That's probably good advice 
DeepMind Neural Network Strength
Sort:
Jan 26, 2020
0
#23
drmrboss wrote:
It is difficut to determine whether CPU engine ( Stockfish) is better or NN engine (Leela ) is better because they run on completely different hardwares.
If you give 128 cores cluster cpu to Stockfish and 1 RTX GPU to Leela , Stockfish will smash leela but 1 core CPU SF will badly lose to 1 RTX GPU.
Currently , 16 cores CPU Stockfish ($800 for CPU only) will equally perform to one 2080 RTX Ti GPU ( $1000 for GPU only).
With current technology and cheaper prize, 2x 2080TI ( $2000 for GPU only) on Leela can run similar performance as Alpha Zero running on 4 TPU in 2017. And also 64 thread AMD 3970X ($2000) can run similar performane on Stockfish 8 hardware that deep mind used.
As a consumer point of view, you can decide whether you can afford to upgrade CPU or GPU or both.
I go for the best CPU I can afford. RTX cards are overpriced due to High Demand by Crypto Currency Traders. The question now is which gives better Chess Performance? Intel or AMD if price are almost equal?
Jan 26, 2020
0
#24
bong711 escribió:
drmrboss wrote:
It is difficut to determine whether CPU engine ( Stockfish) is better or NN engine (Leela ) is better because they run on completely different hardwares.
If you give 128 cores cluster cpu to Stockfish and 1 RTX GPU to Leela , Stockfish will smash leela but 1 core CPU SF will badly lose to 1 RTX GPU.
Currently , 16 cores CPU Stockfish ($800 for CPU only) will equally perform to one 2080 RTX Ti GPU ( $1000 for GPU only).
With current technology and cheaper prize, 2x 2080TI ( $2000 for GPU only) on Leela can run similar performance as Alpha Zero running on 4 TPU in 2017. And also 64 thread AMD 3970X ($2000) can run similar performane on Stockfish 8 hardware that deep mind used.
As a consumer point of view, you can decide whether you can afford to upgrade CPU or GPU or both.
I go for the best CPU I can afford. RTX cards are overpriced due to High Demand by Crypto Currency Traders. The question now is which gives better Chess Performance? Intel or AMD if price are almost equal?
Given SF makes good use of multiple cores, Threadripper is a very good option if you can afford it.
Jan 26, 2020
0
#25
Prometheus_Fuschs wrote:
bong711 escribió:
drmrboss wrote:
It is difficut to determine whether CPU engine ( Stockfish) is better or NN engine (Leela ) is better because they run on completely different hardwares.
If you give 128 cores cluster cpu to Stockfish and 1 RTX GPU to Leela , Stockfish will smash leela but 1 core CPU SF will badly lose to 1 RTX GPU.
Currently , 16 cores CPU Stockfish ($800 for CPU only) will equally perform to one 2080 RTX Ti GPU ( $1000 for GPU only).
With current technology and cheaper prize, 2x 2080TI ( $2000 for GPU only) on Leela can run similar performance as Alpha Zero running on 4 TPU in 2017. And also 64 thread AMD 3970X ($2000) can run similar performane on Stockfish 8 hardware that deep mind used.
As a consumer point of view, you can decide whether you can afford to upgrade CPU or GPU or both.
I go for the best CPU I can afford. RTX cards are overpriced due to High Demand by Crypto Currency Traders. The question now is which gives better Chess Performance? Intel or AMD if price are almost equal?
Given SF makes good use of multiple cores, Threadripper is a very good option if you can afford it.
If my budget is 800-1000 USD which CPU by AMD or Intel?
Seriously....how does one believe he increases his skills by having one of these ridiculous items?
Jan 26, 2020
0
#27
TheBestBeer_Root wrote:
Seriously....how does one believe he increases his skills by having one of these ridiculous items?
It's not about chess improvement. It's about better toys.
Is your ‘claim’ that you’re better because of some program genuine, or deceptive?🧐
toys.... ?
toys as in for cheating??
Jan 26, 2020
0
#31
Car enthusiast buy the best cars they can afford. It doesn't make them better drivers.
ahhh the ego
bong711 wrote:
Prometheus_Fuschs wrote:
bong711 escribió:
drmrboss wrote:
It is difficut to determine whether CPU engine ( Stockfish) is better or NN engine (Leela ) is better because they run on completely different hardwares.
If you give 128 cores cluster cpu to Stockfish and 1 RTX GPU to Leela , Stockfish will smash leela but 1 core CPU SF will badly lose to 1 RTX GPU.
Currently , 16 cores CPU Stockfish ($800 for CPU only) will equally perform to one 2080 RTX Ti GPU ( $1000 for GPU only).
With current technology and cheaper prize, 2x 2080TI ( $2000 for GPU only) on Leela can run similar performance as Alpha Zero running on 4 TPU in 2017. And also 64 thread AMD 3970X ($2000) can run similar performane on Stockfish 8 hardware that deep mind used.
As a consumer point of view, you can decide whether you can afford to upgrade CPU or GPU or both.
I go for the best CPU I can afford. RTX cards are overpriced due to High Demand by Crypto Currency Traders. The question now is which gives better Chess Performance? Intel or AMD if price are almost equal?
