WEBVTT Kind: captions Language: en 00:00:09.510 --> 00:00:15.680 align:start position:0% [Music] 00:00:15.680 --> 00:00:15.690 align:start position:0% [Music] 00:00:15.690 --> 00:00:19.410 align:start position:0% [Music] I'm<00:00:16.690> Greg<00:00:16.990> Colvin<00:00:17.560> I<00:00:17.949> spend<00:00:18.460> my<00:00:18.640> time<00:00:18.670> for 00:00:19.410 --> 00:00:19.420 align:start position:0% I'm Greg Colvin I spend my time for 00:00:19.420 --> 00:00:23.749 align:start position:0% I'm Greg Colvin I spend my time for aetherium<00:00:20.130> working<00:00:21.130> on<00:00:21.279> the<00:00:21.430> virtual<00:00:21.820> machine 00:00:23.749 --> 00:00:23.759 align:start position:0% aetherium working on the virtual machine 00:00:23.759 --> 00:00:26.909 align:start position:0% aetherium working on the virtual machine working<00:00:24.759> on<00:00:24.910> improving<00:00:25.329> its<00:00:25.630> performance<00:00:26.529> and 00:00:26.909 --> 00:00:26.919 align:start position:0% working on improving its performance and 00:00:26.919 --> 00:00:30.890 align:start position:0% working on improving its performance and working<00:00:27.550> on<00:00:27.699> designs<00:00:28.269> for<00:00:29.880> possible 00:00:30.890 --> 00:00:30.900 align:start position:0% working on designs for possible 00:00:30.900 --> 00:00:34.350 align:start position:0% working on designs for possible successors<00:00:31.900> to<00:00:32.200> what<00:00:32.380> we<00:00:32.560> have<00:00:32.829> in<00:00:33.730> order<00:00:34.120> to 00:00:34.350 --> 00:00:34.360 align:start position:0% successors to what we have in order to 00:00:34.360 --> 00:00:37.319 align:start position:0% successors to what we have in order to solve<00:00:34.890> some<00:00:35.890> of<00:00:36.040> the<00:00:36.160> performance<00:00:36.760> problems 00:00:37.319 --> 00:00:37.329 align:start position:0% solve some of the performance problems 00:00:37.329 --> 00:00:41.280 align:start position:0% solve some of the performance problems that<00:00:37.540> were<00:00:37.660> running<00:00:37.899> into<00:00:39.600> if<00:00:40.600> this<00:00:40.960> works 00:00:41.280 --> 00:00:41.290 align:start position:0% that were running into if this works 00:00:41.290 --> 00:00:45.360 align:start position:0% that were running into if this works there<00:00:41.650> will<00:00:41.800> be<00:00:41.830> another<00:00:42.220> slide<00:00:44.040> how<00:00:45.040> about 00:00:45.360 --> 00:00:45.370 align:start position:0% there will be another slide how about 00:00:45.370 --> 00:00:49.380 align:start position:0% there will be another slide how about that<00:00:46.500> the<00:00:47.500> first<00:00:47.770> problem<00:00:48.220> you<00:00:48.250> run<00:00:48.580> into<00:00:48.640> an 00:00:49.380 --> 00:00:49.390 align:start position:0% that the first problem you run into an 00:00:49.390 --> 00:00:53.400 align:start position:0% that the first problem you run into an any<00:00:49.630> sort<00:00:49.990> of<00:00:50.080> optimization<00:00:50.680> work<00:00:52.410> according 00:00:53.400 --> 00:00:53.410 align:start position:0% any sort of optimization work according 00:00:53.410 --> 00:00:55.830 align:start position:0% any sort of optimization work according to<00:00:53.440> my<00:00:53.620> old<00:00:53.830> friend<00:00:54.160> Jerry<00:00:54.430> Schwartz<00:00:54.940> is<00:00:55.240> one 00:00:55.830 --> 00:00:55.840 align:start position:0% to my old friend Jerry Schwartz is one 00:00:55.840 --> 00:00:58.800 align:start position:0% to my old friend Jerry Schwartz is one all<00:00:56.290> benchmarks<00:00:57.100> are<00:00:57.370> bogus<00:00:57.700> you'll<00:00:58.570> never 00:00:58.800 --> 00:00:58.810 align:start position:0% all benchmarks are bogus you'll never 00:00:58.810 --> 00:01:01.230 align:start position:0% all benchmarks are bogus you'll never have<00:00:59.020> a<00:00:59.080> set<00:00:59.380> of<00:00:59.530> benchmarks<00:00:59.710> it<00:01:00.430> actually 00:01:01.230 --> 00:01:01.240 align:start position:0% have a set of benchmarks it actually 00:01:01.240 --> 00:01:04.350 align:start position:0% have a set of benchmarks it actually represents<00:01:01.450> the<00:01:02.200> real<00:01:02.380> world<00:01:02.680> but<00:01:03.640> if<00:01:04.299> you 00:01:04.350 --> 00:01:04.360 align:start position:0% represents the real world but if you 00:01:04.360 --> 00:01:06.270 align:start position:0% represents the real world but if you don't<00:01:04.479> have<00:01:04.869> benchmarks<00:01:05.470> you<00:01:05.650> will<00:01:05.829> just<00:01:06.070> go 00:01:06.270 --> 00:01:06.280 align:start position:0% don't have benchmarks you will just go 00:01:06.280 --> 00:01:08.219 align:start position:0% don't have benchmarks you will just go in<00:01:06.430> circles<00:01:06.700> you'll<00:01:07.240> you'll<00:01:07.570> never<00:01:07.930> make 00:01:08.219 --> 00:01:08.229 align:start position:0% in circles you'll you'll never make 00:01:08.229 --> 00:01:09.050 align:start position:0% in circles you'll you'll never make progress 00:01:09.050 --> 00:01:09.060 align:start position:0% progress 00:01:09.060 --> 00:01:12.540 align:start position:0% progress so<00:01:10.060> the<00:01:10.270> benchmarks<00:01:10.840> I'm<00:01:11.140> working<00:01:11.680> with<00:01:11.950> are<00:01:12.490> a 00:01:12.540 --> 00:01:12.550 align:start position:0% so the benchmarks I'm working with are a 00:01:12.550 --> 00:01:17.250 align:start position:0% so the benchmarks I'm working with are a few<00:01:13.030> algorithm<00:01:14.020> kernels<00:01:14.560> that<00:01:15.490> are<00:01:16.260> relevant 00:01:17.250 --> 00:01:17.260 align:start position:0% few algorithm kernels that are relevant 00:01:17.260 --> 00:01:18.620 align:start position:0% few algorithm kernels that are relevant to<00:01:17.470> what<00:01:17.560> we're<00:01:17.860> doing 00:01:18.620 --> 00:01:18.630 align:start position:0% to what we're doing 00:01:18.630 --> 00:01:22.649 align:start position:0% to what we're doing RC<00:01:19.630> 5s<00:01:20.290> and<00:01:20.440> old<00:01:20.590> and<00:01:21.100> useless<00:01:21.310> cipher<00:01:22.240> but 00:01:22.649 --> 00:01:22.659 align:start position:0% RC 5s and old and useless cipher but 00:01:22.659 --> 00:01:27.000 align:start position:0% RC 5s and old and useless cipher but it's<00:01:23.200> a<00:01:23.380> good<00:01:23.590> example<00:01:23.770> of<00:01:25.020> a<00:01:26.020> cipher<00:01:26.979> that 00:01:27.000 --> 00:01:27.010 align:start position:0% it's a good example of a cipher that 00:01:27.010 --> 00:01:30.840 align:start position:0% it's a good example of a cipher that uses<00:01:27.640> a<00:01:27.670> lot<00:01:27.970> of<00:01:28.000> 32<00:01:28.659> and<00:01:28.900> 64-bit<00:01:29.560> arithmetic<00:01:30.550> a 00:01:30.840 --> 00:01:30.850 align:start position:0% uses a lot of 32 and 64-bit arithmetic a 00:01:30.850 --> 00:01:36.539 align:start position:0% uses a lot of 32 and 64-bit arithmetic a lot<00:01:31.630> of<00:01:31.780> complex<00:01:32.290> logic<00:01:33.210> and<00:01:34.680> Blake<00:01:35.680> to<00:01:35.979> be<00:01:36.250> is 00:01:36.539 --> 00:01:36.549 align:start position:0% lot of complex logic and Blake to be is 00:01:36.549 --> 00:01:39.000 align:start position:0% lot of complex logic and Blake to be is still<00:01:36.700> an<00:01:36.940> important<00:01:37.570> hash<00:01:37.780> function<00:01:38.470> it's 00:01:39.000 --> 00:01:39.010 align:start position:0% still an important hash function it's 00:01:39.010 --> 00:01:44.460 align:start position:0% still an important hash function it's also<00:01:39.310> a<00:01:39.610> lot<00:01:39.940> of<00:01:40.650> 64-bit<00:01:41.650> logic<00:01:43.229> blum<00:01:44.229> blum 00:01:44.460 --> 00:01:44.470 align:start position:0% also a lot of 64-bit logic blum blum 00:01:44.470 --> 00:01:48.180 align:start position:0% also a lot of 64-bit logic blum blum Schaub<00:01:44.950> is<00:01:45.460> a<00:01:45.880> cryptographic<00:01:46.630> rant<00:01:47.620> random 00:01:48.180 --> 00:01:48.190 align:start position:0% Schaub is a cryptographic rant random 00:01:48.190 --> 00:01:50.700 align:start position:0% Schaub is a cryptographic rant random number<00:01:48.520> generator<00:01:49.120> I<00:01:49.540> think<00:01:50.140> it's<00:01:50.290> one<00:01:50.440> of<00:01:50.470> the 00:01:50.700 --> 00:01:50.710 align:start position:0% number generator I think it's one of the 00:01:50.710 --> 00:01:55.140 align:start position:0% number generator I think it's one of the slowest<00:01:51.280> in<00:01:51.460> the<00:01:51.580> world<00:01:52.470> it<00:01:53.470> operates<00:01:54.100> on<00:01:54.490> big 00:01:55.140 --> 00:01:55.150 align:start position:0% slowest in the world it operates on big 00:01:55.150 --> 00:01:59.520 align:start position:0% slowest in the world it operates on big registers<00:01:56.400> so<00:01:57.400> we<00:01:57.549> could<00:01:57.760> use<00:01:57.970> the<00:01:58.240> 256<00:01:59.200> bit 00:01:59.520 --> 00:01:59.530 align:start position:0% registers so we could use the 256 bit 00:01:59.530 --> 00:02:03.660 align:start position:0% registers so we could use the 256 bit registers<00:02:00.190> of<00:02:00.220> the<00:02:00.549> VM<00:02:01.060> effectively<00:02:01.990> an<00:02:02.670> DC 00:02:03.660 --> 00:02:03.670 align:start position:0% registers of the VM effectively an DC 00:02:03.670 --> 00:02:06.810 align:start position:0% registers of the VM effectively an DC mall<00:02:03.990> also<00:02:04.990> can<00:02:05.650> use<00:02:05.860> big<00:02:06.220> registers 00:02:06.810 --> 00:02:06.820 align:start position:0% mall also can use big registers 00:02:06.820 --> 00:02:10.229 align:start position:0% mall also can use big registers effectively<00:02:07.720> and<00:02:08.080> then<00:02:08.920> I<00:02:08.979> have<00:02:09.340> a<00:02:09.369> few<00:02:09.700> tests 00:02:10.229 --> 00:02:10.239 align:start position:0% effectively and then I have a few tests 00:02:10.239 --> 00:02:14.640 align:start position:0% effectively and then I have a few