Is anyone here already using a Ryzen 1800X processor ?

[corres]

[quote="jstanback"]

When I first ran my Ryzen I saw the problem you mention where Windows 10 was using hyperthreads for 4 of the 8 threads instead of using the 4 inactive threads. With SMT turn on, the nodes/sec for 8 threads was much worse than 8 * nodes/sec for 1 thread. I updated the BIOS and then changed to the Windows 10 high-performance power plan and this fixed the problem. Now I get the same nps for 8 cores whether SMT is turned on or off.

John[/quote]

High performance power plan of Windows 10 has effect on the power state (~CPU clock frequency) of CPU and no on the use of logical cores.
I think the new BIOS installed by you caused the advantageous alternation.
What is the number of the new BIOS installed?
I use the No.603 until now for ASUS Prime B350 Plus motherboard.
Maybe I also should change it to a newer one.

[Adam Hair]

Why are you unable to quote properly in ALL your messages? They are hard to read. Also, why you are often stating wrong things? Windows even without affinities set, knows pretty well to set active threads on physical cores, if available. With affinities set, one can test properly with HT=ON as with HT=OFF on the same machine.

Why do you like to provoke often anybody?

Can you see [qoute=""] and [/qoute] marks?

You could go into your profile and set "Always allow BBCode:" to yes. It would make your posts easier for us to read.

At the moment, they look like this:



With BBCode on, your posts would look like this:

[shrapnel]

IN THE CASE OF OVERCLOCKED RYZEN 7 THE EFFECT OF SMT MODE IS VERY SMALL.

BUT WITHOUT OVERCLOCKING SMT MODE (SMT=AUTO) IS EFFECTIVE AND ENHANCE THE POWER OF RYZEN 7.

NOTE: Intel processors like your i7 5960x behaves the same way, so there is no surprising.

Yes, I did read that very carefully, as it is a little difficult to understand your English.
But you are right.
So, in essence, there is no difference in the Hyperthreading of Intel and Ryzen Processors, at least as far as chess is concerned.
HT is useless if the Processor is already overclocked.
So, maybe it is better to overclock CPU, disable SMT and use just 8 REAL Cores for chess, just like Intel.

[jstanback]

I played a couple very short matches between my latest development version of Wasp using 8 threads and the same version using 16 threads and 12 threads. Time control was Game/10s + 167ms. Here are the results:

Wasp(8 threads) vs Wasp(16 threads): 97-104-308
Wasp(8 threads) vs Wasp(12 threads): 125-126-347

So it appears that my program is not making good use of the hyperthreads. I will start fiddling around with lazy SMP to see if I can improve the average search depth for the arasan19 test suite when using more than 8 threads.

I am very happy with the Ryzen processor and inexpensive ASUS B350M motherboard. Same nodes/second for a given clock speed as my i5 and good value for 8 cores & 16 threads. The stock cooler seems sufficient for 3.7 Ghz with Vcore=1.26V (temps are around 60C when running 10 simultaneous matches) and it's perfectly stable.

John

Hi John
Could you run this Test and also the previous one, both with Stockfish and Komodo, and let us know the results ?
I have a hunch that you will get varying results.
Thank You.

Hi Anil,

I'm keeping my computers busy running various tests with Wasp, so don't want to run any long tests with other programs. But here is some data I took for stockfish_8_x64_popcnt.exe with SMT=ON. I ran "go depth 23" from the opening position. I restarted the program for each run to be sure hash was cleared.

Code: Select all

Threads Time     Knps
 1      23.1     1560 
 2       8.5     3063
 4       6.5     6105
 8       7.1    11750
12      11.5    14436 
16       7.4    16946

Note that for one run on a single position the time doesn't decrease with increasing threads which is normal for lazy SMP. But if this same test was run for a bunch of positions, I'm sure the average time would drop as threads increase.

John

[jstanback]

When I first ran my Ryzen I saw the problem you mention where Windows 10 was using hyperthreads for 4 of the 8 threads instead of using the 4 inactive threads. With SMT turn on, the nodes/sec for 8 threads was much worse than 8 * nodes/sec for 1 thread. I updated the BIOS and then changed to the Windows 10 high-performance power plan and this fixed the problem. Now I get the same nps for 8 cores whether SMT is turned on or off.

John

High performance power plan of Windows 10 has effect on the power state (~CPU clock frequency) of CPU and no on the use of logical cores.
I think the new BIOS installed by you caused the advantageous alternation.
What is the number of the new BIOS installed?
I use the No.603 until now for ASUS Prime B350 Plus motherboard.
Maybe I also should change it to a newer one.

Hi Robert,

I was thinking that the power plan did have an effect on use of logical cores, but perhaps I'm mistaken. BTW, I get the same results using high-performance or AMD Ryzen balanced power plan. I haven't tried the Windows balanced plan since I changed the BIOS. I'm have the ASUS uATX motherboard and am using the 0604 BIOS.

