AVX2 optimized SF+NNUE and processor temperature

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

corres
Posts: 3657
Joined: Wed Nov 18, 2015 11:41 am
Location: hungary

Re: AVX2 optimized SF+NNUE and processor temperature

Post by corres »

Gregory Owett wrote: Sun Sep 06, 2020 11:41 am Hi,
Do you keep your machines running all the time? Me, I do it approx. 14-16 h. max per day. I have a Ryzen 3900x, and I use 16 threads (temp. approx. 77-81 ° C, 4000 MHz according to Ryzen Master).
Sometimes (in test time) I used my Ryzen 9 3950x 12-16 hours continuously and at 16 x 4000 MHZ the proce4ssor temperature was ~ 75 grade Celsius, no Ryzen Master but manually setting the BIOS and no SMT.
User avatar
MikeB
Posts: 4889
Joined: Thu Mar 09, 2006 6:34 am
Location: Pen Argyl, Pennsylvania

Re: AVX2 optimized SF+NNUE and processor temperature

Post by MikeB »

Gregory Owett wrote: Sun Sep 06, 2020 11:41 am Hi,
Do you keep your machines running all the time? Me, I do it approx. 14-16 h. max per day. I have a Ryzen 3900x, and I use 16 threads (temp. approx. 77-81 ° C, 4000 MHz according to Ryzen Master).
Pretty much - however, it never goes over 78, that is set as a hard throttle in what they used to called the "BIOS" , but they call it something else now.

Edit: I just looked it up and it's now called the UEFI or the Unified Extensible Firmware Interface. Fwiw, it looks like a BIOS to me - it's the preboot session before the OS boots. Generally speaking, with the Ryzen the fewer cores you have the faster you can run it . If I was just using 16 cores , i could run it @ 4 Ghz all day. ( Note it it does run at 4 Ghz or higher when using 16 or fewer cores. Today my max speed so far is 4.341 Ghz)
Image
Gregory Owett
Posts: 249
Joined: Fri Mar 10, 2006 10:26 am
Location: France

Re: AVX2 optimized SF+NNUE and processor temperature

Post by Gregory Owett »

It would be interesting to know, how long a machine can run continuously without damage. Maybe 3-4 days ?
jdart
Posts: 4367
Joined: Fri Mar 10, 2006 5:23 am
Location: http://www.arasanchess.org

Re: AVX2 optimized SF+NNUE and processor temperature

Post by jdart »

I am running a 3970x with an air cooler (this one: plus 8 case fans), and it is running about 79-81 Celcius under full load. I have run it fully loaded for hours at a time.
User avatar
MikeB
Posts: 4889
Joined: Thu Mar 09, 2006 6:34 am
Location: Pen Argyl, Pennsylvania

Re: AVX2 optimized SF+NNUE and processor temperature

Post by MikeB »

Gregory Owett wrote: Mon Sep 07, 2020 4:28 pm It would be interesting to know, how long a machine can run continuously without damage. Maybe 3-4 days ?
I have run it @ max load for 72 hours with no ill effects. I'm sure it could be run much much longer. The secret is the heat - keeping it near 80C is perfect 0- anything over 85C for an extended period of time - in my view and from own experience - will become problematic. I use 78C as my throttling threshold - since I personally want a big wide buffer from my running temp and the temp which i know firsthand is problematic for my machine - 85C is problematic for my machine - ymmv. One of the reasons why I use 78C is that this was a big purchase for me and I want it to last ten years - should I be bless and live that long.
Image
Albert Silver
Posts: 3019
Joined: Wed Mar 08, 2006 9:57 pm
Location: Rio de Janeiro, Brazil

Re: AVX2 optimized SF+NNUE and processor temperature

Post by Albert Silver »

Gregory Owett wrote: Mon Sep 07, 2020 4:28 pm It would be interesting to know, how long a machine can run continuously without damage. Maybe 3-4 days ?

Depends on your definition of continuously. Does it need to be 100% load? Or an average of at least 50% running a minimum of 23 hours per day? If the latter: years. At the very least that is what my desktops are subject to. The Threadripper 1950X has had this for two years now, and the Ryzen 3900X is almost a year.
"Tactics are the bricks and sticks that make up a game, but positional play is the architectural blueprint."
syzygy
Posts: 5566
Joined: Tue Feb 28, 2012 11:56 pm

Re: AVX2 optimized SF+NNUE and processor temperature

Post by syzygy »

Albert Silver wrote: Mon Sep 07, 2020 4:50 pm
Gregory Owett wrote: Mon Sep 07, 2020 4:28 pm It would be interesting to know, how long a machine can run continuously without damage. Maybe 3-4 days ?