tests of<00:02:10.420> act<00:02:11.549> individual<00:02:12.549> EVM<00:02:13.080> operations<00:02:14.080> and 00:02:14.640 --> 00:02:14.650 align:start position:0% of act individual EVM operations and 00:02:14.650 --> 00:02:18.990 align:start position:0% of act individual EVM operations and those<00:02:14.950> are<00:02:15.220> small<00:02:15.900> EVM<00:02:16.900> assembly<00:02:17.560> programs<00:02:18.220> to 00:02:18.990 --> 00:02:19.000 align:start position:0% those are small EVM assembly programs to 00:02:19.000 --> 00:02:21.420 align:start position:0% those are small EVM assembly programs to try<00:02:19.240> to<00:02:19.300> isolate<00:02:19.980> individual<00:02:20.980> opera 00:02:21.420 --> 00:02:21.430 align:start position:0% try to isolate individual opera 00:02:21.430 --> 00:02:27.270 align:start position:0% try to isolate individual opera ratios<00:02:23.400> and<00:02:24.400> this<00:02:25.359> is<00:02:25.540> a<00:02:25.569> graph<00:02:25.930> of<00:02:26.290> what<00:02:26.890> I<00:02:26.920> got 00:02:27.270 --> 00:02:27.280 align:start position:0% ratios and this is a graph of what I got 00:02:27.280 --> 00:02:34.140 align:start position:0% ratios and this is a graph of what I got with<00:02:27.909> the<00:02:28.090> whole<00:02:28.299> thing<00:02:30.000> and<00:02:32.730> it's<00:02:33.730> a<00:02:33.939> bit 00:02:34.140 --> 00:02:34.150 align:start position:0% with the whole thing and it's a bit 00:02:34.150 --> 00:02:38.520 align:start position:0% with the whole thing and it's a bit complex<00:02:35.129> but<00:02:36.129> what<00:02:36.879> you<00:02:37.060> can<00:02:37.299> see<00:02:37.629> down<00:02:38.109> along 00:02:38.520 --> 00:02:38.530 align:start position:0% complex but what you can see down along 00:02:38.530 --> 00:02:40.830 align:start position:0% complex but what you can see down along the<00:02:38.709> bottom<00:02:38.920> is<00:02:39.400> the<00:02:39.730> different<00:02:40.120> benchmarks 00:02:40.830 --> 00:02:40.840 align:start position:0% the bottom is the different benchmarks 00:02:40.840 --> 00:02:44.910 align:start position:0% the bottom is the different benchmarks and<00:02:41.379> along<00:02:42.099> the<00:02:42.370> right<00:02:42.670> are<00:02:43.269> the<00:02:44.230> first<00:02:44.530> three 00:02:44.910 --> 00:02:44.920 align:start position:0% and along the right are the first three 00:02:44.920 --> 00:02:48.629 align:start position:0% and along the right are the first three are<00:02:45.400> some<00:02:45.790> major<00:02:46.030> clients<00:02:47.010> EVM<00:02:48.010> is<00:02:48.280> our<00:02:48.459> go 00:02:48.629 --> 00:02:48.639 align:start position:0% are some major clients EVM is our go 00:02:48.639 --> 00:02:53.909 align:start position:0% are some major clients EVM is our go client<00:02:49.359> parity<00:02:50.349> is<00:02:50.500> Russ<00:02:51.280> client<00:02:52.379> EVM<00:02:53.379> is<00:02:53.709> the 00:02:53.909 --> 00:02:53.919 align:start position:0% client parity is Russ client EVM is the 00:02:53.919 --> 00:02:59.280 align:start position:0% client parity is Russ client EVM is the C++<00:02:54.450> client<00:02:56.700> the<00:02:57.700> rest<00:02:57.939> aren't<00:02:58.239> clients<00:02:58.959> yet 00:02:59.280 --> 00:02:59.290 align:start position:0% C++ client the rest aren't clients yet 00:02:59.290 --> 00:03:03.140 align:start position:0% C++ client the rest aren't clients yet but<00:02:59.739> EVM<00:03:00.489> to<00:03:00.790> wasm<00:03:01.389> is<00:03:01.720> part<00:03:02.109> of<00:03:02.290> the<00:03:02.530> Azzam 00:03:03.140 --> 00:03:03.150 align:start position:0% but EVM to wasm is part of the Azzam 00:03:03.150 --> 00:03:06.809 align:start position:0% but EVM to wasm is part of the Azzam research<00:03:04.150> it's<00:03:04.959> a<00:03:05.139> program<00:03:05.680> that<00:03:05.709> takes<00:03:06.250> EVM 00:03:06.809 --> 00:03:06.819 align:start position:0% research it's a program that takes EVM 00:03:06.819 --> 00:03:10.459 align:start position:0% research it's a program that takes EVM code<00:03:07.590> translates<00:03:08.590> it<00:03:08.829> into<00:03:08.950> Azzam<00:03:09.669> code<00:03:10.030> and 00:03:10.459 --> 00:03:10.469 align:start position:0% code translates it into Azzam code and 00:03:10.469 --> 00:03:14.729 align:start position:0% code translates it into Azzam code and then<00:03:11.469> I<00:03:11.500> fed<00:03:12.099> that<00:03:12.459> to<00:03:13.209> Google's<00:03:13.629> v8<00:03:14.049> engine 00:03:14.729 --> 00:03:14.739 align:start position:0% then I fed that to Google's v8 engine 00:03:14.739 --> 00:03:19.800 align:start position:0% then I fed that to Google's v8 engine which<00:03:15.010> generates<00:03:15.599> assembly<00:03:16.599> code<00:03:18.329> EVM<00:03:19.329> JIT 00:03:19.800 --> 00:03:19.810 align:start position:0% which generates assembly code EVM JIT 00:03:19.810 --> 00:03:22.080 align:start position:0% which generates assembly code EVM JIT Pavel<00:03:20.409> was<00:03:20.530> just<00:03:20.889> talking<00:03:21.400> about<00:03:21.549> it 00:03:22.080 --> 00:03:22.090 align:start position:0% Pavel was just talking about it 00:03:22.090 --> 00:03:24.479 align:start position:0% Pavel was just talking about it generates<00:03:22.659> assembly<00:03:23.290> code<00:03:23.530> directly<00:03:24.280> from 00:03:24.479 --> 00:03:24.489 align:start position:0% generates assembly code directly from 00:03:24.489 --> 00:03:30.059 align:start position:0% generates assembly code directly from EVM<00:03:25.000> code<00:03:25.389> and<00:03:26.099> then<00:03:27.099> the<00:03:27.370> native<00:03:27.729> C++<00:03:28.470> is<00:03:29.470> I 00:03:30.059 --> 00:03:30.069 align:start position:0% EVM code and then the native C++ is I 00:03:30.069 --> 00:03:34.229 align:start position:0% EVM code and then the native C++ is I rewrote<00:03:30.479> the<00:03:31.479> benchmark<00:03:32.319> programs<00:03:32.949> in<00:03:33.370> C++ 00:03:34.229 --> 00:03:34.239 align:start position:0% rewrote the benchmark programs in C++ 00:03:34.239 --> 00:03:38.429 align:start position:0% rewrote the benchmark programs in C++ instead<00:03:34.900> of<00:03:35.079> solidity<00:03:35.650> or<00:03:36.340> EVM<00:03:36.879> assembly<00:03:37.870> and 00:03:38.429 --> 00:03:38.439 align:start position:0% instead of solidity or EVM assembly and 00:03:38.439 --> 00:03:42.149 align:start position:0% instead of solidity or EVM assembly and compile<00:03:38.979> those<00:03:39.400> to<00:03:40.090> assembly<00:03:40.720> so<00:03:41.680> pretty 00:03:42.149 --> 00:03:42.159 align:start position:0% compile those to assembly so pretty 00:03:42.159 --> 00:03:48.420 align:start position:0% compile those to assembly so pretty clearly<00:03:42.609> the<00:03:43.500> C++<00:03:44.500> wins<00:03:44.859> the<00:03:45.220> race<00:03:46.919> and<00:03:47.919> pretty 00:03:48.420 --> 00:03:48.430 align:start position:0% clearly the C++ wins the race and pretty 00:03:48.430 --> 00:03:51.960 align:start position:0% clearly the C++ wins the race and pretty clearly<00:03:49.799> exponentiation<00:03:50.799> is<00:03:51.310> pretty<00:03:51.609> hard 00:03:51.960 --> 00:03:51.970 align:start position:0% clearly exponentiation is pretty hard 00:03:51.970 --> 00:03:57.179 align:start position:0% clearly exponentiation is pretty hard for<00:03:52.750> everybody<00:03:53.319> which<00:03:53.829> isn't<00:03:54.040> surprising<00:03:56.189> but 00:03:57.179 --> 00:03:57.189 align:start position:0% for everybody which isn't surprising but 00:03:57.189 --> 00:03:59.129 align:start position:0% for everybody which isn't surprising but it's<00:03:57.459> a<00:03:57.579> little<00:03:57.849> concerning<00:03:58.329> there<00:03:58.900> there 00:03:59.129 --> 00:03:59.139 align:start position:0% it's a little concerning there there 00:03:59.139 --> 00:04:03.059 align:start position:0% it's a little concerning there there might<00:03:59.379> actually<00:03:59.650> be<00:04:00.299> possible<00:04:01.319> exploits<00:04:02.319> by 00:04:03.059 --> 00:04:03.069 align:start position:0% might actually be possible exploits by 00:04:03.069 --> 00:04:05.449 align:start position:0% might actually be possible exploits by writing<00:04:03.489> contracts<00:04:04.120> that<00:04:04.329> do<00:04:04.479> exponentiation 00:04:05.449 --> 00:04:05.459 align:start position:0% writing contracts that do exponentiation 00:04:05.459 --> 00:04:08.999 align:start position:0% writing contracts that do exponentiation and<00:04:06.459> get<00:04:06.790> charged<00:04:07.180> only<00:04:07.419> a<00:04:07.629> few<00:04:07.959> gasps<00:04:08.259> but 00:04:08.999 --> 00:04:09.009 align:start position:0% and get charged only a few gasps but 00:04:09.009 --> 00:04:14.009 align:start position:0% and get charged only a few gasps but take<00:04:09.220> a<00:04:09.250> whole<00:04:09.489> lot<00:04:09.699> of<00:04:09.729> time<00:04:10.979> and<00:04:12.479> rc5<00:04:13.479> looks 00:04:14.009 --> 00:04:14.019 align:start position:0% take a whole lot of time and rc5 looks 00:04:14.019 --> 00:04:18.740 align:start position:0% take a whole lot of time and rc5 looks pretty<00:04:14.349> hard<00:04:14.829> because<00:04:15.780> rc5<00:04:16.780> works<00:04:17.739> on<00:04:18.039> dynamic 00:04:18.740 --> 00:04:18.750 align:start position:0% pretty hard because rc5 works on dynamic 00:04:18.750 --> 00:04:22.499 align:start position:0% pretty hard because rc5 works on dynamic shifting<00:04:19.750> and<00:04:20.400> EVM<00:04:21.400> does<00:04:21.729> not<00:04:21.970> yet<00:04:22.240> have<00:04:22.300> a 00:04:22.499 --> 00:04:22.509 align:start position:0% shifting and EVM does not yet have a 00:04:22.509 --> 00:04:25.860 align:start position:0% shifting and EVM does not yet have a shift<00:04:22.870> operator<00:04:23.530> and<00:04:23.919> so<00:04:24.610> it<00:04:24.820> gets<00:04:25.120> imitated 00:04:25.860 --> 00:04:25.870 align:start position:0% shift operator and so it gets imitated 00:04:25.870 --> 00:04:31.190 align:start position:0% shift operator and