John

[corres]

[quote="jstanback"][quote="corres"][quote="jstanback"]

When I first ran my Ryzen I saw the problem you mention where Windows 10 was using hyperthreads for 4 of the 8 threads instead of using the 4 inactive threads. With SMT turn on, the nodes/sec for 8 threads was much worse than 8 * nodes/sec for 1 thread. I updated the BIOS and then changed to the Windows 10 high-performance power plan and this fixed the problem. Now I get the same nps for 8 cores whether SMT is turned on or off.

John[/quote]

High performance power plan of Windows 10 has effect on the power state (~CPU clock frequency) of CPU and no on the use of logical cores.
I think the new BIOS installed by you caused the advantageous alternation.
What is the number of the new BIOS installed?
I use the No.603 until now for ASUS Prime B350 Plus motherboard.
Maybe I also should change it to a newer one.[/quote]

Hi Robert,

I was thinking that the power plan did have an effect on use of logical cores, but perhaps I'm mistaken. BTW, I get the same results using high-performance or AMD Ryzen balanced power plan. I haven't tried the Windows balanced plan since I changed the BIOS. I'm have the ASUS uATX motherboard and am using the 0604 BIOS.

John

[/quote]


Thanks for your answers.

[corres]

[quote="shrapnel"][quote="corres"]IN THE CASE OF OVERCLOCKED RYZEN 7 THE EFFECT OF SMT MODE IS VERY SMALL.

BUT WITHOUT OVERCLOCKING SMT MODE (SMT=AUTO) IS EFFECTIVE AND ENHANCE THE POWER OF RYZEN 7.

NOTE: Intel processors like your i7 5960x behaves the same way, so there is no surprising.

[/quote]

Yes, I did read that very carefully, as it is a little difficult to understand your English.
But you are right.
So, in essence, there is no difference in the Hyperthreading of Intel and Ryzen Processors, at least as far as chess is concerned.
HT is useless if the Processor is already overclocked.
So, maybe it is better to overclock CPU, disable SMT and use just 8 REAL Cores for chess, just like Intel.

[/quote]

Sorry my pure English.
But who wants to understand me that one can get me.

[mjlef]

Code: Select all

  Threads  Knps   Depth   EBF
    1          1447    14.2   1.72
    2          2886    14.9   1.72  
    4          5660    15.9   1.68
    6          8294    16.5   1.71
    8        10806    16.7   1.68
   10       12045    16.7   1.69
   12       13141    16.7   1.71
   14       14149    16.7   1.69
   16       15135    16.7   1.67

Hi John,

It's interesting to me, going from 8 threads to 16 yields a speed increase of almost 5 million nodes per
second but no depth increase. I wonder why that is? I would have expected a small increase, but you
get an identical depth of 16.7 for all threads at 8 and beyond yet the nodes per second is clearly increasing.

regards,
--tom

regards,
--tom

Hi Tom,

With "lazy SMP" search, the number of nodes required to reach a given depth increases as #threads increases. In the case of Wasp, it appears that the additional nodes and additional speed for cores beyond 8 just balance and the depth stays constant. But ask Kai mentioned, the search "widens" with more threads -- which means that different branches are examined and the chance of finding a better move increases. So perhaps Kai is right that 16 threads is best for Wasp. Maybe I'll run a match between 16 threads and 8 threads to see what happens. But it will take quite a while to get enough games....

John

What I said is valid for Stockfish with Lazy SMP, I forgot that the numbers are for another engine.

What Kai says is also true for Komodo. But we do some other things on top of Lazy SMP which could help. My belief is that basic Lazy SMP simply researches a lot of positions, but with the history tables changing over time, search some moves in a different order. And this improves the search.

You can see the same thing by taking any program that does not all hash cutoffs in PV nodes, have it search to a given depth, then immediately research to the same depth. Sometimes the program will find a better move on the second search. I have even considered this a way to use a little extra time if the available time for this move is not enough to search a ply deeper based on average branching factor.

[corres]

[quote="mjlef"]

What Kai says is also true for Komodo. But we do some other things on top of Lazy SMP which could help. My belief is that basic Lazy SMP simply researches a lot of positions, but with the history tables changing over time, search some moves in a different order. And this improves the search.

You can see the same thing by taking any program that does not all hash cutoffs in PV nodes, have it search to a given depth, then immediately research to the same depth. Sometimes the program will find a better move on the second search. I have even considered this a way to use a little extra time if the available time for this move is not enough to search a ply deeper based on average branching factor.

[/quote]

I think there is another effect too: Because the SMP search is not a deterministic one a repeated search enhances the probability to find a better move. To investigate the effect of the increasing of threads to effectiveness of search should be made a lot of matches with fixed depth and different threads. But to this investigation one ought to use physical cores only. Naturally for this more better to make computer-computer matches than engine-engine matches on the same machine.

[Ponti]

Windows 10 is fully supported, as is (I think) 8.1. It's Windows 7 that is not supported, but there are ways around that if you google a bit.

What about Linux ?
Would it be different to run a 64-bit linux-compatible engine with Ryzen ? Would it run faster than running under Windows ?