetta :P.
> Also comparable with x86 task times.
> M1 @ 1600Mhz produces same times as 2700X @ 2200Mhz for WCG tasks.
> Some even faster.
>
>>
>> Andrew.
>
>
eg:
bmaxa@Branimirs-Air nbody % g++ nbody.cpp -O3 -o nbody -std=c++11 -target x86_64-apple-darwin
bmaxa@Branimirs-Air nbody % g++ nbody.cpp -O3 -o nbodyx86 -std=c++11 -target x86_64-apple-darwin
bmaxa@Branimirs-Air nbody % g++ nbody.cpp -O3 -o nbodyxARM -std=c++11
bmaxa@Branimirs-Air nbody % time ./nbodyx86 50000000
-0.169075164
-0.169059907
../nbodyx86 50000000 1.69s user 0.01s system 82% cpu 2.066 total
bmaxa@Branimirs-Air nbody % time ./nbodyx86 50000000
-0.169075164
-0.169059907
../nbodyx86 50000000 1.69s user 0.01s system 99% cpu 1.703 total
bmaxa@Branimirs-Air nbody % time ./nbodyx86 50000000
-0.169075164
-0.169059907
../nbodyx86 50000000 1.69s user 0.01s system 99% cpu 1.698 total
bmaxa@Branimirs-Air nbody % time ./nbodyARM 50000000
-0.169075164
-0.169059907
../nbodyARM 50000000 1.58s user 0.01s system 98% cpu 1.614 total
bmaxa@Branimirs-Air nbody % time ./nbodyARM 50000000
-0.169075164
-0.169059907
../nbodyARM 50000000 1.57s user 0.01s system 99% cpu 1.582 total
bmaxa@Branimirs-Air nbody % time ./nbodyARM 50000000
-0.169075164
-0.169059907
../nbodyARM 50000000 1.57s user 0.00s system 99% cpu 1.580 total
bmaxa@Branimirs-Air nbody % time ./nbodyx86 50000000
-0.169075164
-0.169059907
../nbodyx86 50000000 1.69s user 0.01s system 99% cpu 1.706 total
bmaxa@Branimirs-Air nbody % time ./nbodyx86 50000000
-0.169075164
-0.169059907
../nbodyx86 50000000 1.69s user 0.01s system 99% cpu 1.705 total
bmaxa@Branimirs-Air nbody % time ./nbodyARM 50000000
-0.169075164
-0.169059907
../nbodyARM 50000000 1.57s user 0.01s system 99% cpu 1.584 total
bmaxa@Branimirs-Air nbody % file nbodyx86
nbodyx86: Mach-O 64-bit executable x86_64
bmaxa@Branimirs-Air nbody % file nbodyxARM
nbodyxARM: Mach-O 64-bit executable arm64

--
something dumb

Re: My experience with Apple M1 chip

<pHZEI.6$o_5.5@fx29.iad>

 copy mid

https://www.novabbs.com/devel/article-flat.php?id=18419&group=comp.arch#18419

 copy link   Newsgroups: comp.arch
Path: i2pn2.org!i2pn.org!aioe.org!news.uzoreto.com!news-out.netnews.com!news.alt.net!fdc3.netnews.com!peer01.ams1!peer.ams1.xlned.com!news.xlned.com!peer02.iad!feed-me.highwinds-media.com!news.highwinds-media.com!fx29.iad.POSTED!not-for-mail
Newsgroups: comp.arch
From: branimir...@gmail.com (Branimir Maksimovic)
Subject: Re: My experience with Apple M1 chip
References: <muhEI.26121$P64.11471@fx47.iad> <sbtkp3$1i4$1@gioia.aioe.org>
<13uEI.28078$Vj7.1721@fx46.iad>
<rtudnVCjNr63hnn9nZ2dnUU78cWdnZ2d@supernews.com>
<T6VEI.159$VU3.17@fx46.iad>
<oJWdnayjuspXr3n9nZ2dnUU78SXNnZ2d@supernews.com>
<koZEI.3417$rr3.341@fx34.iad> <ZBZEI.5$o_5.3@fx29.iad>
User-Agent: slrn/1.0.3 (Darwin)
Lines: 89
Message-ID: <pHZEI.6$o_5.5@fx29.iad>
X-Complaints-To: abuse@usenet-news.net
NNTP-Posting-Date: Tue, 06 Jul 2021 14:23:17 UTC
Organization: usenet-news.net
Date: Tue, 06 Jul 2021 14:23:17 GMT
X-Received-Bytes: 4348
 by: Branimir Maksimovic - Tue, 6 Jul 2021 14:23 UTC

