Can somebody compare the AMD Ryzen processors to the intel processors

[Carlos777]

AMD Ryzen 9 3900 (avx2 binary):

SF nnue off: 1399478 nps
SF nnue on: 1830004 nps (!!!)
= using nnue is 131% speed compared to nnue off

That looks too good to be true.

This is what I get on my Ryzen 9 3900. Fresh booted system, offline.

Amazing results.

Could somebody do the same test with the Ryzen 7 3700x?

[MMarco]

Intel Haswell i7-6700HQ (bmi2 binary):

SF nnue off: 1071209 nps
SF nnue on: 932672 nps
= using nnue is 87% speed compared to nnue off

With a 10700, NNUE gets 88% of classical's speed here:

Code: Select all

setoption name Hash value 1024
go movetime 60000
info depth 35 seldepth 45 multipv 1 score cp 27 nodes 97881822 nps 1631336 hashfull 660 tbhits 0 time 60001 pv e2e4 e7e5 g1f3 b8c6 f1b5 g8f6 e1g1 f6e4 f1e1 e4d6 f3e5 f8e7 b5f1 c6e5 e1e5 e8g8 d2d4 e7f6 e5e1 f8e8 c1f4 e8e1 d1e1 d6e8 c2c3 d7d5 f1d3 g7g6 b1d2 e8g7 d2f3 c8g4 f3e5 g4f5 d3f5 g7f5 e1d2 f6e5 d4e5 f5g7 h2h3

setoption name Hash value 1024
setoption name Use NNUE value false
go movetime 60000
info depth 32 seldepth 42 multipv 1 score cp 62 nodes 111021262 nps 1850323 hashfull 738 tbhits 0 time 60001 pv d2d4
bestmove d2d4 ponder e7e6

[MMarco]

AMD Ryzen 9 3900 (avx2 binary):

SF nnue off: 1399478 nps
SF nnue on: 1830004 nps (!!!)
= using nnue is 131% speed compared to nnue off

That looks too good to be true.

This is what I get on my Ryzen 9 3900. Fresh booted system, offline.

The nnue off numbers look low to me. My mobile chip gets 1673 knps while your desktop Ryzen-9 gets only 1399 knps. Something is wrong.

[pohl4711]

AMD Ryzen 9 3900 (avx2 binary):

SF nnue off: 1399478 nps
SF nnue on: 1830004 nps (!!!)
= using nnue is 131% speed compared to nnue off

That looks too good to be true.

This is what I get on my Ryzen 9 3900. Fresh booted system, offline.

The nnue off numbers look low to me. My mobile chip gets 1673 knps while your desktop Ryzen-9 gets only 1399 knps. Something is wrong.

The Ryzen is build in a notebook. So, I set the CPU speed to 95%, to prevent it from overheating, which means constantly 2.9 GHz (switches off the TurboBoost), nonetheless 1 Thread or 24 Threads are used.
Speed measuring of singlethread engines lead to wrong resuls, if the CPU is running at 100%. Because of the TurboBoost.
So, perhaps your results are wrong

[Modern Times]

On the Ryzen 9 5900x I have access to at the moment, I ran the AVX 2 exe as follows:

setoption name Hash value 2048
setoption name Use NNUE value false
go movetime 60000

info depth 31 seldepth 35 multipv 1 score cp 71 nodes 93783263 nps 2336404 hashfull 378 tbhits 0 time 40140 pv d2d4 d7d5 c2c4 e7e6 g1f3 g8f6 b1c3 f8b4 c1d2 e8g8 e2e3 b7b6 c4d5 b4c3 d2c3 d8d5 f1e2 c8b7 e1g1 b8d7 b2b3 a8c8 a1c1 f8d8 h2h3 f6e4 c3b2 c7c5 d4c5 d7c5 d1d4 d5d4 b2d4
info depth 32 currmove d2d4 currmovenumber 1


setoption name Hash value 2048
go movetime 60000

info depth 34 seldepth 46 multipv 1 score cp 22 nodes 131802885 nps 2196678 hashfull 478 tbhits 0 time 60001 pv e2e4 c7c5
bestmove e2e4 ponder c7c5


I kept an eye on the clocks. In both runs it was boosting in the same 4.8-4.9 range, so there was no significant difference and it is not a factor in the comparison.

---NNUE speed 94 % of speed without.

[elpapa]

AMD Ryzen 9 3900 (avx2 binary):

SF nnue off: 1399478 nps
SF nnue on: 1830004 nps (!!!)
= using nnue is 131% speed compared to nnue off

That looks too good to be true.

