|
< Earlier Kibitzing · PAGE 2 OF 3 ·
Later Kibitzing> |
Jul-05-20
 | | AylerKupp: <RandomVisitor> That seems about right. I bought and started to read a book on CUDA and it seems that most of what the host CPU does is (1) transfer some data to the GPU and tell it to start executing its program, (2) wait for the GPU to finish crunching the data, (3) transfer the results from the GPU, (4) prepare for the next cycle. I mentioned the Titan RTX mostly in jest since it costs about $ 2,500 and since for this type of application it seems to produce only about a 10% improvement in performance it's clearly not worth it. Now, if Lc0 is modified to use tensor cores if the GPU supports it, then we could probably see a substantial improvement in performance even with the 2080 Ti. We'll see what happens in the next TCEC. And yes, the TCEC used for TCEC Season 18 most likely had 4 2080 GPUs since the GPU server for TCEC Season 17 had them. |
|
| Jul-05-20 | | Jambow: I was all over it until the first move... I was locked into a knight sack to open up the king, I didn't know which knight or what pawn but why quibble over minor details right? No I wasn't just on the wrong page I was in the wrong book. Excellent puzzle for Nunn perhaps. |
|
| Jul-12-20 | | Predrag3141: I nailed this one! It helps that it was the puzzle of the day exactly one week ago. |
|
| Jul-12-20 | | jith1207: In my calendar, one week equals 10 years.
So, I will take full credit for solving this puzzle. |
|
| Jul-12-20 | | boringplayer: Didn't we see this puzzle recently? |
|
Jul-12-20
 | | perfidious: <boringplayer>, judging from all the kibitzing last Sunday, the answer is: very definitely. |
|
| Jul-12-20 | | Brenin: This one is a test of short-term (one week) memory. |
|
| Jul-12-20 | | malt: This game looks familiar, have gone for 17...f4
18.N:f4 R:f4 |
|
Jul-12-20
| | perfidious: <jith1207: In my calendar, one week equals 10 years....> You will recall Chandler vs Y Gruenfeld, 1979, in which numerous decades have, therefore, passed. |
|
Jul-12-20
 | | OhioChessFan: I still can't begin to get this....man, GM's are good. |
|
| Jul-12-20 | | areknames: <I still can't begin to get this> Haha, sort of my sentiments exactly, cg.com needs to run this puzzle every week for a year or so until we all get it eventually! |
|
Jul-12-20
 | | saffuna: Much too difficult for a Sunday. |
|
Jul-12-20
 | | dorsnikov: Believe it or not I got the first two moves. The rest were too difficult. |
|
| Jul-12-20 | | agb2002: Level 4: 17.?
Capablanca vs E Tennenwurzel, 1911
click for larger view |
|
| Jul-12-20 | | catlover: <agb2020> Very nice puzzle suggestion. |
|
| Jul-12-20 | | unrepentant: Yippeekaiyay, muafugga, I solved it again!!! :) :) twice in the same week for the same puzzle!! I still stumbled on 24. ... Rg4 just like last time.
Give it to me again in 4 days and I'll remember the solution all the way to the resignation. hahaha |
|
| Jul-12-20 | | agb2002: <catlover: <agb2020> Very nice puzzle suggestion.> Thank you! |
|
| Jul-12-20 | | Walter Glattke: 19.gxf3 Qh4 20.h3 Qxf2+ 21.Kh1 Ng3# Try 18.Be4 Bxe4 19.f3 Ne3 |
|
Jul-12-20
 | | Check It Out: I did better than last week. Must mean I'm improving. |
|
Jul-12-20
| | AylerKupp: <Check it Out> Or the POTD's are getting easier. ;-) |
|
| Jul-12-20 | | W Westerlund: What happens after 21. gxf4 ? I don't see it. |
|
Jul-12-20
| | OhioChessFan: <WW> 21...Nxf4 either leads to mate after Qg5+, or White loses the Queen.
click for larger view |
|
| Jul-12-20 | | mel gibson: Why was this puzzle repeated from last Sunday? |
|
Jul-12-20
 | | Phony Benoni: <mel gibson: Why was this puzzle repeated from last Sunday?> It happens from time to time; just some gum-up in the mechanism. What worries me is that occasionally it will be the prelude to more repeats. |
|
Jul-12-20
| | perfidious: <unrepentant>, they keep giving this, I'll be able to give all the variations leading to Black's demise. |
|
 |
|
|
< Earlier Kibitzing · PAGE 2 OF 3 ·
Later Kibitzing> |
Kibitzing
Annotations by 