On 2021-07-06, Branimir Maksimovic <branimir.maksimovic@gmail.com> wrote:
> On 2021-07-06, Branimir Maksimovic <branimir.maksimovic@gmail.com> wrote:
>> On 2021-07-06, aph@littlepinkcloud.invalid <aph@littlepinkcloud.invalid> wrote:
>>> Branimir Maksimovic <branimir.maksimovic@gmail.com> wrote:
>>>> On 2021-07-06, aph@littlepinkcloud.invalid <aph@littlepinkcloud.invalid> wrote:
>>>>>> I guess this laptop without fan would melt with WCG running on all 8 cores :P
>>>>>
>>>>> Yes, exactly: marginal thermal properties on Macbook Air. Macbook Pro
>>>>> seems to be fine.
>>>> Fine in what way? It has fan?
>>>
>>> Yes. That's one of the important differences.
>>
>> It's same chip, with fan it can sustain higher
>> clock with heavy load, then.
>>>
>>>>> By the way, did you even get a M1 release of WCG? I
>>>>> can't find one, just Intel binaries.
>>>>
>>>> Rosetta takes care of that...
>>>
>>> Well, sort of. Rosetta works very well, but it's not native code.
>>>
>>>> Binaries are tweaked for M1 as Rosetta doesn't support AVX.
>>>> There is no performance loss, for what I tested...
>>>
>>> How can you tell what the performance loss is without native binaries?
>>
>> Compiled native and x86 some benchmarks, same some even fasterthroug rosetta :P.
>> Also comparable with x86 task times.
>> M1 @ 1600Mhz produces same times as 2700X @ 2200Mhz for WCG tasks.
>> Some even faster.
>>
>>>
>>> Andrew.
>>
>>
> eg:
> bmaxa@Branimirs-Air nbody % g++ nbody.cpp -O3 -o nbody -std=c++11 -target x86_64-apple-darwin
> bmaxa@Branimirs-Air nbody % g++ nbody.cpp -O3 -o nbodyx86 -std=c++11 -target x86_64-apple-darwin
> bmaxa@Branimirs-Air nbody % g++ nbody.cpp -O3 -o nbodyxARM -std=c++11
> bmaxa@Branimirs-Air nbody % time ./nbodyx86 50000000
> -0.169075164
> -0.169059907
> ./nbodyx86 50000000 1.69s user 0.01s system 82% cpu 2.066 total
> bmaxa@Branimirs-Air nbody % time ./nbodyx86 50000000
> -0.169075164
> -0.169059907
> ./nbodyx86 50000000 1.69s user 0.01s system 99% cpu 1.703 total
> bmaxa@Branimirs-Air nbody % time ./nbodyx86 50000000
> -0.169075164
> -0.169059907
> ./nbodyx86 50000000 1.69s user 0.01s system 99% cpu 1.698 total
> bmaxa@Branimirs-Air nbody % time ./nbodyARM 50000000
> -0.169075164
> -0.169059907
> ./nbodyARM 50000000 1.58s user 0.01s system 98% cpu 1.614 total
> bmaxa@Branimirs-Air nbody % time ./nbodyARM 50000000
> -0.169075164
> -0.169059907
> ./nbodyARM 50000000 1.57s user 0.01s system 99% cpu 1.582 total
> bmaxa@Branimirs-Air nbody % time ./nbodyARM 50000000
> -0.169075164
> -0.169059907
> ./nbodyARM 50000000 1.57s user 0.00s system 99% cpu 1.580 total
> bmaxa@Branimirs-Air nbody % time ./nbodyx86 50000000
> -0.169075164
> -0.169059907
> ./nbodyx86 50000000 1.69s user 0.01s system 99% cpu 1.706 total
> bmaxa@Branimirs-Air nbody % time ./nbodyx86 50000000
> -0.169075164
> -0.169059907
> ./nbodyx86 50000000 1.69s user 0.01s system 99% cpu 1.705 total
> bmaxa@Branimirs-Air nbody % time ./nbodyARM 50000000
> -0.169075164
> -0.169059907
> ./nbodyARM 50000000 1.57s user 0.01s system 99% cpu 1.584 total
> bmaxa@Branimirs-Air nbody % file nbodyx86
> nbodyx86: Mach-O 64-bit executable x86_64
> bmaxa@Branimirs-Air nbody % file nbodyxARM
> nbodyxARM: Mach-O 64-bit executable arm64
>
This program is https://benchmarksgame-team.pages.debian.net/benchmarksgame/program/nbody-gpp-6.html
executes in 5.7 sec on ivy bridge at 3.2
On M1 you don't have to program Neon explicitelly :P