so it gets imitated with<00:04:26.849> exponentiation<00:04:28.289> and<00:04:29.520> in<00:04:30.520> between 00:04:31.190 --> 00:04:31.200 align:start position:0% with exponentiation and in between 00:04:31.200 --> 00:04:34.379 align:start position:0% with exponentiation and in between things<00:04:32.200> are<00:04:32.469> relatively<00:04:32.979> regular 00:04:34.379 --> 00:04:34.389 align:start position:0% things are relatively regular 00:04:34.389 --> 00:04:37.230 align:start position:0% things are relatively regular and<00:04:34.479> the<00:04:35.199> speed<00:04:35.590> you<00:04:36.129> could<00:04:36.280> actually<00:04:36.520> predict 00:04:37.230 --> 00:04:37.240 align:start position:0% and the speed you could actually predict 00:04:37.240 --> 00:04:40.619 align:start position:0% and the speed you could actually predict by<00:04:37.539> the<00:04:37.599> language<00:04:38.939> that<00:04:39.939> the<00:04:40.150> client<00:04:40.539> is 00:04:40.619 --> 00:04:40.629 align:start position:0% by the language that the client is 00:04:40.629 --> 00:04:46.010 align:start position:0% by the language that the client is written<00:04:41.020> in<00:04:41.199> oh<00:04:42.090> where<00:04:43.090> did<00:04:43.180> that<00:04:43.330> come<00:04:43.539> from 00:04:46.010 --> 00:04:46.020 align:start position:0% 00:04:46.020 --> 00:04:53.309 align:start position:0% the<00:04:47.020> next<00:04:47.289> slide<00:04:47.530> I<00:04:51.419> can't<00:04:52.419> see<00:04:52.689> the<00:04:52.840> screen 00:04:53.309 --> 00:04:53.319 align:start position:0% the next slide I can't see the screen 00:04:53.319 --> 00:04:57.390 align:start position:0% the next slide I can't see the screen these<00:04:53.800> glasses<00:04:54.159> on<00:04:55.020> so<00:04:56.020> to<00:04:56.080> simplify<00:04:56.830> it<00:04:56.889> this 00:04:57.390 --> 00:04:57.400 align:start position:0% these glasses on so to simplify it this 00:04:57.400 --> 00:04:59.369 align:start position:0% these glasses on so to simplify it this is<00:04:57.639> looking<00:04:58.030> at<00:04:58.150> one<00:04:58.360> angle 00:04:59.369 --> 00:04:59.379 align:start position:0% is looking at one angle 00:04:59.379 --> 00:05:02.909 align:start position:0% is looking at one angle it's<00:04:59.889> a<00:05:00.039> harmonic<00:05:00.610> being<00:05:00.969> of<00:05:01.360> the<00:05:02.199> performance 00:05:02.909 --> 00:05:02.919 align:start position:0% it's a harmonic being of the performance 00:05:02.919 --> 00:05:06.240 align:start position:0% it's a harmonic being of the performance of<00:05:03.129> each<00:05:03.280> client<00:05:03.939> yeah<00:05:04.689> and<00:05:05.469> it<00:05:05.650> shows<00:05:05.860> pretty 00:05:06.240 --> 00:05:06.250 align:start position:0% of each client yeah and it shows pretty 00:05:06.250 --> 00:05:08.969 align:start position:0% of each client yeah and it shows pretty much<00:05:06.520> the<00:05:07.000> same<00:05:07.240> thing 00:05:08.969 --> 00:05:08.979 align:start position:0% much the same thing 00:05:08.979 --> 00:05:13.559 align:start position:0% much the same thing so<00:05:09.789> clearly<00:05:10.210> the<00:05:10.509> interpreters<00:05:11.729> are<00:05:12.729> not<00:05:13.270> as 00:05:13.559 --> 00:05:13.569 align:start position:0% so clearly the interpreters are not as 00:05:13.569 --> 00:05:16.230 align:start position:0% so clearly the interpreters are not as fast<00:05:13.870> as<00:05:14.349> as<00:05:14.650> going<00:05:15.189> straight<00:05:15.520> to<00:05:15.819> machine 00:05:16.230 --> 00:05:16.240 align:start position:0% fast as as going straight to machine 00:05:16.240 --> 00:05:20.580 align:start position:0% fast as as going straight to machine code<00:05:16.509> by<00:05:16.870> any<00:05:17.110> route<00:05:17.500> yeah<00:05:18.370> and<00:05:18.819> clearly<00:05:19.590> some 00:05:20.580 --> 00:05:20.590 align:start position:0% code by any route yeah and clearly some 00:05:20.590 --> 00:05:22.610 align:start position:0% code by any route yeah and clearly some interpreters<00:05:21.250> are<00:05:21.490> better<00:05:21.699> than<00:05:21.939> others<00:05:22.300> but 00:05:22.610 --> 00:05:22.620 align:start position:0% interpreters are better than others but 00:05:22.620 --> 00:05:25.709 align:start position:0% interpreters are better than others but they've<00:05:23.620> all<00:05:23.770> been<00:05:24.009> good<00:05:24.189> enough<00:05:24.370> so<00:05:24.909> far<00:05:24.939> for 00:05:25.709 --> 00:05:25.719 align:start position:0% they've all been good enough so far for 00:05:25.719 --> 00:05:29.790 align:start position:0% they've all been good enough so far for our<00:05:25.810> purposes<00:05:27.479> and<00:05:28.479> of<00:05:28.810> course<00:05:29.050> I<00:05:29.259> love<00:05:29.469> car 00:05:29.790 --> 00:05:29.800 align:start position:0% our purposes and of course I love car 00:05:29.800 --> 00:05:33.450 align:start position:0% our purposes and of course I love car races<00:05:30.310> as<00:05:30.520> examples<00:05:31.210> last<00:05:32.139> year<00:05:32.650> somebody<00:05:33.099> in 00:05:33.450 --> 00:05:33.460 align:start position:0% races as examples last year somebody in 00:05:33.460 --> 00:05:35.820 align:start position:0% races as examples last year somebody in the<00:05:33.550> audience<00:05:33.699> shouted<00:05:34.539> out<00:05:34.659> that<00:05:35.110> instead<00:05:35.560> of 00:05:35.820 --> 00:05:35.830 align:start position:0% the audience shouted out that instead of 00:05:35.830 --> 00:05:38.159 align:start position:0% the audience shouted out that instead of used<00:05:36.129> in<00:05:36.310> classic<00:05:36.849> cars<00:05:37.150> burning<00:05:37.629> tons<00:05:37.990> of 00:05:38.159 --> 00:05:38.169 align:start position:0% used in classic cars burning tons of 00:05:38.169 --> 00:05:42.409 align:start position:0% used in classic cars burning tons of gasoline<00:05:38.800> I<00:05:39.129> should<00:05:39.879> be<00:05:40.089> using<00:05:40.330> Tesla<00:05:41.259> and 00:05:42.409 --> 00:05:42.419 align:start position:0% gasoline I should be using Tesla and 00:05:42.419 --> 00:05:46.260 align:start position:0% gasoline I should be using Tesla and Tesla's<00:05:43.419> are<00:05:43.569> nice<00:05:44.129> it's<00:05:45.129> George<00:05:45.400> Hallam<00:05:45.879> here 00:05:46.260 --> 00:05:46.270 align:start position:0% Tesla's are nice it's George Hallam here 00:05:46.270 --> 00:05:49.680 align:start position:0% Tesla's are nice it's George Hallam here are<00:05:46.930> you<00:05:47.050> here<00:05:47.229> George<00:05:48.089> George<00:05:49.089> would<00:05:49.360> agree 00:05:49.680 --> 00:05:49.690 align:start position:0% are you here George George would agree 00:05:49.690 --> 00:05:51.899 align:start position:0% are you here George George would agree that<00:05:49.719> rather<00:05:50.289> than<00:05:50.560> a<00:05:50.589> Tesla<00:05:51.250> this<00:05:51.550> would<00:05:51.789> be 00:05:51.899 --> 00:05:51.909 align:start position:0% that rather than a Tesla this would be 00:05:51.909 --> 00:05:53.749 align:start position:0% that rather than a Tesla this would be much<00:05:52.089> cooler<00:05:52.360> this<00:05:52.750> is<00:05:52.930> a<00:05:52.960> 68<00:05:53.589> Mustang 00:05:53.749 --> 00:05:53.759 align:start position:0% much cooler this is a 68 Mustang 00:05:53.759 --> 00:05:57.959 align:start position:0% much cooler this is a 68 Mustang hatchback<00:05:55.770> under<00:05:56.770> the<00:05:56.889> hood<00:05:57.129> is<00:05:57.339> actually 00:05:57.959 --> 00:05:57.969 align:start position:0% hatchback under the hood is actually 00:05:57.969 --> 00:06:01.019 align:start position:0% hatchback under the hood is actually some<00:05:58.479> powerful<00:05:58.810> electric<00:05:59.469> motors<00:05:59.919> and<00:06:00.190> the 00:06:01.019 --> 00:06:01.029 align:start position:0% some powerful electric motors and the 00:06:01.029 --> 00:06:03.890 align:start position:0% some powerful electric motors and the trunk<00:06:01.300> is<00:06:01.449> full<00:06:01.659> of<00:06:01.870> lithium-ion<00:06:02.409> batteries 00:06:03.890 --> 00:06:03.900 align:start position:0% trunk is full of lithium-ion batteries 00:06:03.900 --> 00:06:13.800 align:start position:0% trunk is full of lithium-ion batteries and<00:06:07.500> if<00:06:08.500> this<00:06:08.830> works<00:06:10.650> where's<00:06:11.650> the<00:06:11.949> button<00:06:12.810> we 00:06:13.800 --> 00:06:13.810 align:start position:0% and if this works where's the button we 00:06:13.810 --> 00:06:16.700 align:start position:0% and if this works where's the button we will<00:06:14.020> see<00:06:14.229> how<00:06:14.439> this<00:06:14.650> does<00:06:14.889> against<00:06:15.370> the<00:06:15.460> Tesla 00:06:16.700 --> 00:06:16.710 align:start position:0% will see how this does against the Tesla 00:06:16.710 --> 00:06:21.900 align:start position:0% will see how this does against the Tesla is<00:06:17.710> ready<00:06:18.069> sound<00:06:18.370> just<00:06:18.819> no<00:06:18.969> such<00:06:19.180> luck 00:06:21.900 --> 00:06:21.910 align:start position:0% 00:06:21.910 --> 00:06:28.450 align:start position:0% but<00:06:22.910> there<00:06:23.630> goes<00:06:23.660> the<00:06:24.020> Mustang<00:06:27.250> Tesla<00:06:28.250> doesn't 00:06:28.450 --> 00:06:28.460 align:start position:0% but there goes the Mustang Tesla doesn't 00:06:28.460 --> 00:06:37.450 align:start position:0% but there goes the Mustang Tesla doesn't have<00:06:28.729> a<00:06:28.759> chance 00:06:37.450 --> 00:06:37.460 align:start position:0% 00:06:37.460 --> 00:06:40.629 align:start position:0% it<00:06:38.030> did<00:06:38.240> just<00:06:38.479> missing<00:06:39.080> but<00:06:40.070> after<00:06:40.610> a 00:06:40.629 --> 00:06:40.639 align:start position:0% it did just missing but after a 00:06:40.639 --> 00:06:43.480 align:start position:0% it did just missing but after a quarter-mile<00:06:41.449> it<00:06:41.750> had<00:06:41.930> gotten<00:06:42.139> to<00:06:42.500> 140<00:06:43.280> miles 00:06:43.480 --> 