Given SF makes good use of multiple cores, Threadripper is a very good option if you can afford it.
If my budget is 800-1000 USD which CPU by AMD or Intel?
AMD, of course. Intel is too expensive.
For GPU, go Nvidia RTX, not AMD( AMD GPU dont have NN friendly algorithm, so they may be good for gaming but not for Neural Networks).
Jan 26, 2020
0
#35
drmrboss wrote:
bong711 wrote:
Prometheus_Fuschs wrote:
bong711 escribió:
drmrboss wrote:
It is difficut to determine whether CPU engine ( Stockfish) is better or NN engine (Leela ) is better because they run on completely different hardwares.
If you give 128 cores cluster cpu to Stockfish and 1 RTX GPU to Leela , Stockfish will smash leela but 1 core CPU SF will badly lose to 1 RTX GPU.
Currently , 16 cores CPU Stockfish ($800 for CPU only) will equally perform to one 2080 RTX Ti GPU ( $1000 for GPU only).
With current technology and cheaper prize, 2x 2080TI ( $2000 for GPU only) on Leela can run similar performance as Alpha Zero running on 4 TPU in 2017. And also 64 thread AMD 3970X ($2000) can run similar performane on Stockfish 8 hardware that deep mind used.
As a consumer point of view, you can decide whether you can afford to upgrade CPU or GPU or both.
I go for the best CPU I can afford. RTX cards are overpriced due to High Demand by Crypto Currency Traders. The question now is which gives better Chess Performance? Intel or AMD if price are almost equal?
Given SF makes good use of multiple cores, Threadripper is a very good option if you can afford it.
If my budget is 800-1000 USD which CPU by AMD or Intel?
AMD, of course. Intel is too expensive.
For GPU, go Nvidia RTX, not AMD( AMD GPU dont have NN friendly algorithm, so they may be good for gaming but not for Neural Networks).
Thanks for the information. AMD graphics cards are No No for chess players.
Threadripper? Well, unless you have a vague idea about how to handle the heat, using any 180+ Watts TDP processor for chess analysis sounds like a very bad idea to me. Even with the most efficient watercooling system and plenty of fans on the box, the heat would be too high to handle effectively- notice here that when analysing all used cores run at 100% continously.
pfren wrote:
Threadripper? Well, unless you have a vague idea about how to handle the heat, using any 180+ Watts TDP processor for chess analysis sounds like a very bad idea to me. Even with the most efficient watercooling system and plenty of fans on the box, the heat would be too high to handle effectively- notice here that when analysing all used cores run at 100% continously.
That doesn't make any sense.
If your CPU is rated for a 180 Watts TDP than you know exactly how much cooling effort is required to dissipate that heat. Since you mentioned watercooling, which I use myself, a rough estimation of how much radiator surface area is needed for any given wattage output can be made.
This is the general formula: [Volume LxWxH in mm]
(LxWxH) x 0.00023129193 = Watts dissipated for 10°C delta-T (estimated)
Here an example for a prominent 240/360 rad product:
Jan 27, 2020
0
#38
bong711 escribió:
Prometheus_Fuschs wrote:
bong711 escribió:
drmrboss wrote:
It is difficut to determine whether CPU engine ( Stockfish) is better or NN engine (Leela ) is better because they run on completely different hardwares.
If you give 128 cores cluster cpu to Stockfish and 1 RTX GPU to Leela , Stockfish will smash leela but 1 core CPU SF will badly lose to 1 RTX GPU.
Currently , 16 cores CPU Stockfish ($800 for CPU only) will equally perform to one 2080 RTX Ti GPU ( $1000 for GPU only).
With current technology and cheaper prize, 2x 2080TI ( $2000 for GPU only) on Leela can run similar performance as Alpha Zero running on 4 TPU in 2017. And also 64 thread AMD 3970X ($2000) can run similar performane on Stockfish 8 hardware that deep mind used.
As a consumer point of view, you can decide whether you can afford to upgrade CPU or GPU or both.
I go for the best CPU I can afford. RTX cards are overpriced due to High Demand by Crypto Currency Traders. The question now is which gives better Chess Performance? Intel or AMD if price are almost equal?
Given SF makes good use of multiple cores, Threadripper is a very good option if you can afford it.
If my budget is 800-1000 USD which CPU by AMD or Intel?
I'd go with AMD, however, I'm not that well informed.
Jan 27, 2020
0
#39
pfren escribió:
Threadripper? Well, unless you have a vague idea about how to handle the heat, using any 180+ Watts TDP processor for chess analysis sounds like a very bad idea to me. Even with the most efficient watercooling system and plenty of fans on the box, the heat would be too high to handle effectively- notice here that when analysing all used cores run at 100% continously.
You are certainly overstating it, a good noctua fan and good ventilation can certainly deal with the heat, then again, it's not like intel chips are any cooler at those core counts.
What I think is terrible about engines is you’re losing your entire focus that you should be working to improve, while moving and finding the proper placement each piece.
Using those has you more on such reliance, where relying on your own positioning, that you’re mindset is that you will find your proper placement each move, should be one’s main standpoint, progressing more and more by doing so and actually playing chess and not directed where best is your move, that you apparently (because how programmed), think using engines are wise stumps your increased skills the game.