--
something dumb

Re: My experience with Apple M1 chip

<19dcc459-6eb5-4191-a186-c50d12ed347fn@googlegroups.com>

 copy mid

https://www.novabbs.com/devel/article-flat.php?id=18420&group=comp.arch#18420

 copy link   Newsgroups: comp.arch
X-Received: by 2002:a37:b6c5:: with SMTP id g188mr19698324qkf.92.1625585335359;
Tue, 06 Jul 2021 08:28:55 -0700 (PDT)
X-Received: by 2002:a4a:3305:: with SMTP id q5mr5486534ooq.83.1625585335035;
Tue, 06 Jul 2021 08:28:55 -0700 (PDT)
Path: i2pn2.org!i2pn.org!weretis.net!feeder8.news.weretis.net!proxad.net!feeder1-2.proxad.net!209.85.160.216.MISMATCH!news-out.google.com!nntp.google.com!postnews.google.com!google-groups.googlegroups.com!not-for-mail
Newsgroups: comp.arch
Date: Tue, 6 Jul 2021 08:28:54 -0700 (PDT)
In-Reply-To: <memo.20210706143219.6784R@jgd.cix.co.uk>
Injection-Info: google-groups.googlegroups.com; posting-host=199.203.251.52; posting-account=ow8VOgoAAAAfiGNvoH__Y4ADRwQF1hZW
NNTP-Posting-Host: 199.203.251.52
References: <T6VEI.159$VU3.17@fx46.iad> <memo.20210706143219.6784R@jgd.cix.co.uk>
User-Agent: G2/1.0
MIME-Version: 1.0
Message-ID: <19dcc459-6eb5-4191-a186-c50d12ed347fn@googlegroups.com>
Subject: Re: My experience with Apple M1 chip
From: already5...@yahoo.com (Michael S)
Injection-Date: Tue, 06 Jul 2021 15:28:55 +0000
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
 by: Michael S - Tue, 6 Jul 2021 15:28 UTC

On Tuesday, July 6, 2021 at 4:32:22 PM UTC+3, John Dallman wrote:
> In article <T6VEI.159$VU3...@fx46.iad>, branimir....@gmail.com
> (Branimir Maksimovic) wrote:
>
> > Rosetta takes care of that...
> > Binaries are tweaked for M1 as Rosetta doesn't support AVX.
> > There is no performance loss, for what I tested...
> Rosetta typically runs Intel macOS binaries on an M1 as fast as an Intel
> Mac. However, native M1 binaries are usually considerably faster than
> that. The M1 is really quite a good microprocessor.
>
> I'm building the same software and running the same tests: an Intel-based
> Macmini8,1 takes about 12.5 hours for the job that an M1-based Macmini9,1
> does in a little over 7 hours.
>
> John

8.1 and 9.1 sound like Mac OS versions.
What is actual HW in each case?
I can believe that M1@3.2 GHz/Rosetta is able to run x64 software as fast as i3-8100B but have trouble believing that it could match i7-8700B either in single thread or in multithread throughput. Unless, of course, absolute majority of run time spent in native libraries.

Re: My experience with Apple M1 chip

<xXB*b5roy@news.chiark.greenend.org.uk>

 copy mid

https://www.novabbs.com/devel/article-flat.php?id=18422&group=comp.arch#18422

 copy link   Newsgroups: comp.arch
Path: i2pn2.org!i2pn.org!paganini.bofh.team!newsfeed.xs3.de!nntp-feed.chiark.greenend.org.uk!ewrotcd!.POSTED!not-for-mail
From: theom+n...@chiark.greenend.org.uk (Theo)
Newsgroups: comp.arch
Subject: Re: My experience with Apple M1 chip
Date: 06 Jul 2021 16:38:47 +0100 (BST)
Organization: University of Cambridge, England
Lines: 35
Message-ID: <xXB*b5roy@news.chiark.greenend.org.uk>
References: <T6VEI.159$VU3.17@fx46.iad> <memo.20210706143219.6784R@jgd.cix.co.uk> <19dcc459-6eb5-4191-a186-c50d12ed347fn@googlegroups.com>
NNTP-Posting-Host: chiark.greenend.org.uk
X-Trace: chiark.greenend.org.uk 1625585929 3293 212.13.197.229 (6 Jul 2021 15:38:49 GMT)
X-Complaints-To: abuse@chiark.greenend.org.uk
NNTP-Posting-Date: Tue, 6 Jul 2021 15:38:49 +0000 (UTC)
User-Agent: tin/1.8.3-20070201 ("Scotasay") (UNIX) (Linux/3.16.0-7-amd64 (x86_64))
Originator: theom@chiark.greenend.org.uk ([212.13.197.229])
 by: Theo - Tue, 6 Jul 2021 15:38 UTC

Michael S <already5chosen@yahoo.com> wrote:
> On Tuesday, July 6, 2021 at 4:32:22 PM UTC+3, John Dallman wrote:
> > I'm building the same software and running the same tests: an Intel-based
> > Macmini8,1 takes about 12.5 hours for the job that an M1-based Macmini9,1
> > does in a little over 7 hours.
> >
>
> 8.1 and 9.1 sound like Mac OS versions.