00:06:43.490 align:start position:0% quarter-mile it had gotten to 140 miles 00:06:43.490 --> 00:06:45.510 align:start position:0% quarter-mile it had gotten to 140 miles an<00:06:43.729> hour 00:06:45.510 --> 00:06:45.520 align:start position:0% an hour 00:06:45.520 --> 00:06:51.820 align:start position:0% an hour that's<00:06:46.520> a<00:06:46.699> lot<00:06:49.449> so<00:06:50.449> what<00:06:50.810> what<00:06:51.139> keeps<00:06:51.500> those 00:06:51.820 --> 00:06:51.830 align:start position:0% that's a lot so what what keeps those 00:06:51.830 --> 00:06:54.790 align:start position:0% that's a lot so what what keeps those interpreters<00:06:52.759> from<00:06:53.060> reaching<00:06:53.690> native<00:06:54.289> speed 00:06:54.790 --> 00:06:54.800 align:start position:0% interpreters from reaching native speed 00:06:54.800 --> 00:06:57.820 align:start position:0% interpreters from reaching native speed and<00:06:55.270> the<00:06:56.270> first<00:06:56.479> answer<00:06:56.990> is<00:06:57.020> their 00:06:57.820 --> 00:06:57.830 align:start position:0% and the first answer is their 00:06:57.830 --> 00:06:59.560 align:start position:0% and the first answer is their interpreters<00:06:58.580> so<00:06:59.030> they've<00:06:59.180> got<00:06:59.330> that 00:06:59.560 --> 00:06:59.570 align:start position:0% interpreters so they've got that 00:06:59.570 --> 00:07:02.170 align:start position:0% interpreters so they've got that overhead<00:07:00.199> you<00:07:01.039> can<00:07:01.070> work<00:07:01.490> hard<00:07:01.849> you<00:07:02.150> can 00:07:02.170 --> 00:07:02.180 align:start position:0% overhead you can work hard you can 00:07:02.180 --> 00:07:06.040 align:start position:0% overhead you can work hard you can reduce<00:07:02.720> the<00:07:02.990> overhead<00:07:03.699> but<00:07:04.699> generally<00:07:05.509> you 00:07:06.040 --> 00:07:06.050 align:start position:0% reduce the overhead but generally you 00:07:06.050 --> 00:07:09.189 align:start position:0% reduce the overhead but generally you can't<00:07:06.349> do<00:07:06.560> better<00:07:06.770> than<00:07:06.979> about<00:07:07.690> three<00:07:08.690> three 00:07:09.189 --> 00:07:09.199 align:start position:0% can't do better than about three three 00:07:09.199 --> 00:07:11.860 align:start position:0% can't do better than about three three or<00:07:09.320> four<00:07:09.620> to<00:07:09.800> one<00:07:10.070> compared<00:07:11.060> to<00:07:11.240> native<00:07:11.419> code 00:07:11.860 --> 00:07:11.870 align:start position:0% or four to one compared to native code 00:07:11.870 --> 00:07:16.499 align:start position:0% or four to one compared to native code um<00:07:12.940> for<00:07:13.940> our<00:07:14.270> particular<00:07:14.949> interpreter<00:07:15.949> the 00:07:16.499 --> 00:07:16.509 align:start position:0% um for our particular interpreter the 00:07:16.509 --> 00:07:20.320 align:start position:0% um for our particular interpreter the 256<00:07:17.509> bit<00:07:17.810> registers<00:07:18.500> slowed<00:07:18.800> us<00:07:19.009> down<00:07:19.330> because 00:07:20.320 --> 00:07:20.330 align:start position:0% 256 bit registers slowed us down because 00:07:20.330 --> 00:07:25.300 align:start position:0% 256 bit registers slowed us down because real<00:07:20.900> hardware<00:07:21.560> has<00:07:22.240> you<00:07:23.240> know<00:07:23.360> 32<00:07:24.139> 64<00:07:24.800> bit 00:07:25.300 --> 00:07:25.310 align:start position:0% real hardware has you know 32 64 bit 00:07:25.310 --> 00:07:28.659 align:start position:0% real hardware has you know 32 64 bit registers<00:07:25.970> and<00:07:26.360> the<00:07:27.260> unconstrained<00:07:28.039> control 00:07:28.659 --> 00:07:28.669 align:start position:0% registers and the unconstrained control 00:07:28.669 --> 00:07:34.529 align:start position:0% registers and the unconstrained control flow<00:07:29.060> hurts<00:07:29.720> us<00:07:29.930> a<00:07:30.110> lot<00:07:30.349> and<00:07:30.680> I'll<00:07:30.800> get<00:07:31.039> to<00:07:31.070> that 00:07:34.529 --> 00:07:34.539 align:start position:0% 00:07:34.539 --> 00:07:38.860 align:start position:0% the<00:07:35.539> 256<00:07:36.470> Hertz<00:07:36.770> if<00:07:37.430> you<00:07:37.580> remember<00:07:38.240> you<00:07:38.750> know 00:07:38.860 --> 00:07:38.870 align:start position:0% the 256 Hertz if you remember you know 00:07:38.870 --> 00:07:44.830 align:start position:0% the 256 Hertz if you remember you know grade<00:07:39.229> school<00:07:39.530> math<00:07:41.440> adding<00:07:42.639> multiplying<00:07:43.840> you 00:07:44.830 --> 00:07:44.840 align:start position:0% grade school math adding multiplying you 00:07:44.840 --> 00:07:47.140 align:start position:0% grade school math adding multiplying you know<00:07:44.960> two<00:07:45.260> numbers<00:07:45.710> is<00:07:45.949> pretty<00:07:46.220> easy<00:07:46.520> you<00:07:46.970> can 00:07:47.140 --> 00:07:47.150 align:start position:0% know two numbers is pretty easy you can 00:07:47.150 --> 00:07:51.640 align:start position:0% know two numbers is pretty easy you can do<00:07:47.270> it<00:07:47.300> in<00:07:47.539> your<00:07:47.659> head<00:07:48.729> if<00:07:49.729> you<00:07:50.330> have<00:07:50.590> you<00:07:51.590> know 00:07:51.640 --> 00:07:51.650 align:start position:0% do it in your head if you have you know 00:07:51.650 --> 00:07:54.159 align:start position:0% do it in your head if you have you know four<00:07:52.190> numbers<00:07:52.639> suddenly<00:07:53.539> it<00:07:53.659> gets<00:07:53.810> a<00:07:54.020> lot 00:07:54.159 --> 00:07:54.169 align:start position:0% four numbers suddenly it gets a lot 00:07:54.169 --> 00:07:57.580 align:start position:0% four numbers suddenly it gets a lot harder<00:07:54.229> and<00:07:54.889> it's<00:07:55.729> quadratic<00:07:56.599> so<00:07:57.139> it<00:07:57.349> gets 00:07:57.580 --> 00:07:57.590 align:start position:0% harder and it's quadratic so it gets 00:07:57.590 --> 00:08:02.589 align:start position:0% harder and it's quadratic so it gets worse<00:07:57.800> and<00:07:58.009> worse<00:07:59.620> 256<00:08:00.620> squared<00:08:01.479> there's<00:08:02.479> a 00:08:02.589 --> 00:08:02.599 align:start position:0% worse and worse 256 squared there's a 00:08:02.599 --> 00:08:09.550 align:start position:0% worse and worse 256 squared there's a lot<00:08:05.110> control<00:08:06.110> flow<00:08:07.090> the<00:08:08.090> jump<00:08:08.419> operator<00:08:08.630> an 00:08:09.550 --> 00:08:09.560 align:start position:0% lot control flow the jump operator an 00:08:09.560 --> 00:08:13.749 align:start position:0% lot control flow the jump operator an EVM<00:08:10.130> I<00:08:10.639> mean<00:08:11.860> you<00:08:12.860> know<00:08:12.949> go<00:08:13.190> twos<00:08:13.490> are 00:08:13.749 --> 00:08:13.759 align:start position:0% EVM I mean you know go twos are 00:08:13.759 --> 00:08:16.300 align:start position:0% EVM I mean you know go twos are considered<00:08:14.240> harmful<00:08:14.659> but<00:08:15.110> at<00:08:15.229> least<00:08:15.349> you<00:08:16.190> know 00:08:16.300 --> 00:08:16.310 align:start position:0% considered harmful but at least you know 00:08:16.310 --> 00:08:20.020 align:start position:0% considered harmful but at least you know if<00:08:16.580> you<00:08:16.759> say<00:08:17.000> go<00:08:17.270> to<00:08:17.380> label<00:08:18.380> it<00:08:18.620> will<00:08:19.159> go<00:08:19.340> to<00:08:19.400> one 00:08:20.020 --> 00:08:20.030 align:start position:0% if you say go to label it will go to one 00:08:20.030 --> 00:08:24.850 align:start position:0% if you say go to label it will go to one in<00:08:20.270> exactly<00:08:20.810> one<00:08:21.139> label<00:08:21.919> in<00:08:22.130> the<00:08:23.060> EVM<00:08:23.590> you<00:08:24.590> say 00:08:24.850 --> 00:08:24.860 align:start position:0% in exactly one label in the EVM you say 00:08:24.860 --> 00:08:29.529 align:start position:0% in exactly one label in the EVM you say go<00:08:25.130> to<00:08:26.169> whatever's<00:08:27.169> on<00:08:27.380> the<00:08:27.560> stack 00:08:29.529 --> 00:08:29.539 align:start position:0% go to whatever's on the stack 00:08:29.539 --> 00:08:32.079 align:start position:0% go to whatever's on the stack so<00:08:30.169> there's<00:08:30.380> no<00:08:30.590> way<00:08:30.770> of<00:08:30.979> knowing<00:08:31.430> often 00:08:32.079 --> 00:08:32.089 align:start position:0% so there's no way of knowing often 00:08:32.089 --> 00:08:36.300 align:start position:0% so there's no way of knowing often statically<00:08:33.130> where<00:08:34.130> it's<00:08:34.339> going<00:08:34.640> to<00:08:34.789> go 00:08:36.300 --> 00:08:36.310 align:start position:0% statically where it's going to go 00:08:36.310 --> 00:08:38.949 align:start position:0% statically where it's going to go so<00:08:37.310> you<00:08:37.399> could<00:08:37.699> have<00:08:37.729> a<00:08:37.880> nice<00:08:38.149> little<00:08:38.389> program 00:08:38.949 --> 00:08:38.959 align:start position:0% so you could have a nice little program 00:08:38.959 --> 00:08:42.250 align:start position:0% so you could have a nice little program like<00:08:39.260> this<00:08:39.589> and<00:08:40.099> F<00:08:40.459> calls<00:08:40.969> G<00:08:41.419> and<00:08:41.719> H<00:08:41.899> and 00:08:42.250 --> 00:08:42.260 align:start position:0% like this and F calls G and H and 00:08:42.260 --> 00:08:45.340 align:start position:0% like this and F calls G and H and returns<00:08:42.649> and<00:08:43.159> it<00:08:43.250> calls<00:08:43.459> AI<00:08:43.729> and<00:08:44.000> returns<00:08:44.420> etc 00:08:45.340 --> 00:08:45.350 align:start position:0% returns and it calls AI and returns etc 00:08:45.350 --> 00:08:48.460 align:start position:0% returns and it calls AI and returns etc nice<00:08:46.339> clean<00:08:46.730> structure<00:08:47.360> no<00:08:47.660> trouble<00:08:48.139> no 00:08:48.460 --> 00:08:48.470 align:start position:0% nice clean structure no trouble no 00:08:48.470 --> 00:08:52.470 align:start position:0% nice clean structure no trouble no