Can someone with a 6 core i7 do me a favor?

[Leto]

Hi if you have a 6 core i7 (12 total threads via HT technology) could you do me a big favor? Could you tell me the node speeds and current depth on start position at 1 minute of analysis time with Stockfish 5 SSE4.2 when running on these cores? Let me know whether you are running on overclocked machine or not.

Here's my run on my AMD FX8300 at stock 3.3ghz, 1 minute analysis time:
1cpu= d24 1161k
2cpu= d25 2018k 1.74x from 1cpu
4cpu= d26 3805k 1.89x from 2cpu (3.28x from 1cpu)
6cpu= d28 5226k 1.37x from 4cpu (4.5x from 1cpu)
8cpu= d28 6468k 1.24x from 6cpu (1.7x from 4cpu) (5.57x from 1cpu)

[zullil]

Hi if you have a 6 core i7 (12 total threads via HT technology) could you do me a big favor? Could you tell me the node speeds and current depth on start position at 1 minute of analysis time with Stockfish 5 SSE4.2 when running on these cores? Let me know whether you are running on overclocked machine or not.

Here's my run on my AMD FX8300 at stock 3.3ghz, 1 minute analysis time:
1cpu= d24 1161k
2cpu= d25 2018k 1.74x from 1cpu
4cpu= d26 3805k 1.89x from 2cpu (3.28x from 1cpu)
6cpu= d28 5226k 1.37x from 4cpu (4.5x from 1cpu)
8cpu= d28 6468k 1.24x from 6cpu (1.7x from 4cpu) (5.57x from 1cpu)

You might want to specify what hash size you are using. Also, is TurboBoost active or disabled? Maybe mention the OS too.

I don't have an i7, so I can't help.

[Leto]

You might want to specify what hash size you are using. Also, is TurboBoost active or disabled? Maybe mention the OS too.

I don't have an i7, so I can't help.

Hash size= 509mb. Turbo disabled, Win 7 x64.

[Joerg Oster]

Hi if you have a 6 core i7 (12 total threads via HT technology) could you do me a big favor? Could you tell me the node speeds and current depth on start position at 1 minute of analysis time with Stockfish 5 SSE4.2 when running on these cores? Let me know whether you are running on overclocked machine or not.

Here's my run on my AMD FX8300 at stock 3.3ghz, 1 minute analysis time:
1cpu= d24 1161k
2cpu= d25 2018k 1.74x from 1cpu
4cpu= d26 3805k 1.89x from 2cpu (3.28x from 1cpu)
6cpu= d28 5226k 1.37x from 4cpu (4.5x from 1cpu)
8cpu= d28 6468k 1.24x from 6cpu (1.7x from 4cpu) (5.57x from 1cpu)

No i7, but here are my numbers on my i5-4570 @3.2GHz, LinuxMint 17, up to 4 cores:
(SF5 from abrok.eu)

1cpu= d26 nps 1593801
2cpu= d26 nps 3149604 1.97x from 1cpu
4cpu= d28 nps 6050553 1.92x from 2cpu (3.79x from 1cpu)

[syzygy]

Hash size= 509mb.

Do you mean 512?

[kgburcham]

[D] rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq -


Stockfish 5 64 SSE4.2 (8192 MB)

26/35 0:21 +0.27 1.e4 e6 2.d4 d5 3.Nc3 Bb4 4.Bd3 dxe4 5.Bxe4 Nf6 6.Bd3 Nc6 7.Nf3 O-O (190.080.103) 8846
27/35 0:33 +0.21 1.d4 Nf6 2.Nf3 d5 3.c4 e6 4.Nc3 Be7 5.Bg5 O-O 6.e3 Nbd7 7.cxd5 exd5 (310.889.669) 9176
28/37 0:50 +0.27++ 1.d4 Nf6 2.c4 e6 3.Nc3 Bb4 4.e3 O-O 5.Bd3 Bxc3+ 6.bxc3 Nc6 7.Nf3 (475.847.509) 9340
28/37 0:59 +0.17 1.d4 Nf6 2.c4 e6 3.Nc3 Bb4 4.Bd2 O-O 5.Nf3 c5 6.a3 Bxc3 7.Bxc3 cxd4 (556.479.010) 9398
29/37 1:06 +0.24++ 1.d4 Nf6 2.c4 e6 3.Nf3 Bb4+ 4.Bd2 Nc6 5.Bxb4 Nxb4 6.a3 Nc6 7.e3 O-O 8.Nc3 d5 (626.027.490) 9415

[Leto]

Nice. Can you do that with 1 core? That way I can see what speedup ratios you're getting. Also if you don't mind could you do a run with 8 threads, and then 12 threads? You'd probably have to enable HT in bios.

[daws]

i7 4930k OC @4.4ghz hyper-threading off Arena GUI

1 core d25/32 1896k nps

2 core d26/32 3713k nps

4 core d28/38 7035k nps

6 core d28/38 9989k nps

[Leto]

i7 4930k OC @4.4ghz hyper-threading off Arena GUI

1 core d25/32 1896k nps

2 core d26/32 3713k nps

4 core d28/38 7035k nps

6 core d28/38 9989k nps

1.89 speedup going from 2 to 4 cores, 3.71 speedup going from 1 to 4, and 5.26 speedup going from 1 to 6 cores, quite impressive. Seems far more efficient than my FX-8300, at least up to 6 cores.

[Modern Times]

Here's my run on my AMD FX8300 at stock 3.3ghz, 1 minute analysis time:
1cpu= d24 1161k
2cpu= d25 2018k 1.74x from 1cpu
4cpu= d26 3805k 1.89x from 2cpu (3.28x from 1cpu)
6cpu= d28 5226k 1.37x from 4cpu (4.5x from 1cpu)
8cpu= d28 6468k 1.24x from 6cpu (1.7x from 4cpu) (5.57x from 1cpu)

AMD Phenom II X6 1100T @3.6GHz

1cpu= d26/32 1348k
2cpu= d26/34 2607k 1.93x from 1cpu
4cpu= d28/35 5053k 1.93x from 2cpu (3.75x from 1cpu)
6cpu= d29/34 7247k 1.43x from 4cpu (5.38x from 1cpu)

Even if I clocked the X6 down to 3.3, the kn/s on 6 threads would still be higher than on your 8350 on 8 threads. I prefer the older X6 architecture.