No, they're hardware versions. (there's a comma in them, not a dot)

> What is actual HW in each case?

Macmini8,1
https://en.wikipedia.org/wiki/Mac_Mini#Fourth_generation_(Space_Gray)_2

Macmini9,1
https://en.wikipedia.org/wiki/Mac_Mini#Fifth_generation_(Apple_silicon)_2

although the Intel CPU is configurable.

> I can believe that M1@3.2 GHz/Rosetta is able to run x64 software as fast
> as i3-8100B but have trouble believing that it could match i7-8700B either
> in single thread or in multithread throughput. Unless, of course,
> absolute majority of run time spent in native libraries.

It's doing it by binary decompilation and recompilation, so it is possible
to do it with small or even negative overhead once that process is done.
The post upthread with times showed that process took 0.3 seconds on the
first run, and then after that the code was on par with native performance.

I'm surprised the recompile time was that quick, but it sounds like wasn't a
complicated binary in that instance.

Theo

Re: My experience with Apple M1 chip

<sc1to8$1cp$1@gal.iecc.com>

 copy mid

https://www.novabbs.com/devel/article-flat.php?id=18424&group=comp.arch#18424

 copy link   Newsgroups: comp.arch
Path: i2pn2.org!i2pn.org!weretis.net!feeder6.news.weretis.net!xmission!usenet.csail.mit.edu!news.iecc.com!.POSTED.news.iecc.com!not-for-mail
From: joh...@taugh.com (John Levine)
Newsgroups: comp.arch
Subject: Re: My experience with Apple M1 chip
Date: Tue, 6 Jul 2021 15:44:08 -0000 (UTC)
Organization: Taughannock Networks
Message-ID: <sc1to8$1cp$1@gal.iecc.com>
References: <T6VEI.159$VU3.17@fx46.iad> <memo.20210706143219.6784R@jgd.cix.co.uk> <19dcc459-6eb5-4191-a186-c50d12ed347fn@googlegroups.com>
Injection-Date: Tue, 6 Jul 2021 15:44:08 -0000 (UTC)
Injection-Info: gal.iecc.com; posting-host="news.iecc.com:2001:470:1f07:1126:0:676f:7373:6970";
logging-data="1433"; mail-complaints-to="abuse@iecc.com"
In-Reply-To: <T6VEI.159$VU3.17@fx46.iad> <memo.20210706143219.6784R@jgd.cix.co.uk> <19dcc459-6eb5-4191-a186-c50d12ed347fn@googlegroups.com>
Cleverness: some
X-Newsreader: trn 4.0-test77 (Sep 1, 2010)
Originator: johnl@iecc.com (John Levine)
 by: John Levine - Tue, 6 Jul 2021 15:44 UTC

According to Michael S <already5chosen@yahoo.com>:
>> I'm building the same software and running the same tests: an Intel-based
>> Macmini8,1 takes about 12.5 hours for the job that an M1-based Macmini9,1
>> does in a little over 7 hours.
>
>8.1 and 9.1 sound like Mac OS versions.

No, they are hardware model identifiers. See https://support.apple.com/en-us/HT201894

The 8.1 is an Intel Core i3 running at 3 or 3.6 GHz, the 9.1 is the M1.

--
Regards,
John Levine, johnl@taugh.com, Primary Perpetrator of "The Internet for Dummies",
Please consider the environment before reading this e-mail. https://jl.ly