trouble<00:08:48.500> to<00:08:49.100> understand<00:08:50.230> static<00:08:51.230> analysis 00:08:52.470 --> 00:08:52.480 align:start position:0% trouble to understand static analysis 00:08:52.480 --> 00:08:55.600 align:start position:0% trouble to understand static analysis anything<00:08:53.480> no<00:08:53.720> trouble<00:08:54.310> what's<00:08:55.310> it<00:08:55.459> actually 00:08:55.600 --> 00:08:55.610 align:start position:0% anything no trouble what's it actually 00:08:55.610 --> 00:08:58.329 align:start position:0% anything no trouble what's it actually looked<00:08:56.120> like<00:08:56.209> to<00:08:56.510> the<00:08:56.660> EVM<00:08:57.050> that's<00:08:58.040> what<00:08:58.250> it 00:08:58.329 --> 00:08:58.339 align:start position:0% looked like to the EVM that's what it 00:08:58.339 --> 00:09:01.780 align:start position:0% looked like to the EVM that's what it looks<00:08:58.550> like<00:08:58.730> to<00:08:58.940> the<00:08:59.120> EVM<00:09:00.490> so<00:09:01.490> if<00:09:01.639> you're 00:09:01.780 --> 00:09:01.790 align:start position:0% looks like to the EVM so if you're 00:09:01.790 --> 00:09:03.699 align:start position:0% looks like to the EVM so if you're trying<00:09:02.029> to<00:09:02.180> do<00:09:02.300> formal<00:09:02.750> analysis<00:09:03.350> if<00:09:03.560> you're 00:09:03.699 --> 00:09:03.709 align:start position:0% trying to do formal analysis if you're 00:09:03.709 --> 00:09:05.590 align:start position:0% trying to do formal analysis if you're trying<00:09:03.949> to<00:09:04.040> write<00:09:04.190> a<00:09:04.220> compiler<00:09:05.060> if<00:09:05.480> you're 00:09:05.590 --> 00:09:05.600 align:start position:0% trying to write a compiler if you're 00:09:05.600 --> 00:09:09.579 align:start position:0% trying to write a compiler if you're trying<00:09:05.839> to<00:09:05.959> do<00:09:06.079> anything<00:09:06.320> with<00:09:06.709> it<00:09:07.690> again<00:09:08.690> the 00:09:09.579 --> 00:09:09.589 align:start position:0% trying to do anything with it again the 00:09:09.589 --> 00:09:12.009 align:start position:0% trying to do anything with it again the number<00:09:09.920> of<00:09:09.980> paths<00:09:10.339> goes<00:09:10.730> up<00:09:11.019> quadratically 00:09:12.009 --> 00:09:12.019 align:start position:0% number of paths goes up quadratically 00:09:12.019 --> 00:09:16.300 align:start position:0% number of paths goes up quadratically and<00:09:12.589> you're<00:09:12.800> in<00:09:12.949> trouble<00:09:14.529> because<00:09:15.529> pretty 00:09:16.300 --> 00:09:16.310 align:start position:0% and you're in trouble because pretty 00:09:16.310 --> 00:09:18.689 align:start position:0% and you're in trouble because pretty much<00:09:16.430> if<00:09:17.060> you<00:09:17.269> can't<00:09:17.540> do<00:09:17.720> it<00:09:17.870> in<00:09:17.990> linear<00:09:18.199> time 00:09:18.689 --> 00:09:18.699 align:start position:0% much if you can't do it in linear time 00:09:18.699 --> 00:09:22.449 align:start position:0% much if you can't do it in linear time or<00:09:19.699> at<00:09:19.910> least<00:09:20.089> n<00:09:20.329> log<00:09:20.570> n<00:09:20.839> time<00:09:21.079> on<00:09:21.769> the 00:09:22.449 --> 00:09:22.459 align:start position:0% or at least n log n time on the 00:09:22.459 --> 00:09:25.300 align:start position:0% or at least n log n time on the blockchain<00:09:22.810> you<00:09:23.810> know<00:09:23.899> that<00:09:24.260> deployment<00:09:25.040> time 00:09:25.300 --> 00:09:25.310 align:start position:0% blockchain you know that deployment time 00:09:25.310 --> 00:09:30.759 align:start position:0% blockchain you know that deployment time or<00:09:25.639> at<00:09:25.880> runtime<00:09:27.100> you<00:09:28.100> can't<00:09:28.370> do<00:09:28.670> it<00:09:29.380> so<00:09:30.380> how<00:09:30.709> do 00:09:30.759 --> 00:09:30.769 align:start position:0% or at runtime you can't do it so how do 00:09:30.769 --> 00:09:33.189 align:start position:0% or at runtime you can't do it so how do we<00:09:30.920> do<00:09:31.070> better 00:09:33.189 --> 00:09:33.199 align:start position:0% we do better 00:09:33.199 --> 00:09:38.050 align:start position:0% we do better well<00:09:34.160> EVM<00:09:34.850> JIT<00:09:35.300> is<00:09:36.050> already<00:09:36.560> doing<00:09:36.829> better<00:09:37.639> I 00:09:38.050 --> 00:09:38.060 align:start position:0% well EVM JIT is already doing better I 00:09:38.060 --> 00:09:40.269 align:start position:0% well EVM JIT is already doing better I won't<00:09:38.389> back<00:09:38.630> up<00:09:38.990> but<00:09:39.440> if<00:09:39.649> you<00:09:39.800> look<00:09:40.010> at<00:09:40.160> the 00:09:40.269 --> 00:09:40.279 align:start position:0% won't back up but if you look at the 00:09:40.279 --> 00:09:43.870 align:start position:0% won't back up but if you look at the slide<00:09:41.199> the<00:09:42.199> EVM<00:09:42.529> JIT<00:09:42.829> is<00:09:43.100> actually<00:09:43.610> pretty 00:09:43.870 --> 00:09:43.880 align:start position:0% slide the EVM JIT is actually pretty 00:09:43.880 --> 00:09:46.840 align:start position:0% slide the EVM JIT is actually pretty close<00:09:44.269> to<00:09:44.660> the<00:09:44.779> native<00:09:45.139> speed<00:09:45.589> you<00:09:46.399> know<00:09:46.519> it 00:09:46.840 --> 00:09:46.850 align:start position:0% close to the native speed you know it 00:09:46.850 --> 00:09:50.379 align:start position:0% close to the native speed you know it does<00:09:47.319> very<00:09:48.319> well<00:09:48.649> on<00:09:49.040> the<00:09:49.310> wide<00:09:49.519> arithmetic 00:09:50.379 --> 00:09:50.389 align:start position:0% does very well on the wide arithmetic 00:09:50.389 --> 00:09:54.309 align:start position:0% does very well on the wide arithmetic and<00:09:50.690> not<00:09:51.230> so<00:09:51.589> well<00:09:51.860> uh<00:09:52.250> narrow<00:09:52.699> arithmetic<00:09:53.600> and 00:09:54.309 --> 00:09:54.319 align:start position:0% and not so well uh narrow arithmetic and 00:09:54.319 --> 00:09:59.910 align:start position:0% and not so well uh narrow arithmetic and complex<00:09:55.100> logic<00:09:56.620> but<00:09:57.620> it's<00:09:57.769> a<00:09:57.889> very<00:09:58.100> good<00:09:58.339> chip 00:09:59.910 --> 00:09:59.920 align:start position:0% complex logic but it's a very good chip 00:09:59.920 --> 00:10:02.620 align:start position:0% complex logic but it's a very good chip I've<00:10:00.920> told<00:10:01.279> Pavel<00:10:01.790> he<00:10:01.970> gets<00:10:02.180> to<00:10:02.360> be<00:10:02.389> the 00:10:02.620 --> 00:10:02.630 align:start position:0% I've told Pavel he gets to be the 00:10:02.630 --> 00:10:04.960 align:start position:0% I've told Pavel he gets to be the electric<00:10:03.110> Fox<00:10:03.410> he's<00:10:03.889> tired<00:10:04.190> of<00:10:04.459> these<00:10:04.579> little 00:10:04.960 --> 00:10:04.970 align:start position:0% electric Fox he's tired of these little 00:10:04.970 --> 00:10:07.420 align:start position:0% electric Fox he's tired of these little you<00:10:05.899> know<00:10:05.990> three<00:10:06.319> and<00:10:06.350> four<00:10:06.709> letter<00:10:07.040> names 00:10:07.420 --> 00:10:07.430 align:start position:0% you know three and four letter names 00:10:07.430 --> 00:10:09.939 align:start position:0% you know three and four letter names that<00:10:07.790> don't<00:10:07.970> mean<00:10:08.149> anything<00:10:08.709> this<00:10:09.709> is<00:10:09.769> a 00:10:09.939 --> 00:10:09.949 align:start position:0% that don't mean anything this is a 00:10:09.949 --> 00:10:13.030 align:start position:0% that don't mean anything this is a racing<00:10:10.880> team<00:10:11.120> out<00:10:11.329> of<00:10:11.360> Latvia<00:10:12.019> that's<00:10:12.769> a 00:10:13.030 --> 00:10:13.040 align:start position:0% racing team out of Latvia that's a 00:10:13.040 --> 00:10:17.800 align:start position:0% racing team out of Latvia that's a completely<00:10:13.490> electric<00:10:14.560> dragster<00:10:16.000> and<00:10:17.000> here<00:10:17.660> it 00:10:17.800 --> 00:10:17.810 align:start position:0% completely electric dragster and here it 00:10:17.810 --> 00:10:25.079 align:start position:0% completely electric dragster and here it is<00:10:18.310> winning<00:10:19.310> the<00:10:20.410> European<00:10:21.410> world<00:10:22.399> record 00:10:25.079 --> 00:10:25.089 align:start position:0% 00:10:25.089 --> 00:10:29.500 align:start position:0% there<00:10:26.089> it<00:10:26.300> goes 00:10:29.500 --> 00:10:29.510 align:start position:0% 00:10:29.510 --> 00:10:34.430 align:start position:0% drag<00:10:30.510> races<00:10:30.960> are<00:10:31.110> fast<00:10:32.690> it's<00:10:33.690> not<00:10:33.720> impressive 00:10:34.430 --> 00:10:34.440 align:start position:0% drag races are fast it's not impressive 00:10:34.440 --> 00:10:38.590 align:start position:0% drag races are fast it's not impressive is<00:10:34.620> it<00:10:35.030> it's<00:10:36.030> over 00:10:38.590 --> 00:10:38.600 align:start position:0% 00:10:38.600 --> 00:10:44.960 align:start position:0% hello<00:10:40.670> 275<00:10:41.670> miles<00:10:41.970> an<00:10:42.240> hour<00:10:42.420> just<00:10:43.790> a<00:10:44.790> few 00:10:44.960 --> 00:10:44.970 align:start position:0% hello 275 miles an hour just a few 00:10:44.970 --> 00:10:52.060 align:start position:0% hello 275 miles an hour just a few seconds<00:10:46.190> love<00:10:47.190> these<00:10:47.400> things<00:10:50.210> so<00:10:51.210> we've<00:10:51.450> got 00:10:52.060 --> 00:10:52.070 align:start position:0% seconds love these things so we've got 00:10:52.070 --> 00:10:55.040 align:start position:0% seconds love these things so we've got we've<00:10:53.070> got<00:10:53.130> two<00:10:53.490> research<00:10:53.970> programs<00:10:54.660> who'd 00:10:55.040 --> 00:10:55.050 align:start position:0% we've got two research programs who'd 00:10:55.050 --> 00:10:59.660 align:start position:0% we've got two research programs who'd have<00:10:55.560> been<00:10:55.680> going<00:10:56.630> on<00:10:57.630> how<00:10:58.560> to<00:10:58.620> improve<00:10:59.160> things 