Depends on your definition of continuously. Does it need to be 100% load? Or an average of at least 50% running a minimum of 23 hours per day? If the latter: years. At the very least that is what my desktops are subject to. The Threadripper 1950X has had this for two years now, and the Ryzen 3900X is almost a year.
I would be very disappointed in AMD or Intel if their chips could not run at 100% load continuously for many years in a properly cooled and non-overclocked system, with core temperatures high but within spec.
cucumber
Posts: 144
Joined: Sun Oct 14, 2018 8:21 pm
Full name: JSmith

Re: AVX2 optimized SF+NNUE and processor temperature

Post by cucumber »

syzygy wrote: Tue Sep 08, 2020 12:08 am
Albert Silver wrote: Mon Sep 07, 2020 4:50 pm
Gregory Owett wrote: Mon Sep 07, 2020 4:28 pm It would be interesting to know, how long a machine can run continuously without damage. Maybe 3-4 days ?

Depends on your definition of continuously. Does it need to be 100% load? Or an average of at least 50% running a minimum of 23 hours per day? If the latter: years. At the very least that is what my desktops are subject to. The Threadripper 1950X has had this for two years now, and the Ryzen 3900X is almost a year.
I would be very disappointed in AMD or Intel if their chips could not run at 100% load continuously for many years in a properly cooled and non-overclocked system, with core temperatures high but within spec.
Likewise. This is what chips are designed for. Running at the same temperature 24/7 is pretty much optimal as long as that temperature is on [0, 80] C. There are far fewer thermal cycles (though modern packaging is resilient enough that thermal cycles are rarely a big deal), voltage spikes, etc., to cause any problems.
Leo
Posts: 1080
Joined: Fri Sep 16, 2016 6:55 pm
Location: USA/Minnesota
Full name: Leo Anger

Re: AVX2 optimized SF+NNUE and processor temperature

Post by Leo »

syzygy wrote: Tue Sep 08, 2020 12:08 am
Albert Silver wrote: Mon Sep 07, 2020 4:50 pm
Gregory Owett wrote: Mon Sep 07, 2020 4:28 pm It would be interesting to know, how long a machine can run continuously without damage. Maybe 3-4 days ?

Depends on your definition of continuously. Does it need to be 100% load? Or an average of at least 50% running a minimum of 23 hours per day? If the latter: years. At the very least that is what my desktops are subject to. The Threadripper 1950X has had this for two years now, and the Ryzen 3900X is almost a year.
I would be very disappointed in AMD or Intel if their chips could not run at 100% load continuously for many years in a properly cooled and non-overclocked system, with core temperatures high but within spec.
I very rarely hear of a chip burning up by either company. No need to be disappointed.
Advanced Micro Devices fan.
corres
Posts: 3657
Joined: Wed Nov 18, 2015 11:41 am
Location: hungary

Re: AVX2 optimized SF+NNUE and processor temperature

Post by corres »

Leo wrote: Tue Sep 08, 2020 3:52 am
syzygy wrote: Tue Sep 08, 2020 12:08 am
Albert Silver wrote: Mon Sep 07, 2020 4:50 pm
Gregory Owett wrote: Mon Sep 07, 2020 4:28 pm It would be interesting to know, how long a machine can run continuously without damage. Maybe 3-4 days ?

Depends on your definition of continuously. Does it need to be 100% load? Or an average of at least 50% running a minimum of 23 hours per day? If the latter: years. At the very least that is what my desktops are subject to. The Threadripper 1950X has had this for two years now, and the Ryzen 3900X is almost a year.
I would be very disappointed in AMD or Intel if their chips could not run at 100% load continuously for many years in a properly cooled and non-overclocked system, with core temperatures high but within spec.
I very rarely hear of a chip burning up by either company. No need to be disappointed.
At the earlier time processors have no defense against overheating and at least on pictures we can see burned processors (typically earlier AMD processors) what lead to the loss of guarantee. In general a good air cooler give better defense against overheating because water boiling up at 100 degrees Celsius already.
Note
My best-buying was an Intel processor, a Sandy Bridge i5 2500K with 3200 MHz base clock frequency what I overclocked to 4000 MHz for full (4) cores and I am using it with a relative weak air cooler for eight years without any issue. I used it 10-16 hours per day. The machine is my favorite, because at present this is my only one machine for running Windows 7 and Windows XP.