Re: My experience with Apple M1 chip

<2l%EI.8$gE.2@fx21.iad>

 copy mid

https://www.novabbs.com/devel/article-flat.php?id=18427&group=comp.arch#18427

 copy link   Newsgroups: comp.arch
Path: i2pn2.org!i2pn.org!aioe.org!feeder1.feed.usenet.farm!feed.usenet.farm!news-out.netnews.com!news.alt.net!fdc3.netnews.com!peer03.ams1!peer.ams1.xlned.com!news.xlned.com!peer01.iad!feed-me.highwinds-media.com!news.highwinds-media.com!fx21.iad.POSTED!not-for-mail
Newsgroups: comp.arch
From: branimir...@gmail.com (Branimir Maksimovic)
Subject: Re: My experience with Apple M1 chip
References: <T6VEI.159$VU3.17@fx46.iad>
<memo.20210706143219.6784R@jgd.cix.co.uk>
<19dcc459-6eb5-4191-a186-c50d12ed347fn@googlegroups.com>
User-Agent: slrn/1.0.3 (Darwin)
Lines: 27
Message-ID: <2l%EI.8$gE.2@fx21.iad>
X-Complaints-To: abuse@usenet-news.net
NNTP-Posting-Date: Tue, 06 Jul 2021 16:15:58 UTC
Organization: usenet-news.net
Date: Tue, 06 Jul 2021 16:15:58 GMT
X-Received-Bytes: 1929
 by: Branimir Maksimovic - Tue, 6 Jul 2021 16:15 UTC

On 2021-07-06, Michael S <already5chosen@yahoo.com> wrote:
> On Tuesday, July 6, 2021 at 4:32:22 PM UTC+3, John Dallman wrote:
>> In article <T6VEI.159$VU3...@fx46.iad>, branimir....@gmail.com
>> (Branimir Maksimovic) wrote:
>>
>> > Rosetta takes care of that...
>> > Binaries are tweaked for M1 as Rosetta doesn't support AVX.
>> > There is no performance loss, for what I tested...
>> Rosetta typically runs Intel macOS binaries on an M1 as fast as an Intel
>> Mac. However, native M1 binaries are usually considerably faster than
>> that. The M1 is really quite a good microprocessor.
>>
>> I'm building the same software and running the same tests: an Intel-based
>> Macmini8,1 takes about 12.5 hours for the job that an M1-based Macmini9,1
>> does in a little over 7 hours.
>>
>> John
>
> 8.1 and 9.1 sound like Mac OS versions.
> What is actual HW in each case?
> I can believe that M1@3.2 GHz/Rosetta is able to run x64 software as fast as i3-8100B but have trouble believing that it could match i7-8700B either in single thread or in multithread throughput. Unless, of course, absolute majority of run time spent in native libraries.
You don't count that M1 is ~25-33% faster single core then any x86 :P
>

--
something dumb

Re: My experience with Apple M1 chip

<80ac50a0-dde2-4a66-b09c-62663cd5b4aan@googlegroups.com>

 copy mid

https://www.novabbs.com/devel/article-flat.php?id=18431&group=comp.arch#18431

 copy link   Newsgroups: comp.arch
X-Received: by 2002:ac8:5949:: with SMTP id 9mr16171060qtz.222.1625593038593;
Tue, 06 Jul 2021 10:37:18 -0700 (PDT)
X-Received: by 2002:a05:6808:a19:: with SMTP id n25mr15273856oij.0.1625593038396;
Tue, 06 Jul 2021 10:37:18 -0700 (PDT)
Path: i2pn2.org!i2pn.org!weretis.net!feeder8.news.weretis.net!proxad.net!feeder1-2.proxad.net!209.85.160.216.MISMATCH!news-out.google.com!nntp.google.com!postnews.google.com!google-groups.googlegroups.com!not-for-mail
Newsgroups: comp.arch
Date: Tue, 6 Jul 2021 10:37:18 -0700 (PDT)
In-Reply-To: <2l%EI.8$gE.2@fx21.iad>
Injection-Info: google-groups.googlegroups.com; posting-host=199.203.251.52; posting-account=ow8VOgoAAAAfiGNvoH__Y4ADRwQF1hZW
NNTP-Posting-Host: 199.203.251.52
References: <T6VEI.159$VU3.17@fx46.iad> <memo.20210706143219.6784R@jgd.cix.co.uk>
<19dcc459-6eb5-4191-a186-c50d12ed347fn@googlegroups.com> <2l%EI.8$gE.2@fx21.iad>
User-Agent: G2/1.0
MIME-Version: 1.0
Message-ID: <80ac50a0-dde2-4a66-b09c-62663cd5b4aan@googlegroups.com>
Subject: Re: My experience with Apple M1 chip
From: already5...@yahoo.com (Michael S)
Injection-Date: Tue, 06 Jul 2021 17:37:18 +0000
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
 by: Michael S - Tue, 6 Jul 2021 17:37 UTC