00:10:59.660 --> 00:10:59.670 align:start position:0% have been going on how to improve things 00:10:59.670 --> 00:11:04.670 align:start position:0% have been going on how to improve things they've<00:11:00.570> been<00:11:00.660> nicknamed<00:11:01.410> EVM<00:11:02.280> 1.5<00:11:03.180> and<00:11:03.810> EVM 00:11:04.670 --> 00:11:04.680 align:start position:0% they've been nicknamed EVM 1.5 and EVM 00:11:04.680 --> 00:11:08.390 align:start position:0% they've been nicknamed EVM 1.5 and EVM 2.0<00:11:06.530> which<00:11:07.530> doesn't<00:11:07.950> really<00:11:08.070> mean<00:11:08.370> anything 00:11:08.390 --> 00:11:08.400 align:start position:0% 2.0 which doesn't really mean anything 00:11:08.400 --> 00:11:12.310 align:start position:0% 2.0 which doesn't really mean anything except<00:11:09.390> those<00:11:09.570> are<00:11:09.780> the<00:11:09.900> nicknames<00:11:10.350> yeah 00:11:12.310 --> 00:11:12.320 align:start position:0% except those are the nicknames yeah 00:11:12.320 --> 00:11:19.790 align:start position:0% except those are the nicknames yeah and<00:11:14.300> for<00:11:15.300> 1.5<00:11:17.270> that's<00:11:18.270> that's<00:11:19.050> a<00:11:19.260> suggestion 00:11:19.790 --> 00:11:19.800 align:start position:0% and for 1.5 that's that's a suggestion 00:11:19.800 --> 00:11:24.410 align:start position:0% and for 1.5 that's that's a suggestion to<00:11:20.370> extend<00:11:21.080> the<00:11:22.080> current<00:11:22.380> EVM<00:11:23.130> by<00:11:23.940> adding 00:11:24.410 --> 00:11:24.420 align:start position:0% to extend the current EVM by adding 00:11:24.420 --> 00:11:30.140 align:start position:0% to extend the current EVM by adding adding<00:11:25.140> new<00:11:25.320> op<00:11:25.530> codes<00:11:25.950> and<00:11:27.020> requirements<00:11:29.150> so 00:11:30.140 --> 00:11:30.150 align:start position:0% adding new op codes and requirements so 00:11:30.150 --> 00:11:33.830 align:start position:0% adding new op codes and requirements so we<00:11:30.540> forbid<00:11:31.290> those<00:11:32.220> unconstrained<00:11:33.120> jumps<00:11:33.690> you 00:11:33.830 --> 00:11:33.840 align:start position:0% we forbid those unconstrained jumps you 00:11:33.840 --> 00:11:36.610 align:start position:0% we forbid those unconstrained jumps you just<00:11:34.110> we<00:11:34.620> will<00:11:34.830> not<00:11:35.010> allow<00:11:35.130> you<00:11:35.250> to<00:11:35.460> do<00:11:35.760> that 00:11:36.610 --> 00:11:36.620 align:start position:0% just we will not allow you to do that 00:11:36.620 --> 00:11:41.870 align:start position:0% just we will not allow you to do that and<00:11:37.880> we<00:11:38.990> we<00:11:39.990> then<00:11:40.230> have<00:11:40.500> to<00:11:40.830> provide<00:11:41.550> a<00:11:41.580> way<00:11:41.850> to 00:11:41.870 --> 00:11:41.880 align:start position:0% and we we then have to provide a way to 00:11:41.880 --> 00:11:44.210 align:start position:0% and we we then have to provide a way to do<00:11:42.300> the<00:11:42.510> things<00:11:42.780> that<00:11:42.960> you<00:11:43.260> otherwise<00:11:43.560> do<00:11:44.040> with 00:11:44.210 --> 00:11:44.220 align:start position:0% do the things that you otherwise do with 00:11:44.220 --> 00:11:47.660 align:start position:0% do the things that you otherwise do with those<00:11:44.370> jumps<00:11:45.090> so<00:11:45.810> there's<00:11:46.140> up<00:11:46.410> there's<00:11:47.100> up 00:11:47.660 --> 00:11:47.670 align:start position:0% those jumps so there's up there's up 00:11:47.670 --> 00:11:52.700 align:start position:0% those jumps so there's up there's up codes<00:11:48.090> for<00:11:48.330> subroutines<00:11:49.730> and<00:11:50.730> then<00:11:51.510> we've<00:11:52.500> got 00:11:52.700 --> 00:11:52.710 align:start position:0% codes for subroutines and then we've got 00:11:52.710 --> 00:11:55.370 align:start position:0% codes for subroutines and then we've got to<00:11:52.860> get<00:11:52.980> away<00:11:53.100> from<00:11:53.310> having<00:11:53.970> nothing<00:11:54.150> but<00:11:54.390> 256 00:11:55.370 --> 00:11:55.380 align:start position:0% to get away from having nothing but 256 00:11:55.380 --> 00:11:59.000 align:start position:0% to get away from having nothing but 256 bit<00:11:55.620> registers<00:11:56.330> so<00:11:57.330> we've<00:11:57.570> got<00:11:57.870> up<00:11:58.170> codes<00:11:58.650> for 00:11:59.000 --> 00:11:59.010 align:start position:0% bit registers so we've got up codes for 00:11:59.010 --> 00:12:02.510 align:start position:0% bit registers so we've got up codes for native<00:11:59.370> scalars<00:11:59.970> and<00:12:00.540> op<00:12:01.020> codes<00:12:01.410> for<00:12:01.860> sim<00:12:02.280> D 00:12:02.510 --> 00:12:02.520 align:start position:0% native scalars and op codes for sim D 00:12:02.520 --> 00:12:05.420 align:start position:0% native scalars and op codes for sim D because<00:12:03.300> real<00:12:04.140> hardware<00:12:04.740> has<00:12:04.980> all<00:12:05.190> this 00:12:05.420 --> 00:12:05.430 align:start position:0% because real hardware has all this 00:12:05.430 --> 00:12:09.350 align:start position:0% because real hardware has all this silicon<00:12:06.000> devoted<00:12:06.720> to<00:12:07.410> sim<00:12:07.770> d<00:12:07.980> registers<00:12:08.610> and 00:12:09.350 --> 00:12:09.360 align:start position:0% silicon devoted to sim d registers and 00:12:09.360 --> 00:12:12.290 align:start position:0% silicon devoted to sim d registers and if<00:12:09.750> you<00:12:09.900> go<00:12:10.080> to<00:12:10.140> google<00:12:10.620> and<00:12:10.890> type<00:12:11.060> sim<00:12:12.060> D 00:12:12.290 --> 00:12:12.300 align:start position:0% if you go to google and type sim D 00:12:12.300 --> 00:12:16.970 align:start position:0% if you go to google and type sim D crypto<00:12:13.070> you<00:12:14.070> get<00:12:14.400> a<00:12:14.430> lot<00:12:14.610> of<00:12:14.730> results<00:12:15.770> so<00:12:16.770> it 00:12:16.970 --> 00:12:16.980 align:start position:0% crypto you get a lot of results so it 00:12:16.980 --> 00:12:18.830 align:start position:0% crypto you get a lot of results so it would<00:12:17.130> be<00:12:17.250> useful<00:12:17.460> to<00:12:17.760> make<00:12:18.090> that<00:12:18.360> hardware 00:12:18.830 --> 00:12:18.840 align:start position:0% would be useful to make that hardware 00:12:18.840 --> 00:12:23.450 align:start position:0% would be useful to make that hardware available<00:12:19.940> and<00:12:20.940> at<00:12:21.690> deployment<00:12:22.320> time<00:12:22.680> there's 00:12:23.450 --> 00:12:23.460 align:start position:0% available and at deployment time there's 00:12:23.460 --> 00:12:26.210 align:start position:0% available and at deployment time there's a<00:12:23.520> validation<00:12:24.360> phase<00:12:24.750> that<00:12:25.560> goes<00:12:25.770> through<00:12:26.100> the 00:12:26.210 --> 00:12:26.220 align:start position:0% a validation phase that goes through the 00:12:26.220 --> 00:12:28.580 align:start position:0% a validation phase that goes through the code<00:12:26.460> and<00:12:26.490> make<00:12:27.030> sure<00:12:27.360> that<00:12:27.660> it<00:12:27.780> actually<00:12:28.260> does 00:12:28.580 --> 00:12:28.590 align:start position:0% code and make sure that it actually does 00:12:28.590 --> 00:12:33.610 align:start position:0% code and make sure that it actually does follow<00:12:29.460> the<00:12:29.760> rules<00:12:31.250> what<00:12:32.250> a<00:12:32.280> concept 00:12:33.610 --> 00:12:33.620 align:start position:0% follow the rules what a concept 00:12:33.620 --> 00:12:36.340 align:start position:0% follow the rules what a concept and<00:12:34.620> then<00:12:35.490> 2.0 00:12:36.340 --> 00:12:36.350 align:start position:0% and then 2.0 00:12:36.350 --> 00:12:39.740 align:start position:0% and then 2.0 [Music] 00:12:39.740 --> 00:12:39.750 align:start position:0% 00:12:39.750 --> 00:12:43.220 align:start position:0% well<00:12:40.589> gee<00:12:41.009> it<00:12:41.459> provides<00:12:41.879> opcodes<00:12:42.449> first<00:12:42.930> got 00:12:43.220 --> 00:12:43.230 align:start position:0% well gee it provides opcodes first got 00:12:43.230 --> 00:12:46.280 align:start position:0% well gee it provides opcodes first got your<00:12:43.470> control<00:12:43.740> flow<00:12:44.129> you<00:12:44.910> know<00:12:45.029> it's<00:12:45.629> stricter 00:12:46.280 --> 00:12:46.290 align:start position:0% your control flow you know it's stricter 00:12:46.290 --> 00:12:49.400 align:start position:0% your control flow you know it's stricter than<00:12:46.680> 1.5<00:12:47.579> it<00:12:48.089> actually<00:12:48.899> looks<00:12:49.139> like<00:12:49.230> a 00:12:49.400 --> 00:12:49.410 align:start position:0% than 1.5 it actually looks like a 00:12:49.410 --> 00:12:52.340 align:start position:0% than 1.5 it actually looks like a high-level<00:12:49.949> language<00:12:50.310> with<00:12:51.000> if-else<00:12:51.480> and 00:12:52.340 --> 00:12:52.350 align:start position:0% high-level language with if-else and 00:12:52.350 --> 00:12:56.600 align:start position:0% high-level language with if-else and such<00:12:53.420> and<00:12:54.420> it<00:12:54.660> provides<00:12:55.139> opcodes<00:12:55.800> for<00:12:56.250> native 00:12:56.600 --> 00:12:56.610 align:start position:0% such and it provides opcodes for native 00:12:56.610 --> 00:13:00.379 align:start position:0% such and it provides opcodes for native scalars<00:12:57.319> it's<00:12:58.319> Cindy<00:12:58.829> provoke<00:12:59.310> the<00:12:59.970> Cindy's 00:13:00.379 --> 00:13:00.389 align:start position:0% scalars it's Cindy provoke the Cindy's 00:13:00.389 --> 00:13:02.240 align:start position:0% scalars it's Cindy provoke the Cindy's coming<00:13:00.629> later<00:13:01.139> but<00:13:01.560> there's<00:13:01.769> a<00:13:01.829> Cindy 00:13:02.240 --> 00:13:02.250 align:start position:0% coming later but there's a Cindy 00:13:02.250 --> 00:13:05.389 align:start position:0% coming later but there's a Cindy proposal<00:13:02.970> and<00:13:03.300> it<00:13:03.839> also<00:13:04.050> has<00:13:04.439> a<00:13:04.470> validation 