This is what I get on my Ryzen 9 3900. Fresh booted system, offline.

Amazing results.

Could somebody do the same test with the Ryzen 7 3700x?

3700x, manual overclock @4300 MHz
sf 13 avx2

1 thread, go depth 28
nnue off: 1,980,523 nps
nnue on: 1,826,595 nps (92%)

16 threads, go depth 28
nnue off: 21,893,048 nps
nnue on: 19,614,930 nps (90%)

[Carlos777]

AMD Ryzen 9 3900 (avx2 binary):

SF nnue off: 1399478 nps
SF nnue on: 1830004 nps (!!!)
= using nnue is 131% speed compared to nnue off

That looks too good to be true.

This is what I get on my Ryzen 9 3900. Fresh booted system, offline.

Amazing results.

Could somebody do the same test with the Ryzen 7 3700x?

3700x, manual overclock @4300 MHz
sf 13 avx2

1 thread, go depth 28
nnue off: 1,980,523 nps
nnue on: 1,826,595 nps (92%)

16 threads, go depth 28
nnue off: 21,893,048 nps
nnue on: 19,614,930 nps (90%)

Nice result. I am thinking about getting a 3700x, but I don't want to OC it. How would it do it without OC?
At that frequency, do you need a special cooling system for the CPU?

[mclane]

I am considering to buy a commodore c64.

[MMarco]

AMD Ryzen 9 3900 (avx2 binary):

SF nnue off: 1399478 nps
SF nnue on: 1830004 nps (!!!)
= using nnue is 131% speed compared to nnue off

That looks too good to be true.

This is what I get on my Ryzen 9 3900. Fresh booted system, offline.

The nnue off numbers look low to me. My mobile chip gets 1673 knps while your desktop Ryzen-9 gets only 1399 knps. Something is wrong.

The Ryzen is build in a notebook. So, I set the CPU speed to 95%, to prevent it from overheating, which means constantly 2.9 GHz (switches off the TurboBoost), nonetheless 1 Thread or 24 Threads are used.
Speed measuring of singlethread engines lead to wrong resuls, if the CPU is running at 100%. Because of the TurboBoost.
So, perhaps your results are wrong

Seriously?

You're the only one here with a ratio far above 100% amongs a few users with an AMD processor. MikeB got 92%. I get 91%, Elpapa 90%, Modern Times 94%. You get 131% and you think that perhaps other's results are wrong?

When I run my machine in base clock I get a ratio of 91.9% = 1212842/1319436:

Code: Select all

setoption name Hash value 1024
go movetime 60000
info depth 33 seldepth 44 multipv 1 score cp 25 nodes 72771790 nps 1212842 hashfull 511 tbhits 0 time 60001 pv e2e4 e7e5 g1f3 b8c6 f1b5 g8f6 e1g1 f6e4 f1e1 e4d6 f3e5 f8e7 b5f1 c6e5 e1e5 e8g8 d2d4 e7f6 e5e1 f8e8 c1f4 e8e1 d1e1 d6e8 c2c3 d7d6 b1d2 c8f5 e1d1 c7c6 a2a4 d6d5 d2f3 e8c7 f4e5 c7e6 a4a5

setoption name Hash value 1024
setoption name Use NNUE value false
go movetime 60000
info depth 31 seldepth 41 multipv 1 score cp 50 nodes 79167534 nps 1319436 hashfull 577 tbhits 0 time 60001 pv d2d4 e7e6 c2c4 g8f6 b1c3 f8b4 e2e3 e8g8 f1d3 d7d5 g1f3 d5c4 d3c4 b7b6 e1g1 c8b7 c1d2 b8d7 c4d3 c7c5 a1c1 b4c3 b2c3 a8c8 h2h3 h7h6 c3c4 f6e4
bestmove d2d4 ponder e7e6

Maybe it is time for you to consider the hypothesis that your own results could be flawed.

[elpapa]

Nice result. I am thinking about getting a 3700x, but I don't want to OC it. How would it do it without OC?
At that frequency, do you need a special cooling system for the CPU?

It's not really an overclock as I see it, more like a constant frequency which I prefer. The CPU is stable at 1.225V so it doesn't need any special cooling, but the stock cooler made some noise under heavy load. I bought an Arctic Freezer eSports Duo and changed the fan curves and now I can't hear the fans at all.

I can run the same sf benches later with stock settings if you want, but I think it's about 10% faster with current settings, including memory OC.