On Tuesday, July 6, 2021 at 7:16:01 PM UTC+3, Branimir Maksimovic wrote:
> On 2021-07-06, Michael S <already...@yahoo.com> wrote:
> > On Tuesday, July 6, 2021 at 4:32:22 PM UTC+3, John Dallman wrote:
> >> In article <T6VEI.159$VU3...@fx46.iad>, branimir....@gmail.com
> >> (Branimir Maksimovic) wrote:
> >>
> >> > Rosetta takes care of that...
> >> > Binaries are tweaked for M1 as Rosetta doesn't support AVX.
> >> > There is no performance loss, for what I tested...
> >> Rosetta typically runs Intel macOS binaries on an M1 as fast as an Intel
> >> Mac. However, native M1 binaries are usually considerably faster than
> >> that. The M1 is really quite a good microprocessor.
> >>
> >> I'm building the same software and running the same tests: an Intel-based
> >> Macmini8,1 takes about 12.5 hours for the job that an M1-based Macmini9,1
> >> does in a little over 7 hours.
> >>
> >> John
> >
> > 8.1 and 9.1 sound like Mac OS versions.
> > What is actual HW in each case?
> > I can believe that M1@3.2 GHz/Rosetta is able to run x64 software as fast as i3-8100B but have trouble believing that it could match i7-8700B either in single thread or in multithread throughput. Unless, of course, absolute majority of run time spent in native libraries.
> You don't count that M1 is ~25-33% faster single core then any x86 :P

I took it into account.

Besides, while it's true for x86 CPUs in prev-gen Mac-Mini it's not true for *any* x86.
M1 is slower than top Zen3 bins and about the same or a little slower than top Comet Lake.
Probably somewhat slower than top Tiger Lake, but that comparison is rather close.
Probably, measurably slower than top Rocket Lake, but I didn't look at Rocket Lake closely.

> >
>
>
> --
> something dumb

Re: My experience with Apple M1 chip

<SJGdnUhI6OG5N3n9nZ2dnUU7-ffNnZ2d@giganews.com>

 copy mid

https://www.novabbs.com/devel/article-flat.php?id=18433&group=comp.arch#18433

 copy link   Newsgroups: comp.arch
Path: i2pn2.org!i2pn.org!aioe.org!news.uzoreto.com!tr2.eu1.usenetexpress.com!feeder.usenetexpress.com!tr1.iad1.usenetexpress.com!border1.nntp.dca1.giganews.com!nntp.giganews.com!buffer1.nntp.dca1.giganews.com!news.giganews.com.POSTED!not-for-mail
NNTP-Posting-Date: Tue, 06 Jul 2021 14:08:19 -0500
Newsgroups: comp.arch
Subject: Re: My experience with Apple M1 chip
References: <T6VEI.159$VU3.17@fx46.iad> <19dcc459-6eb5-4191-a186-c50d12ed347fn@googlegroups.com> <2l%EI.8$gE.2@fx21.iad> <80ac50a0-dde2-4a66-b09c-62663cd5b4aan@googlegroups.com>
Organization: provalid.com
X-Newsreader: trn 4.0-test76 (Apr 2, 2001)
From: keg...@provalid.com (Kent Dickey)
Originator: kegs@provalid.com (Kent Dickey)
Message-ID: <SJGdnUhI6OG5N3n9nZ2dnUU7-ffNnZ2d@giganews.com>
Date: Tue, 06 Jul 2021 14:08:20 -0500
Lines: 33
X-Usenet-Provider: http://www.giganews.com
X-Trace: sv3-gORqvU0CRBHo4uLB+aZFausnUTRZJQC0xOSO4oSqYgRIP85bGiJbx0+cfqFpHiqp6JuAT3Ig/zxwLIQ!9ecbih7SLzpyiSxfRwRNlkkpyI8UZzY0hCZn6BHsinbIz4/e8YzJiu/WdIp8zC0x+5aT/E1UNdU=
X-Complaints-To: abuse@giganews.com
X-DMCA-Notifications: http://www.giganews.com/info/dmca.html
X-Abuse-and-DMCA-Info: Please be sure to forward a copy of ALL headers
X-Abuse-and-DMCA-Info: Otherwise we will be unable to process your complaint properly
X-Postfilter: 1.3.40
X-Original-Bytes: 2871
 by: Kent Dickey - Tue, 6 Jul 2021 19:08 UTC

In article <80ac50a0-dde2-4a66-b09c-62663cd5b4aan@googlegroups.com>,
Michael S <already5chosen@yahoo.com> wrote:
>On Tuesday, July 6, 2021 at 7:16:01 PM UTC+3, Branimir Maksimovic wrote:
>> On 2021-07-06, Michael S <already...@yahoo.com> wrote:
>> > I can believe that M1@3.2 GHz/Rosetta is able to run x64 software as
>fast as i3-8100B but have trouble believing that it could match i7-8700B
>either in single thread or in multithread throughput. Unless, of course,
>absolute majority of run time spent in native libraries.
>> You don't count that M1 is ~25-33% faster single core then any x86 :P
>
>I took it into account.
>
>Besides, while it's true for x86 CPUs in prev-gen Mac-Mini it's not true
>for *any* x86.
>M1 is slower than top Zen3 bins and about the same or a little slower
>than top Comet Lake.
>Probably somewhat slower than top Tiger Lake, but that comparison is
>rather close.
>Probably, measurably slower than top Rocket Lake, but I didn't look at
>Rocket Lake closely.