00:13:05.389 --> 00:13:05.399 align:start position:0% proposal and it also has a validation 00:13:05.399 --> 00:13:08.210 align:start position:0% proposal and it also has a validation phase<00:13:05.730> where<00:13:06.209> it<00:13:06.329> validates<00:13:06.930> control<00:13:07.889> flow 00:13:08.210 --> 00:13:08.220 align:start position:0% phase where it validates control flow 00:13:08.220 --> 00:13:12.679 align:start position:0% phase where it validates control flow and<00:13:08.550> stack<00:13:08.819> discipline<00:13:09.480> and<00:13:09.750> type<00:13:10.290> safety<00:13:11.689> so 00:13:12.679 --> 00:13:12.689 align:start position:0% and stack discipline and type safety so 00:13:12.689 --> 00:13:14.929 align:start position:0% and stack discipline and type safety so they're<00:13:12.930> they're<00:13:13.199> very<00:13:13.410> similar<00:13:13.800> at<00:13:14.310> that<00:13:14.699> at 00:13:14.929 --> 00:13:14.939 align:start position:0% they're they're very similar at that at 00:13:14.939 --> 00:13:20.540 align:start position:0% they're they're very similar at that at that<00:13:15.180> level<00:13:18.290> so<00:13:19.290> we've<00:13:19.500> got<00:13:19.709> some<00:13:19.980> technically 00:13:20.540 --> 00:13:20.550 align:start position:0% that level so we've got some technically 00:13:20.550 --> 00:13:24.559 align:start position:0% that level so we've got some technically very<00:13:20.850> very<00:13:21.269> similar<00:13:21.540> proposals<00:13:22.350> um<00:13:23.360> they<00:13:24.360> both 00:13:24.559 --> 00:13:24.569 align:start position:0% very very similar proposals um they both 00:13:24.569 --> 00:13:28.819 align:start position:0% very very similar proposals um they both provide<00:13:25.699> for<00:13:26.699> for<00:13:27.089> very<00:13:27.269> fast<00:13:27.680> compilations<00:13:28.680> a 00:13:28.819 --> 00:13:28.829 align:start position:0% provide for for very fast compilations a 00:13:28.829 --> 00:13:33.590 align:start position:0% provide for for very fast compilations a native<00:13:29.160> code<00:13:30.920> it<00:13:31.920> could<00:13:32.339> be<00:13:32.579> done<00:13:32.819> as<00:13:33.029> the<00:13:33.209> JIT 00:13:33.590 --> 00:13:33.600 align:start position:0% native code it could be done as the JIT 00:13:33.600 --> 00:13:39.590 align:start position:0% native code it could be done as the JIT yeah<00:13:34.529> we've<00:13:35.310> come<00:13:35.339> to<00:13:35.850> realize<00:13:38.600> etherium 00:13:39.590 --> 00:13:39.600 align:start position:0% yeah we've come to realize etherium 00:13:39.600 --> 00:13:43.869 align:start position:0% yeah we've come to realize etherium cannot<00:13:40.290> do<00:13:40.529> jets<00:13:40.889> there<00:13:41.819> they<00:13:42.300> are<00:13:42.449> actually 00:13:43.869 --> 00:13:43.879 align:start position:0% cannot do jets there they are actually 00:13:43.879 --> 00:13:49.189 align:start position:0% cannot do jets there they are actually exploitable<00:13:44.879> that<00:13:45.750> is<00:13:46.850> if<00:13:47.850> you<00:13:48.059> find<00:13:48.389> or<00:13:48.839> write 00:13:49.189 --> 00:13:49.199 align:start position:0% exploitable that is if you find or write 00:13:49.199 --> 00:13:54.889 align:start position:0% exploitable that is if you find or write a<00:13:49.559> contract<00:13:51.589> which<00:13:52.589> takes<00:13:53.189> a<00:13:53.430> long<00:13:53.670> time<00:13:53.899> to 00:13:54.889 --> 00:13:54.899 align:start position:0% a contract which takes a long time to 00:13:54.899 --> 00:13:59.509 align:start position:0% a contract which takes a long time to compile<00:13:55.559> with<00:13:56.129> the<00:13:56.309> JIT<00:13:57.110> but<00:13:58.110> requires<00:13:59.069> very 00:13:59.509 --> 00:13:59.519 align:start position:0% compile with the JIT but requires very 00:13:59.519 --> 00:14:04.340 align:start position:0% compile with the JIT but requires very little<00:13:59.819> gas<00:14:00.089> to<00:14:00.600> run<00:14:02.329> and<00:14:03.329> you<00:14:04.050> start 00:14:04.340 --> 00:14:04.350 align:start position:0% little gas to run and you start 00:14:04.350 --> 00:14:09.350 align:start position:0% little gas to run and you start hammering<00:14:05.120> those<00:14:06.139> contracts<00:14:07.519> you<00:14:08.519> can<00:14:08.910> you 00:14:09.350 --> 00:14:09.360 align:start position:0% hammering those contracts you can you 00:14:09.360 --> 00:14:11.749 align:start position:0% hammering those contracts you can you can<00:14:09.569> do<00:14:09.720> a<00:14:09.750> really<00:14:10.139> nice<00:14:10.319> you<00:14:11.040> know<00:14:11.160> do<00:14:11.459> s 00:14:11.749 --> 00:14:11.759 align:start position:0% can do a really nice you know do s 00:14:11.759 --> 00:14:14.809 align:start position:0% can do a really nice you know do s attack<00:14:12.410> so<00:14:13.410> if<00:14:13.740> you're<00:14:13.889> going<00:14:14.069> to<00:14:14.370> do<00:14:14.579> any 00:14:14.809 --> 00:14:14.819 align:start position:0% attack so if you're going to do any 00:14:14.819 --> 00:14:17.629 align:start position:0% attack so if you're going to do any compiling<00:14:15.389> you've<00:14:15.839> got<00:14:16.050> to<00:14:16.230> do<00:14:16.410> it<00:14:16.620> up<00:14:16.829> front 00:14:17.629 --> 00:14:17.639 align:start position:0% compiling you've got to do it up front 00:14:17.639 --> 00:14:27.470 align:start position:0% compiling you've got to do it up front at<00:14:17.970> deployment<00:14:18.509> time<00:14:20.509> and<00:14:25.699> Marvin's<00:14:26.699> notion 00:14:27.470 --> 00:14:27.480 align:start position:0% at deployment time and Marvin's notion 00:14:27.480 --> 00:14:33.049 align:start position:0% at deployment time and Marvin's notion of<00:14:28.009> transpilers<00:14:29.809> I<00:14:30.809> think<00:14:31.589> is<00:14:31.769> very<00:14:32.189> important 00:14:33.049 --> 00:14:33.059 align:start position:0% of transpilers I think is very important 00:14:33.059 --> 00:14:35.889 align:start position:0% of transpilers I think is very important here<00:14:33.480> um 00:14:35.889 --> 00:14:35.899 align:start position:0% here um 00:14:35.899 --> 00:14:41.509 align:start position:0% here um 1.5<00:14:36.899> could<00:14:37.559> be<00:14:37.680> transpiled<00:14:38.309> to<00:14:38.610> 2.0<00:14:39.410> yeah<00:14:40.519> 2.0 00:14:41.509 --> 00:14:41.519 align:start position:0% 1.5 could be transpiled to 2.0 yeah 2.0 00:14:41.519 --> 00:14:46.369 align:start position:0% 1.5 could be transpiled to 2.0 yeah 2.0 can<00:14:41.759> be<00:14:41.879> transpiled<00:14:42.509> to<00:14:42.779> 1.5<00:14:43.819> 1.5<00:14:44.819> or<00:14:45.389> 2.0<00:14:46.079> can 00:14:46.369 --> 00:14:46.379 align:start position:0% can be transpiled to 1.5 1.5 or 2.0 can 00:14:46.379 --> 00:14:49.370 align:start position:0% can be transpiled to 1.5 1.5 or 2.0 can be<00:14:46.529> transpiled<00:14:47.100> to<00:14:47.399> 1.0<00:14:48.089> either<00:14:48.809> of<00:14:49.079> them<00:14:49.259> can 00:14:49.370 --> 00:14:49.380 align:start position:0% be transpiled to 1.0 either of them can 00:14:49.380 --> 00:14:51.689 align:start position:0% be transpiled to 1.0 either of them can be<00:14:49.559> transferred<00:14:50.100> to<00:14:50.130> the<00:14:50.279> JVM 00:14:51.689 --> 00:14:51.699 align:start position:0% be transferred to the JVM 00:14:51.699 --> 00:14:55.949 align:start position:0% be transferred to the JVM you<00:14:52.269> can<00:14:52.480> make<00:14:52.629> up<00:14:52.839> new<00:14:53.079> ones<00:14:53.290> pretty<00:14:54.220> much<00:14:54.959> you 00:14:55.949 --> 00:14:55.959 align:start position:0% you can make up new ones pretty much you 00:14:55.959 --> 00:14:58.409 align:start position:0% you can make up new ones pretty much you know<00:14:56.019> you<00:14:56.259> can<00:14:56.529> compile<00:14:56.949> any<00:14:57.429> VM<00:14:57.730> you<00:14:57.759> wanted 00:14:58.409 --> 00:14:58.419 align:start position:0% know you can compile any VM you wanted 00:14:58.419 --> 00:15:03.299 align:start position:0% know you can compile any VM you wanted to<00:14:58.899> some<00:14:59.139> other<00:14:59.410> VM<00:15:00.449> and<00:15:01.739> he<00:15:02.739> also<00:15:02.889> has<00:15:03.279> a 00:15:03.299 --> 00:15:03.309 align:start position:0% to some other VM and he also has a 00:15:03.309 --> 00:15:06.239 align:start position:0% to some other VM and he also has a notion<00:15:03.669> of<00:15:03.999> gas<00:15:04.239> injection<00:15:04.839> so<00:15:05.829> you<00:15:05.889> put 00:15:06.239 --> 00:15:06.249 align:start position:0% notion of gas injection so you put 00:15:06.249 --> 00:15:08.099 align:start position:0% notion of gas injection so you put little<00:15:06.279> pieces<00:15:06.639> of<00:15:07.059> code<00:15:07.329> into<00:15:07.660> the<00:15:07.929> right 00:15:08.099 --> 00:15:08.109 align:start position:0% little pieces of code into the right 00:15:08.109 --> 00:15:13.289 align:start position:0% little pieces of code into the right places<00:15:08.470> to<00:15:09.009> count<00:15:09.249> the<00:15:09.429> gas<00:15:10.679> and<00:15:12.119> what<00:15:13.119> this 00:15:13.289 --> 00:15:13.299 align:start position:0% places to count the gas and what this 00:15:13.299 --> 00:15:17.329 align:start position:0% places to count the gas and what this means<00:15:13.600> to<00:15:14.079> me<00:15:14.410> is<00:15:14.850> it<00:15:15.850> doesn't<00:15:16.269> matter<00:15:16.689> what 00:15:17.329 --> 00:15:17.339 align:start position:0% means to me is it doesn't matter what 00:15:17.339 --> 00:15:22.369 align:start position:0% means to me is it doesn't matter what execution<00:15:18.339> engine<00:15:19.059> a<00:15:19.329> client<00:15:19.989> chooses 00:15:22.369 --> 00:15:22.379 align:start position:0% execution engine a client chooses 00:15:22.379 --> 00:15:27.210 align:start position:0% execution engine a client chooses because<00:15:23.639> on<00:15:24.639> the<00:15:25.179> blockchain<00:15:25.470> you<00:15:26.470> can<00:15:26.679> put<00:15:26.980> a 