I have a Mac Mini M1, and it seems fast--very fast for some workloads (hard to
predict branches, or working set in the 100-200KB range). It is not the
fastest CPU on the planet, but it likely is the fastest laptop CPU. At < 10W
at the AC plug it compares pretty favorably to 60W CPUs. If you have a
relatively short benchmark (say, one file, C or C++, can be run from the Unix
command line, doesn't require me to install anything else, should run in less
than 5 minutes), I can compile it and run it for you, and then you can compare
those results to any system you like. I don't think comparing optimized AVX
is going to be useful, but simple integer or floating point algorithms would
be best.

Kent

Re: My experience with Apple M1 chip

<d34FI.336$0N5.13@fx06.iad>

 copy mid

https://www.novabbs.com/devel/article-flat.php?id=18436&group=comp.arch#18436

 copy link   Newsgroups: comp.arch
Path: i2pn2.org!i2pn.org!weretis.net!feeder6.news.weretis.net!4.us.feeder.erje.net!2.eu.feeder.erje.net!feeder.erje.net!newsreader4.netcologne.de!news.netcologne.de!peer03.ams1!peer.ams1.xlned.com!news.xlned.com!peer03.iad!feed-me.highwinds-media.com!news.highwinds-media.com!fx06.iad.POSTED!not-for-mail
Newsgroups: comp.arch
From: branimir...@gmail.com (Branimir Maksimovic)
Subject: Re: My experience with Apple M1 chip
References: <T6VEI.159$VU3.17@fx46.iad>
<memo.20210706143219.6784R@jgd.cix.co.uk>
<19dcc459-6eb5-4191-a186-c50d12ed347fn@googlegroups.com>
<2l%EI.8$gE.2@fx21.iad>
<80ac50a0-dde2-4a66-b09c-62663cd5b4aan@googlegroups.com>
User-Agent: slrn/1.0.3 (Darwin)
Lines: 47
Message-ID: <d34FI.336$0N5.13@fx06.iad>
X-Complaints-To: abuse@usenet-news.net
NNTP-Posting-Date: Tue, 06 Jul 2021 21:38:17 UTC
Organization: usenet-news.net
Date: Tue, 06 Jul 2021 21:38:17 GMT
X-Received-Bytes: 2957
 by: Branimir Maksimovic - Tue, 6 Jul 2021 21:38 UTC

On 2021-07-06, Michael S <already5chosen@yahoo.com> wrote:
> On Tuesday, July 6, 2021 at 7:16:01 PM UTC+3, Branimir Maksimovic wrote:
>> On 2021-07-06, Michael S <already...@yahoo.com> wrote:
>> > On Tuesday, July 6, 2021 at 4:32:22 PM UTC+3, John Dallman wrote:
>> >> In article <T6VEI.159$VU3...@fx46.iad>, branimir....@gmail.com
>> >> (Branimir Maksimovic) wrote:
>> >>
>> >> > Rosetta takes care of that...
>> >> > Binaries are tweaked for M1 as Rosetta doesn't support AVX.
>> >> > There is no performance loss, for what I tested...
>> >> Rosetta typically runs Intel macOS binaries on an M1 as fast as an Intel
>> >> Mac. However, native M1 binaries are usually considerably faster than
>> >> that. The M1 is really quite a good microprocessor.
>> >>
>> >> I'm building the same software and running the same tests: an Intel-based
>> >> Macmini8,1 takes about 12.5 hours for the job that an M1-based Macmini9,1
>> >> does in a little over 7 hours.
>> >>
>> >> John
>> >
>> > 8.1 and 9.1 sound like Mac OS versions.
>> > What is actual HW in each case?
>> > I can believe that M1@3.2 GHz/Rosetta is able to run x64 software as fast as i3-8100B but have trouble believing that it could match i7-8700B either in single thread or in multithread throughput. Unless, of course, absolute majority of run time spent in native libraries.
>> You don't count that M1 is ~25-33% faster single core then any x86 :P
>
> I took it into account.
>
> Besides, while it's true for x86 CPUs in prev-gen Mac-Mini it's not true for *any* x86.
> M1 is slower than top Zen3 bins and about the same or a little slower than top Comet Lake.
> Probably somewhat slower than top Tiger Lake, but that comparison is rather close.
> Probably, measurably slower than top Rocket Lake, but I didn't look at Rocket Lake closely.

Top Intel hardly can achieve 3200 points in scimark2, mine M1 yesterday 4000+ ;)
That is pretty good measurement of single thread performance.
Show me scimark2 score of Top Zen3 please.
I can bet is not more then 3200 ;P
Tell you I didn't believe, but had to buy M1 to try myself :P
>
>> >
>>
>>
>> --
>> something dumb

--
something dumb

Re: My experience with Apple M1 chip

<A74FI.172$Yv3.30@fx41.iad>

 copy mid

https://www.novabbs.com/devel/article-flat.php?id=18437&group=comp.arch#18437

 copy link   Newsgroups: comp.arch
Path: i2pn2.org!i2pn.org!weretis.net!feeder8.news.weretis.net!feeder5.feed.usenet.farm!feeder1.feed.usenet.farm!feed.usenet.farm!peer03.ams4!peer.am4.highwinds-media.com!peer02.iad!feed-me.highwinds-media.com!news.highwinds-media.com!fx41.iad.POSTED!not-for-mail
Newsgroups: comp.arch
From: branimir...@gmail.com (Branimir Maksimovic)
Subject: Re: My experience with Apple M1 chip
References: <T6VEI.159$VU3.17@fx46.iad>
<19dcc459-6eb5-4191-a186-c50d12ed347fn@googlegroups.com>
<2l%EI.8$gE.2@fx21.iad>
<80ac50a0-dde2-4a66-b09c-62663cd5b4aan@googlegroups.com>
<SJGdnUhI6OG5N3n9nZ2dnUU7-ffNnZ2d@giganews.com>
User-Agent: slrn/1.0.3 (Darwin)
Lines: 52
Message-ID: <A74FI.172$Yv3.30@fx41.iad>
X-Complaints-To: abuse@usenet-news.net
NNTP-Posting-Date: Tue, 06 Jul 2021 21:42:56 UTC
Organization: usenet-news.net
Date: Tue, 06 Jul 2021 21:42:56 GMT
X-Received-Bytes: 2873
 by: Branimir Maksimovic - Tue, 6 Jul 2021 21:42 UTC

On 2021-07-06, Kent Dickey <kegs@provalid.com> wrote:
> In article <80ac50a0-dde2-4a66-b09c-62663cd5b4aan@googlegroups.com>,
> Michael S <already5chosen@yahoo.com> wrote:
>>On Tuesday, July 6, 2021 at 7:16:01 PM UTC+3, Branimir Maksimovic wrote:
>>> On 2021-07-06, Michael S <already...@yahoo.com> wrote:
>>> > I can believe that M1@3.2 GHz/Rosetta is able to run x64 software as
>>fast as i3-8100B but have trouble believing that it could match i7-8700B
>>either in single thread or in multithread throughput. Unless, of course,
>>absolute majority of run time spent in native libraries.
>>> You don't count that M1 is ~25-33% faster single core then any x86 :P
>>
>>I took it into account.
>>
>>Besides, while it's true for x86 CPUs in prev-gen Mac-Mini it's not true
>>for *any* x86.
>>M1 is slower than top Zen3 bins and about the same or a little slower
>>than top Comet Lake.
>>Probably somewhat slower than top Tiger Lake, but that comparison is
>>rather close.
>>Probably, measurably slower than top Rocket Lake, but I didn't look at
>>Rocket Lake closely.
>
> I have a Mac Mini M1, and it seems fast--very fast for some workloads (hard to
> predict branches,

Yes, my assembler doesn't work as well as on x86 ;p
likes loop unrolling very much :p
Got 10% just coknverting suboutine in macro and calling several times :P

or working set in the 100-200KB range). It is not the
> fastest CPU on the planet, but it likely is the fastest laptop CPU.

Sorry can't gree. Blows away x86 alright.

At < 10W
> at the AC plug it compares pretty favorably to 60W CPUs. If you have a
> relatively short benchmark (say, one file, C or C++, can be run from the Unix
> command line, doesn't require me to install anything else, should run in less
> than 5 minutes), I can compile it and run it for you, and then you can compare
> those results to any system you like. I don't think comparing optimized AVX
> is going to be useful, but simple integer or floating point algorithms would
> be best.

It can compare with optimised AVX alright :P

>
> Kent

--
something dumb

Pages:123
server_pubkey.txt

rocksolid light 0.9.7
clearnet tor