00:15:27.210 --> 00:15:27.220 align:start position:0% because on the blockchain you can put a 00:15:27.220 --> 00:15:35.039 align:start position:0% because on the blockchain you can put a contract<00:15:27.989> that<00:15:28.989> translates<00:15:32.819> into<00:15:34.049> that 00:15:35.039 --> 00:15:35.049 align:start position:0% contract that translates into that 00:15:35.049 --> 00:15:38.159 align:start position:0% contract that translates into that execution<00:15:35.679> engine<00:15:36.249> so<00:15:36.999> these<00:15:37.149> are<00:15:37.419> these<00:15:37.899> can 00:15:38.159 --> 00:15:38.169 align:start position:0% execution engine so these are these can 00:15:38.169 --> 00:15:41.220 align:start position:0% execution engine so these are these can be<00:15:38.319> completely<00:15:38.799> independent<00:15:39.220> choices<00:15:39.939> and<00:15:40.779> we 00:15:41.220 --> 00:15:41.230 align:start position:0% be completely independent choices and we 00:15:41.230 --> 00:15:43.439 align:start position:0% be completely independent choices and we could<00:15:41.470> actually<00:15:41.679> choose<00:15:42.279> to<00:15:42.309> support<00:15:42.850> a 00:15:43.439 --> 00:15:43.449 align:start position:0% could actually choose to support a 00:15:43.449 --> 00:15:47.720 align:start position:0% could actually choose to support a number<00:15:44.019> of<00:15:44.259> VMS<00:15:45.100> if<00:15:45.369> we<00:15:45.669> wanted<00:15:46.059> to<00:15:46.119> and 00:15:47.720 --> 00:15:47.730 align:start position:0% number of VMS if we wanted to and 00:15:47.730 --> 00:15:49.889 align:start position:0% number of VMS if we wanted to and independent<00:15:48.730> parties<00:15:49.149> could<00:15:49.480> decide<00:15:49.869> to 00:15:49.889 --> 00:15:49.899 align:start position:0% independent parties could decide to 00:15:49.899 --> 00:15:53.220 align:start position:0% independent parties could decide to support<00:15:50.799> a<00:15:50.949> different<00:15:51.129> VM<00:15:51.699> but<00:15:52.600> a<00:15:52.629> transpiler 00:15:53.220 --> 00:15:53.230 align:start position:0% support a different VM but a transpiler 00:15:53.230 --> 00:16:03.869 align:start position:0% support a different VM but a transpiler on<00:15:53.619> the<00:15:53.769> blockchain<00:15:54.040> and<00:15:54.929> away<00:15:55.929> they<00:15:56.139> go<00:16:02.879> gee 00:16:03.869 --> 00:16:03.879 align:start position:0% on the blockchain and away they go gee 00:16:03.879 --> 00:16:08.729 align:start position:0% on the blockchain and away they go gee I'm<00:16:04.059> almost<00:16:04.389> out<00:16:04.569> of<00:16:04.629> slides<00:16:07.410> and<00:16:08.410> there's<00:16:08.589> no 00:16:08.729 --> 00:16:08.739 align:start position:0% I'm almost out of slides and there's no 00:16:08.739 --> 00:16:13.499 align:start position:0% I'm almost out of slides and there's no sound<00:16:09.039> that's<00:16:09.669> really<00:16:10.029> too<00:16:10.209> bad<00:16:11.850> so<00:16:12.850> what<00:16:13.239> what 00:16:13.499 --> 00:16:13.509 align:start position:0% sound that's really too bad so what what 00:16:13.509 --> 00:16:15.840 align:start position:0% sound that's really too bad so what what is<00:16:13.720> the<00:16:13.989> big<00:16:14.199> deal<00:16:14.379> about<00:16:14.439> native<00:16:15.129> performance 00:16:15.840 --> 00:16:15.850 align:start position:0% is the big deal about native performance 00:16:15.850 --> 00:16:19.109 align:start position:0% is the big deal about native performance and<00:16:16.410> what's<00:16:17.410> the<00:16:17.589> big<00:16:17.769> deal<00:16:17.980> about<00:16:18.009> being<00:16:18.399> lean 00:16:19.109 --> 00:16:19.119 align:start position:0% and what's the big deal about being lean 00:16:19.119 --> 00:16:22.949 align:start position:0% and what's the big deal about being lean and<00:16:19.359> mean<00:16:19.539> and<00:16:19.839> close<00:16:20.259> to<00:16:20.529> the<00:16:20.649> metal<00:16:21.089> well<00:16:22.089> we 00:16:22.949 --> 00:16:22.959 align:start position:0% and mean and close to the metal well we 00:16:22.959 --> 00:16:25.529 align:start position:0% and mean and close to the metal well we saw<00:16:23.199> the<00:16:23.259> electric<00:16:23.859> dragsters<00:16:24.489> here's<00:16:25.149> like<00:16:25.480> a 00:16:25.529 --> 00:16:25.539 align:start position:0% saw the electric dragsters here's like a 00:16:25.539 --> 00:16:28.889 align:start position:0% saw the electric dragsters here's like a real<00:16:26.079> top<00:16:26.679> fuel<00:16:27.100> dragster<00:16:27.449> they<00:16:28.449> run<00:16:28.720> on<00:16:28.869> a 00:16:28.889 --> 00:16:28.899 align:start position:0% real top fuel dragster they run on a 00:16:28.899 --> 00:16:35.659 align:start position:0% real top fuel dragster they run on a mixture<00:16:29.410> of<00:16:29.589> diesel<00:16:30.189> fuel<00:16:30.459> and<00:16:32.009> nitro<00:16:33.009> methane 00:16:35.659 --> 00:16:35.669 align:start position:0% 00:16:35.669 --> 00:16:42.210 align:start position:0% and<00:16:39.629> just<00:16:40.629> about<00:16:40.929> a<00:16:40.989> month<00:16:41.259> and<00:16:41.529> a<00:16:41.649> half<00:16:41.679> ago 00:16:42.210 --> 00:16:42.220 align:start position:0% and just about a month and a half ago 00:16:42.220 --> 00:16:45.210 align:start position:0% and just about a month and a half ago not<00:16:43.059> too<00:16:43.419> far<00:16:43.720> from<00:16:44.049> Ming's<00:16:44.379> place<00:16:44.859> in 00:16:45.210 --> 00:16:45.220 align:start position:0% not too far from Ming's place in 00:16:45.220 --> 00:16:54.150 align:start position:0% not too far from Ming's place in Michigan<00:16:46.139> this<00:16:47.139> guy<00:16:48.809> got<00:16:49.809> the<00:16:49.839> world<00:16:50.019> record 00:16:54.150 --> 00:16:54.160 align:start position:0% 00:16:54.160 --> 00:17:00.130 align:start position:0% and<00:16:55.160> it's<00:16:55.460> over<00:16:57.640> 338<00:16:58.640> miles<00:16:58.970> per<00:16:59.390> hour<00:16:59.600> and<00:17:00.080> a 00:17:00.130 --> 00:17:00.140 align:start position:0% and it's over 338 miles per hour and a 00:17:00.140 --> 00:17:04.600 align:start position:0% and it's over 338 miles per hour and a quarter<00:17:00.350> mile<00:17:01.180> these<00:17:02.180> guys<00:17:02.480> pull<00:17:03.020> about<00:17:03.890> 5<00:17:04.550> or 00:17:04.600 --> 00:17:04.610 align:start position:0% quarter mile these guys pull about 5 or 00:17:04.610 --> 00:17:08.050 align:start position:0% quarter mile these guys pull about 5 or 6<00:17:04.880> G's<00:17:05.600> which<00:17:06.560> is<00:17:06.770> about<00:17:07.069> the<00:17:07.339> same<00:17:07.640> as<00:17:07.910> an 00:17:08.050 --> 00:17:08.060 align:start position:0% 6 G's which is about the same as an 00:17:08.060 --> 00:17:10.449 align:start position:0% 6 G's which is about the same as an astronaut<00:17:08.780> taking<00:17:09.589> off<00:17:09.800> in<00:17:10.069> the<00:17:10.100> space 00:17:10.449 --> 00:17:10.459 align:start position:0% astronaut taking off in the space 00:17:10.459 --> 00:17:20.319 align:start position:0% astronaut taking off in the space shuttle<00:17:16.870> but<00:17:18.189> there's<00:17:19.189> a<00:17:19.310> problem<00:17:19.550> with<00:17:19.819> going 00:17:20.319 --> 00:17:20.329 align:start position:0% shuttle but there's a problem with going 00:17:20.329 --> 00:17:23.980 align:start position:0% shuttle but there's a problem with going fully<00:17:20.600> native<00:17:21.069> in<00:17:22.069> the<00:17:22.610> save<00:17:23.030> world<00:17:23.360> we<00:17:23.720> call 00:17:23.980 --> 00:17:23.990 align:start position:0% fully native in the save world we call 00:17:23.990 --> 00:17:34.680 align:start position:0% fully native in the save world we call it<00:17:24.400> undefined<00:17:25.400> behavior<00:17:29.560> hello 00:17:34.680 --> 00:17:34.690 align:start position:0% 00:17:34.690 --> 00:17:43.530 align:start position:0% no<00:17:35.690> that's<00:17:36.290> the<00:17:36.530> less<00:17:37.600> come<00:17:38.600> here<00:17:40.630> okay 00:17:43.530 --> 00:17:43.540 align:start position:0% no that's the less come here okay 00:17:43.540 --> 00:17:50.530 align:start position:0% no that's the less come here okay play<00:17:44.620> play<00:17:46.060> there<00:17:47.060> it<00:17:47.180> goes<00:17:49.120> what's<00:17:50.120> going<00:17:50.420> on 00:17:50.530 --> 00:17:50.540 align:start position:0% play play there it goes what's going on 00:17:50.540 --> 00:17:55.169 align:start position:0% play play there it goes what's going on my<00:17:50.750> screen 00:17:55.169 --> 00:17:55.179 align:start position:0% 00:17:55.179 --> 00:18:03.269 align:start position:0% we'll<00:17:56.109> try<00:17:56.409> it<00:17:56.440> again<00:18:00.960> dudu<00:18:01.960> the<00:18:02.950> driver 00:18:03.269 --> 00:18:03.279 align:start position:0% we'll try it again dudu the driver 00:18:03.279 --> 00:18:09.599 align:start position:0% we'll try it again dudu the driver actually<00:18:04.089> walked<00:18:04.330> away<00:18:04.919> unscratched<00:18:05.919> I<00:18:08.609> love 00:18:09.599 --> 00:18:09.609 align:start position:0% actually walked away unscratched I love 00:18:09.609 --> 00:18:24.419 align:start position:0% actually walked away unscratched I love this<00:18:09.759> guy<00:18:10.029> we'll<00:18:10.629> give<00:18:10.779> you<00:18:10.899> one<00:18:11.080> more<00:18:22.259> and<00:18:23.429> it 00:18:24.419 --> 00:18:24.429 align:start position:0% this guy we'll give you one more and it 00:18:24.429 --> 00:18:36.709 align:start position:0% this guy we'll give you one more and it turns<00:18:24.669> out<00:18:24.820> I'm<00:18:24.999> done<00:18:25.210> so<00:18:26.080> I<00:18:33.119> knew<00:18:34.119> it 00:18:36.709 --> 00:18:36.719 align:start position:0% 00:18:36.719 --> 00:18:46.820 align:start position:0% audiences<00:18:37.719> love<00:18:37.979> explosions 00:18:46.820 --> 00:18:46.830 align:start position:0% 00:18:46.830 --> 00:18:54.410 align:start position:0% [Music]